Asymmetric whole eye movement during non-contact tonometry: independent validation and implications for corneal biomechanical parameters

Asymmetric whole eye movement during non-contact tonometry: independent validation and implications for corneal biomechanical parameters | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Asymmetric whole eye movement during non-contact tonometry: independent validation and implications for corneal biomechanical parameters Daniela Oehring, Phillip Buckhurst, Christine Purslow, Hetal Buckhurst This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9226844/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 7 You are reading this latest preprint version Abstract Purpose The Corvis ST non-contact tonometer calculates whole eye movement (WEM) to isolate corneal deflection from globe displacement, assuming symmetrical retraction. This study aimed to validate the Corvis ST WEM parameter against an independent high-speed camera (HSC) system, quantify nasal-temporal asymmetry in globe retraction, and develop a correction method for identified asymmetries. Methods In this prospective cross-sectional study, 68 healthy subjects (mean age 22.65 ± 4.14 years) underwent Corvis ST measurement. In a subset of 20 participants, globe movement was simultaneously recorded using an orthogonal HSC system (3,030 fps). Raw displacement matrices were analysed to decompose the ocular response into translational and rotational components. Twelve WEM separation methods were compared using peripheral root-mean-square residuals. Results Robust regression achieved the lowest peripheral residuals (4.74 ± 2.97 µm), representing an 87.3% improvement over traditional edge-averaging (36.33 ± 18.76 µm, p < 0.001) and was optimal for 95.3% of measurements. HSC validation confirmed moderate-to-good reproducibility (ICC = 0.608 for amplitude and 0.720 for time). Significant nasal-temporal asymmetry was observed: nasal displacement exceeded temporal by 44–141 µm (N:T ratio 1.18–1.69, p < 0.001), with greater asymmetry in left eyes (N:T = 1.69 ± 0.34) than right eyes (N:T = 1.18 ± 0.20, p < 0.001). Globe retraction exhibited biphasic dynamics (slow phase 6.42 ± 3.17 ms, peak WEM at 21.07 ± 0.51 ms). Corneal curvature (SimK) predicted rotation amplitude in left eyes only (β=−0.47, p = 0.005). Deformation bounce was prevalent (63.0%), predominantly severe (55.9%) Conclusions The Corvis ST accurately measures overall globe displacement, but consistent nasal-temporal asymmetry introduces rotational error into standard WEM correction. An exploratory robust regression approach improved DCR parameter accuracy in this cohort, with potential implications for biomechanically corrected IOP, keratoconus screening indices, and clinical use of WEM as an orbital tissue biomarker. These correction findings require confirmation in larger, prospectively powered samples. Ocular Biomechanics Corvis ST Whole Eye Movement Non-Contact Tonometry Globe Retraction Nasal-Temporal Asymmetry Figures Figure 1 Figure 2 Figure 3 Figure 4 Key Points ⋅ Independent high-speed camera validation confirms that the Corvis ST WEM parameter accurately measures overall globe displacement during non-contact tonometry. ⋅ Consistent nasal-temporal asymmetry in globe retraction (N:T ratio 1.18–1.69) introduces a rotational artefact not accounted for by current WEM correction algorithms. ⋅ Exploratory robust regression correction reduced WEM-attributable error in dynamic corneal response parameters, but requires prospective confirmation in larger samples. Introduction Ocular biomechanics, the study of the eye's response to mechanical forces, provides fundamental insights into the pathophysiology of numerous sight-threatening conditions [1, 2]. The mechanical properties of the corneoscleral shell are critical factors in the development, diagnosis, and management of diseases, including keratoconus, post-refractive surgery ectasia, and glaucoma [3]. Of particular clinical importance is intraocular pressure (IOP) assessment, which remains the only modifiable risk factor for glaucoma [4]. However, all clinical tonometry methods are indirect, measuring the force required to deform the cornea rather than true intraocular pressure. This measurement is profoundly influenced by corneal biomechanical properties, creating a complex interplay where a stiffer cornea leads to IOP overestimation and a more compliant cornea to underestimation [5]. Given that glaucoma management decisions often depend on IOP changes of 1–2 mmHg, measurement inaccuracies can significantly impact risk stratification and treatment efficacy [4]. The Corneal Visualisation Scheimpflug Technology (Corvis ST; Oculus Optikgeräte GmbH, Wetzlar, Germany) employs a high-speed Scheimpflug camera capturing images of dynamic corneal response (DCR) to a precisely metered air puff [6]. A critical analytical step is isolating true corneal deflection from confounding motion of the entire globe. During the air puff, the globe translates posteriorly into the orbit; this is termed whole eye movement (WEM) [6]. The Corvis ST software subtracts WEM from total displacement measured at the corneal apex to derive the pure corneal deflection curve. This correction is essential, as failure to account for WEM leads to substantial overestimation of corneal deflection, systematically skewing all derived DCR parameters [7–10]. The proprietary algorithm calculates WEM by averaging the displacement of the two outermost points of its 8 mm horizontal scan (at positions c = ± 4 mm, where c is horizontal corneal position relative to the corneal apex) for each time frame [7]. This method is predicated on a fundamental, yet unvalidated, assumption: that posterior globe retraction is symmetrical across the horizontal meridian. While WEM has been increasingly utilised to explore orbital soft tissue properties in conditions like thyroid eye disease [11–13] and to investigate relationships with axial length in myopia [14], the validity of this symmetrical assumption has not been independently verified. Prior work has provided qualitative observations of asymmetric globe behaviour. Boszczyk et al. [15] noted a tendency for the globe to rotate nasally during Corvis ST measurements, though without quantification. Jannesari et al. [16–18] established through inverse modelling that separating corneal deformation from globe movement is theoretically essential for accurate estimation of material properties, though no practical correction method was proposed. Computational studies by Makarem et al. [19] predicted that asymmetric orbital support would produce nasal rotation during air-puff loading, but their finite element models could not fully replicate clinically observed patterns, and they explicitly called for further investigation. This study was designed to address these gaps through a multifaceted investigation. The objectives were (1) to compare methods for separating WEM from corneal deformation and identify the optimal approach; (2) to externally validate the Corvis ST WEM parameter against a synchronised, orthogonal high-speed camera system; (3) to quantify nasal-temporal asymmetry in globe retraction and assess its impact on derived DCR parameters. We hypothesised that globe retraction during Corvis ST tonometry is accurately measured but asymmetrically distributed, leading to rotational error in standard WEM-corrected deflection analysis. Materials and Methods This cross-sectional observational study analysed whole eye movement (WEM) during non-contact tonometry using Corvis ST (Oculus Optikgeräte GmbH, Wetzlar, Germany). The study was conducted at the University of Plymouth and received ethical approval from the Ethics and Integrity Committee (ref. no. 13/14-222). All participants provided written informed consent in accordance with the Declaration of Helsinki. This study is reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines for cross-sectional studies (Online Resource 3). Healthy adult volunteers aged 18–40 years were recruited from the University of Plymouth student and staff population. Inclusion criteria were: best-corrected visual acuity ≥6/9, no history of ocular surgery, no corneal pathology, and no contact lens wear within 24 hours of measurement. Exclusion criteria included glaucoma, keratoconus or other ectatic disorders, connective tissue disorders affecting ocular or orbital biomechanics (e.g., Ehlers-Danlos syndrome, Marfan syndrome), and inability to fixate during measurement. Sixty-eight participants were enrolled. Both eyes of each participant were measured once, yielding 136 Corvis ST measurements (68 OD, 68 OS) for the main WEM analysis. A subset of 20 participants underwent three consecutive left-eye measurements with simultaneous high-speed camera (HSC) recording for external validation and reproducibility assessment. A two-minute interval was maintained between measurements to allow tissue recovery [17]. For HSC validation, the mean of three consecutive measurements was used for each participant (n=20). For reproducibility analysis, all 60 repeated OS measurements were analysed. For the main WEM analysis, only the first OS measurement was included to maintain independence. Sample size was calculated based on equivalence testing methodology [20], expecting standard deviation of 0.083 mm for WEM [11], with α=0.05 and β=0.10. Corvis ST measurement protocol: The Corvis ST is a non-contact tonometer that captures 140 Scheimpflug images at 4,330 frames per second during a 32.3 ms air-puff cycle. Each horizontal cross-sectional image spans 8 mm of corneal width with 572 spatial points (resolution ~14 µm). The device records anterior and posterior corneal surface positions in each frame [21, 22]. A trained examiner performed all measurements during a single session. The examiner excluded measurements with quality flags indicating poor alignment or incomplete capture. Detailed instrument specifications and spatial mapping parameters are provided in Online Resource 2. Air-puff characteristics : The air-puff profile was characterised in a separate calibration study [23]: peak pressure 96.4±1.4 mmHg at 14.8 ms, total duration 21.5±0.3 ms, full width at half maximum (FWHM) 15.2 ms, affected corneal area 45.2 mm². HSC validation: A Corvis ST tonometer and HSC system (MotionBLITZ Director2 Kit, Mikrotron GmbH; Unterschleissheim, Germany) were positioned orthogonally. The subject's left eye was located at the intersection of both instruments' optical axes. The HSC was fitted with a Tokina AT-X PRO M 100mm macro lens, configured at 3,030 frames per second with 560×320 pixel resolution. A custom-designed infrared ring illuminator (48 LEDs, 17.2V, 5.5A) provided optimal imaging luminance. Light intensity at the corneal plane was 0.01±0.012 lx. A direct current generator prevented flickering. Synchronisation was achieved by recording the audible 'click' from the Corvis ST joystick via a microphone positioned adjacent to it, serving as the HSC trigger signal. Data extraction and processing Raw Scheimpflug data were exported from the Corvis ST (.csv format) and processed using custom Python software (version 3.10). Anterior and posterior corneal surface positions were extracted as matrices of dimension 140 (frames)×572 (spatial points). The neutral axis (corneal midplane) was calculated as the arithmetic mean of anterior and posterior surfaces. All surface positions were converted from millimetres to micrometres (×1000) for subsequent analysis. Edge preprocessing: Prior to WEM separation, peripheral (edge) regions were pre-processed to remove artefacts. Outliers were detected using a median absolute deviation (MAD) approach: values deviating more than 4.0 times the local standard deviation (estimated from MAD×1.4826) from the local median (5×5 window) were flagged. Flagged values were replaced with the local median. Edge detection quality was classified as good (outlier fraction <1%), marginal (1–5%), or fallback (≥5%). Boundary detection: The peripheral region suitable for WEM estimation was identified using a two-pass adaptive algorithm. Signal-to-noise ratio (SNR) was computed for each edge region; SNR≥10 indicated good quality, ≥5 marginal quality, and <5 required fallback estimation. Edge width (number of columns used) averaged 0.75±0.15 mm (53–54 columns). WEM separation methods Twelve methods for separating WEM from true corneal deformation were implemented and compared: (1) Traditional edge-averaging: Mean displacement of peripheral columns (equivalent to Corvis ST built-in correction) [24]; (2) Linear regression: Models WEM as translation plus tilt; displacement and tilt smoothed using Savitzky-Golay filter (window 11 frames, polynomial order 2) [14, 24]; (3) Robust regression: Huber regression on all peripheral points, resistant to outliers [25]; (4) Adaptive edge: Dynamically determines edge width based on deformation boundary detection; (5) Cross-correlation alignment: Minimises peripheral RMS difference between each frame and baseline [26]; (6) Frequency-domain filtering: Low-pass filter (50 Hz cutoff) to isolate slow WEM from rapid corneal deformation [27]; (7) Viscoelastic modelling: Kelvin-Voigt model of globe response to measured air-puff force [23]; (8) Physiological viscoelastic: Two-component model with separation at 80 Hz; (9) Adaptive-viscoelastic hybrid: Combines methods 4 and 7 (10) Adaptive-robust hybrid: Combines methods 4 and 3; (11) Dynamic selection: Evaluates all methods per measurement, selects lowest peripheral RMS residual; (12) Arc-corrected methods: Account for corneal curvature in peripheral displacement estimation. All methods (except traditional) model WEM as a combination of translation and rotation: WEM(t, x)=d(t) + θ(t) · x, where d(t) is posterior displacement (µm), θ(t) is tilt (µm/mm), and x is the spatial position (mm). Corneal deformation was then calculated as: deformation(t, x)=[raw(t, x) − raw(0, x)] − WEM(t, x). Method performance was evaluated using root-mean-square (RMS) residuals in peripheral regions where true corneal deformation is minimal. The optimal method for each measurement was identified; robust regression was optimal for 95.3% of measurements and was adopted for all subsequent analyses. Full mathematical derivations for all separation methods are provided in Online Resource 1. Parameter calculation Corvis-equivalent parameters (Figure 1) : First applanation (A1) was identified as the first negative-to-positive zero-crossing in corneal curvature during the loading phase (5–12 ms). Highest concavity (HC) was the frame with maximum inward deformation. Second applanation (A2) was the first positive-to-negative curvature zero-crossing during recovery (HC+2ms to 28ms). Sub-frame interpolation was applied for temporal precision. Velocity profiles were computed using central differences on Savitzky-Golay smoothed deformation (window 9, order 3), converted to m/s. Radius of curvature at HC was estimated via linearised circle fitting on the central 3 mm region. Globe translation and rotation: Globe movement was decomposed into translational and rotational components using peripheral displacements at the nasal (c=−4 mm) and temporal (c=+4 mm) boundaries, where c denotes horizontal corneal position relative to the corneal apex (negative=nasal, positive=temporal). Translation was defined as the mean of nasal and temporal displacements: Translation(t)=[d N (t) + d T (t)] / 2. Rotation angle was derived from the differential displacement: Rotation(t)=arctan[(d T (t) − d N (t)) / 7.59 mm], where 7.59 mm is the span between peripheral measurement positions. Positive rotation indicates nasal-ward tilt (temporal side displacing more posteriorly). The nasal:temporal (N:T) ratio was calculated as peak nasal displacement divided by peak temporal displacement. Temporal dynamics (Figure 1) : WEM temporal phases were characterised as follows. Movement onset was defined as the first frame exceeding 5% of peak displacement. The slow-fast transition was identified when velocity crossed 20% of maximum retraction velocity. Slow-loading phase duration was measured from onset to transition; fast-loading phase duration from transition to peak amplitude. Following peak displacement, the globe returned anteriorly. Recovery metrics included percentage return to baseline and time to 50% and 90% recovery. Peak retraction and protraction velocities were recorded, along with asymmetry ratios comparing retraction and protraction characteristics. Velocity oscillations: High-frequency oscillations in WEM velocity were quantified during the slow loading phase (5–16 ms) and recovery phase (peak to end). The velocity signal was smoothed using a Savitzky-Golay filter (window=7, order=3) and detrended to isolate oscillatory components. Oscillation count was determined by identifying zero-crossings in the detrended velocity signal. Oscillation amplitude was calculated as the mean absolute peak-to-trough amplitude. Dominant frequency was estimated using Welch power spectral density analysis within the 10–200 Hz band. Measurements were classified as low (5 oscillations) based on total oscillation count. Deformation bounce (Figure 1) : Bounce events, transient velocity reductions or reversals during corneal loading, were detected using two algorithms. Type 1 (reversal) identified local extrema in the smoothed deformation profile (scipy find_peaks, minimum prominence 1.5µm). Type 2 (plateau) identified velocity ratio |v(t)|/running maximum5≤20µm), severe (max amplitude>20µm), or multiple (n>3) based on the number and amplitude of events. Crater geometry: The air-puff-induced deformation crater was characterised at the frame of maximum corneal indentation. Crater centre was determined as the weighted mean position, using deformation magnitude as weights. Crater offset from the corneal centre was calculated and classified as nasal, temporal, or central (offset<0.2 mm). Crater extent was defined as the region where deformation exceeded 10% of maximum depth. Extent asymmetry was quantified by comparing the spread toward nasal versus temporal peripheries. Corneal inclination was assessed from the baseline (pre-air-puff) corneal profile by calculating the angle between peripheral edge heights. Outlier detection and data cleaning: A hierarchical outlier detection approach was applied, to remove artefacts from lashes or eyelids: (1) IQR-based detection with values beyond Q1 − 2.0×IQR or Q3 + 2.0×IQR; (2) Z-score detection with absolute z-score > 3.0; (3) Validity range checks for physiologically implausible values (e.g., IOP 40 mmHg). Outliers were set to missing (NaN) rather than excluded, preserving measurement-level data. The hierarchy proceeded from foundational parameters to derived parameters to prevent cascading outlier flags. Statistical analysis All analyses were performed using Python 3.10 with scipy (version 1.11), numpy (1.24), and pandas (2.0). Continuous variables are presented as mean±standard deviation (SD) with 95% confidence intervals (CI) based on the t-distribution, or as median with interquartile range (IQR) for non-normally distributed data. Categorical variables are presented as counts with percentages and Wilson 95% CI. Normality was assessed using both Shapiro-Wilk (n≤5000) and D'Agostino-Pearson (n≥20) tests; data were considered normally distributed only if both tests yielded p>0.05. Paired comparisons used paired t-tests for normally distributed differences or Wilcoxon signed-rank tests otherwise. Independent comparisons used Student's or Welch's t-test (based on Levene's test for variance equality) for normal data, or Mann-Whitney U test for non-normal data. Effect sizes are reported as Cohen's d for independent comparisons and Cohen's d z for paired comparisons. Interpretation followed conventional thresholds: small (0.2), medium (0.5), large (0.8). Holm-Bonferroni correction was applied for pairwise method comparisons. For correlations with ocular parameters, p<0.05 was used as the significance threshold given the exploratory nature; no further correction for multiple comparisons was applied to these analyses, and readers should interpret nominally significant associations accordingly. Spearman rank correlations were used for associations between WEM parameters and ocular characteristics. Multiple linear regression using simultaneous entry of all predictors was used to identify predictors of rotation amplitude, with variance inflation factors (VIF) assessed for multicollinearity (VIF > 10 indicating concern). Per-eye models were also fitted to examine whether predictors differed between right and left eyes. The regression analyses are considered exploratory and hypothesis-generating. For reproducibility analysis, within-subject coefficient of variation (CV) and intraclass correlation coefficient (ICC, two-way random, single measures) were calculated. ICC values were interpreted following Koo and Li (2016): below 0.50 poor, 0.50–0.75 moderate, 0.75–0.90 good, above 0.90 excellent. All analyses were performed using custom Python code. Complete source code, parameter documentation, and reproducibility scripts are available at upon request. The analysis pipeline generated 285 output parameters per measurement, organised by category: Corvis device (40), WEM and derived (81), geometry (12), crater dynamics (15), bounce detection (11), and other calculated parameters (126). Results The sample was predominantly female (70.1%; Table 1). Normality testing (Shapiro-Wilk) indicated non-normal distributions for age (W=0.830, p < 0.001) and IOP (W=0.957, p=0.006), while CCT (p=0.533) was normally distributed. Table 1. Participant characteristics by sex Parameter Total (n=68) Female (n=48) Male (n=20) p-value Age (years) 22.65±4.14 22.33±3.99 23.60±4.48 0.550 CCT (µm) 541.7±38.8 543.4±38.9 538.2±38.9 0.600 IOP (mmHg) 13.68±1.87 13.52±1.81 14.02±1.98 0.272 CCT=central corneal thickness; IOP=intraocular pressure; Values are mean±SD. p-values from Mann-Whitney U test. WEM separation method comparison: Lower RMS residuals indicate superior separation of globe movement from corneal deflection in peripheral regions where air-puff-induced deformation is minimal (Table 2, Figure 2). Dynamic selection and viscoelastic modelling achieved equivalently optimal performance with no significant difference between them (Holm-adjusted p<0.99). Both significantly outperformed all other approaches (all Holm-adjusted p<0.001). The viscoelastic method achieved an 87.3% reduction in RMS residuals compared to traditional edge-averaging (effect size d=1.69, p<0.001), the method currently employed by Corvis ST. Frequency-domain filtering performed poorest, significantly worse than traditional edge-averaging (p<0.001). Three distinct performance clusters emerged: optimal performers (dynamic, viscoelastic, adaptive-viscoelastic, robust regression, adaptive-robust), intermediate performers (cross-correlation, adaptive edge, linear regression, arc-corrected), and poor performers (viscoelastic-physiological, traditional, frequency-based). Notably, arc-correction, which accounts for corneal curvature rather than assuming linear peripheral geometry, performed slightly worse than uncorrected linear regression (mean difference +0.11 µm, d=0.73, p<0.001), with linear regression outperforming arc-correction in 79.5% of measurements. This validates that the linear spatial approximation is adequate for the ±4 mm Corvis ST measurement window. Per-measurement analysis identified robust regression as optimal in 95.3% of cases, with adaptive-viscoelastic optimal in the remainder. Comparison of dynamic versus fixed method selection showed negligible practical improvement (Cohen's d=0.16), with only 0.7% of measurements benefiting from dynamic selection. 6 participants (8.8%) had different optimal methods assigned to each eye (all had robust regression for OD and adaptive-viscoelastic for OS). This suggests method selection may be influenced by eye-specific characteristics. Table 2. Peripheral RMS residuals (µm) by separation method (n=68, 136 measurements). # Method Mean ± SD Median (IQR) Range (Min-Max) 95% CI 1 Dynamic selection 4.54±2.90 3.82 (1.89) 0.85-20.19 4.09-5.11 2 Viscoelastic (Kelvin-Voigt) 4.60±2.90 3.82 (1.89) 0.85-20.19 4.09-5.12 3 Adaptive-viscoelastic 4.61±2.99 3.82 (1.98) 0.85-20.19 4.10-5.12 4 Robust regression 4.74±2.97 3.96 (1.93) 0.86-20.63 4.22-5.26 5 Adaptive-robust 4.75±2.97 3.96 (2.01) 0.86-20.63 4.23-5.27 6 Cross-correlation 5.15±3.34 4.17 (1.91) 1.93-25.05 4.56-5.73 7 Adaptive edge 5.15±3.33 4.18 (1.94) 1.93-25.05 4.56-5.73 8 Linear regression 5.41±4.09 4.18 (1.94) 1.93-29.19 4.69-6.13 9 Arc-corrected 5.52 ± 4.07 4.36 (1.89) 1.94–29.19 4.81–6.23 10 Viscoelastic (physiological) 13.43±3.57 12.95 (3.79) 6.64-28.49 12.80-14.06 11 Traditional edge-averaging 36.33±18.76 33.43 (23.87) 7.86-105.51 33.03-39.62 12 Frequency-domain filtering 46.70±10.03 48.27 (9.46) 14.89-66.21 44.94-48.46 Methods ranked by mean RMS residual (ascending). IQR = interquartile range. Peripheral edge detection and crater geometry: The peripheral regions used for WEM estimation and the air-puff-induced deformation crater were characterised to assess measurement quality and spatial symmetry (Table 3). Edge detection quality was good in 64.7% of measurements, with the peripheral region averaging 0.75 mm in width (~54 Scheimpflug columns). The deformation crater was highly symmetric at HC (N:T extent ratio=1.00±0.02) and centrally positioned (99.3% within ±0.2mm of centre), confirming uniform air-puff loading. Crater expansion (8.23ms) exceeded retraction duration (6.35ms), while retraction velocity exceeded expansion velocity (3.81 vs 3.19mm/ms). Table 3. Edge detection and crater characteristics (n=136) Parameter Value Edge detection Detection method: noise floor / fallback 135 (99.3%) / 1 (0.7%) Quality: good / marginal / fallback 88 (64.7%) / 39 (28.7%) / 9 (6.6%) Edge width, mean ± SD 0.75 ± 0.15 mm (53.5 ± 10.8 columns) Edge width, range 0.14–1.23 mm (10–88 columns) Edge residual RMS, mean ± SD 4.59 ± 2.90 µm Crater geometry (at HC) Crater width 5.65 ± 0.17 mm (5.20–6.19) Nasal extent 2.81 ± 0.09 mm (2.61–3.03) Temporal extent 2.83 ± 0.08 mm (2.59–3.03) Extent ratio (N:T) 1.00 ± 0.02 (0.81–1.04) Maximum deformation 907.1 ± 81.8 µm (705.5–1104.5) Crater position Crater offset 0.07 ± 0.06 mm (−0.09–0.22) Apex offset 0.19 ± 0.23 mm (−0.39–0.80) Corneal inclination 1.87 ± 1.84° (−2.81–6.59) Crater dynamics Maximum crater width 5.70 ± 0.16 mm (5.32–6.18) Crater onset 6.91 ± 0.22 ms (6.24–7.62) Crater duration 14.59 ± 0.58 ms (13.17–16.17) Expansion duration 8.23 ± 0.98 ms (5.78–10.86) Retraction duration 6.35 ± 1.10 ms (3.01–9.01) Peak expansion rate 3.19 ± 0.19 mm/ms (2.66–3.58) Peak retraction rate 3.81 ± 0.19 mm/ms (3.37–4.38) Values are mean ± SD (range) unless otherwise specified. HC = highest concavity; N:T = nasal:temporal. External validation against HSC: Peak amplitude measured by HSC (373.0±129.0 µm) was numerically larger than Corvis ST-derived values (307±68.5 µm), though this difference did not reach statistical significance (p=0.064) (Figure 3). Time to peak differed significantly between systems, with HSC detecting peak amplitude approximately 1.7 ms earlier than Corvis ST (19.3 vs 21.1ms, p=0.006). The temporal discrepancy was attributed to different triggering mechanisms. Linear regression indicated a consistent 3.6ms delay. Applying this correction improved temporal coverage from 56% to 83%, confirming the fundamental validity of Corvis ST temporal tracking. Analysis of three consecutive measurements (n=20) showed no significant difference between repeated measurements (Friedman χ²=1.83, df=2, p=0.401). Peak amplitude demonstrated moderate reliability with ICC(2,1)=0.608 (95% CI [0.408, 0.808], F=6.53, p=0.006), though the wide confidence interval spanning poor to good categories reflects the small repeatability subset (n=20); within-subject coefficient of variation (CV) was 24.6%. Time to peak showed moderate-to-good reliability with ICC(2,1)=0.720 (95% CI [0.520, 0.920]) and within-subject CV of 6.3%. Nasal-Temporal asymmetry: Contrary to the Corvis ST assumption of symmetric peripheral displacement, nasal retraction consistently exceeded temporal retraction (Table 4). Despite symmetric crater geometry (N:T extent ratio=1.00), globe retraction was markedly asymmetric, suggesting orbital tissue properties rather than non-uniform loading. Asymmetry was more pronounced in OS than OD (N:T ratio 1.69±0.34 vs 1.18±0.20, d=−1.83). Classification: nasal-dominant (N:T>1.2) 72.4%, symmetric (0.8–1.2) 25.2%, temporal-dominant (<0.8) 2.4%. Globe movement decomposed into translation (282.4±56.5µm) and rotation (1.00±0.44°, peak −0.98±0.50°). Negative rotation indicates nasal-ward tilt. Translation did not differ between eyes (p=0.136); rotation was significantly greater in OS (1.23±0.43° vs 0.78±0.33°, d=−1.19). Table 4. Nasal-temporal asymmetry by laterality Parameter OD OS p d Asymmetry Nasal amplitude (µm) 305.0±82.0 362.5±73.8 <0.001 −0.74 Temporal amplitude (µm) 257.3±51.6 221.0±51.7 <0.001 0.70 N:T ratio 1.18±0.20 1.69±0.34 <0.001 −1.83 N−T difference (µm) 44.3±52.9 140.6±66.4 <0.001 −1.61 Movement components Translation amplitude (µm) 275.7±55.6 289.2±57.0 0.136 −0.24 Rotation angle (°) 0.78±0.33 1.23±0.43 1.0=nasal predominance. Negative rotation=nasal-ward tilt. Overall: translation 282.4±56.5µm, rotation amplitude 1.00±0.44°, peak rotation −0.98±0.50°, reversals 1.02±2.56 (median 0). Temporal dynamics of globe movement: WEM exhibited biphasic retraction (slow phase 6.42±3.17ms, fast phase 10.56±2.88ms) followed by faster protraction (velocity asymmetry ratio 1.27±0.35) (Table 5). Maximum WEM velocity occurred near peak corneal deformation (15.87±1.97ms vs 15.85±0.41ms); peak WEM displacement lagged by 4.61±0.60ms. Only 5.5% of eyes returned to within ±10µm of baseline; mean residual displacement was −65.7±63.9µm. Temporal dynamics were broadly similar between eyes, though paired testing revealed a significant laterality difference in duration asymmetry ratio (p=0.004) and peak timing (p=0.040). Table 5. WEM temporal dynamics Parameter Mean±SD Median (IQR) Range Retraction Slow phase duration (ms) 6.42±3.17 6.93 (4.27–8.78) 0.00–13.86 Fast phase duration (ms) 10.56±2.88 9.93 (8.32–12.24) 4.85–18.02 Max velocity (µm/ms) −48.54±16.27 −46.12 (−59.46––36.77) −94.11––11.97 Protraction Slow phase duration (ms) 0.97±0.97 0.69 (0.00–1.62) 0.00–3.70 Fast phase duration (ms) 10.08±0.92 10.16 (9.47–10.86) 7.39–12.24 Max velocity (µm/ms) 37.76±8.23 38.27 (33.16–43.27) 16.72–58.46 Peak and recovery Peak amplitude (µm) 282.35±56.63 286.41 (243.92–323.85) 135.72–411.55 Time to peak (ms) 21.07±0.51 21.02 (20.79–21.37) 19.87–22.87 Time to 50% recovery (ms) 6.62±0.94 - - Time to 90% recovery (ms) 11.20±1.90 - - Final residual (µm) −65.7±63.9 - - Negative velocities=retraction (posterior); positive=protraction (anterior). WEM oscillation and corneal deflection bounce: High-frequency velocity WEM oscillations averaged 3.65±1.38 per measurement, predominantly during loading (4.29±1.45) versus recovery (1.02±0.96); 29.1% showed no recovery oscillations (Table 6). Classification: low (5) 23.6%. Corneal deflection bounce, transient slowing or reversal before HC, occurred in 63.0% of measurements (Table 6). Two types: reversal (Type 1, local extremum) and plateau (Type 2, flattening without reversal). Among bounces: 31.3% reversal only, 52.5% plateau only, 16.3% both. Severity: none 37.0%, moderate 5.5%, severe 55.9%, multiple 1.6%. Table 6. Oscillation and bounce characteristics Parameter Mean±SD Median (IQR) Range Oscillations Total count 3.6±1.4 4 (3–4) 1–8 Loading phase count 4.3±1.5 4 (3–5) 1–8 Recovery phase count 1.0±0.9 1 (0–1) 0–5 Amplitude (µm/ms) 11.45±5.51 10.11 (7.42–15.17) 2.25–26.38 Bounce First bounce time (ms) 13.96±0.86 14.09 (13.63–14.39) 11.32–17.09 Max amplitude (µm) 35.14±35.73 33.31 (0.00–55.03) 0.00–139.48 Duration (ms) 0.69±0.65 0.69 (0.00–1.16) 0.00–2.54 Event count 0.8±0.7 1 (0–1) 0–3 Reversals (Type 1) 0.3±0.5 0 (0–1) 0–2 Plateaus (Type 2) 0.5±0.6 0 (0–1) 0–2 Velocity variability (µm/ms) 1.248±0.026 1.247 (1.23–1.26) 1.18–1.32 Most oscillation and bounce parameters did not differ between eyes. Significant laterality effects: recovery oscillations higher in OS (1.19±1.03 vs 0.84±0.86, p=0.030, d=−0.37); first bounce earlier in OS (13.79±0.87ms vs 14.11±0.81ms, p=0.026, d=0.29). Eyes with lower IOP, faster A1, and greater deformation exhibited more variable pre-HC velocity. Greater deformation associated with earlier bounce onset (Table 7). Table 7. Bounce correlations with ocular parameters Parameter Ocular parameter ρ p Velocity variability IOP −0.375 <0.001 Velocity variability A1 time −0.374 <0.001 Velocity variability HC DefAmp +0.335 0.001 First bounce time HC DefAmp −0.299 0.004 Aetiology of asymmetric globe movement: Bivariate correlations with rotation amplitude (Table 8): apex offset (r=−0.251, p=0.004), SimK 3mm (r=−0.245, p=0.025), X position (r=0.234, p=0.033); non-significant: ZonalK 7mm (r=−0.217, p=0.060), CCT (r=−0.165, p=0.116), crater offset (r=−0.112, p=0.212), Z position (r=−0.033, p=0.756), IOP (r=−0.010, p=0.928). Apex offset and X position were collinear (r=−0.996). SimK 3mm was the only significant predictor: steeper corneas exhibited greater rotation. Per-eye models: OD (R²=0.083, Adj R²=−0.060, F=0.58, p=0.716) not significant; OS (R²=0.281, Adj R²=0.189, F=3.05, p=0.020) significant with SimK 3mm (β=−0.47, p=0.005) and Z position (β=−0.74, p=0.054) as predictors. This asymmetric predictive relationship may contribute to greater rotation in left eyes. Table 8. Multivariate predictors of rotation amplitude Predictor β SE p VIF Intercept 3.54 1.05 0.001 - SimK 3mm −0.29 0.13 0.024 3.57 Apex offset (mm) 2.63 1.42 0.067 52.3 Corneal inclination (°) −0.31 0.16 0.056 59.3 Z position (mm) −0.32 0.30 0.295 1.29 Crater offset (mm) 0.29 0.86 0.735 2.11 Model: R²=0.118, Adj R²=0.061, F(5,77)=2.06, p=0.080. Impact on DCR parameters: WEM correction resulted in systematic DCR parameter differences versus standard Corvis ST output (Table 9, Figure 4). Clinical significance using predefined thresholds: A1 time unaffected (0% exceeding 0.5ms), HC time exceeded threshold in 46.6% (mean |diff| 0.63ms), A2 time in 98.9% (mean |diff| 0.81ms), HC DefAmp in 8.8% (mean |diff| 19.8µm), HC radius in 100% (mean |diff| 3.03mm). Table 2. DCR parameters: Corvis ST vs WEM-corrected Parameter Corvis Corrected Diff (95% CI) p dz >Threshold A1 time (ms) 7.17±0.22 7.18±0.33 −0.008 (−0.043–0.028) 0.660 −0.05 0% A1 velocity (m/s) −0.17±0.01 −0.21±0.04 0.047 (0.041–0.054) <0.001 1.50 - A1 deformation 0.13±0.01 mm 237.0±21.8 µm - <0.001 −10.87 - HC time (ms) 16.49±0.38 15.87±0.41 0.619 (0.519–0.719) <0.001 1.31 46.6% HC DefAmp 1.11±0.09 mm 1088.2±90.3 µm - <0.001 −12.05 8.8% HC radius (mm) 6.68±0.65 3.66±0.20 3.025 (2.915–3.135) <0.001 5.79 100% HC peak dist (mm) 5.12±0.21 7.22±0.18 −2.105 (−2.151–−2.060) <0.001 −9.71 - A2 time (ms) 22.12±0.34 21.32±0.39 0.806 (0.775–0.837) <0.001 5.44 98.9% A2 velocity (m/s) −0.36±0.05 0.31±0.03 −0.661 (−0.676–−0.647) <0.001 −9.53 - A2 deformation 0.40±0.06 mm 494.4±53.6 µm - <0.001 −9.22 - dz=standardised paired difference. Thresholds: timing 0.5ms, HC DefAmp 50µm, HC radius 0.5mm. Clinical significance 95% CIs: A1 time 0.0–4.1%, HC time 36.5–56.9%, A2 time 94.0–99.8%, HC DefAmp 4.5–16.4%, HC radius 95.9–100%. Mean |diff|: A1 0.13ms, HC 0.63ms, A2 0.81ms, DefAmp 19.77µm, radius 3.03mm. Discussion To our knowledge, this study provides the first independent validation of Corvis ST whole eye movement against an orthogonal reference system. The central finding, that globe retraction during air-puff tonometry exhibits consistent nasal-temporal asymmetry generating a rotational artefact, challenges the fundamental assumption underlying all current WEM correction algorithms. Several observations warrant detailed consideration: the relative performance advantage of robust regression over viscoelastic modelling in this exploratory analysis, despite the latter’s theoretical appeal; the pronounced laterality effect with left eyes showing 1.6-fold greater rotation than right; the tight temporal coupling between WEM velocity and corneal deformation peak; the unexpectedly high prevalence of deformation bounce; and the laterality-specific predictive relationship between corneal curvature and rotation amplitude. The finding that robust regression yielded the lowest residuals in 95.3% of measurements in this exploratory comparison, outperforming viscoelastic modelling despite its explicit incorporation of orbital tissue mechanics, requires explanation. It should be noted that the overall regression model was borderline non-significant (p = 0.080, adjusted R² = 0.061), indicating limited explained variance; these results are hypothesis-generating rather than confirmatory. Viscoelastic models assume homogeneous tissue properties and uniform boundary conditions, which are often violated in the complex orbital environment due to anisotropic structures and the need to model orbital fat as viscoelastic [28]. The retrobulbar space contains fat lobules of varying size separated by fibrous septa, extraocular muscles with non-uniform insertion geometries, and the optic nerve sheath complex with its own viscoelastic characteristics [29]. These structural heterogeneities create spatially variable impedance to globe displacement that cannot be captured by lumped-parameter models. Robust regression succeeds precisely because it makes minimal assumptions. By down-weighting outlying peripheral measurements, which arise from localised tissue inhomogeneities, edge detection artefacts at the limbus, or transient scleral deformation, the Huber estimator identifies the dominant linear trend representing bulk globe motion without requiring explicit tissue modelling. The 87.3% reduction in peripheral RMS residuals compared to traditional edge-averaging reflects the magnitude of error introduced by the symmetric translation assumption: at peak retraction, the standard algorithm attributes rotational displacement to corneal deformation, systematically contaminating the deflection curve. The poor performance of frequency-domain filtering (worse than traditional edge-averaging) is instructive. Low-pass filtering assumes that globe movement occupies a distinct frequency band separable from corneal deformation, but both processes share overlapping spectral content during the rapid loading phase (5–16 ms). The corneal apex velocity during early indentation (~ 0.15–0.20 m/s) generates frequency components indistinguishable from globe acceleration, rendering spectral separation ineffective. The consistent nasal predominance in peripheral displacement (N:T ratio 1.18–1.69) indicates that the temporal corneal periphery retracts further posteriorly than the nasal, generating globe rotation toward the nose. This asymmetry persisted despite symmetric crater geometry (N:T extent ratio 1.00 ± 0.02), definitively localising its origin to orbital rather than corneal mechanics. The explanation lies in differential orbital constraint mechanics. The medial orbital wall, though comprising the paper-thin lamina papyracea of the ethmoid, is functionally stiffened by the medial rectus and its associated connective tissue pulley system [29], creating substantial resistance to medial globe displacement. MRI studies demonstrate that orbital fat immediately behind the eye follows globe displacement by approximately one-half, with regional differences in tissue firmness, fat around the optic nerve behaving almost like a fluid while fat in other orbital regions is more firm [30]. Under rapid loading, this differential tissue compliance, combined with the asymmetric geometry of the orbital soft tissue support system, may permit greater temporal than nasal globe excursion. Makarem et al. [19] modelled asymmetric orbital support but could not validate their nasal rotation predictions clinically, suggesting that the biomechanical basis of this asymmetry remains incompletely understood. When the air puff delivers its ~ 60 mN peak force [23] to the central cornea, the globe accelerates posteriorly. The medial orbit presents higher impedance, while the lateral orbit offers lower impedance. The globe follows the path of least resistance, rotating nasalward as the temporal side displaces more than the nasal. This is analogous to pushing a sphere embedded asymmetrically in foam: it rotates toward the stiffer side. The measured rotation angles (0.78° OD, 1.23° OS) correspond to peripheral displacement differentials of 44–141 µm over the 7.6 mm measurement span. These values align remarkably well with finite element predictions by Makarem et al. [19], who modelled asymmetric orbital support but could not validate their predictions clinically. Our data provides that validation while revealing a laterality effect that their symmetric models could not predict. The 58% greater rotation in left eyes (1.23° vs 0.78°) represents the study's most unexpected finding. Several mechanisms warrant consideration: Anatomical asymmetry : True left-right differences in orbital anatomy exist but are typically subtle. The dominant eye may have marginally different extraocular muscle tone or orbital fat distribution, though published evidence is limited [32, 33]. We did not assess our cohort for ocular dominance, precluding direct evaluation. Measurement sequence effects : In our protocol, eye side was chosen randomly; however, sequential measurements could induce subtle fatigue, altered blink patterns, or accommodation changes affecting orbital tissue tension. We did not find a significant difference in rotation by measurement order. Nasal bridge geometry : The Corvis ST positions the air-puff nozzle along the instrument's optical axis, which aligns with the corneal apex. For OS measurements, the nozzle approaches from the patient's right side, placing it closer to the nasal bridge. This geometric asymmetry could create subtle airflow perturbations or influence head positioning. The air-puff pressure field is not perfectly symmetric [23]; interaction with the nasal bridge for OS measurements but not for OD measurements could generate differential loading. Fixation asymmetry : During OS measurement, the right eye fixates on the instrument housing or examiner. This adducted position of the fellow eye tenses the medial rectus via Hering's law of equal innervation, potentially altering orbital mechanics [33]. For OD measurement, OS fixates in relative abduction, creating different baseline muscle tension states. The finding that SimK predicted rotation amplitude significantly for OS (R²=0.28, p = 0.02) but not OD (p = 0.72) provides a crucial clue. In left eyes, steeper corneas exhibited greater rotation (β=−0.47). Corneal curvature influences the spatial distribution of air-puff force: steeper corneas concentrate the pressure profile over a smaller apical area, potentially creating more eccentric loading if the apex is even slightly decentred. The laterality-specific nature of this relationship suggests that whatever factor distinguishes OS from OD (nasal bridge interaction, fixation state, anatomical asymmetry) interacts with corneal geometry to amplify rotational response. Several methodological caveats attend the multivariate analysis. The high variance inflation factors for apex offset and X position (VIF 52.3 and 59.3, respectively) reflect near-perfect collinearity between these predictors (r = − 0.996), inflating standard errors and rendering their individual coefficients unreliable. We retained both to demonstrate this structural redundancy rather than for independent inference; the model’s overall fit (noting p = 0.080) and SimK’s independent predictive value (VIF 1.4) are unaffected by this collinearity. Future studies should select one spatial predictor a priori or employ dimensionality reduction to avoid multicollinearity. The biphasic WEM pattern, slow initial retraction followed by rapid acceleration, provides direct insight into orbital tissue mechanics that has not previously been characterised in vivo. The slow phase represents the "toe region" of orbital tissue stress-strain behaviour. During this period, collagen fibres in Tenon's capsule, check ligaments, and intermuscular septum progressively recruit from their crimped resting configuration. Simultaneously, orbital fat undergoes consolidation as interlobular fluid redistributes. The system exhibits high compliance during this phase because deformation occurs primarily through geometric rearrangement rather than material strain. The transition to the fast phase (mean 6.4 ms post-onset) marks exhaustion of slack and engagement of material stiffness. Now the globe accelerates against the elastic resistance of stretched connective tissues and compressed fat. The 11-fold increase in displacement rate (from ~ 4 µm/ms during slow phase to ~ 49 µm/ms peak velocity) reflects this dramatic compliance change. This biphasic behaviour reflects the composite viscoelasticity of orbital soft tissues arising from both solid matrix properties and interstitial fluid flow [34]. Our WEM separation employed a Kelvin-Voigt formulation, which achieved residuals comparable to purely mathematical fitting (robust regression), suggesting performance is limited by measurement noise rather than model fidelity. Standard linear solid models [35], which incorporate stress relaxation via a second time constant, would not improve separation quality but may offer more physically interpretable parameters, particularly for characterising the incomplete elastic recovery observed at measurement end (residual displacement − 65.7 ± 63.9µm). Whether such parameters provide clinical discriminative value warrants future investigation. However, the transition timing provides quantitative constraints previously unavailable for orbital tissues. The ~ 6.4 ms slow phase duration, combined with the ~ 60 mN peak air-puff force, implies an initial orbital compliance of approximately 0.5–1.0 mm/N before tissue engagement. This value could inform patient-specific modelling for orbital surgery planning or thyroid eye disease assessment. The near-simultaneous occurrence of maximum WEM velocity and peak corneal deformation (15.84 ± 1.96 ms vs 15.85 ± 0.41 ms) reveals tight mechanical coupling that has significant implications for DCR parameter interpretation. At the moment of maximum corneal indentation, the rate of volume displacement into the anterior chamber peaks. By conservation of mass, this aqueous displacement must be accommodated, either by posterior bowing of the iris-lens diaphragm, compression of the vitreous, or posterior translation of the entire globe. The simultaneity we observe indicates that globe retraction is the primary accommodation mechanism, not iris-lens compliance as sometimes assumed. The subsequent 4.61 ms lag between deformation peak and WEM peak represents momentum-driven coasting: once corneal recovery begins (reducing the driving force), the globe continues posteriorly due to inertia before orbital tissue elasticity reverses its motion. This inertial overshoot is clinically relevant because it means WEM continues increasing after the cornea has begun recovering; the standard assumption that WEM and corneal deformation can be independently subtracted is an oversimplification. The phase relationship also explains why A2 parameters showed the largest correction effects (98.9% exceeding clinical threshold). At second applanation (~ 22 ms), WEM is near its peak and changing rapidly; small errors in WEM estimation propagate maximally to A2-derived parameters. A1 parameters, occurring during the slow WEM phase when displacement is minimal (~ 7 ms), are relatively immune. The high prevalence of bounce events (63% of measurements) and their association with lower IOP (ρ=−0.38) suggest involvement of intraocular pressure dynamics that merit closer examination. Type 1 bounce (reversal) requires momentary force reversal, the cornea briefly moves anteriorly against the air-puff direction. The only plausible mechanism is a transient IOP spike creating outward force exceeding residual air-puff pressure. During rapid corneal indentation, aqueous is displaced posteriorly, transiently elevating vitreous pressure. If this pressure wave reflects from the posterior sclera and returns to the cornea faster than the air-puff force decays, it could cause momentary anterior deflection. Type 2 bounce (plateau) represents dramatic velocity reduction without reversal, the cornea slows to near-zero velocity before resuming indentation. This pattern suggests transient equilibrium between air-puff force and internal resistance, potentially from the iris-lens diaphragm reaching its posterior excursion limit before aqueous can redistribute further. The negative correlation between bounce characteristics and IOP is mechanistically consistent: lower IOP implies a more compliant system with greater capacity for internal fluid redistribution, creating conditions favouring pressure wave reflection and transient equilibria. Higher IOP eyes have stiffer behaviour with less internal compliance, resulting in monotonic indentation. The earlier bounce onset in left eyes (13.79 vs 14.11 ms, p = 0.026) parallels the greater rotation in OS, suggesting a common underlying factor, possibly the same fixation or geometric asymmetry that amplifies rotation also affects the timing of internal pressure equilibration. The clinical significance extends beyond improved DCR accuracy to fundamental questions about biomechanical phenotyping. IOP measurement : The Corvis ST calculates deflection amplitude by subtracting WEM from total deformation; this deflection-derived signal then feeds into DCR parameters. The bIOP algorithm employs a polynomial function of these DCR parameters [36], meaning WEM errors propagate indirectly: incorrect WEM separation yields incorrect deflection, which contaminates the DCR inputs to bIOP. The asymmetric rotational artefact we document (N − T difference 44–141µm) would differentially affect nasal versus temporal deflection estimates, though the clinical magnitude of this effect on bIOP requires further investigation. Keratoconus detection : The TBI index employs a random forest classifier incorporating multiple DCR parameters; although exact feature weights are unpublished, A2 velocity and A2 length are established discriminators between subclinical keratoconus and normal eyes [38, 39]. In contrast, CBI relies on A1-phase and stiffness parameters without explicit A2 terms [39]. Our finding that A2 parameters show the largest WEM correction effects (98.9% exceeding clinical threshold) suggests TBI performance could be affected by asymmetric WEM, whereas CBI may be relatively robust. Whether correction improves or degrades diagnostic performance requires prospective evaluation; current algorithms were trained on uncorrected data and may have implicitly compensated for systematic error. Orbital biomarker applications : The emerging use of WEM to assess orbital tissue stiffness in thyroid eye disease [13, 40] and diabetes [40] assumes that WEM reflects bulk orbital compliance. Our demonstration of substantial rotational contamination suggests these applications may be measuring a composite of compliance and rotational susceptibility rather than pure translation. Correction could improve specificity for orbital tissue assessment, or the rotational component itself could prove diagnostically informative. Asymmetric fibrosis in TED might alter N:T ratios differently than symmetric fat expansion. The principal strength is the novel orthogonal validation design enabling ground-truth assessment of WEM accuracy. The systematic comparison of twelve separation methods provides definitive guidance for future implementations. Comprehensive temporal characterisation, including bounce analysis, reveals orbital biomechanics inaccessible by other clinical methods. Limitations include the absence of an a priori power calculation; the sample size (n = 68 for main analysis, n = 20 for validation subset) was determined by logistical constraints rather than formal power estimation, rendering the correction model findings exploratory. Additionally, the young, healthy, predominantly Caucasian cohort limits generalisability, as orbital fat volume and tissue properties change with age and vary ethnically[19, 41], potentially affecting asymmetry magnitude and patterns. The mechanistic explanation for laterality differences remains speculative; definitive resolution requires systematic manipulation of proposed factors (measurement order, fixation targets, head position sensors). The HSC frame rate (3,030 fps) was lower than Corvis ST (4,330 fps), limiting the temporal resolution of validation comparisons, though this affects precision rather than accuracy. Finally, the correction algorithm was developed and evaluated in the same cohort; external validation is essential before clinical implementation. Future priorities include validation in thyroid eye disease and glaucoma populations where WEM parameters have established clinical relevance, evaluation of correction effects on CBI/TBI diagnostic performance, and prospective confirmation of the regression model in adequately powered samples. Correlation with orbital MRI would test anatomical hypotheses regarding fat distribution and asymmetry origins. Resolution of the laterality effect requires systematic studies manipulating measurement order, fixation conditions, and head positioning. Declarations Funding Oculus Optikgeräte GmbH partially supported this research by providing a high-speed camera system and the Corvis ST instrument. Competing interests The authors declare no financial or non-financial competing interests related to this work. Ethics approval This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the University of Plymouth Ethics and Integrity Committee (ref. no. 13/14-222). Consent to participate Written informed consent was obtained from all individual participants included in the study. Consent to publish Not applicable. Availability of data and materials The datasets generated and analysed during the current study are available from the corresponding author on reasonable request. Code availability The Python code for the novel correction algorithm is available from the corresponding author on reasonable request. Authors' contributions DO: Conceptualisation, Methodology, Software, Validation, Formal Analysis, Investigation, Resources, Data Curation, Writing. Original Draft, Writing. Review and Editing, Visualisation, Supervision, Project Administration, Funding Acquisition. PB, HB and CP: Writing. Review and Editing. All authors read and approved the final manuscript and agree to be accountable for all aspects of the work. Acknowledgments The authors wish to acknowledge the colleagues and research assistants who contributed to data collection for this study. References Kling, S. and F. Hafezi, Corneal biomechanics - a review. Ophthalmic Physiol Opt, 2017. 37 (3): p. 240–252. Roberts, C.J. and W.J. Dupps, Jr., Biomechanics of corneal ectasia and biomechanical treatments. J Cataract Refract Surg, 2014. 40 (6): p. 991–8. Ambrósio, R., I. Ramos, and A. Luz, Dynamic ultra high speed Scheimpflug imaging for assessing corneal biomechanical properties. Revista Brasileira de Oftalmologia, 2013. 72 : p. 99–102. European Glaucoma, S., Terminology and Guidelines for Glaucoma . 5 ed. 2020: PubliComm. Liu, J. and C.J. Roberts, Influence of corneal biomechanical properties on intraocular pressure measurement: quantitative analysis. J Cataract Refract Surg, 2005. 31 (1): p. 146–55. Oculus Optikgeräte Gmb, H., Corvis ST Operating Instructions. 2020. Koprowski, R., Automatic method of analysis and measurement of additional parameters of corneal deformation in the Corvis tonometer. Biomed Eng Online, 2014. 13 : p. 150. Koprowski, R., R. Ambrosio, Jr., and S. Reisdorf, Scheimpflug camera in the quantitative assessment of reproducibility of high-speed corneal deformation during intraocular pressure measurement. J Biophotonics, 2015. 8 (11-12): p. 968–78. Aoki, S., et al., The effect of air pulse-driven whole eye motion on the association between corneal hysteresis and glaucomatous visual field progression. Scientific Reports, 2018. 8 (1). Zhang, D., et al., Exploring the Biomechanical Properties of the Human Cornea In Vivo Based on Corvis ST. Frontiers in Bioengineering and Biotechnology, 2021. 9 . Hwang, H.S., et al., A novel method for quantifying the biomechanical parameters of orbital soft tissue using a corneal dynamic scheimpflug analyser: a retrospective study. BMC Ophthalmol, 2019. 19 (1): p. 53. Leszczynska, A., et al., Measurement of Orbital Biomechanical Properties in Patients with Thyroid Orbitopathy Using the Dynamic Scheimpflug Analyzer (Corvis ST). Curr Eye Res, 2018. 43 (3): p. 289–292. Vellara, H.R., et al., In vivo ocular biomechanical compliance in thyroid eye disease. Br J Ophthalmol, 2017. 101 (8): p. 1076–1079. Vellara, H.R., et al., Quantitative Analysis of Corneal Energy Dissipation and Corneal and Orbital Deformation in Response to an Air-Pulse in Healthy Eyes. Invest Ophthalmol Vis Sci, 2015. 56 (11): p. 6941–7. Boszczyk, A., H. Kasprzak, and A. Jozwik, Eye retraction and rotation during Corvis ST 'air puff' intraocular pressure measurement and its quantitative analysis. Ophthalmic Physiol Opt, 2017. 37 (3): p. 253–262. Jannesari, M., et al., Assessment of corneal and fatty tissues biomechanical response in dynamic tonometry tests by using inverse models. Acta Bioeng Biomech, 2018. 20 (1): p. 39–48. Jannesari, M., P. Mosaddegh, and M. Kadkhodaei, Numerical and clinical investigation on the material model of the cornea in Corvis tonometry tests. Mechanics of Time-Dependent Materials, 2019. 23 : p. 373–384. Jannesari, M. and H.T. Kasprzak, In-vivo Calibration of Corneal Biomechanical Properties by Virtual Fields Method. Research Square (Research Square), 2023. Makarem, A., et al., Assessment of age-related change of the ocular support system. Front Bioeng Biotechnol, 2023. 11 : p. 1146828. Flight, L. and S.A. Julious, Practical guide to sample size calculations: non-inferiority and equivalence trials. Pharm Stat, 2016. 15 (1): p. 80–9. Mlyniuk, P., et al., Probing biomechanical properties of the cornea with air-puff-based techniques – an overview. Advanced Optical Technologies, 2021. 10 (6): p. 375–391. Wang, X., et al., Assessment of corneal biomechanics, tonometry and pachymetry with the Corvis ST in myopia. Scientific Reports, 2021. 11 (1). Oehring, D., et al., Experimental Assessment of the Spatial and Temporal Distribution of Non-Contact Tonometer Airflows. Applied Sciences, 2021. Koprowski, R., et al., Selected parameters of the corneal deformation in the Corvis tonometer. Biomed Eng Online, 2014. 13 : p. 55. Matković, K., et al., Quantitative Externalization of Visual Data Analysis Results Using Local Regression Models , in Lecture notes in computer science . 2017, Springer Science+Business Media. p. 199–218. Stefano, V.S.D., et al., Live human assessment of depth-dependent corneal displacements with swept-source optical coherence elastography. PLoS ONE, 2018. 13 (12). Koprowski, R., et al., Quantitative assessment of responses of the eyeball based on data from the Corvis tonometer. Comput Biol Med, 2015. 58 : p. 91–100. Jafari, S., et al., Finite element model of ocular adduction with unconstrained globe translation. Biomech Model Mechanobiol, 2024. 23 (2): p. 601–614. Demer, J.L., Mechanics of the orbita. Dev Ophthalmol, 2007. 40 : p. 132–57. Hotte, G.J., et al., Visualization of Sliding and Deformation of Orbital Fat During Eye Rotation. Transl Vis Sci Technol, 2016. 5 (4): p. 9. Cui, H. and Z. Zhang, Interocular symmetry and asymmetry in ocular health and disease. Front Med (Lausanne), 2025. 12 : p. 1626329. Smith, E.A., et al., Orbital volume changes during growth and development in human children assessed using cone beam computed tomography. Head Face Med, 2022. 18 (1): p. 8. King, W.M., Binocular coordination of eye movements--Hering's Law of equal innervation or uniocular control? Eur J Neurosci, 2011. 33 (11): p. 2139–46. Suh, J.K. and M.R. DiSilvestro, Biphasic Poroviscoelastic Behavior of Hydrated Biological Soft Tissue. Journal of Applied Mechanics, 1999. 66 (2): p. 528–535. Ahmed, H.M., N.M. Salem, and W. Al-Atabany, Human cornea thermo-viscoelastic behavior modelling using standard linear solid model. BMC Ophthalmol, 2023. 23 (1): p. 250. Eliasy, A., et al., Ex-vivo experimental validation of biomechanically-corrected intraocular pressure measurements on human eyes using the CorVis ST. Exp Eye Res, 2018. 175 : p. 98–102. Koc, M., et al., Biomechanical Analysis of Subclinical Keratoconus With Normal Topographic, Topometric, and Tomographic Findings. J Refract Surg, 2019. 35 (4): p. 247–252. Asroui, L., et al., Biomechanical Evaluation of Topographically and Tomographically Normal Fellow Eyes of Patients With Keratoconus. J Refract Surg, 2022. 38 (5): p. 318–325. Vinciguerra, R., et al., Detection of Keratoconus With a New Biomechanical Index. J Refract Surg, 2016. 32 (12): p. 803–810. Ohn, K., Effect of diabetes mellitus on corneal biomechanical parameters. Medicine, 2022. 101 : p. e30248. Regensburg, N.I., et al., Age and gender-specific reference values of orbital fat and muscle volumes in Caucasians. Br J Ophthalmol, 2011. 95 (12): p. 1660–3. Additional Declarations No competing interests reported. Supplementary Files STROBEChecklist.docx SupplementaryMethodsOnlineResource2.docx SupplementaryMethodsOnlineResource1.docx Cite Share Download PDF Status: Under Review Version 1 posted Reviews received at journal 18 May, 2026 Reviewers agreed at journal 08 May, 2026 Reviewers agreed at journal 08 May, 2026 Reviewers invited by journal 26 Mar, 2026 Editor assigned by journal 26 Mar, 2026 Submission checks completed at journal 25 Mar, 2026 First submitted to journal 25 Mar, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9226844","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":612782140,"identity":"ff8320ea-56ed-4c3f-b99f-03668a89d3bb","order_by":0,"name":"Daniela Oehring","email":"data:image/png;base64,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","orcid":"","institution":"University of Plymouth","correspondingAuthor":true,"prefix":"","firstName":"Daniela","middleName":"","lastName":"Oehring","suffix":""},{"id":612782141,"identity":"97de84e0-dd77-4ef3-bc49-33bfa6981909","order_by":1,"name":"Phillip Buckhurst","email":"","orcid":"","institution":"University of Plymouth","correspondingAuthor":false,"prefix":"","firstName":"Phillip","middleName":"","lastName":"Buckhurst","suffix":""},{"id":612782142,"identity":"096b2b6b-2a97-47ee-ac58-6e3d71d6db4d","order_by":2,"name":"Christine Purslow","email":"","orcid":"","institution":"Cardiff University","correspondingAuthor":false,"prefix":"","firstName":"Christine","middleName":"","lastName":"Purslow","suffix":""},{"id":612782143,"identity":"77ecccb0-e597-46e6-86d7-d951b3ee264a","order_by":3,"name":"Hetal Buckhurst","email":"","orcid":"","institution":"University of Plymouth","correspondingAuthor":false,"prefix":"","firstName":"Hetal","middleName":"","lastName":"Buckhurst","suffix":""}],"badges":[],"createdAt":"2026-03-25 20:09:58","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9226844/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9226844/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":105786456,"identity":"6d76ee66-faf7-4c34-8d3b-06b846f9429f","added_by":"auto","created_at":"2026-03-31 06:45:38","extension":"jpeg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":490859,"visible":true,"origin":"","legend":"\u003cp\u003eConceptual schematic of temporal dynamics during Corvis ST air-puff tonometry. The figure illustrates the relationship between whole eye movement (WEM, globe retraction) and corneal deformation over the ~30 ms measurement cycle. Time progresses vertically downward. WEM exhibits biphasic behavio\u003cstrong\u003eu\u003c/strong\u003er: an initial slow phase (toe region, ~0–7 ms) representing orbital tissue recruitment, followed by a fast phase of rapid globe retraction. Corneal deformation shows characteristic landmarks: A1 (first applanation), a transient bounce (deceleration or brief reversal in deformation velocity), highest concavity (HC), and A2 (second applanation). The horizontal separation between the two curves at any timepoint represents the true corneal deflection after WEM correction. Maximum amplitude occurs at HC. The recovery phase (protraction) shows the globe returning toward its original position. Understanding these coupled dynamics is essential for accurate biomechanical parameter extraction, as uncorrected WEM contaminates corneal deformation measurements.\u003c/p\u003e","description":"","filename":"Figure1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9226844/v1/77266b344a98f51f8bc93b8c.jpeg"},{"id":105904312,"identity":"1f9f3a3f-f620-40b9-8d6a-1260f012385e","added_by":"auto","created_at":"2026-04-01 10:07:17","extension":"jpeg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":444319,"visible":true,"origin":"","legend":"\u003cp\u003eForest plot of residual WEM artefact magnitude across 13 correction methods. Mean peripheral edge RMS residual (µm) with 95% confidence intervals for each WEM separation method, evaluated across 136 Corvis ST measurements from 68 subjects. Methods are ordered by performance (lowest residual = best correction). The dashed line at zero represents ideal correction with no residual WEM in the corneal deflection signal. The top-performing Dynamic Selection meta-method (green). The Traditional Edge Average (grey) represents the Corvis ST device's built-in approach. The uncorrected baseline (160.16 ± 41.60 µm) quantifies the full WEM artefact magnitude prior to any correction. Nine methods achieved comparable residuals between 4.5–5.5 µm, representing a ~97% reduction from the uncorrected artefact.\u003c/p\u003e","description":"","filename":"Figure2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9226844/v1/2a0e4f23f6f73fc6d0f8f2f5.jpeg"},{"id":105786461,"identity":"1959f2b3-239f-4ef6-8f31-51135765712d","added_by":"auto","created_at":"2026-03-31 06:45:38","extension":"jpeg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":315192,"visible":true,"origin":"","legend":"\u003cp\u003eValidation of Corvis ST-derived whole eye movement against high-speed camera reference (n=20). (a) Temporal profile of globe movement measured by high-speed camera (HSC, dark) and Corvis ST (light), showing mean ± SD. Three phases are delineated: initial phase (white), retraction (blue), and protraction (green). (b) Comparison of phase-specific parameters between measurement systems. Violin plots show distribution with median (dashed) and quartiles (dotted). Top row: velocities; bottom row: phase durations. (c) Peak amplitude (top) and time to peak (bottom) comparison. Amplitude showed no significant difference between systems (p=0.064), while time to peak differed significantly, with HSC detecting peak ~1.7ms earlier than Corvis ST (p=0.006). *p\u0026lt;0.05, **p\u0026lt;0.01, ns=not significant. HSC=high-speed camera.\u003c/p\u003e","description":"","filename":"Figure3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9226844/v1/c5601330b55caef2420d47c0.jpeg"},{"id":105786460,"identity":"2f08f7da-f77d-4a4a-ad6f-ef50b9006108","added_by":"auto","created_at":"2026-03-31 06:45:38","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":1238420,"visible":true,"origin":"","legend":"\u003cp\u003eDecomposition of whole eye movement (WEM) correction error propagation to dynamic corneal response (DCR) parameters. Sankey diagram illustrating how three components of WEM correction error flow through intermediate biomechanical pathway categories to individual DCR parameters. Error is expressed as percentage deviation from Corvis ST device reference values. Left nodes\u0026nbsp;represent three hierarchical error components:\u0026nbsp;Translation: the effect of correcting whole-eye translational movement, computed as |%Δ\u003csub\u003euncorrected\u0026nbsp;\u003c/sub\u003e− %Δ\u003csub\u003etraditional\u003c/sub\u003e| for each parameter;\u0026nbsp;Tilt: the additional effect of correcting globe tilt and per-eye optimisation, computed as |%Δ\u003csub\u003etraditional\u003c/sub\u003e−%Δ\u003csub\u003edynamic\u003c/sub\u003e|;\u0026nbsp;Refinement: sensitivity to the specific correction algorithm, computed as the range of %Δ across all 11 fixed correction methods (excluding uncorrected and dynamic). Middle nodes\u0026nbsp;group parameters into three biomechanical pathway categories:\u0026nbsp;\u003cem\u003eApex Deformation\u003c/em\u003e\u0026nbsp;(central corneal deflection depth),\u0026nbsp;\u003cem\u003eEvent Detection\u003c/em\u003e\u0026nbsp;(temporal identification of first applanation, highest concavity, and second applanation), and\u0026nbsp;\u003cem\u003eSpatial Profile\u003c/em\u003e\u0026nbsp;(spatial extent and shape of the deformation response). Right nodes\u0026nbsp;show individual DCR parameters; edge width is proportional to effect magnitude. The 25 parameters with direct Corvis ST device equivalents are shown. The three error components are not strictly additive due to non-linear interactions between WEM effects.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e","description":"","filename":"Figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-9226844/v1/1668e7834745eaf7fd399833.png"},{"id":106401744,"identity":"7b748590-b3cb-45f5-bb43-a4ca961511da","added_by":"auto","created_at":"2026-04-08 09:09:26","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":3736676,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9226844/v1/9b898fc6-fb02-4b52-b032-79a4785f45b7.pdf"},{"id":105904093,"identity":"85a8bafc-6396-4bf8-b91e-9d4398847af0","added_by":"auto","created_at":"2026-04-01 10:03:58","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":19520,"visible":true,"origin":"","legend":"","description":"","filename":"STROBEChecklist.docx","url":"https://assets-eu.researchsquare.com/files/rs-9226844/v1/d6d2a584a3d820149c0a0276.docx"},{"id":105786458,"identity":"b9b7afdb-d5bf-4feb-a87c-917bd9e42172","added_by":"auto","created_at":"2026-03-31 06:45:38","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":26759,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryMethodsOnlineResource2.docx","url":"https://assets-eu.researchsquare.com/files/rs-9226844/v1/7c54b1e428f07c2f9212f18b.docx"},{"id":105786462,"identity":"0469ec96-5a6f-4631-80e6-53211ff3730e","added_by":"auto","created_at":"2026-03-31 06:45:38","extension":"docx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":1026864,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryMethodsOnlineResource1.docx","url":"https://assets-eu.researchsquare.com/files/rs-9226844/v1/1195e208f8b20c6903a3f343.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Asymmetric whole eye movement during non-contact tonometry: independent validation and implications for corneal biomechanical parameters","fulltext":[{"header":"Key Points","content":"\u003cp\u003e\u0026sdot; Independent high-speed camera validation confirms that the Corvis ST WEM parameter accurately measures overall globe displacement during non-contact tonometry.\u003c/p\u003e\u003cp\u003e\u0026sdot; Consistent nasal-temporal asymmetry in globe retraction (N:T ratio 1.18\u0026ndash;1.69) introduces a rotational artefact not accounted for by current WEM correction algorithms.\u003c/p\u003e\u003cp\u003e\u0026sdot; Exploratory robust regression correction reduced WEM-attributable error in dynamic corneal response parameters, but requires prospective confirmation in larger samples.\u003c/p\u003e"},{"header":"Introduction","content":"\u003cp\u003eOcular biomechanics, the study of the eye's response to mechanical forces, provides fundamental insights into the pathophysiology of numerous sight-threatening conditions [1, 2]. The mechanical properties of the corneoscleral shell are critical factors in the development, diagnosis, and management of diseases, including keratoconus, post-refractive surgery ectasia, and glaucoma [3]. Of particular clinical importance is intraocular pressure (IOP) assessment, which remains the only modifiable risk factor for glaucoma [4]. However, all clinical tonometry methods are indirect, measuring the force required to deform the cornea rather than true intraocular pressure. This measurement is profoundly influenced by corneal biomechanical properties, creating a complex interplay where a stiffer cornea leads to IOP overestimation and a more compliant cornea to underestimation [5]. Given that glaucoma management decisions often depend on IOP changes of 1\u0026ndash;2 mmHg, measurement inaccuracies can significantly impact risk stratification and treatment efficacy [4].\u003c/p\u003e \u003cp\u003eThe Corneal Visualisation Scheimpflug Technology (Corvis ST; Oculus Optikger\u0026auml;te GmbH, Wetzlar, Germany) employs a high-speed Scheimpflug camera capturing images of dynamic corneal response (DCR) to a precisely metered air puff [6]. A critical analytical step is isolating true corneal deflection from confounding motion of the entire globe. During the air puff, the globe translates posteriorly into the orbit; this is termed whole eye movement (WEM) [6]. The Corvis ST software subtracts WEM from total displacement measured at the corneal apex to derive the pure corneal deflection curve. This correction is essential, as failure to account for WEM leads to substantial overestimation of corneal deflection, systematically skewing all derived DCR parameters [7\u0026ndash;10].\u003c/p\u003e \u003cp\u003eThe proprietary algorithm calculates WEM by averaging the displacement of the two outermost points of its 8 mm horizontal scan (at positions c\u0026thinsp;=\u0026thinsp;\u0026plusmn;\u0026thinsp;4 mm, where c is horizontal corneal position relative to the corneal apex) for each time frame [7]. This method is predicated on a fundamental, yet unvalidated, assumption: that posterior globe retraction is symmetrical across the horizontal meridian. While WEM has been increasingly utilised to explore orbital soft tissue properties in conditions like thyroid eye disease [11\u0026ndash;13] and to investigate relationships with axial length in myopia [14], the validity of this symmetrical assumption has not been independently verified.\u003c/p\u003e \u003cp\u003ePrior work has provided qualitative observations of asymmetric globe behaviour. Boszczyk et al. [15] noted a tendency for the globe to rotate nasally during Corvis ST measurements, though without quantification. Jannesari et al. [16\u0026ndash;18] established through inverse modelling that separating corneal deformation from globe movement is theoretically essential for accurate estimation of material properties, though no practical correction method was proposed. Computational studies by Makarem et al. [19] predicted that asymmetric orbital support would produce nasal rotation during air-puff loading, but their finite element models could not fully replicate clinically observed patterns, and they explicitly called for further investigation.\u003c/p\u003e \u003cp\u003eThis study was designed to address these gaps through a multifaceted investigation. The objectives were (1) to compare methods for separating WEM from corneal deformation and identify the optimal approach; (2) to externally validate the Corvis ST WEM parameter against a synchronised, orthogonal high-speed camera system; (3) to quantify nasal-temporal asymmetry in globe retraction and assess its impact on derived DCR parameters. We hypothesised that globe retraction during Corvis ST tonometry is accurately measured but asymmetrically distributed, leading to rotational error in standard WEM-corrected deflection analysis.\u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cp\u003eThis cross-sectional observational study analysed whole eye movement (WEM) during non-contact tonometry using Corvis ST (Oculus Optikger\u0026auml;te GmbH, Wetzlar, Germany). The study was conducted at the University of Plymouth and received ethical approval from the Ethics and Integrity Committee (ref. no. 13/14-222). All participants provided written informed consent in accordance with the Declaration of Helsinki. This study is reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines for cross-sectional studies (Online Resource 3).\u003c/p\u003e\n\u003cp\u003eHealthy adult volunteers aged 18\u0026ndash;40 years were recruited from the University of Plymouth student and staff population. Inclusion criteria were: best-corrected visual acuity \u0026ge;6/9, no history of ocular surgery, no corneal pathology, and no contact lens wear within 24 hours of measurement. Exclusion criteria included glaucoma, keratoconus or other ectatic disorders, connective tissue disorders affecting ocular or orbital biomechanics (e.g., Ehlers-Danlos syndrome, Marfan syndrome), and inability to fixate during measurement.\u003c/p\u003e\n\u003cp\u003eSixty-eight participants were enrolled. Both eyes of each participant were measured once, yielding 136 Corvis ST measurements (68 OD, 68 OS) for the main WEM analysis. A subset of 20 participants underwent three consecutive left-eye measurements with simultaneous high-speed camera (HSC) recording for external validation and reproducibility assessment. A two-minute interval was maintained between measurements to allow tissue recovery [17].\u003c/p\u003e\n\u003cp\u003eFor HSC validation, the mean of three consecutive measurements was used for each participant (n=20). For reproducibility analysis, all 60 repeated OS measurements were analysed. For the main WEM analysis, only the first OS measurement was included to maintain independence. Sample size was calculated based on equivalence testing methodology [20], expecting standard deviation of 0.083 mm for WEM [11], with \u0026alpha;=0.05 and \u0026beta;=0.10.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCorvis ST measurement protocol:\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eThe Corvis ST is a non-contact tonometer that captures 140 Scheimpflug images at 4,330 frames per second during a 32.3 ms air-puff cycle. Each horizontal cross-sectional image spans 8 mm of corneal width with 572 spatial points (resolution ~14 \u0026micro;m). The device records anterior and posterior corneal surface positions in each frame\u0026nbsp;[21, 22]. A trained examiner performed all measurements during a single session. The examiner excluded measurements with quality flags indicating poor alignment or incomplete capture. Detailed instrument specifications and spatial mapping parameters are provided in Online Resource 2.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAir-puff characteristics\u003c/strong\u003e: The air-puff profile was characterised in a separate calibration study [23]: peak pressure 96.4\u0026plusmn;1.4 mmHg at 14.8 ms, total duration 21.5\u0026plusmn;0.3 ms, full width at half maximum (FWHM) 15.2 ms, affected corneal area 45.2 mm\u0026sup2;.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eHSC validation:\u003c/strong\u003e A Corvis ST tonometer and HSC system (MotionBLITZ Director2 Kit, Mikrotron GmbH; Unterschleissheim, Germany) were positioned orthogonally. The subject\u0026apos;s left eye was located at the intersection of both instruments\u0026apos; optical axes. The HSC was fitted with a Tokina AT-X PRO M 100mm macro lens, configured at 3,030 frames per second with 560\u0026times;320 pixel resolution. A custom-designed infrared ring illuminator (48 LEDs, 17.2V, 5.5A) provided optimal imaging luminance. Light intensity at the corneal plane was 0.01\u0026plusmn;0.012 lx. A direct current generator prevented flickering. Synchronisation was achieved by recording the audible \u0026apos;click\u0026apos; from the Corvis ST joystick via a microphone positioned adjacent to it, serving as the HSC trigger signal.\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eData extraction and processing\u003c/h2\u003e\n\u003cp\u003eRaw Scheimpflug data were exported from the Corvis ST (.csv format) and processed using custom Python software (version 3.10). Anterior and posterior corneal surface positions were extracted as matrices of dimension 140 (frames)\u0026times;572 (spatial points). The neutral axis (corneal midplane) was calculated as the arithmetic mean of anterior and posterior surfaces. All surface positions were converted from millimetres to micrometres (\u0026times;1000) for subsequent analysis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEdge preprocessing:\u003c/strong\u003e Prior to WEM separation, peripheral (edge) regions were pre-processed to remove artefacts. Outliers were detected using a median absolute deviation (MAD) approach: values deviating more than 4.0 times the local standard deviation (estimated from MAD\u0026times;1.4826) from the local median (5\u0026times;5 window) were flagged. Flagged values were replaced with the local median. Edge detection quality was classified as good (outlier fraction \u0026lt;1%), marginal (1\u0026ndash;5%), or fallback (\u0026ge;5%).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eBoundary detection:\u003c/strong\u003e The peripheral region suitable for WEM estimation was identified using a two-pass adaptive algorithm. Signal-to-noise ratio (SNR) was computed for each edge region; SNR\u0026ge;10 indicated good quality, \u0026ge;5 marginal quality, and \u0026lt;5 required fallback estimation. Edge width (number of columns used) averaged 0.75\u0026plusmn;0.15 mm (53\u0026ndash;54 columns).\u003c/p\u003e\n\u003ch2\u003eWEM separation methods\u003c/h2\u003e\n\u003cp\u003eTwelve methods for separating WEM from true corneal deformation were implemented and compared: (1) Traditional edge-averaging: Mean displacement of peripheral columns (equivalent to Corvis ST built-in correction) [24]; (2) Linear regression: Models WEM as translation plus tilt; displacement and tilt smoothed using Savitzky-Golay filter (window 11 frames, polynomial order 2) [14, 24]; (3) Robust regression: Huber regression on all peripheral points, resistant to outliers [25]; (4) \u0026nbsp;Adaptive edge: Dynamically determines edge width based on deformation boundary detection; (5) Cross-correlation alignment: Minimises peripheral RMS difference between each frame and baseline [26]; (6) Frequency-domain filtering: Low-pass filter (50 Hz cutoff) to isolate slow WEM from rapid corneal deformation [27]; (7) Viscoelastic modelling: Kelvin-Voigt model of globe response to measured air-puff force [23]; (8) Physiological viscoelastic: Two-component model with separation at 80 Hz; (9) Adaptive-viscoelastic hybrid: Combines methods 4 and 7 (10) Adaptive-robust hybrid: Combines methods 4 and 3; (11) Dynamic selection: Evaluates all methods per measurement, selects lowest peripheral RMS residual; (12) Arc-corrected methods: Account for corneal curvature in peripheral displacement estimation.\u003c/p\u003e\n\u003cp\u003eAll methods (except traditional) model WEM as a combination of translation and rotation: WEM(t, x)=d(t) + \u0026theta;(t) \u0026middot; x, where d(t) is posterior displacement (\u0026micro;m), \u0026theta;(t) is tilt (\u0026micro;m/mm), and x is the spatial position (mm). Corneal deformation was then calculated as: deformation(t, x)=[raw(t, x) \u0026minus; raw(0, x)] \u0026minus; WEM(t, x). Method performance was evaluated using root-mean-square (RMS) residuals in peripheral regions where true corneal deformation is minimal. The optimal method for each measurement was identified; robust regression was optimal for 95.3% of measurements and was adopted for all subsequent analyses. Full mathematical derivations for all separation methods are provided in Online Resource 1.\u003c/p\u003e\n\u003ch2\u003eParameter calculation\u003c/h2\u003e\n\u003cp\u003e\u003cstrong\u003eCorvis-equivalent parameters\u0026nbsp;\u003c/strong\u003e(Figure 1)\u003cstrong\u003e:\u003c/strong\u003e First applanation (A1) was identified as the first negative-to-positive zero-crossing in corneal curvature during the loading phase (5\u0026ndash;12 ms). Highest concavity (HC) was the frame with maximum inward deformation. Second applanation (A2) was the first positive-to-negative curvature zero-crossing during recovery (HC+2ms to 28ms). Sub-frame interpolation was applied for temporal precision. Velocity profiles were computed using central differences on Savitzky-Golay smoothed deformation (window 9, order 3), converted to m/s. Radius of curvature at HC was estimated via linearised circle fitting on the central 3 mm region.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eGlobe translation and rotation:\u003c/strong\u003e Globe movement was decomposed into translational and rotational components using peripheral displacements at the nasal (c=\u0026minus;4 mm) and temporal (c=+4 mm) boundaries, where c denotes horizontal corneal position relative to the corneal apex (negative=nasal, positive=temporal). Translation was defined as the mean of nasal and temporal displacements: Translation(t)=[d\u003csub\u003eN\u003c/sub\u003e(t) + d\u003csub\u003eT\u003c/sub\u003e(t)] / 2. Rotation angle was derived from the differential displacement: Rotation(t)=arctan[(d\u003csub\u003eT\u003c/sub\u003e(t) \u0026minus; d\u003csub\u003eN\u003c/sub\u003e(t)) / 7.59 mm], where 7.59 mm is the span between peripheral measurement positions. Positive rotation indicates nasal-ward tilt (temporal side displacing more posteriorly). The nasal:temporal (N:T) ratio was calculated as peak nasal displacement divided by peak temporal displacement.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTemporal dynamics\u0026nbsp;\u003c/strong\u003e(Figure 1)\u003cstrong\u003e:\u003c/strong\u003e WEM temporal phases were characterised as follows. Movement onset was defined as the first frame exceeding 5% of peak displacement. The slow-fast transition was identified when velocity crossed 20% of maximum retraction velocity. Slow-loading phase duration was measured from onset to transition; fast-loading phase duration from transition to peak amplitude. Following peak displacement, the globe returned anteriorly. Recovery metrics included percentage return to baseline and time to 50% and 90% recovery. Peak retraction and protraction velocities were recorded, along with asymmetry ratios comparing retraction and protraction characteristics.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eVelocity oscillations:\u003c/strong\u003e High-frequency oscillations in WEM velocity were quantified during the slow loading phase (5\u0026ndash;16 ms) and recovery phase (peak to end). The velocity signal was smoothed using a Savitzky-Golay filter (window=7, order=3) and detrended to isolate oscillatory components. Oscillation count was determined by identifying zero-crossings in the detrended velocity signal. Oscillation amplitude was calculated as the mean absolute peak-to-trough amplitude. Dominant frequency was estimated using Welch power spectral density analysis within the 10\u0026ndash;200 Hz band. Measurements were classified as low (\u0026lt;3 oscillations), moderate (3\u0026ndash;5 oscillations), or high (\u0026gt;5 oscillations) based on total oscillation count.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDeformation bounce\u0026nbsp;\u003c/strong\u003e(Figure 1)\u003cstrong\u003e:\u003c/strong\u003e Bounce events, transient velocity reductions or reversals during corneal loading, were detected using two algorithms. Type 1 (reversal) identified local extrema in the smoothed deformation profile (scipy find_peaks, minimum prominence 1.5\u0026micro;m). Type 2 (plateau) identified velocity ratio |v(t)|/running maximum\u0026lt;0.25, indicating dramatic slowing without direction reversal. Bounce severity was classified as none, mild (max amplitude\u0026le;5\u0026micro;m), moderate (max amplitude\u0026gt;5\u0026le;20\u0026micro;m), severe (max amplitude\u0026gt;20\u0026micro;m), or multiple (n\u0026gt;3) based on the number and amplitude of events.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCrater geometry:\u003c/strong\u003e The air-puff-induced deformation crater was characterised at the frame of maximum corneal indentation. Crater centre was determined as the weighted mean position, using deformation magnitude as weights. Crater offset from the corneal centre was calculated and classified as nasal, temporal, or central (offset\u0026lt;0.2 mm). Crater extent was defined as the region where deformation exceeded 10% of maximum depth. Extent asymmetry was quantified by comparing the spread toward nasal versus temporal peripheries. Corneal inclination was assessed from the baseline (pre-air-puff) corneal profile by calculating the angle between peripheral edge heights.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eOutlier detection and data cleaning:\u003c/strong\u003e A hierarchical outlier detection approach was applied, to remove artefacts from lashes or eyelids: (1) IQR-based detection with values beyond Q1 \u0026minus; 2.0\u0026times;IQR or Q3 + 2.0\u0026times;IQR; (2) Z-score detection with absolute z-score \u0026gt; 3.0; (3) Validity range checks for physiologically implausible values (e.g., IOP \u0026lt;5 or \u0026gt;40 mmHg). Outliers were set to missing (NaN) rather than excluded, preserving measurement-level data. The hierarchy proceeded from foundational parameters to derived parameters to prevent cascading outlier flags.\u003c/p\u003e\n\u003ch2\u003eStatistical analysis\u003c/h2\u003e\n\u003cp\u003eAll analyses were performed using Python 3.10 with scipy (version 1.11), numpy (1.24), and pandas (2.0). Continuous variables are presented as mean\u0026plusmn;standard deviation (SD) with 95% confidence intervals (CI) based on the t-distribution, or as median with interquartile range (IQR) for non-normally distributed data. Categorical variables are presented as counts with percentages and Wilson 95% CI. Normality was assessed using both Shapiro-Wilk (n\u0026le;5000) and D\u0026apos;Agostino-Pearson (n\u0026ge;20) tests; data were considered normally distributed only if both tests yielded p\u0026gt;0.05. Paired comparisons used paired t-tests for normally distributed differences or Wilcoxon signed-rank tests otherwise. Independent comparisons used Student\u0026apos;s or Welch\u0026apos;s t-test (based on Levene\u0026apos;s test for variance equality) for normal data, or Mann-Whitney U test for non-normal data. Effect sizes are reported as Cohen\u0026apos;s d for independent comparisons and Cohen\u0026apos;s d\u003csub\u003ez\u003c/sub\u003e for paired comparisons. Interpretation followed conventional thresholds: small (0.2), medium (0.5), large (0.8). Holm-Bonferroni correction was applied for pairwise method comparisons. For correlations with ocular parameters, p\u0026lt;0.05 was used as the significance threshold given the exploratory nature; no further correction for multiple comparisons was applied to these analyses, and readers should interpret nominally significant associations accordingly. Spearman rank correlations were used for associations between WEM parameters and ocular characteristics. Multiple linear regression using simultaneous entry of all predictors was used to identify predictors of rotation amplitude, with variance inflation factors (VIF) assessed for multicollinearity (VIF \u0026gt; 10 indicating concern). Per-eye models were also fitted to examine whether predictors differed between right and left eyes. The regression analyses are considered exploratory and hypothesis-generating. For reproducibility analysis, within-subject coefficient of variation (CV) and intraclass correlation coefficient (ICC, two-way random, single measures) were calculated. ICC values were interpreted following Koo and Li (2016): below 0.50 poor, 0.50\u0026ndash;0.75 moderate, 0.75\u0026ndash;0.90 good, above 0.90 excellent.\u003c/p\u003e\n\u003cp\u003eAll analyses were performed using custom Python code. Complete source code, parameter documentation, and reproducibility scripts are available at upon request. The analysis pipeline generated 285 output parameters per measurement, organised by category: Corvis device (40), WEM and derived (81), geometry (12), crater dynamics (15), bounce detection (11), and other calculated parameters (126).\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eThe sample was predominantly female (70.1%; Table 1). Normality testing (Shapiro-Wilk) indicated non-normal distributions for age (W=0.830, p \u0026lt; 0.001) and IOP (W=0.957, p=0.006), while CCT (p=0.533) was normally distributed.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 1. Participant characteristics by sex\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eParameter\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTotal (n=68)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eFemale (n=48)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMale (n=20)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ep-value\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003eAge (years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22px;\"\u003e\n \u003cp\u003e22.65\u0026plusmn;4.14\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22px;\"\u003e\n \u003cp\u003e22.33\u0026plusmn;3.99\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22px;\"\u003e\n \u003cp\u003e23.60\u0026plusmn;4.48\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003e0.550\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003eCCT (\u0026micro;m)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22px;\"\u003e\n \u003cp\u003e541.7\u0026plusmn;38.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22px;\"\u003e\n \u003cp\u003e543.4\u0026plusmn;38.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22px;\"\u003e\n \u003cp\u003e538.2\u0026plusmn;38.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003e0.600\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003eIOP (mmHg)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22px;\"\u003e\n \u003cp\u003e13.68\u0026plusmn;1.87\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22px;\"\u003e\n \u003cp\u003e13.52\u0026plusmn;1.81\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22px;\"\u003e\n \u003cp\u003e14.02\u0026plusmn;1.98\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003e0.272\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cem\u003eCCT=central corneal thickness; IOP=intraocular pressure;\u003c/em\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003cem\u003eValues are mean\u0026plusmn;SD. p-values from Mann-Whitney U test.\u003c/em\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eWEM separation method comparison:\u0026nbsp;\u003c/strong\u003eLower RMS residuals indicate superior separation of globe movement from corneal deflection in peripheral regions where air-puff-induced deformation is minimal (Table 2, Figure 2). Dynamic selection and viscoelastic modelling achieved equivalently optimal performance with no significant difference between them (Holm-adjusted p\u0026lt;0.99). Both significantly outperformed all other approaches (all Holm-adjusted p\u0026lt;0.001). The viscoelastic method achieved an 87.3% reduction in RMS residuals compared to traditional edge-averaging (effect size d=1.69, p\u0026lt;0.001), the method currently employed by Corvis ST. Frequency-domain filtering performed poorest, significantly worse than traditional edge-averaging (p\u0026lt;0.001). Three distinct performance clusters emerged: optimal performers (dynamic, viscoelastic, adaptive-viscoelastic, robust regression, adaptive-robust), intermediate performers (cross-correlation, adaptive edge, linear regression, arc-corrected), and poor performers (viscoelastic-physiological, traditional, frequency-based). Notably, arc-correction, which accounts for corneal curvature rather than assuming linear peripheral geometry, performed slightly worse than uncorrected linear regression (mean difference +0.11 \u0026micro;m, d=0.73, p\u0026lt;0.001), with linear regression outperforming arc-correction in 79.5% of measurements. This validates that the linear spatial approximation is adequate for the \u0026plusmn;4 mm Corvis ST measurement window. Per-measurement analysis identified robust regression as optimal in 95.3% of cases, with adaptive-viscoelastic optimal in the remainder. Comparison of dynamic versus fixed method selection showed negligible practical improvement (Cohen\u0026apos;s d=0.16), with only 0.7% of measurements benefiting from dynamic selection. 6 participants (8.8%) had different optimal methods assigned to each eye (all had robust regression for OD and adaptive-viscoelastic for OS). This suggests method selection may be influenced by eye-specific characteristics.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 2. Peripheral RMS residuals (\u0026micro;m) by separation method (n=68, 136 measurements).\u0026nbsp;\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 4px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e#\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 35px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMethod\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMean \u0026plusmn; SD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMedian (IQR)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eRange (Min-Max)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 12px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e95% CI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 4px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 35px;\"\u003e\n \u003cp\u003eDynamic selection\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e4.54\u0026plusmn;2.90\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e3.82 (1.89)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003e0.85-20.19\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 12px;\"\u003e\n \u003cp\u003e4.09-5.11\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 4px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 35px;\"\u003e\n \u003cp\u003eViscoelastic (Kelvin-Voigt)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e4.60\u0026plusmn;2.90\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e3.82 (1.89)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003e0.85-20.19\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 12px;\"\u003e\n \u003cp\u003e4.09-5.12\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 4px;\"\u003e\n \u003cp\u003e3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 35px;\"\u003e\n \u003cp\u003eAdaptive-viscoelastic\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e4.61\u0026plusmn;2.99\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e3.82 (1.98)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003e0.85-20.19\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 12px;\"\u003e\n \u003cp\u003e4.10-5.12\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 4px;\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 35px;\"\u003e\n \u003cp\u003eRobust regression\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e4.74\u0026plusmn;2.97\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e3.96 (1.93)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003e0.86-20.63\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 12px;\"\u003e\n \u003cp\u003e4.22-5.26\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 4px;\"\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 35px;\"\u003e\n \u003cp\u003eAdaptive-robust\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e4.75\u0026plusmn;2.97\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e3.96 (2.01)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003e0.86-20.63\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 12px;\"\u003e\n \u003cp\u003e4.23-5.27\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 4px;\"\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 35px;\"\u003e\n \u003cp\u003eCross-correlation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e5.15\u0026plusmn;3.34\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e4.17 (1.91)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003e1.93-25.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 12px;\"\u003e\n \u003cp\u003e4.56-5.73\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 4px;\"\u003e\n \u003cp\u003e7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 35px;\"\u003e\n \u003cp\u003eAdaptive edge\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e5.15\u0026plusmn;3.33\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e4.18 (1.94)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003e1.93-25.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 12px;\"\u003e\n \u003cp\u003e4.56-5.73\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 4px;\"\u003e\n \u003cp\u003e8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 35px;\"\u003e\n \u003cp\u003eLinear regression\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e5.41\u0026plusmn;4.09\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e4.18 (1.94)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003e1.93-29.19\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 12px;\"\u003e\n \u003cp\u003e4.69-6.13\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 4px;\"\u003e\n \u003cp\u003e9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 35px;\"\u003e\n \u003cp\u003eArc-corrected\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e5.52 \u0026plusmn; 4.07\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e4.36 (1.89)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003e1.94\u0026ndash;29.19\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 12px;\"\u003e\n \u003cp\u003e4.81\u0026ndash;6.23\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 4px;\"\u003e\n \u003cp\u003e10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 35px;\"\u003e\n \u003cp\u003eViscoelastic (physiological)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e13.43\u0026plusmn;3.57\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e12.95 (3.79)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003e6.64-28.49\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 12px;\"\u003e\n \u003cp\u003e12.80-14.06\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 4px;\"\u003e\n \u003cp\u003e11\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 35px;\"\u003e\n \u003cp\u003eTraditional edge-averaging\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e36.33\u0026plusmn;18.76\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e33.43 (23.87)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003e7.86-105.51\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 12px;\"\u003e\n \u003cp\u003e33.03-39.62\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 4px;\"\u003e\n \u003cp\u003e12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 35px;\"\u003e\n \u003cp\u003eFrequency-domain filtering\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e46.70\u0026plusmn;10.03\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e48.27 (9.46)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003e14.89-66.21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 12px;\"\u003e\n \u003cp\u003e44.94-48.46\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cem\u003eMethods ranked by mean RMS residual (ascending). IQR\u003c/em\u003e\u003cem\u003e=\u003c/em\u003e\u003cem\u003einterquartile range.\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePeripheral edge detection and crater geometry:\u0026nbsp;\u003c/strong\u003eThe peripheral regions used for WEM estimation and the air-puff-induced deformation crater were characterised to assess measurement quality and spatial symmetry (Table 3). Edge detection quality was good in 64.7% of measurements, with the peripheral region averaging 0.75 mm in width (~54 Scheimpflug columns). The deformation crater was highly symmetric at HC (N:T extent ratio=1.00\u0026plusmn;0.02) and centrally positioned (99.3% within \u0026plusmn;0.2mm of centre), confirming uniform air-puff loading. Crater expansion (8.23ms) exceeded retraction duration (6.35ms), while retraction velocity exceeded expansion velocity (3.81 vs 3.19mm/ms).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 3. Edge detection and crater characteristics (n=136)\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 40px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eParameter\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eValue\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eEdge detection\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 40px;\"\u003e\n \u003cp\u003eDetection method: noise floor / fallback\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e135 (99.3%) / 1 (0.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 40px;\"\u003e\n \u003cp\u003eQuality: good / marginal / fallback\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e88 (64.7%) / 39 (28.7%) / 9 (6.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 40px;\"\u003e\n \u003cp\u003eEdge width, mean \u0026plusmn; SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e0.75 \u0026plusmn; 0.15 mm (53.5 \u0026plusmn; 10.8 columns)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 40px;\"\u003e\n \u003cp\u003eEdge width, range\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e0.14\u0026ndash;1.23 mm (10\u0026ndash;88 columns)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 40px;\"\u003e\n \u003cp\u003eEdge residual RMS, mean \u0026plusmn; SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e4.59 \u0026plusmn; 2.90 \u0026micro;m\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCrater geometry (at HC)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 40px;\"\u003e\n \u003cp\u003eCrater width\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e5.65 \u0026plusmn; 0.17 mm (5.20\u0026ndash;6.19)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 40px;\"\u003e\n \u003cp\u003eNasal extent\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e2.81 \u0026plusmn; 0.09 mm (2.61\u0026ndash;3.03)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 40px;\"\u003e\n \u003cp\u003eTemporal extent\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e2.83 \u0026plusmn; 0.08 mm (2.59\u0026ndash;3.03)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 40px;\"\u003e\n \u003cp\u003eExtent ratio (N:T)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e1.00 \u0026plusmn; 0.02 (0.81\u0026ndash;1.04)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 40px;\"\u003e\n \u003cp\u003eMaximum deformation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e907.1 \u0026plusmn; 81.8 \u0026micro;m (705.5\u0026ndash;1104.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCrater position\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 40px;\"\u003e\n \u003cp\u003eCrater offset\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e0.07 \u0026plusmn; 0.06 mm (\u0026minus;0.09\u0026ndash;0.22)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 40px;\"\u003e\n \u003cp\u003eApex offset\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e0.19 \u0026plusmn; 0.23 mm (\u0026minus;0.39\u0026ndash;0.80)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 40px;\"\u003e\n \u003cp\u003eCorneal inclination\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e1.87 \u0026plusmn; 1.84\u0026deg; (\u0026minus;2.81\u0026ndash;6.59)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCrater dynamics\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 40px;\"\u003e\n \u003cp\u003eMaximum crater width\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e5.70 \u0026plusmn; 0.16 mm (5.32\u0026ndash;6.18)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 40px;\"\u003e\n \u003cp\u003eCrater onset\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e6.91 \u0026plusmn; 0.22 ms (6.24\u0026ndash;7.62)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 40px;\"\u003e\n \u003cp\u003eCrater duration\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e14.59 \u0026plusmn; 0.58 ms (13.17\u0026ndash;16.17)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 40px;\"\u003e\n \u003cp\u003eExpansion duration\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e8.23 \u0026plusmn; 0.98 ms (5.78\u0026ndash;10.86)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 40px;\"\u003e\n \u003cp\u003eRetraction duration\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e6.35 \u0026plusmn; 1.10 ms (3.01\u0026ndash;9.01)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 40px;\"\u003e\n \u003cp\u003ePeak expansion rate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e3.19 \u0026plusmn; 0.19 mm/ms (2.66\u0026ndash;3.58)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 40px;\"\u003e\n \u003cp\u003ePeak retraction rate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e3.81 \u0026plusmn; 0.19 mm/ms (3.37\u0026ndash;4.38)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cem\u003eValues are mean \u0026plusmn; SD (range) unless otherwise specified. HC\u003c/em\u003e\u003cem\u003e=\u003c/em\u003e\u003cem\u003ehighest concavity; N:T\u003c/em\u003e\u003cem\u003e=\u003c/em\u003e\u003cem\u003enasal:temporal.\u003c/em\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eExternal validation against HSC:\u0026nbsp;\u003c/strong\u003ePeak amplitude measured by HSC (373.0\u0026plusmn;129.0 \u0026micro;m) was numerically larger than Corvis ST-derived values (307\u0026plusmn;68.5 \u0026micro;m), though this difference did not reach statistical significance (p=0.064) (Figure 3). Time to peak differed significantly between systems, with HSC detecting peak amplitude approximately 1.7 ms earlier than Corvis ST (19.3 vs 21.1ms, p=0.006). The temporal discrepancy was attributed to different triggering mechanisms. Linear regression indicated a consistent 3.6ms delay. Applying this correction improved temporal coverage from 56% to 83%, confirming the fundamental validity of Corvis ST temporal tracking.\u003c/p\u003e\n\u003cp\u003eAnalysis of three consecutive measurements (n=20) showed no significant difference between repeated measurements (Friedman \u0026chi;\u0026sup2;=1.83, df=2, p=0.401). Peak amplitude demonstrated moderate reliability with ICC(2,1)=0.608 (95% CI [0.408, 0.808], F=6.53, p=0.006), though the wide confidence interval spanning poor to good categories reflects the small repeatability subset (n=20); within-subject coefficient of variation (CV) was 24.6%. Time to peak showed moderate-to-good reliability with ICC(2,1)=0.720 (95% CI [0.520, 0.920]) and within-subject CV of 6.3%.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eNasal-Temporal asymmetry:\u0026nbsp;\u003c/strong\u003eContrary to the Corvis ST assumption of symmetric peripheral displacement, nasal retraction consistently exceeded temporal retraction (Table 4). Despite symmetric crater geometry (N:T extent ratio=1.00), globe retraction was markedly asymmetric, suggesting orbital tissue properties rather than non-uniform loading. Asymmetry was more pronounced in OS than OD (N:T ratio 1.69\u0026plusmn;0.34 vs 1.18\u0026plusmn;0.20, d=\u0026minus;1.83). Classification: nasal-dominant (N:T\u0026gt;1.2) 72.4%, symmetric (0.8\u0026ndash;1.2) 25.2%, temporal-dominant (\u0026lt;0.8) 2.4%. Globe movement decomposed into translation (282.4\u0026plusmn;56.5\u0026micro;m) and rotation (1.00\u0026plusmn;0.44\u0026deg;, peak \u0026minus;0.98\u0026plusmn;0.50\u0026deg;). Negative rotation indicates nasal-ward tilt. Translation did not differ between eyes (p=0.136); rotation was significantly greater in OS (1.23\u0026plusmn;0.43\u0026deg; vs 0.78\u0026plusmn;0.33\u0026deg;, d=\u0026minus;1.19).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 4. Nasal-temporal asymmetry by laterality\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 34px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eParameter\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 15px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eOD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eOS\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 16px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ep\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 15px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ed\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 34px;\"\u003e\n \u003cp\u003eAsymmetry\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 15px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 16px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 15px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 34px;\"\u003e\n \u003cp\u003eNasal amplitude (\u0026micro;m)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 15px;\"\u003e\n \u003cp\u003e305.0\u0026plusmn;82.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e362.5\u0026plusmn;73.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 16px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 15px;\"\u003e\n \u003cp\u003e\u0026minus;0.74\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 34px;\"\u003e\n \u003cp\u003eTemporal amplitude (\u0026micro;m)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 15px;\"\u003e\n \u003cp\u003e257.3\u0026plusmn;51.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e221.0\u0026plusmn;51.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 16px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 15px;\"\u003e\n \u003cp\u003e0.70\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 34px;\"\u003e\n \u003cp\u003eN:T ratio\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 15px;\"\u003e\n \u003cp\u003e1.18\u0026plusmn;0.20\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e1.69\u0026plusmn;0.34\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 16px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 15px;\"\u003e\n \u003cp\u003e\u0026minus;1.83\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 34px;\"\u003e\n \u003cp\u003eN\u0026minus;T difference (\u0026micro;m)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 15px;\"\u003e\n \u003cp\u003e44.3\u0026plusmn;52.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e140.6\u0026plusmn;66.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 16px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 15px;\"\u003e\n \u003cp\u003e\u0026minus;1.61\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 34px;\"\u003e\n \u003cp\u003eMovement components\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 15px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 16px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 15px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 34px;\"\u003e\n \u003cp\u003eTranslation amplitude (\u0026micro;m)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 15px;\"\u003e\n \u003cp\u003e275.7\u0026plusmn;55.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e289.2\u0026plusmn;57.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 16px;\"\u003e\n \u003cp\u003e0.136\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 15px;\"\u003e\n \u003cp\u003e\u0026minus;0.24\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 34px;\"\u003e\n \u003cp\u003eRotation angle (\u0026deg;)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 15px;\"\u003e\n \u003cp\u003e0.78\u0026plusmn;0.33\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e1.23\u0026plusmn;0.43\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 16px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 15px;\"\u003e\n \u003cp\u003e\u0026minus;1.19\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cem\u003eN:T=nasal/temporal amplitude; values\u0026gt;1.0=nasal predominance. Negative rotation=nasal-ward tilt. Overall: translation 282.4\u0026plusmn;56.5\u0026micro;m, rotation amplitude 1.00\u0026plusmn;0.44\u0026deg;, peak rotation \u0026minus;0.98\u0026plusmn;0.50\u0026deg;, reversals 1.02\u0026plusmn;2.56 (median 0).\u003c/em\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTemporal dynamics of globe movement:\u0026nbsp;\u003c/strong\u003eWEM exhibited biphasic retraction (slow phase 6.42\u0026plusmn;3.17ms, fast phase 10.56\u0026plusmn;2.88ms) followed by faster protraction (velocity asymmetry ratio 1.27\u0026plusmn;0.35) (Table 5). Maximum WEM velocity occurred near peak corneal deformation (15.87\u0026plusmn;1.97ms vs 15.85\u0026plusmn;0.41ms); peak WEM displacement lagged by 4.61\u0026plusmn;0.60ms. Only 5.5% of eyes returned to within \u0026plusmn;10\u0026micro;m of baseline; mean residual displacement was \u0026minus;65.7\u0026plusmn;63.9\u0026micro;m. Temporal dynamics were broadly similar between eyes, though paired testing revealed a significant laterality difference in duration asymmetry ratio (p=0.004) and peak timing (p=0.040).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 5. WEM temporal dynamics\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eParameter\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMean\u0026plusmn;SD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMedian (IQR)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eRange\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eRetraction\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003eSlow phase duration (ms)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e6.42\u0026plusmn;3.17\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e6.93 (4.27\u0026ndash;8.78)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e0.00\u0026ndash;13.86\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003eFast phase duration (ms)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e10.56\u0026plusmn;2.88\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e9.93 (8.32\u0026ndash;12.24)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e4.85\u0026ndash;18.02\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003eMax velocity (\u0026micro;m/ms)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e\u0026minus;48.54\u0026plusmn;16.27\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e\u0026minus;46.12 (\u0026minus;59.46\u0026ndash;\u0026ndash;36.77)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e\u0026minus;94.11\u0026ndash;\u0026ndash;11.97\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eProtraction\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003eSlow phase duration (ms)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e0.97\u0026plusmn;0.97\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e0.69 (0.00\u0026ndash;1.62)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e0.00\u0026ndash;3.70\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003eFast phase duration (ms)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e10.08\u0026plusmn;0.92\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e10.16 (9.47\u0026ndash;10.86)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e7.39\u0026ndash;12.24\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003eMax velocity (\u0026micro;m/ms)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e37.76\u0026plusmn;8.23\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e38.27 (33.16\u0026ndash;43.27)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e16.72\u0026ndash;58.46\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003ePeak and recovery\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003ePeak amplitude (\u0026micro;m)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e282.35\u0026plusmn;56.63\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e286.41 (243.92\u0026ndash;323.85)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e135.72\u0026ndash;411.55\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003eTime to peak (ms)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e21.07\u0026plusmn;0.51\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e21.02 (20.79\u0026ndash;21.37)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e19.87\u0026ndash;22.87\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003eTime to 50% recovery (ms)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e6.62\u0026plusmn;0.94\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003eTime to 90% recovery (ms)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e11.20\u0026plusmn;1.90\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003eFinal residual (\u0026micro;m)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e\u0026minus;65.7\u0026plusmn;63.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cem\u003eNegative velocities=retraction (posterior); positive=protraction (anterior).\u003c/em\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eWEM oscillation and corneal deflection bounce:\u0026nbsp;\u003c/strong\u003eHigh-frequency velocity WEM oscillations averaged 3.65\u0026plusmn;1.38 per measurement, predominantly during loading (4.29\u0026plusmn;1.45) versus recovery (1.02\u0026plusmn;0.96); 29.1% showed no recovery oscillations (Table 6). Classification: low (\u0026lt;3) 20.5%, moderate (3\u0026ndash;5) 55.9%, high (\u0026gt;5) 23.6%.\u003c/p\u003e\n\u003cp\u003eCorneal deflection bounce, transient slowing or reversal before HC, occurred in 63.0% of measurements (Table 6). Two types: reversal (Type 1, local extremum) and plateau (Type 2, flattening without reversal). Among bounces: 31.3% reversal only, 52.5% plateau only, 16.3% both. Severity: none 37.0%, moderate 5.5%, severe 55.9%, multiple 1.6%.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 6. Oscillation and bounce characteristics\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eParameter\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMean\u0026plusmn;SD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMedian (IQR)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eRange\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eOscillations\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003eTotal count\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e3.6\u0026plusmn;1.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e4 (3\u0026ndash;4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e1\u0026ndash;8\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003eLoading phase count\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e4.3\u0026plusmn;1.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e4 (3\u0026ndash;5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e1\u0026ndash;8\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003eRecovery phase count\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e1.0\u0026plusmn;0.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e1 (0\u0026ndash;1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e0\u0026ndash;5\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003eAmplitude (\u0026micro;m/ms)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e11.45\u0026plusmn;5.51\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e10.11 (7.42\u0026ndash;15.17)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e2.25\u0026ndash;26.38\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eBounce\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003eFirst bounce time (ms)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e13.96\u0026plusmn;0.86\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e14.09 (13.63\u0026ndash;14.39)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e11.32\u0026ndash;17.09\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003eMax amplitude (\u0026micro;m)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e35.14\u0026plusmn;35.73\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e33.31 (0.00\u0026ndash;55.03)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e0.00\u0026ndash;139.48\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003eDuration (ms)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e0.69\u0026plusmn;0.65\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e0.69 (0.00\u0026ndash;1.16)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e0.00\u0026ndash;2.54\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003eEvent count\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e0.8\u0026plusmn;0.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e1 (0\u0026ndash;1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e0\u0026ndash;3\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003eReversals (Type 1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e0.3\u0026plusmn;0.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e0 (0\u0026ndash;1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e0\u0026ndash;2\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003ePlateaus (Type 2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e0.5\u0026plusmn;0.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e0 (0\u0026ndash;1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e0\u0026ndash;2\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003eVelocity variability (\u0026micro;m/ms)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e1.248\u0026plusmn;0.026\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e1.247 (1.23\u0026ndash;1.26)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e1.18\u0026ndash;1.32\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eMost oscillation and bounce parameters did not differ between eyes. Significant laterality effects: recovery oscillations higher in OS (1.19\u0026plusmn;1.03 vs 0.84\u0026plusmn;0.86, p=0.030, d=\u0026minus;0.37); first bounce earlier in OS (13.79\u0026plusmn;0.87ms vs 14.11\u0026plusmn;0.81ms, p=0.026, d=0.29).\u003c/p\u003e\n\u003cp\u003eEyes with lower IOP, faster A1, and greater deformation exhibited more variable pre-HC velocity. Greater deformation associated with earlier bounce onset (Table 7).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 7. Bounce correlations with ocular parameters\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eParameter\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eOcular parameter\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026rho;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ep\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003eVelocity variability\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003eIOP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e\u0026minus;0.375\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003eVelocity variability\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003eA1 time\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e\u0026minus;0.374\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003eVelocity variability\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003eHC DefAmp\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e+0.335\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26px;\"\u003e\n \u003cp\u003eFirst bounce time\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003eHC DefAmp\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 24px;\"\u003e\n \u003cp\u003e\u0026minus;0.299\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e0.004\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cstrong\u003eAetiology of asymmetric globe movement:\u003c/strong\u003e Bivariate correlations with rotation amplitude (Table 8): apex offset (r=\u0026minus;0.251, p=0.004), SimK 3mm (r=\u0026minus;0.245, p=0.025), X position (r=0.234, p=0.033); non-significant: ZonalK 7mm (r=\u0026minus;0.217, p=0.060), CCT (r=\u0026minus;0.165, p=0.116), crater offset (r=\u0026minus;0.112, p=0.212), Z position (r=\u0026minus;0.033, p=0.756), IOP (r=\u0026minus;0.010, p=0.928). Apex offset and X position were collinear (r=\u0026minus;0.996). SimK 3mm was the only significant predictor: steeper corneas exhibited greater rotation. Per-eye models: OD (R\u0026sup2;=0.083, Adj R\u0026sup2;=\u0026minus;0.060, F=0.58, p=0.716) not significant; OS (R\u0026sup2;=0.281, Adj R\u0026sup2;=0.189, F=3.05, p=0.020) significant with SimK 3mm (\u0026beta;=\u0026minus;0.47, p=0.005) and Z position (\u0026beta;=\u0026minus;0.74, p=0.054) as predictors. This asymmetric predictive relationship may contribute to greater rotation in left eyes.\u003c/p\u003e\n\u003cp\u003eTable 8. Multivariate predictors of rotation amplitude\u0026nbsp;\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePredictor\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 23px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026beta;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSE\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ep\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 16px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eVIF\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 22px;\"\u003e\n \u003cp\u003eIntercept\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003e3.54\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003e1.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 16px;\"\u003e\n \u003cp\u003e\u003cem\u003e0.001\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 16px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 22px;\"\u003e\n \u003cp\u003eSimK 3mm\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003e\u0026minus;0.29\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003e0.13\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 16px;\"\u003e\n \u003cp\u003e\u003cem\u003e0.024\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 16px;\"\u003e\n \u003cp\u003e3.57\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 22px;\"\u003e\n \u003cp\u003eApex offset (mm)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003e2.63\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003e1.42\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 16px;\"\u003e\n \u003cp\u003e0.067\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 16px;\"\u003e\n \u003cp\u003e52.3\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 22px;\"\u003e\n \u003cp\u003eCorneal inclination (\u0026deg;)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003e\u0026minus;0.31\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003e0.16\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 16px;\"\u003e\n \u003cp\u003e0.056\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 16px;\"\u003e\n \u003cp\u003e59.3\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 22px;\"\u003e\n \u003cp\u003eZ position (mm)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003e\u0026minus;0.32\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003e0.30\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 16px;\"\u003e\n \u003cp\u003e0.295\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 16px;\"\u003e\n \u003cp\u003e1.29\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 22px;\"\u003e\n \u003cp\u003eCrater offset (mm)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003e0.29\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003e0.86\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 16px;\"\u003e\n \u003cp\u003e0.735\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 16px;\"\u003e\n \u003cp\u003e2.11\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cem\u003eModel: R\u0026sup2;=0.118, Adj R\u0026sup2;=0.061, F(5,77)=2.06, p=0.080.\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eImpact on DCR parameters:\u0026nbsp;\u003c/strong\u003eWEM correction resulted in systematic DCR parameter differences versus standard Corvis ST output (Table 9, Figure 4). Clinical significance using predefined thresholds: A1 time unaffected (0% exceeding 0.5ms), HC time exceeded threshold in 46.6% (mean |diff| 0.63ms), A2 time in 98.9% (mean |diff| 0.81ms), HC DefAmp in 8.8% (mean |diff| 19.8\u0026micro;m), HC radius in 100% (mean |diff| 3.03mm).\u003c/p\u003e\n\u003cp\u003eTable 2. DCR parameters: Corvis ST vs WEM-corrected\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\" class=\"fr-table-selection-hover\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eParameter\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCorvis\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCorrected\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eDiff (95% CI)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ep\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e\u003cstrong\u003edz\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026gt;Threshold\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003eA1 time (ms)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e7.17\u0026plusmn;0.22\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e7.18\u0026plusmn;0.33\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e\u0026minus;0.008\u003c/p\u003e\n \u003cp\u003e(\u0026minus;0.043\u0026ndash;0.028)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6px;\"\u003e\n \u003cp\u003e0.660\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e\u0026minus;0.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003eA1 velocity (m/s)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e\u0026minus;0.17\u0026plusmn;0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e\u0026minus;0.21\u0026plusmn;0.04\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e0.047\u003c/p\u003e\n \u003cp\u003e(0.041\u0026ndash;0.054)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e1.50\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003eA1 deformation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e0.13\u0026plusmn;0.01 mm\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e237.0\u0026plusmn;21.8 \u0026micro;m\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e\u0026minus;10.87\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003eHC time (ms)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e16.49\u0026plusmn;0.38\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e15.87\u0026plusmn;0.41\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e0.619\u003c/p\u003e\n \u003cp\u003e(0.519\u0026ndash;0.719)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e1.31\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e46.6%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003eHC DefAmp\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e1.11\u0026plusmn;0.09 mm\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e1088.2\u0026plusmn;90.3 \u0026micro;m\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e\u0026minus;12.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e8.8%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003eHC radius (mm)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e6.68\u0026plusmn;0.65\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e3.66\u0026plusmn;0.20\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e3.025\u003c/p\u003e\n \u003cp\u003e(2.915\u0026ndash;3.135)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e5.79\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e100%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003eHC peak dist (mm)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e5.12\u0026plusmn;0.21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e7.22\u0026plusmn;0.18\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e\u0026minus;2.105\u003c/p\u003e\n \u003cp\u003e(\u0026minus;2.151\u0026ndash;\u0026minus;2.060)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e\u0026minus;9.71\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003eA2 time (ms)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e22.12\u0026plusmn;0.34\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e21.32\u0026plusmn;0.39\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e0.806\u003c/p\u003e\n \u003cp\u003e(0.775\u0026ndash;0.837)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e5.44\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e98.9%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003eA2 velocity (m/s)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e\u0026minus;0.36\u0026plusmn;0.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e0.31\u0026plusmn;0.03\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e\u0026minus;0.661\u003c/p\u003e\n \u003cp\u003e(\u0026minus;0.676\u0026ndash;\u0026minus;0.647)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e\u0026minus;9.53\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003eA2 deformation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e0.40\u0026plusmn;0.06 mm\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e494.4\u0026plusmn;53.6 \u0026micro;m\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e\u0026minus;9.22\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cem\u003edz=standardised paired difference. Thresholds: timing 0.5ms, HC DefAmp 50\u0026micro;m, HC radius 0.5mm. Clinical significance 95% CIs: A1 time 0.0\u0026ndash;4.1%, HC time 36.5\u0026ndash;56.9%, A2 time 94.0\u0026ndash;99.8%, HC DefAmp 4.5\u0026ndash;16.4%, HC radius 95.9\u0026ndash;100%. Mean |diff|: A1 0.13ms, HC 0.63ms, A2 0.81ms, DefAmp 19.77\u0026micro;m, radius 3.03mm.\u003c/em\u003e\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eTo our knowledge, this study provides the first independent validation of Corvis ST whole eye movement against an orthogonal reference system. The central finding, that globe retraction during air-puff tonometry exhibits consistent nasal-temporal asymmetry generating a rotational artefact, challenges the fundamental assumption underlying all current WEM correction algorithms. Several observations warrant detailed consideration: the relative performance advantage of robust regression over viscoelastic modelling in this exploratory analysis, despite the latter\u0026rsquo;s theoretical appeal; the pronounced laterality effect with left eyes showing 1.6-fold greater rotation than right; the tight temporal coupling between WEM velocity and corneal deformation peak; the unexpectedly high prevalence of deformation bounce; and the laterality-specific predictive relationship between corneal curvature and rotation amplitude.\u003c/p\u003e \u003cp\u003eThe finding that robust regression yielded the lowest residuals in 95.3% of measurements in this exploratory comparison, outperforming viscoelastic modelling despite its explicit incorporation of orbital tissue mechanics, requires explanation. It should be noted that the overall regression model was borderline non-significant (p\u0026thinsp;=\u0026thinsp;0.080, adjusted R\u0026sup2; = 0.061), indicating limited explained variance; these results are hypothesis-generating rather than confirmatory. Viscoelastic models assume homogeneous tissue properties and uniform boundary conditions, which are often violated in the complex orbital environment due to anisotropic structures and the need to model orbital fat as viscoelastic [28]. The retrobulbar space contains fat lobules of varying size separated by fibrous septa, extraocular muscles with non-uniform insertion geometries, and the optic nerve sheath complex with its own viscoelastic characteristics [29]. These structural heterogeneities create spatially variable impedance to globe displacement that cannot be captured by lumped-parameter models.\u003c/p\u003e \u003cp\u003eRobust regression succeeds precisely because it makes minimal assumptions. By down-weighting outlying peripheral measurements, which arise from localised tissue inhomogeneities, edge detection artefacts at the limbus, or transient scleral deformation, the Huber estimator identifies the dominant linear trend representing bulk globe motion without requiring explicit tissue modelling. The 87.3% reduction in peripheral RMS residuals compared to traditional edge-averaging reflects the magnitude of error introduced by the symmetric translation assumption: at peak retraction, the standard algorithm attributes rotational displacement to corneal deformation, systematically contaminating the deflection curve.\u003c/p\u003e \u003cp\u003eThe poor performance of frequency-domain filtering (worse than traditional edge-averaging) is instructive. Low-pass filtering assumes that globe movement occupies a distinct frequency band separable from corneal deformation, but both processes share overlapping spectral content during the rapid loading phase (5\u0026ndash;16 ms). The corneal apex velocity during early indentation (~\u0026thinsp;0.15\u0026ndash;0.20 m/s) generates frequency components indistinguishable from globe acceleration, rendering spectral separation ineffective.\u003c/p\u003e \u003cp\u003eThe consistent nasal predominance in peripheral displacement (N:T ratio 1.18\u0026ndash;1.69) indicates that the temporal corneal periphery retracts further posteriorly than the nasal, generating globe rotation toward the nose. This asymmetry persisted despite symmetric crater geometry (N:T extent ratio 1.00\u0026thinsp;\u0026plusmn;\u0026thinsp;0.02), definitively localising its origin to orbital rather than corneal mechanics.\u003c/p\u003e \u003cp\u003eThe explanation lies in differential orbital constraint mechanics. The medial orbital wall, though comprising the paper-thin lamina papyracea of the ethmoid, is functionally stiffened by the medial rectus and its associated connective tissue pulley system [29], creating substantial resistance to medial globe displacement. MRI studies demonstrate that orbital fat immediately behind the eye follows globe displacement by approximately one-half, with regional differences in tissue firmness, fat around the optic nerve behaving almost like a fluid while fat in other orbital regions is more firm [30]. Under rapid loading, this differential tissue compliance, combined with the asymmetric geometry of the orbital soft tissue support system, may permit greater temporal than nasal globe excursion. Makarem et al. [19] modelled asymmetric orbital support but could not validate their nasal rotation predictions clinically, suggesting that the biomechanical basis of this asymmetry remains incompletely understood.\u003c/p\u003e \u003cp\u003eWhen the air puff delivers its\u0026thinsp;~\u0026thinsp;60 mN peak force [23] to the central cornea, the globe accelerates posteriorly. The medial orbit presents higher impedance, while the lateral orbit offers lower impedance. The globe follows the path of least resistance, rotating nasalward as the temporal side displaces more than the nasal. This is analogous to pushing a sphere embedded asymmetrically in foam: it rotates toward the stiffer side.\u003c/p\u003e \u003cp\u003eThe measured rotation angles (0.78\u0026deg; OD, 1.23\u0026deg; OS) correspond to peripheral displacement differentials of 44\u0026ndash;141 \u0026micro;m over the 7.6 mm measurement span. These values align remarkably well with finite element predictions by Makarem et al. [19], who modelled asymmetric orbital support but could not validate their predictions clinically. Our data provides that validation while revealing a laterality effect that their symmetric models could not predict.\u003c/p\u003e \u003cp\u003eThe 58% greater rotation in left eyes (1.23\u0026deg; vs 0.78\u0026deg;) represents the study's most unexpected finding. Several mechanisms warrant consideration:\u003c/p\u003e \u003cp\u003e \u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eAnatomical asymmetry\u003c/span\u003e: True left-right differences in orbital anatomy exist but are typically subtle. The dominant eye may have marginally different extraocular muscle tone or orbital fat distribution, though published evidence is limited [32, 33]. We did not assess our cohort for ocular dominance, precluding direct evaluation.\u003c/p\u003e \u003cp\u003e \u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eMeasurement sequence effects\u003c/span\u003e: In our protocol, eye side was chosen randomly; however, sequential measurements could induce subtle fatigue, altered blink patterns, or accommodation changes affecting orbital tissue tension. We did not find a significant difference in rotation by measurement order.\u003c/p\u003e \u003cp\u003e \u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eNasal bridge geometry\u003c/span\u003e: The Corvis ST positions the air-puff nozzle along the instrument's optical axis, which aligns with the corneal apex. For OS measurements, the nozzle approaches from the patient's right side, placing it closer to the nasal bridge. This geometric asymmetry could create subtle airflow perturbations or influence head positioning. The air-puff pressure field is not perfectly symmetric [23]; interaction with the nasal bridge for OS measurements but not for OD measurements could generate differential loading.\u003c/p\u003e \u003cp\u003e \u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eFixation asymmetry\u003c/span\u003e: During OS measurement, the right eye fixates on the instrument housing or examiner. This adducted position of the fellow eye tenses the medial rectus via Hering's law of equal innervation, potentially altering orbital mechanics [33]. For OD measurement, OS fixates in relative abduction, creating different baseline muscle tension states.\u003c/p\u003e \u003cp\u003eThe finding that SimK predicted rotation amplitude significantly for OS (R\u0026sup2;=0.28, p\u0026thinsp;=\u0026thinsp;0.02) but not OD (p\u0026thinsp;=\u0026thinsp;0.72) provides a crucial clue. In left eyes, steeper corneas exhibited greater rotation (β=\u0026minus;0.47). Corneal curvature influences the spatial distribution of air-puff force: steeper corneas concentrate the pressure profile over a smaller apical area, potentially creating more eccentric loading if the apex is even slightly decentred. The laterality-specific nature of this relationship suggests that whatever factor distinguishes OS from OD (nasal bridge interaction, fixation state, anatomical asymmetry) interacts with corneal geometry to amplify rotational response.\u003c/p\u003e \u003cp\u003eSeveral methodological caveats attend the multivariate analysis. The high variance inflation factors for apex offset and X position (VIF 52.3 and 59.3, respectively) reflect near-perfect collinearity between these predictors (r\u0026thinsp;=\u0026thinsp;\u0026minus;\u0026thinsp;0.996), inflating standard errors and rendering their individual coefficients unreliable. We retained both to demonstrate this structural redundancy rather than for independent inference; the model\u0026rsquo;s overall fit (noting p\u0026thinsp;=\u0026thinsp;0.080) and SimK\u0026rsquo;s independent predictive value (VIF 1.4) are unaffected by this collinearity. Future studies should select one spatial predictor a priori or employ dimensionality reduction to avoid multicollinearity.\u003c/p\u003e \u003cp\u003eThe biphasic WEM pattern, slow initial retraction followed by rapid acceleration, provides direct insight into orbital tissue mechanics that has not previously been characterised in vivo.\u003c/p\u003e \u003cp\u003eThe slow phase represents the \"toe region\" of orbital tissue stress-strain behaviour. During this period, collagen fibres in Tenon's capsule, check ligaments, and intermuscular septum progressively recruit from their crimped resting configuration. Simultaneously, orbital fat undergoes consolidation as interlobular fluid redistributes. The system exhibits high compliance during this phase because deformation occurs primarily through geometric rearrangement rather than material strain.\u003c/p\u003e \u003cp\u003eThe transition to the fast phase (mean 6.4 ms post-onset) marks exhaustion of slack and engagement of material stiffness. Now the globe accelerates against the elastic resistance of stretched connective tissues and compressed fat. The 11-fold increase in displacement rate (from ~\u0026thinsp;4 \u0026micro;m/ms during slow phase to ~\u0026thinsp;49 \u0026micro;m/ms peak velocity) reflects this dramatic compliance change.\u003c/p\u003e \u003cp\u003eThis biphasic behaviour reflects the composite viscoelasticity of orbital soft tissues arising from both solid matrix properties and interstitial fluid flow [34]. Our WEM separation employed a Kelvin-Voigt formulation, which achieved residuals comparable to purely mathematical fitting (robust regression), suggesting performance is limited by measurement noise rather than model fidelity. Standard linear solid models [35], which incorporate stress relaxation via a second time constant, would not improve separation quality but may offer more physically interpretable parameters, particularly for characterising the incomplete elastic recovery observed at measurement end (residual displacement\u0026thinsp;\u0026minus;\u0026thinsp;65.7\u0026thinsp;\u0026plusmn;\u0026thinsp;63.9\u0026micro;m). Whether such parameters provide clinical discriminative value warrants future investigation.\u003c/p\u003e \u003cp\u003eHowever, the transition timing provides quantitative constraints previously unavailable for orbital tissues. The ~\u0026thinsp;6.4 ms slow phase duration, combined with the ~\u0026thinsp;60 mN peak air-puff force, implies an initial orbital compliance of approximately 0.5\u0026ndash;1.0 mm/N before tissue engagement. This value could inform patient-specific modelling for orbital surgery planning or thyroid eye disease assessment.\u003c/p\u003e \u003cp\u003eThe near-simultaneous occurrence of maximum WEM velocity and peak corneal deformation (15.84\u0026thinsp;\u0026plusmn;\u0026thinsp;1.96 ms vs 15.85\u0026thinsp;\u0026plusmn;\u0026thinsp;0.41 ms) reveals tight mechanical coupling that has significant implications for DCR parameter interpretation.\u003c/p\u003e \u003cp\u003eAt the moment of maximum corneal indentation, the rate of volume displacement into the anterior chamber peaks. By conservation of mass, this aqueous displacement must be accommodated, either by posterior bowing of the iris-lens diaphragm, compression of the vitreous, or posterior translation of the entire globe. The simultaneity we observe indicates that globe retraction is the primary accommodation mechanism, not iris-lens compliance as sometimes assumed.\u003c/p\u003e \u003cp\u003eThe subsequent 4.61 ms lag between deformation peak and WEM peak represents momentum-driven coasting: once corneal recovery begins (reducing the driving force), the globe continues posteriorly due to inertia before orbital tissue elasticity reverses its motion. This inertial overshoot is clinically relevant because it means WEM continues increasing after the cornea has begun recovering; the standard assumption that WEM and corneal deformation can be independently subtracted is an oversimplification.\u003c/p\u003e \u003cp\u003eThe phase relationship also explains why A2 parameters showed the largest correction effects (98.9% exceeding clinical threshold). At second applanation (~\u0026thinsp;22 ms), WEM is near its peak and changing rapidly; small errors in WEM estimation propagate maximally to A2-derived parameters. A1 parameters, occurring during the slow WEM phase when displacement is minimal (~\u0026thinsp;7 ms), are relatively immune.\u003c/p\u003e \u003cp\u003eThe high prevalence of bounce events (63% of measurements) and their association with lower IOP (ρ=\u0026minus;0.38) suggest involvement of intraocular pressure dynamics that merit closer examination.\u003c/p\u003e \u003cp\u003eType 1 bounce (reversal) requires momentary force reversal, the cornea briefly moves anteriorly against the air-puff direction. The only plausible mechanism is a transient IOP spike creating outward force exceeding residual air-puff pressure. During rapid corneal indentation, aqueous is displaced posteriorly, transiently elevating vitreous pressure. If this pressure wave reflects from the posterior sclera and returns to the cornea faster than the air-puff force decays, it could cause momentary anterior deflection.\u003c/p\u003e \u003cp\u003eType 2 bounce (plateau) represents dramatic velocity reduction without reversal, the cornea slows to near-zero velocity before resuming indentation. This pattern suggests transient equilibrium between air-puff force and internal resistance, potentially from the iris-lens diaphragm reaching its posterior excursion limit before aqueous can redistribute further.\u003c/p\u003e \u003cp\u003eThe negative correlation between bounce characteristics and IOP is mechanistically consistent: lower IOP implies a more compliant system with greater capacity for internal fluid redistribution, creating conditions favouring pressure wave reflection and transient equilibria. Higher IOP eyes have stiffer behaviour with less internal compliance, resulting in monotonic indentation.\u003c/p\u003e \u003cp\u003eThe earlier bounce onset in left eyes (13.79 vs 14.11 ms, p\u0026thinsp;=\u0026thinsp;0.026) parallels the greater rotation in OS, suggesting a common underlying factor, possibly the same fixation or geometric asymmetry that amplifies rotation also affects the timing of internal pressure equilibration.\u003c/p\u003e \u003cp\u003eThe clinical significance extends beyond improved DCR accuracy to fundamental questions about biomechanical phenotyping.\u003c/p\u003e \u003cp\u003e \u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eIOP measurement\u003c/span\u003e: The Corvis ST calculates deflection amplitude by subtracting WEM from total deformation; this deflection-derived signal then feeds into DCR parameters. The bIOP algorithm employs a polynomial function of these DCR parameters [36], meaning WEM errors propagate indirectly: incorrect WEM separation yields incorrect deflection, which contaminates the DCR inputs to bIOP. The asymmetric rotational artefact we document (N\u0026thinsp;\u0026minus;\u0026thinsp;T difference 44\u0026ndash;141\u0026micro;m) would differentially affect nasal versus temporal deflection estimates, though the clinical magnitude of this effect on bIOP requires further investigation.\u003c/p\u003e \u003cp\u003e \u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eKeratoconus detection\u003c/span\u003e: The TBI index employs a random forest classifier incorporating multiple DCR parameters; although exact feature weights are unpublished, A2 velocity and A2 length are established discriminators between subclinical keratoconus and normal eyes [38, 39]. In contrast, CBI relies on A1-phase and stiffness parameters without explicit A2 terms [39]. Our finding that A2 parameters show the largest WEM correction effects (98.9% exceeding clinical threshold) suggests TBI performance could be affected by asymmetric WEM, whereas CBI may be relatively robust. Whether correction improves or degrades diagnostic performance requires prospective evaluation; current algorithms were trained on uncorrected data and may have implicitly compensated for systematic error.\u003c/p\u003e \u003cp\u003e \u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eOrbital biomarker applications\u003c/span\u003e: The emerging use of WEM to assess orbital tissue stiffness in thyroid eye disease [13, 40] and diabetes [40] assumes that WEM reflects bulk orbital compliance. Our demonstration of substantial rotational contamination suggests these applications may be measuring a composite of compliance and rotational susceptibility rather than pure translation. Correction could improve specificity for orbital tissue assessment, or the rotational component itself could prove diagnostically informative. Asymmetric fibrosis in TED might alter N:T ratios differently than symmetric fat expansion.\u003c/p\u003e \u003cp\u003eThe principal strength is the novel orthogonal validation design enabling ground-truth assessment of WEM accuracy. The systematic comparison of twelve separation methods provides definitive guidance for future implementations. Comprehensive temporal characterisation, including bounce analysis, reveals orbital biomechanics inaccessible by other clinical methods.\u003c/p\u003e \u003cp\u003eLimitations include the absence of an a priori power calculation; the sample size (n\u0026thinsp;=\u0026thinsp;68 for main analysis, n\u0026thinsp;=\u0026thinsp;20 for validation subset) was determined by logistical constraints rather than formal power estimation, rendering the correction model findings exploratory. Additionally, the young, healthy, predominantly Caucasian cohort limits generalisability, as orbital fat volume and tissue properties change with age and vary ethnically[19, 41], potentially affecting asymmetry magnitude and patterns. The mechanistic explanation for laterality differences remains speculative; definitive resolution requires systematic manipulation of proposed factors (measurement order, fixation targets, head position sensors). The HSC frame rate (3,030 fps) was lower than Corvis ST (4,330 fps), limiting the temporal resolution of validation comparisons, though this affects precision rather than accuracy. Finally, the correction algorithm was developed and evaluated in the same cohort; external validation is essential before clinical implementation.\u003c/p\u003e \u003cp\u003eFuture priorities include validation in thyroid eye disease and glaucoma populations where WEM parameters have established clinical relevance, evaluation of correction effects on CBI/TBI diagnostic performance, and prospective confirmation of the regression model in adequately powered samples. Correlation with orbital MRI would test anatomical hypotheses regarding fat distribution and asymmetry origins. Resolution of the laterality effect requires systematic studies manipulating measurement order, fixation conditions, and head positioning.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eFunding\u003c/h2\u003e\n\u003cp\u003eOculus Optikger\u0026auml;te GmbH partially supported this research by providing a high-speed camera system and the Corvis ST instrument.\u003c/p\u003e\n\u003ch2\u003eCompeting interests\u003c/h2\u003e\n\u003cp\u003eThe authors declare no financial or non-financial competing interests related to this work.\u003c/p\u003e\n\u003ch2\u003eEthics approval\u003c/h2\u003e\n\u003cp\u003eThis study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the University of Plymouth Ethics and Integrity Committee (ref. no. 13/14-222).\u003c/p\u003e\n\u003ch2\u003eConsent to participate\u003c/h2\u003e\n\u003cp\u003eWritten informed consent was obtained from all individual participants included in the study.\u003c/p\u003e\n\u003ch2\u003eConsent to publish\u003c/h2\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003ch2\u003eAvailability of data and materials\u003c/h2\u003e\n\u003cp\u003eThe datasets generated and analysed during the current study are available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003ch2\u003eCode availability\u003c/h2\u003e\n\u003cp\u003eThe Python code for the novel correction algorithm is available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003ch2\u003eAuthors\u0026apos; contributions\u003c/h2\u003e\n\u003cp\u003eDO: Conceptualisation, Methodology, Software, Validation, Formal Analysis, Investigation, Resources, Data Curation, Writing. Original Draft, Writing. Review and Editing, Visualisation, Supervision, Project Administration, Funding Acquisition.\u003c/p\u003e\n\u003cp\u003ePB, HB and CP: Writing. Review and Editing.\u003c/p\u003e\n\u003cp\u003eAll authors read and approved the final manuscript and agree to be accountable for all aspects of the work.\u003c/p\u003e\n\u003ch2\u003eAcknowledgments\u003c/h2\u003e\n\u003cp\u003eThe authors wish to acknowledge the colleagues and research assistants who contributed to data collection for this study.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eKling, S. and F. Hafezi, \u003cem\u003eCorneal biomechanics - a review.\u003c/em\u003e Ophthalmic Physiol Opt, 2017. \u003cstrong\u003e37\u003c/strong\u003e(3): p. 240\u0026ndash;252.\u003c/li\u003e\n\u003cli\u003eRoberts, C.J. and W.J. Dupps, Jr., \u003cem\u003eBiomechanics of corneal ectasia and biomechanical treatments.\u003c/em\u003e J Cataract Refract Surg, 2014. \u003cstrong\u003e40\u003c/strong\u003e(6): p. 991\u0026ndash;8.\u003c/li\u003e\n\u003cli\u003eAmbr\u0026oacute;sio, R., I. Ramos, and A. Luz, \u003cem\u003eDynamic ultra high speed Scheimpflug imaging for assessing corneal biomechanical properties.\u003c/em\u003e Revista Brasileira de Oftalmologia, 2013. \u003cstrong\u003e72\u003c/strong\u003e: p. 99\u0026ndash;102.\u003c/li\u003e\n\u003cli\u003eEuropean Glaucoma, S., \u003cem\u003eTerminology and Guidelines for Glaucoma\u003c/em\u003e. 5 ed. 2020: PubliComm.\u003c/li\u003e\n\u003cli\u003eLiu, J. and C.J. Roberts, \u003cem\u003eInfluence of corneal biomechanical properties on intraocular pressure measurement: quantitative analysis.\u003c/em\u003e J Cataract Refract Surg, 2005. \u003cstrong\u003e31\u003c/strong\u003e(1): p. 146\u0026ndash;55.\u003c/li\u003e\n\u003cli\u003eOculus Optikger\u0026auml;te Gmb, H., \u003cem\u003eCorvis ST Operating Instructions.\u003c/em\u003e 2020.\u003c/li\u003e\n\u003cli\u003eKoprowski, R., \u003cem\u003eAutomatic method of analysis and measurement of additional parameters of corneal deformation in the Corvis tonometer.\u003c/em\u003e Biomed Eng Online, 2014. \u003cstrong\u003e13\u003c/strong\u003e: p. 150.\u003c/li\u003e\n\u003cli\u003eKoprowski, R., R. Ambrosio, Jr., and S. Reisdorf, \u003cem\u003eScheimpflug camera in the quantitative assessment of reproducibility of high-speed corneal deformation during intraocular pressure measurement.\u003c/em\u003e J Biophotonics, 2015. \u003cstrong\u003e8\u003c/strong\u003e(11-12): p. 968\u0026ndash;78.\u003c/li\u003e\n\u003cli\u003eAoki, S., et al., \u003cem\u003eThe effect of air pulse-driven whole eye motion on the association between corneal hysteresis and glaucomatous visual field progression.\u003c/em\u003e Scientific Reports, 2018. \u003cstrong\u003e8\u003c/strong\u003e(1).\u003c/li\u003e\n\u003cli\u003eZhang, D., et al., \u003cem\u003eExploring the Biomechanical Properties of the Human Cornea In Vivo Based on Corvis ST.\u003c/em\u003e Frontiers in Bioengineering and Biotechnology, 2021. \u003cstrong\u003e9\u003c/strong\u003e.\u003c/li\u003e\n\u003cli\u003eHwang, H.S., et al., \u003cem\u003eA novel method for quantifying the biomechanical parameters of orbital soft tissue using a corneal dynamic scheimpflug analyser: a retrospective study.\u003c/em\u003e BMC Ophthalmol, 2019. \u003cstrong\u003e19\u003c/strong\u003e(1): p. 53.\u003c/li\u003e\n\u003cli\u003eLeszczynska, A., et al., \u003cem\u003eMeasurement of Orbital Biomechanical Properties in Patients with Thyroid Orbitopathy Using the Dynamic Scheimpflug Analyzer (Corvis ST).\u003c/em\u003e Curr Eye Res, 2018. \u003cstrong\u003e43\u003c/strong\u003e(3): p. 289\u0026ndash;292.\u003c/li\u003e\n\u003cli\u003eVellara, H.R., et al., \u003cem\u003eIn vivo ocular biomechanical compliance in thyroid eye disease.\u003c/em\u003e Br J Ophthalmol, 2017. \u003cstrong\u003e101\u003c/strong\u003e(8): p. 1076\u0026ndash;1079.\u003c/li\u003e\n\u003cli\u003eVellara, H.R., et al., \u003cem\u003eQuantitative Analysis of Corneal Energy Dissipation and Corneal and Orbital Deformation in Response to an Air-Pulse in Healthy Eyes.\u003c/em\u003e Invest Ophthalmol Vis Sci, 2015. \u003cstrong\u003e56\u003c/strong\u003e(11): p. 6941\u0026ndash;7.\u003c/li\u003e\n\u003cli\u003eBoszczyk, A., H. Kasprzak, and A. Jozwik, \u003cem\u003eEye retraction and rotation during Corvis ST \u0026apos;air puff\u0026apos; intraocular pressure measurement and its quantitative analysis.\u003c/em\u003e Ophthalmic Physiol Opt, 2017. \u003cstrong\u003e37\u003c/strong\u003e(3): p. 253\u0026ndash;262.\u003c/li\u003e\n\u003cli\u003eJannesari, M., et al., \u003cem\u003eAssessment of corneal and fatty tissues biomechanical response in dynamic tonometry tests by using inverse models.\u003c/em\u003e Acta Bioeng Biomech, 2018. \u003cstrong\u003e20\u003c/strong\u003e(1): p. 39\u0026ndash;48.\u003c/li\u003e\n\u003cli\u003eJannesari, M., P. Mosaddegh, and M. Kadkhodaei, \u003cem\u003eNumerical and clinical investigation on the material model of the cornea in Corvis tonometry tests.\u003c/em\u003e Mechanics of Time-Dependent Materials, 2019. \u003cstrong\u003e23\u003c/strong\u003e: p. 373\u0026ndash;384.\u003c/li\u003e\n\u003cli\u003eJannesari, M. and H.T. Kasprzak, \u003cem\u003eIn-vivo Calibration of Corneal Biomechanical Properties by Virtual Fields Method.\u003c/em\u003e Research Square (Research Square), 2023.\u003c/li\u003e\n\u003cli\u003eMakarem, A., et al., \u003cem\u003eAssessment of age-related change of the ocular support system.\u003c/em\u003e Front Bioeng Biotechnol, 2023. \u003cstrong\u003e11\u003c/strong\u003e: p. 1146828.\u003c/li\u003e\n\u003cli\u003eFlight, L. and S.A. Julious, \u003cem\u003ePractical guide to sample size calculations: non-inferiority and equivalence trials.\u003c/em\u003e Pharm Stat, 2016. \u003cstrong\u003e15\u003c/strong\u003e(1): p. 80\u0026ndash;9.\u003c/li\u003e\n\u003cli\u003eMlyniuk, P., et al., \u003cem\u003eProbing biomechanical properties of the cornea with air-puff-based techniques \u0026ndash; an overview.\u003c/em\u003e Advanced Optical Technologies, 2021. \u003cstrong\u003e10\u003c/strong\u003e(6): p. 375\u0026ndash;391.\u003c/li\u003e\n\u003cli\u003eWang, X., et al., \u003cem\u003eAssessment of corneal biomechanics, tonometry and pachymetry with the Corvis ST in myopia.\u003c/em\u003e Scientific Reports, 2021. \u003cstrong\u003e11\u003c/strong\u003e(1).\u003c/li\u003e\n\u003cli\u003eOehring, D., et al., \u003cem\u003eExperimental Assessment of the Spatial and Temporal Distribution of Non-Contact Tonometer Airflows.\u003c/em\u003e Applied Sciences, 2021.\u003c/li\u003e\n\u003cli\u003eKoprowski, R., et al., \u003cem\u003eSelected parameters of the corneal deformation in the Corvis tonometer.\u003c/em\u003e Biomed Eng Online, 2014. \u003cstrong\u003e13\u003c/strong\u003e: p. 55.\u003c/li\u003e\n\u003cli\u003eMatković, K., et al., \u003cem\u003eQuantitative Externalization of Visual Data Analysis Results Using Local Regression Models\u003c/em\u003e, in \u003cem\u003eLecture notes in computer science\u003c/em\u003e. 2017, Springer Science+Business Media. p. 199\u0026ndash;218.\u003c/li\u003e\n\u003cli\u003eStefano, V.S.D., et al., \u003cem\u003eLive human assessment of depth-dependent corneal displacements with swept-source optical coherence elastography.\u003c/em\u003e PLoS ONE, 2018. \u003cstrong\u003e13\u003c/strong\u003e(12).\u003c/li\u003e\n\u003cli\u003eKoprowski, R., et al., \u003cem\u003eQuantitative assessment of responses of the eyeball based on data from the Corvis tonometer.\u003c/em\u003e Comput Biol Med, 2015. \u003cstrong\u003e58\u003c/strong\u003e: p. 91\u0026ndash;100.\u003c/li\u003e\n\u003cli\u003eJafari, S., et al., \u003cem\u003eFinite element model of ocular adduction with unconstrained globe translation.\u003c/em\u003e Biomech Model Mechanobiol, 2024. \u003cstrong\u003e23\u003c/strong\u003e(2): p. 601\u0026ndash;614.\u003c/li\u003e\n\u003cli\u003eDemer, J.L., \u003cem\u003eMechanics of the orbita.\u003c/em\u003e Dev Ophthalmol, 2007. \u003cstrong\u003e40\u003c/strong\u003e: p. 132\u0026ndash;57.\u003c/li\u003e\n\u003cli\u003eHotte, G.J., et al., \u003cem\u003eVisualization of Sliding and Deformation of Orbital Fat During Eye Rotation.\u003c/em\u003e Transl Vis Sci Technol, 2016. \u003cstrong\u003e5\u003c/strong\u003e(4): p. 9.\u003c/li\u003e\n\u003cli\u003eCui, H. and Z. Zhang, \u003cem\u003eInterocular symmetry and asymmetry in ocular health and disease.\u003c/em\u003e Front Med (Lausanne), 2025. \u003cstrong\u003e12\u003c/strong\u003e: p. 1626329.\u003c/li\u003e\n\u003cli\u003eSmith, E.A., et al., \u003cem\u003eOrbital volume changes during growth and development in human children assessed using cone beam computed tomography.\u003c/em\u003e Head Face Med, 2022. \u003cstrong\u003e18\u003c/strong\u003e(1): p. 8.\u003c/li\u003e\n\u003cli\u003eKing, W.M., \u003cem\u003eBinocular coordination of eye movements--Hering\u0026apos;s Law of equal innervation or uniocular control?\u003c/em\u003e Eur J Neurosci, 2011. \u003cstrong\u003e33\u003c/strong\u003e(11): p. 2139\u0026ndash;46.\u003c/li\u003e\n\u003cli\u003eSuh, J.K. and M.R. DiSilvestro, \u003cem\u003eBiphasic Poroviscoelastic Behavior of Hydrated Biological Soft Tissue.\u003c/em\u003e Journal of Applied Mechanics, 1999. \u003cstrong\u003e66\u003c/strong\u003e(2): p. 528\u0026ndash;535.\u003c/li\u003e\n\u003cli\u003eAhmed, H.M., N.M. Salem, and W. Al-Atabany, \u003cem\u003eHuman cornea thermo-viscoelastic behavior modelling using standard linear solid model.\u003c/em\u003e BMC Ophthalmol, 2023. \u003cstrong\u003e23\u003c/strong\u003e(1): p. 250.\u003c/li\u003e\n\u003cli\u003eEliasy, A., et al., \u003cem\u003eEx-vivo experimental validation of biomechanically-corrected intraocular pressure measurements on human eyes using the CorVis ST.\u003c/em\u003e Exp Eye Res, 2018. \u003cstrong\u003e175\u003c/strong\u003e: p. 98\u0026ndash;102.\u003c/li\u003e\n\u003cli\u003eKoc, M., et al., \u003cem\u003eBiomechanical Analysis of Subclinical Keratoconus With Normal Topographic, Topometric, and Tomographic Findings.\u003c/em\u003e J Refract Surg, 2019. \u003cstrong\u003e35\u003c/strong\u003e(4): p. 247\u0026ndash;252.\u003c/li\u003e\n\u003cli\u003eAsroui, L., et al., \u003cem\u003eBiomechanical Evaluation of Topographically and Tomographically Normal Fellow Eyes of Patients With Keratoconus.\u003c/em\u003e J Refract Surg, 2022. \u003cstrong\u003e38\u003c/strong\u003e(5): p. 318\u0026ndash;325.\u003c/li\u003e\n\u003cli\u003eVinciguerra, R., et al., \u003cem\u003eDetection of Keratoconus With a New Biomechanical Index.\u003c/em\u003e J Refract Surg, 2016. \u003cstrong\u003e32\u003c/strong\u003e(12): p. 803\u0026ndash;810.\u003c/li\u003e\n\u003cli\u003eOhn, K., \u003cem\u003eEffect of diabetes mellitus on corneal biomechanical parameters.\u003c/em\u003e Medicine, 2022. \u003cstrong\u003e101\u003c/strong\u003e: p. e30248.\u003c/li\u003e\n\u003cli\u003eRegensburg, N.I., et al., \u003cem\u003eAge and gender-specific reference values of orbital fat and muscle volumes in Caucasians.\u003c/em\u003e Br J Ophthalmol, 2011. \u003cstrong\u003e95\u003c/strong\u003e(12): p. 1660\u0026ndash;3. \u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"ophthalmic-and-physiological-optics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"Learn more about [Ophthalmic and Physiological Optics](https://link.springer.com/journal/44402)","snPcode":"44402","submissionUrl":"https://submission.springernature.com/new-submission/44402/3?","title":"Ophthalmic and Physiological Optics","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Open","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Ocular Biomechanics, Corvis ST, Whole Eye Movement, Non-Contact Tonometry, Globe Retraction, Nasal-Temporal Asymmetry","lastPublishedDoi":"10.21203/rs.3.rs-9226844/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9226844/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003ePurpose\u003c/h2\u003e \u003cp\u003eThe Corvis ST non-contact tonometer calculates whole eye movement (WEM) to isolate corneal deflection from globe displacement, assuming symmetrical retraction. This study aimed to validate the Corvis ST WEM parameter against an independent high-speed camera (HSC) system, quantify nasal-temporal asymmetry in globe retraction, and develop a correction method for identified asymmetries.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eIn this prospective cross-sectional study, 68 healthy subjects (mean age 22.65\u0026thinsp;\u0026plusmn;\u0026thinsp;4.14 years) underwent Corvis ST measurement. In a subset of 20 participants, globe movement was simultaneously recorded using an orthogonal HSC system (3,030 fps). Raw displacement matrices were analysed to decompose the ocular response into translational and rotational components. Twelve WEM separation methods were compared using peripheral root-mean-square residuals.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eRobust regression achieved the lowest peripheral residuals (4.74\u0026thinsp;\u0026plusmn;\u0026thinsp;2.97 \u0026micro;m), representing an 87.3% improvement over traditional edge-averaging (36.33\u0026thinsp;\u0026plusmn;\u0026thinsp;18.76 \u0026micro;m, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and was optimal for 95.3% of measurements. HSC validation confirmed moderate-to-good reproducibility (ICC\u0026thinsp;=\u0026thinsp;0.608 for amplitude and 0.720 for time). Significant nasal-temporal asymmetry was observed: nasal displacement exceeded temporal by 44\u0026ndash;141 \u0026micro;m (N:T ratio 1.18\u0026ndash;1.69, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), with greater asymmetry in left eyes (N:T\u0026thinsp;=\u0026thinsp;1.69\u0026thinsp;\u0026plusmn;\u0026thinsp;0.34) than right eyes (N:T\u0026thinsp;=\u0026thinsp;1.18\u0026thinsp;\u0026plusmn;\u0026thinsp;0.20, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Globe retraction exhibited biphasic dynamics (slow phase 6.42\u0026thinsp;\u0026plusmn;\u0026thinsp;3.17 ms, peak WEM at 21.07\u0026thinsp;\u0026plusmn;\u0026thinsp;0.51 ms). Corneal curvature (SimK) predicted rotation amplitude in left eyes only (β=\u0026minus;0.47, p\u0026thinsp;=\u0026thinsp;0.005). Deformation bounce was prevalent (63.0%), predominantly severe (55.9%)\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eThe Corvis ST accurately measures overall globe displacement, but consistent nasal-temporal asymmetry introduces rotational error into standard WEM correction. An exploratory robust regression approach improved DCR parameter accuracy in this cohort, with potential implications for biomechanically corrected IOP, keratoconus screening indices, and clinical use of WEM as an orbital tissue biomarker. These correction findings require confirmation in larger, prospectively powered samples.\u003c/p\u003e","manuscriptTitle":"Asymmetric whole eye movement during non-contact tonometry: independent validation and implications for corneal biomechanical parameters","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-03-31 06:45:33","doi":"10.21203/rs.3.rs-9226844/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"editorInvitedReview","content":"","date":"2026-05-18T04:34:41+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"311496590080254263300917783964203772753","date":"2026-05-08T23:41:43+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"81297725289781228211805198428126818172","date":"2026-05-08T21:59:51+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-03-26T15:32:42+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-03-26T15:30:52+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-03-25T22:55:01+00:00","index":"","fulltext":""},{"type":"submitted","content":"Ophthalmic and Physiological Optics","date":"2026-03-25T19:53:38+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"ophthalmic-and-physiological-optics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"Learn more about [Ophthalmic and Physiological Optics](https://link.springer.com/journal/44402)","snPcode":"44402","submissionUrl":"https://submission.springernature.com/new-submission/44402/3?","title":"Ophthalmic and Physiological Optics","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Open","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"d516fe76-87dc-44df-b1a0-575975a6a12a","owner":[],"postedDate":"March 31st, 2026","published":true,"recentEditorialEvents":[{"type":"editorInvitedReview","content":"","date":"2026-05-18T04:34:41+00:00","index":27,"fulltext":""},{"type":"reviewerAgreed","content":"311496590080254263300917783964203772753","date":"2026-05-08T23:41:43+00:00","index":25,"fulltext":""},{"type":"reviewerAgreed","content":"81297725289781228211805198428126818172","date":"2026-05-08T21:59:51+00:00","index":24,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-03-31T06:45:33+00:00","versionOfRecord":[],"versionCreatedAt":"2026-03-31 06:45:33","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9226844","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9226844","identity":"rs-9226844","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}