Evaluation of an artificial intelligence model for identification of obstructive hydrocephalus on computed tomography of the head

Evaluation of an artificial intelligence model for identification of obstructive hydrocephalus on computed tomography of the head | 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 Article Evaluation of an artificial intelligence model for identification of obstructive hydrocephalus on computed tomography of the head Ankita Ghatak, Isabella Newbury-Chaet, Sarah Mercaldo, John Chin, and 11 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5487343/v2 This work is licensed under a CC BY 4.0 License Status: Posted Version 2 posted You are reading this latest preprint version Show more versions Abstract Introduction: Obstructive hydrocephalus is a critical radiographic finding requiring emergent treatment. Its identification on head computed tomography (CT) by an artificial intelligence (AI) model could facilitate sooner life-saving interventions, although there are common co-occurring findings including intracranial hemorrhage that can confound this interpretation. This study assessed the accuracy of an AI model (Annalise Enterprise CTB) at identifying obstructive hydrocephalus including in the presence or absence of other findings. Methods This retrospective cohort included 177 thin (≤ 1.5mm) series and 194 thick (> 1.5 and ≤ 5mm) series from 200 non-contrast head CT cases. These cases were obtained from patients aged ≥ 18 years at 5 hospitals in the United States. Each case was interpreted independently by up to three neuroradiologists. Each series was then interpreted by the AI model. Results The AI model performed with area under the curve 0.988 (95% confidence interval (CI): 0.971 to 0.998) on thin series and 0.986 (95% CI: 0.969 to 0.997) on thick series. These results were broadly maintained in subgroups for the presence or absence of intracranial hemorrhage, parenchymal abnormality and ventricular drain, and across demographic and scanner manufacturer subgroups. Conclusions The AI model accurately identified obstructive hydrocephalus in this dataset. Its performance in subgroup analyses reflected its robustness. Health sciences/Neurology/Neurological disorders/Hydrocephalus Health sciences/Medical research/Translational research Health sciences/Health care/Medical imaging/Brain imaging Health sciences/Health care/Medical imaging/Tomography/Computed tomography Figures Figure 1 Figure 2 Figure 3 Introduction Hydrocephalus reflects the dilatation of all or part of the cerebrospinal-fluid-containing spaces within the brain. 1 It is commonly categorized into obstructive and non-obstructive mechanisms, with the former potentially reflecting a neurologic emergency given the presence of increased intracranial pressure. Common causes include intracranial hemorrhage, especially subarachnoid hemorrhage and intraventricular hemorrhage, and masses, particularly intraventricular, cerebellar or pineal tumors. 2 Its presence may necessitate the emergent placement of a ventricular drain to reduce intracranial pressure in addition to the management of the underlying cause. Artificial intelligence (AI) has already been applied to many emergent neuroradiologic findings especially large vessel occlusion and intracranial hemorrhage identification. 3 Its key application has been to triage suspected abnormal radiology cases so that they can be interpreted sooner by radiologists and other clinicians, and so that time-critical interventions can commence. 4 Such software-as-medical-devices are often referred to as computer assisted triage devices (CADt). It has been proposed that AI could similarly be used to identify obstructive hydrocephalus on head CT and this study assessed the performance of an AI model by comparing its outputs to consensus neuroradiologist interpretations. The model was provided separately with thin (≤ 1.5mm) and thick (> 1.5 and ≤ 5mm) axial series from each case so that the performance on different slice thicknesses could be calculated. Its performance was also evaluated across multiple subgroups to assess its robustness including in the presence or absence of common co-occurring findings. Methods Study design This retrospective standalone model performance study was conducted using radiology cases from five hospitals within the Mass General Brigham (MGB) network between 2016 and 2022 using similar methods to previously published studies about intracranial hemorrhage, mass effect and vasogenic edema identification. 5,6 It was approved by the MGB Institutional Review Board with waiver of informed consent. It was conducted in accordance with relevant guidelines and regulations including the Health Insurance Portability and Accountability Act (HIPAA). This manuscript follows the Standards for Reporting Diagnostic Accuracy (STARD 2015) reporting guideline. Case selection The cohort was selected in a consecutive manner based on review of original radiology reports. The cohort size for each of the positive and negative cases was based on powering calculations as described in the statistical analysis section below. The positive cases were identified through a natural language processing search engine (Nuance mPower Clinical Analytics) followed by manual report review by an American Board of Radiology (ABR)-certified neuroradiologist. The negative cases were identified by taking the next negative case acquired on the same CT scanner after each positive case to avoid temporal and technical bias. The cohort considered all CT head cases performed at a hospital including inpatient and outpatient settings; there were no limitations on the original CT head clinical indication. The CT head cases were obtained from patients at least 18 years of age. The CT head cases were taken from unique patients; only the first CT head from a given patient was included. All cases were deidentified and underwent an image quality review by an ABR-certified neuroradiologist. The relevant series for the model interpretations were selected at the same time as described under the series selection section below. The review was performed using the FDA-cleared eUnity image visualization software (Version 6 or higher) and an internal web-based annotation system that utilized the REDCap electronic data capture tools hosted at MGB. 7,8 Series selection The model was provided with a single selected series at the time of model inference. These series were non-contrast thin (≤ 1.5mm) and thick (> 1.5 and ≤ 5mm) axial series for each CT head case. The series were selected such that the thin series was the thinnest available series ≤ 1.5mm; the thick series was randomized between the thinnest and thickest available series > 1.5 and ≤ 5mm to ensure representation of slice thicknesses across the entire range. After series selection, a DICOM metadata review was additionally performed to ensure that the slice thickness was within the appropriate range and that there was a consistent slice interval (with tolerance of ± 0.2mm). Ground truth interpretations Ground truth interpretations were performed by up to three ABR-certified neuroradiologists. They answered whether obstructive hydrocephalus was “Present” or “Absent”. The definition of obstructive hydrocephalus was “Enlargement of one of more ventricles due to obstruction”. They provided their interpretations independently, without access to the original radiology reports and in different worklist orders. They used the same image visualization software and annotation system as was used in the image quality review. They had access to the entire CT head case (i.e., were not restricted to the series selected for model inference). For determining consensus interpretations, a “2 + 1” strategy was used: the first two neuroradiologists interpreted every case and a third neuroradiologist then interpreted cases with discrepant interpretations. The first two neuroradiologists also identified the presence of the co-existing findings or abnormalities of intracranial hemorrhage; parenchymal abnormality including ischemia, mass, cyst, encephalomalacia; and ventricular drain including extraventricular drain, ventriculoperitoneal shunt. These findings were considered present for a case if either of the neuroradiologists interpreted them as being present. Model inference The evaluated AI model was version 3.1.0 of the Annalise Enterprise CTB Triage Trauma device. It is the same AI model used by the Annalise Enterprise (CTB module) device, which is commercially available in some non-US markets and whose development has been previously described. 9 In brief, it consists of an ensemble of five neural networks with three heads: one for classification, one for left-right localization and one for segmentation. It can identify 130 different radiological findings and was trained on over 200,000 CT head cases, which were each labelled by at least three radiologists. The Annalise Enterprise CTB Triage Trauma device only provides binary classification outputs about the identification of findings, which is consistent with US Food and Drug Administration (FDA) regulations for CADt devices. The model was installed at MGB for use in this study and received only the DICOM-formatted CT head cases. It outputted a classification score between 0 and 1 for obstructive hydrocephalus. A binary output could be derived using a prespecified default operating point (0.185900). Statistical analysis The statistical analysis was performed in R (version 4.0.2) on the full analysis set. The predefined endpoints included the area under the receiver operating characteristic curve (AUC) for the identification of obstructive hydrocephalus for each of thin and thick series. The AUCs were calculated using the consensus annotations and the classification scores from the AI model. The predefined endpoints also included the sensitivity and specificity at the operating point. They were calculated by comparing the binary model output at the operating point with the consensus annotations (i.e., by calculating the number of true positive, false negative, true negative and false positive cases). The AUCs, sensitivities and specificities were calculated as exploratory analyses for the subgroups of sex, age groups, ethnicity, race and CT scanner manufacturer. They were also calculated as exploratory analyses for subgroups based on the presence or absence of the co-existing findings or abnormalities. The positive predictive values (PPVs) and negative predictive values (NPVs) were calculated as exploratory analyses at different levels of assumed prevalence. The demographic parameters and scanner manufacturer were derived from clinical databases or DICOM fields for each radiology case; any missing data were treated as “Unavailable” and no data were imputed. All CIs were calculated using bootstrapped intervals with 2,000 resamples. The sample size was powered based on preliminary model results at a balanced operating point to ensure the lower bound of the 95% CI for sensitivity was > 80% and for specificity was > 80%. Results The cohort of CT head cases included 181 thin series and 200 thick series (Fig. 1 ). The model successfully performed inference on 177 (97.8%) thin series and 194 (97.0%) thick series. The final analysis set of thin series cases included 77 (43.5%) positive cases and 100 (56.5%) negative cases; there were 98 (55.4%) women and 79 (44.6%) men; the mean (± SD) age was 60.8 (± 18.8) years (Table 1 ). The final analysis set of thick series cases included 87 (44.8%) positive cases and 107 (55.2%) negative cases; there were 106 (54.6%) women and 88 (45.4%) men; the mean (± SD) age was 61.0 (± 18.8) years. Table 1 Demographic and technical breakdown of CT head cases. Thin series Thick series Positive Negative Positive Negative Total 77 100 87 107 Sex Female 44 (57.1%) 54 (54.0%) 51 (58.6%) 55 (51.4%) Male 33 (42.9%) 46 (46.0%) 36 (41.4%) 52 (48.6%) Age ≤ 65 years 45 (58.4%) 49 (49.0%) 49 (56.3%) 52 (48.6%) > 65 years 32 (41.6%) 51 (51.0%) 38 (43.7%) 55 (51.4%) Mean ± SD (years) 57.9 ± 19.9 63.0 ± 17.7 58.5 ± 19.8 63.1 ± 17.8 Ethnicity Hispanic 7 (9.1%) 10 (10.0%) 7 (8.0%) 10 (9.3%) Not Hispanic 63 (81.8%) 81 (81.0%) 73 (83.9%) 87 (81.3%) Unavailable 7 (9.1%) 9 (9.0%) 7 (8.0%) 10 (9.3%) Race Asian 5 (6.5%) 1 (1.0%) 7 (8.0%) 1 (0.9%) Black or African American 4 (5.2%) 9 (9.0%) 4 (4.6%) 11 (10.3%) Native Hawaiian or Other Pacific Islander 0 (0.0%) 1 (1.0%) 0 (0.0%) 1 (0.9%) White 58 (75.3%) 77 (77.0%) 66 (75.9%) 82 (76.6%) Other 7 (9.1%) 4 (4.0%) 7 (8.0%) 4 (3.7%) Two or more 0 (0.0%) 2 (2.0%) 0 (0.0%) 2 (1.9%) Declined 1 (1.3%) 1 (1.0%) 1 (1.1%) 1 (0.9%) Unavailable 2 (2.6%) 5 (5.0%) 2 (2.3%) 5 (4.7%) Manufacturer GE Healthcare 8 (10.4%) 19 (19.0%) 16 (18.4%) 24 (22.4%) NeuroLogica 1 (1.3%) 1 (1.0%) 2 (2.3%) 1 (0.9%) Siemens 66 (85.7%) 78 (78.0%) 66 (75.9%) 79 (73.8%) Toshiba 2 (2.6%) 2 (2.0%) 3 (3.4%) 3 (2.8%) The model performed with an AUC of 0.988 (95% CI: 0.971 to 0.998) on thin series and 0.986 (95% CI: 0.969 to 0.997) on thick series (Table 2 and Fig. 2 ). These results corresponded to a sensitivity of 93.5% (95% CI: 88.3 to 98.7%) and specificity of 95.0% (95% CI: 90.0 to 99.0%) on thin series, and a sensitivity of 93.1% (95% CI: 87.4 to 97.7%) and specificity of 95.3% (95% CI: 90.7 to 99.1%) on thick series. All demographic and scanner manufacturer subgroups had an AUC of at least 0.95 for both thin series (Supplementary Table 1) and thick series (Supplementary Table 2). The sensitivities and specificities of subgroups were at least 80% whenever there were at least 5 positive or negative cases respectively for the subgroup. Table 2 Model performance summary. Metric Thin series Thick series Positive N 77 87 Negative N 100 107 AUC (95% CI) 0.988 (0.971 to 0.998) 0.986 (0.969 to 0.997) Sensitivity (95% CI) 93.5% (88.3 to 98.7) 93.1% (87.4 to 97.7) Specificity (95% CI) 95.0% (90.0 to 99.0) 95.3% (90.7 to 99.1) The PPVs and NPVs were calculated at different prevalences with recognition that the clinical implications of positive and negative model outputs are also dependent on the prevalence of a finding (Table 3 ). For an assumed prevalence of 0.10, the PPV was 67.5% (95% CI: 51.6 to 91.3%) and the NPV was 99.2% (95% CI: 98.6 to 99.8%) on thin series, and the PPV was 68.9% (95% CI: 53.1 to 91.4%) and the NPV was 99.2% (95% CI: 98.5 to 99.7%) on thick series. Table 3 PPVs and NPVs at different levels of assumed prevalence. Thin series Thick series Assumed prevalence PPV (95% CI) NPV (95% CI) PPV (95% CI) NPV (95% CI) 0.05 49.6% (33.6 to 83.2) 99.6% (99.3 to 99.9) 51.2% (35.0 to 83.4) 99.6% (99.3 to 99.9) 0.10 67.5% (51.6 to 91.3) 99.2% (98.6 to 99.8) 68.9% (53.1 to 91.4) 99.2% (98.5 to 99.7) 0.15 76.7% (62.9 to 94.3) 98.8% (97.8 to 99.8) 77.9% (64.3 to 94.4) 98.7% (97.7 to 99.6) 0.20 82.4% (70.6 to 95.9) 98.3% (96.8 to 99.6) 83.3% (71.8 to 96.0) 98.2% (96.8 to 99.4) A potential mimic finding for hydrocephalus is ventriculomegaly in the setting of age-related atrophy. The age subgroups were therefore separated further into 18 to 49 years, 50 to 64 years, 65 to 79 years and ≥ 80 years (Table 4 ). The model did not appear to be impacted by age-related atrophy, with the model correctly interpreting all cases from patients aged ≥ 80 years. The model did show decreased sensitivities for the 50–64 years age group for both the thin and thick series (82.4% (95% CI: 64.7 to 100.0%) and 84.2% (95% CI: 68.4 to 100.0%) respectively); in both situations, the decrease in these sensitivities reflected only 3 false negative cases. Table 4 Model performance for age subgroups. Age Positive N Negative N AUC (95% CI) Sensitivity (95% CI) Specificity (95% CI) Thin Series 18–49 years 24 23 0.996 (0.978 to 1.000) 95.8% (87.5 to 100.0) 87.0% (73.9 to 100.0) 50–64 years 17 25 0.979 (0.922 to 1.000) 82.4% (64.7 to 100.0) 96.0% (88.0 to 100.0) 65–79 years 26 37 0.986 (0.953 to 1.000) 96.2% (88.5 to 100.0) 97.3% (91.9 to 100.0) ≥ 80 years 10 15 1.000 (1.000 to 1.000) 100.0% (100.0 to 100.0) 100.0% (100.0 to 100.0) Thick Series 18–49 years 26 25 0.995 (0.978 to 1.000) 96.2% (88.5 to 100.0) 88.0% (72.0 to 100.0) 50–64 years 19 26 0.978 (0.919 to 1.000) 84.2% (68.4 to 100.0) 96.2% (88.5 to 100.0) 65–79 years 30 40 0.988 (0.955 to 1.000) 93.3% (83.3 to 100.0) 97.5% (92.5 to 100.0) ≥ 80 years 12 16 1.000 (1.000 to 1.000) 100.0% (100.0 to 100.0) 100.0% (100.0 to 100.0) The model was also assessed in the presence or absence of intracranial hemorrhage, parenchymal abnormality and ventricular drain to ensure it was identifying obstructive hydrocephalus rather than these co-occurring findings (Table 5 ). Intracranial hemorrhage is one of many causes of obstructive hydrocephalus; the model achieved sensitivity of at least 88.5% both in its presence and absence, suggesting the model identified obstructive hydrocephalus regardless of whether the cause was intracranial hemorrhage. A parenchymal abnormality was present on almost all positive cases and the subgroup analysis was most relevant in ensuring that the model did not produce false positive interpretations when one was present; the model maintained a specificity of at least 90.0% both in the presence and absence of a parenchymal abnormality. A ventricular drain is a treatment for obstructive hydrocephalus and a hypothetic non-specific model could learn to identify it rather than obstructive hydrocephalus during training; this model demonstrated sensitivity of at least 82.4% when it was present and at least 95.2% when it was absent. Table 5 Model performance for co-existing findings or abnormalities subgroups. Age Positive N Negative N AUC (95% CI) Sensitivity (95% CI) Specificity (95% CI) Thin Series Intracranial hemorrhage Present 51 32 0.978 (0.939 to 0.999) 96.1% (90.2 to 100.0) 90.6% (81.2 to 100.0) Absent 26 68 0.995 (0.983 to 1.000) 88.5% (73.1 to 100.0) 97.1% (92.6 to 100.0) Parenchymal abnormality Present 72 40 0.980 (0.947 to 0.997) 93.1% (86.1 to 98.6) 90.0% (80.0 to 97.5) Absent 5 60 0.997 (0.970 to 1.000) 100.0% (100.0 to 100.0) 98.3% (95.0 to 100.0) Ventricular drain Present 15 4 1.000 (1.000 to 1.000) 86.7% (66.7 to 100.0) 100.0% (100.0 to 100.0) Absent 62 96 0.987 (0.970 to 0.998) 95.2% (88.7 to 100.0) 94.8% (89.6 to 99.0) Thick Series Intracranial hemorrhage Present 59 33 0.975 (0.932 to 0.998) 94.9% (88.1 to 100.0) 90.9% (81.8 to 100.0) Absent 28 74 0.993 (0.977 to 1.000) 89.3% (75.0 to 100.0) 97.3% (93.2 to 100.0) Parenchymal abnormality Present 81 40 0.976 (0.937 to 0.997) 92.6% (86.4 to 97.5) 90.0% (80.0 to 97.5) Absent 6 67 0.998 (0.978 to 1.000) 100.0% (100.0 to 100.0) 98.5 (95.5 to 100.0) Ventricular drain Present 17 4 1.000 (1.000 to 1.000) 82.4% (64.7 to 100.0) 100.0% (100.0 to 100.0) Absent 70 103 0.986 (0.969 to 0.998) 95.7% (90.0 to 100.0) 95.1% (91.2 to 99.0) Discussion This retrospective diagnostic study assessed the performance of an AI model in identifying obstructive hydrocephalus on head CT. The model achieved an AUC of 0.988 on thin series and 0.986 on thick series, and sensitivities and specificities of at least 93.1% at the pre-defined operating point for both series. These results were mostly maintained across demographic subgroups including sex, age, ethnicity and race, and scanner manufacturer subgroups. These results were also consistent with the expected performance of CADt devices through the US FDA and this model has subsequently received 510(k) clearance through the US FDA as the first CADt device that identifies obstructive hydrocephalus. 10 A key consideration with radiologic AI models is ensuring that they are identifying the intended finding and not a mimic or co-occurring finding. While hopefully the radiology and data science communities are more aware of this potential pitfall now, previous studies have shown that AI models may learn incorrect “shortcuts” for the detection of findings including non-imaging features like radiologist annotations on abnormal cases. 11,12 Obstructive hydrocephalus is a finding that has many mimics and common co-occurring findings. Here we showed that the model performance was not impacted in older age subgroups, suggesting age-related atrophy was not leading to false positive case identification. We similarly showed that the model identified obstructive hydrocephalus regardless of the presence or absence of the co-occurring findings of intracranial hemorrhage, parenchymal abnormality and ventricular drain. Intracranial hemorrhage is a common cause of obstructive hydrocephalus; we considered the maintenance of sensitivity in the presence or absence of intracranial hemorrhage to show that the model could identify obstructive hydrocephalus from both hemorrhagic and non-hemorrhagic causes. A parenchymal abnormality can likewise reflect a cause of obstructive hydrocephalus; however, almost all cases that were positive for obstructive hydrocephalus had a parenchymal abnormality, which limited the conclusions from this subgroup analysis. At the same time, there were many cases with a parenchymal abnormality that were negative for obstructive hydrocephalus; we considered the maintenance of a specificity of 90% among these cases to suggest that the model was identifying obstructive hydrocephalus rather than an underlying parenchymal abnormality causing obstructive hydrocephalus. The model performed similarly in the presence or absence of a ventricular drain; we were reassured that it identified obstructive hydrocephalus in the absence of a ventricular drain especially as this clinical scenario is more likely to reflect patients who have not yet received treatment for obstructive hydrocephalus. A takeaway of having a robust model is that it could play a key role in the clinical environment. CADt devices aim to aid in the prioritization and triage of cases with suspected time sensitive findings with an ultimate goal of enabling management to commence sooner. 13 Obstructive hydrocephalus is a critical finding with management steps including the possible placement of a ventricular drain. This model therefore has the potential to positively impact patient care and it will be important to formally evaluate its effectiveness moving forward. The main limitation of this study is that it was a retrospective study outside of the clinical workflow. The effectiveness of the model will ultimately be best assessed in a prospective manner within the clinical environment. An initial assessment could consider key metrics before and after the implementation of this model. From a radiologic endpoint perspective, these metrics could include changes in the time to interpretation and time to notification of the responding clinician for cases with obstructive hydrocephalus. From a broader clinical endpoint perspective, these metrics could consider time to intervention for patients with obstructive hydrocephalus, percentage of patients who receive an intervention, and morbidity or mortality rates. Conclusion This diagnostic study assessed an AI model that accurately detected obstructive hydrocephalus. It maintained this performance across many subgroup analyses suggesting its robustness. Declarations Access to data and data analysis JMH had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. AG and SFM performed the statistical analyses. Competing Interests This study was funded by Annalise-AI. Annalise-AI were involved in the design and conduct of the study; preparation, review, and approval of the manuscript; and decision to submit the manuscript for publication. Annalise-AI were not involved in the collection, management, analysis, and interpretation of the data. AG, INC, SFM, JKC, MAH, EL, ALM, ASS, KB, JC, WM, SP, SR, BCB, JMH are employees of Mass General Brigham and/or Massachusetts General Hospital, which had received institutional funding from Annalise-AI for the study. Funding This study was funded by Annalise-AI. Annalise-AI were involved in the design and conduct of the study; preparation, review, and approval of the manuscript; and decision to submit the manuscript for publication. Annalise-AI were not involved in the collection, management, analysis, and interpretation of the data. AG, INC, SFM, JKC, MAH, EL, ALM, ASS, KB, JC, WM, SP, SR, BCB, JMH are employees of Mass General Brigham and/or Massachusetts General Hospital, which had received institutional funding from Annalise-AI for the study. Author Contribution The authors contributed to the paper as follows: study conception and design: INC, SFM, JKC, WAM, BCB, JMH; data collection: AG, INC, SFM, JKC, MAH, EL, ALM, ASS, BCB, JMH; ground truth radiologist interpretations: KB, JC, WAM, SP, SR; statistical analysis: AG, SFM; draft manuscript preparation: AG, INC, SFM, JMH. All authors reviewed and approved the manuscript. Acknowledgement The authors thank the broader Mass General Brigham Data Science Office and Annalise teams for their assistance with this project. Data Availability The data from this study include protected health information and are not openly available. Components of the data may be available from the corresponding author upon reasonable request. References 1. Matson, D. D. Hydrocephalus. N Engl J Med 271 , 1360–1362, doi:10.1056/NEJM196412242712608 (1964). 2. Wijdicks, E. F. M. The practice of emergency and critical care neurology . Second edition. edn, (Oxford University Press, 2016). 3. Chandrabhatla, A. S. et al. Artificial Intelligence and Machine Learning in the Diagnosis and Management of Stroke: A Narrative Review of United States Food and Drug Administration-Approved Technologies. J Clin Med 12 , doi:10.3390/jcm12113755 (2023). 4. Rajpurkar, P. & Lungren, M. P. The Current and Future State of AI Interpretation of Medical Images. N Engl J Med 388 , 1981–1990, doi:10.1056/NEJMra2301725 (2023). 5. Hillis, J. M. et al. Evaluation of an Artificial Intelligence Model for Identification of Intracranial Hemorrhage Subtypes on Computed Tomography of the Head. Stroke: Vascular and Interventional Neurology 4 , e001223, doi:10.1161/SVIN.123.001223 (2024). 6. Newbury-Chaet, I. et al. Evaluation of an Artificial Intelligence Model for Identification of Mass Effect and Vasogenic Edema on CT of the Head. AJNR Am J Neuroradiol 45 , 1528–1535, doi:10.3174/ajnr.A8358 (2024). 7. Harris, P. A. et al. The REDCap consortium: Building an international community of software platform partners. J Biomed Inform 95 , 103208, doi:10.1016/j.jbi.2019.103208 (2019). 8. Harris, P. A. et al. Research electronic data capture (REDCap)--a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform 42 , 377–381, doi:10.1016/j.jbi.2008.08.010 (2009). 9. Buchlak, Q. et al. Effects of a comprehensive brain computed tomography deep-learning model on radiologist detection accuracy: a multireader, multicase study. (2022). 10. US Food and Drug Administration. K231094 (Annalise Enterprise CTB Triage - OH) , (2023). 11. DeGrave, A. J., Janizek, J. D. & Lee, S. I. AI for radiographic COVID-19 detection selects shortcuts over signal. medRxiv , doi:10.1101/2020.09.13.20193565 (2020). 12. Zech, J. R. et al. Variable generalization performance of a deep learning model to detect pneumonia in chest radiographs: A cross-sectional study. PLoS Med 15 , e1002683, doi:10.1371/journal.pmed.1002683 (2018). 13. U.S. Department of Health and Human Services. 21CFR 892.2080 Radiological computer aided triage and notification software , (2020). Supplemental Tables Supplemental Table 1 : Demographic and scanner manufacturer subgroup performance for identifying obstructive hydrocephalus in thin series. Thin Series (Operating Point 0. 0.1859) Positive N Negative N AUC (95% CI) Sensitivity (95% CI) Specificity (95% CI) Overall 77 100 0.988 (0.971 to 0.998) 93.5% (88.3 to 98.7) 95.0% (90.0 to 99.0) Sex Female 44 54 0.978 (0.950 to 0.997) 90.9% (81.8 to 97.7) 94.4% (88.9 to 100.0) Male 33 46 0.999 (0.994 to 1.000) 97.0% (90.9 to 100.0) 95.7% (89.1 to 100.0) Age ≤ 65 years 45 49 0.986 (0.962 to 0.998) 91.1% (82.2 to 100.0) 91.8% (83.7 to 98.0) > 65 years 32 51 0.991 (0.967 to 1.000) 96.9% (90.6 to 100.0) 98.0% (94.1 to 100.0) Ethnicity Hispanic 7 10 1.000 (1.000 to 1.000) 100.0% (100.0 to 100.0) 100.0% (100.0 to 100.0) Not Hispanic 63 81 0.984 (0.962 to 0.997) 92.1% (84.1 to 98.4) 95.1% (90.1 to 98.8) Unavailable 7 9 1.000 (1.000 to 1.000) 100.0% (100.0 to 100.0) 88.9% (66.7 to 100.0) Race Asian 5 1 1.000 (1.000 to 1.000) 100.0% (100.0 to 100.0) 100.0% (100.0 to 100.0) Black or African American 4 9 1.000 (0.833 to 1.000) 75.0% (25.0 to 100.0) 88.9% (66.7 to 100.0) Native Hawaiian or Other Pacific Islander 0 1 - - 100.0% (100.0 to 100.0) White 58 77 0.986 (0.966 to 0.999) 93.1% (86.2 to 98.3) 96.1% (90.9 to 100.0) Other 7 4 1.000 (1.000 to 1.000) 100.0% (100.0 to 100.0) 100.0% (100.0 to 100.0) Two or more 0 2 - - 100.0% (100.0 to 100.0) Declined 1 1 1.000 (1.000 to 1.000) 100.0% (100.0 to 100.0) 100.0% (100.0 to 100.0) Unavailable 2 5 1.000 (1.000 to 1.000) 100.0% (100.0 to 100.0) 80.0% (40.0 to 100.0) Manufacturer GE Healthcare 8 19 1.000 (0.961 to 1.000) 100.0% (100.0 to 100.0) 84.2% (68.4 to 100.0) NeuroLogica 1 1 1.000 (1.000 to 1.000) 0.0% (0.0. to 0.0) 100.0% (100.0 to 100.0) Siemens 66 78 0.991 (0.973 to 1.000) 93.9% (87.9 to 98.5) 97.4% (93.6 to 100.0) Toshiba 2 2 1.000 (1.000 to 1.000) 100.0% (100.0 to 100.0) 100.0% (100.0 to 100.0) Supplemental Table 2 : Demographic and scanner manufacturer subgroup performance for identifying obstructive hydrocephalus in thick series. Thick Series (Operating Point 0. 0.1859) Positive N Negative N AUC (95% CI) Sensitivity (95% CI) Specificity (95% CI) Overall 87 107 0.986 (0.969 to 0.997) 93.1% (87.4 to 97.7) 95.3% (90.7 to 99.1) Sex Female 51 55 0.979 (0.949 to 0.998) 90.2% (80.4 to 98.0) 94.5% (87.3 to 100.0) Male 36 52 0.997 (0.987 to 1.000) 97.2% (91.7 to 100.0) 96.2% (90.4 to 100.0) Age ≤ 65 years 49 52 0.984 (0.958 to 0.998) 91.8% (83.7 to 98.0) 92.3% (84.6 to 98.1) > 65 years 38 55 0.989 (0.961 to 1.000) 94.7% (86.8 to 100.0) 98.2% (94.5 to 100.0) Ethnicity Hispanic 7 10 1.000 (1.000 to 1.000) 100.0% (100.0 to 100.0) 100.0% (100.0 to 100.0) Not Hispanic 73 87 0.984 (0.962 to 0.998) 91.8% (84.9 to 97.3) 95.4% (90.8 to 98.9) Unavailable 7 10 1.000 (1.000 to 1.000) 100.0% (100.0 to 100.0) 90.0% (70.0 to 100.0) Race Asian 7 1 1.000 (1.000 to 1.000) 100.0% (100.0 to 100.0) 100.0% (100.0 to 100.0) Black or African American 4 11 0.955 (0.795 to 1.000) 75.0% (25.0 to 100.0) 90.9% (72.7 to 100.0) Native Hawaiian or Other Pacific Islander 0 1 - - 100.0% (100.0 to 100.0) White 66 82 0.986 (0.965 to 0.999) 92.4% (84.8 to 98.5) 96.3% (91.5 to 100.0) Other 7 4 1.000 (1.000 to 1.000) 100.0% (100.0 to 100.0) 100.0% (100.0 to 100.0) Two or more 0 2 - - 100.0% (100.0 to 100.0) Declined 1 1 1.000 (1.000 to 1.000) 100.0% (100.0 to 100.0) 100.0% (100.0 to 100.0) Unavailable 2 5 1.000 (1.000 to 1.000) 100.0% (100.0 to 100.0) 80.0% (40.0 to 100.0) Manufacturer GE Healthcare 16 24 0.982 (0.927 to 1.000) 100.0% (100.0 to 100.0) 87.5% (75.0 to 100.0) NeuroLogica 2 1 1.000 (1.000 to 1.000) 0.0% (0.0 to 0.0) 100.0% (100.0 to 100.0) Siemens 66 79 0.991 (0.972 to 1.000) 93.9% (87.9 to 98.5) 97.5% (93.7 to 100.0) Toshiba 3 3 1.000 (1.000 to 1.000) 100.0% (100.0 to 100.0) 100.0% (100.0 to 100.0) Additional Declarations Competing interest reported. This study was funded by Annalise-AI. Annalise-AI were involved in the design and conduct of the study; preparation, review, and approval of the manuscript; and decision to submit the manuscript for publication. Annalise-AI were not involved in the collection, management, analysis, and interpretation of the data. AG, INC, SFM, JKC, MAH, EL, ALM, ASS, KB, JC, WM, SP, SR, BCB, JMH are employees of Mass General Brigham and/or Massachusetts General Hospital, which had received institutional funding from Annalise-AI for the study. Supplementary Files STARD2015Checklist.docx Cite Share Download PDF Status: Posted Version 2 posted You are reading this latest preprint version Show more versions 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-5487343","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":387223336,"identity":"d0f6056f-ee56-47fd-8f8e-bcf3bcc498ff","order_by":0,"name":"Ankita Ghatak","email":"","orcid":"","institution":"Mass General Brigham AI","correspondingAuthor":false,"prefix":"","firstName":"Ankita","middleName":"","lastName":"Ghatak","suffix":""},{"id":387223337,"identity":"43579693-df08-43e3-ba18-a17f9c6aeec7","order_by":1,"name":"Isabella Newbury-Chaet","email":"","orcid":"","institution":"Mass General Brigham AI","correspondingAuthor":false,"prefix":"","firstName":"Isabella","middleName":"","lastName":"Newbury-Chaet","suffix":""},{"id":387223338,"identity":"4fc91cf5-fe1a-45d6-ab19-d63dd4eb252b","order_by":2,"name":"Sarah Mercaldo","email":"","orcid":"","institution":"Mass General Brigham AI","correspondingAuthor":false,"prefix":"","firstName":"Sarah","middleName":"","lastName":"Mercaldo","suffix":""},{"id":387223339,"identity":"189fa50b-cb03-4ebf-ab90-8f2a027c9232","order_by":3,"name":"John Chin","email":"","orcid":"","institution":"Mass General Brigham AI","correspondingAuthor":false,"prefix":"","firstName":"John","middleName":"","lastName":"Chin","suffix":""},{"id":387223340,"identity":"8f07bf96-ba9d-4cf3-8c1d-64c03d603239","order_by":4,"name":"Madeleine Halle","email":"","orcid":"","institution":"Mass General Brigham AI","correspondingAuthor":false,"prefix":"","firstName":"Madeleine","middleName":"","lastName":"Halle","suffix":""},{"id":387223341,"identity":"a73b154c-1a1d-497e-b388-6bbe7d46ff06","order_by":5,"name":"Eric L'Italien","email":"","orcid":"","institution":"Mass General Brigham AI","correspondingAuthor":false,"prefix":"","firstName":"Eric","middleName":"","lastName":"L'Italien","suffix":""},{"id":387223342,"identity":"fdbd127c-83c8-4d89-afea-cbc10b7b46a0","order_by":6,"name":"Ashley MacDonald","email":"","orcid":"","institution":"Mass General Brigham AI","correspondingAuthor":false,"prefix":"","firstName":"Ashley","middleName":"","lastName":"MacDonald","suffix":""},{"id":387223343,"identity":"2231db3e-b7d1-406c-a435-e788a420e1e5","order_by":7,"name":"Alex Schultz","email":"","orcid":"","institution":"Mass General Brigham AI","correspondingAuthor":false,"prefix":"","firstName":"Alex","middleName":"","lastName":"Schultz","suffix":""},{"id":387223344,"identity":"39c061c3-4735-42bf-b6d0-08f42b131c4d","order_by":8,"name":"Karen Buch","email":"","orcid":"","institution":"Massachusetts General Hospital","correspondingAuthor":false,"prefix":"","firstName":"Karen","middleName":"","lastName":"Buch","suffix":""},{"id":387223345,"identity":"9b5a3490-4f26-42cf-81bb-1dd813905dcb","order_by":9,"name":"John Conklin","email":"","orcid":"","institution":"Massachusetts General Hospital","correspondingAuthor":false,"prefix":"","firstName":"John","middleName":"","lastName":"Conklin","suffix":""},{"id":387223346,"identity":"85081552-4d7d-4590-9110-ba3e7c1d9ca0","order_by":10,"name":"William Mehan","email":"","orcid":"","institution":"Massachusetts General Hospital","correspondingAuthor":false,"prefix":"","firstName":"William","middleName":"","lastName":"Mehan","suffix":""},{"id":387223347,"identity":"f907f0d3-957b-4081-a4bb-fe7171f1af1a","order_by":11,"name":"Stuart Pomerantz","email":"","orcid":"","institution":"Massachusetts General Hospital","correspondingAuthor":false,"prefix":"","firstName":"Stuart","middleName":"","lastName":"Pomerantz","suffix":""},{"id":387223348,"identity":"3a659313-2420-404f-a81d-a52511c062af","order_by":12,"name":"Sandra Rincon","email":"","orcid":"","institution":"Massachusetts General Hospital","correspondingAuthor":false,"prefix":"","firstName":"Sandra","middleName":"","lastName":"Rincon","suffix":""},{"id":387223349,"identity":"f9f8d88e-59c2-474a-ac65-19f5b2cf75ee","order_by":13,"name":"Bernardo Bizzo","email":"","orcid":"","institution":"Massachusetts General Hospital","correspondingAuthor":false,"prefix":"","firstName":"Bernardo","middleName":"","lastName":"Bizzo","suffix":""},{"id":387223350,"identity":"37b71340-9a4d-4c1c-b97a-5d18391d2d73","order_by":14,"name":"James Michael Hillis","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABJ0lEQVRIiWNgGAWjYBACxgYwdYCHjx3I/MDAkMAggZAxwKuFjZmxgXEGMVqg4AADGzMDAzMPMVqY208nPvzBcEeGjZm5+bNtTm0eg3Tzs888NfdkG9ibt0lgc1hP7mZjHoZnIIe1SeduO17MIHPMeDbPsWLjBp5jZVi1NORuk2ZgOAzWwpy77Vhig0SCMeMMtgQgI8cMq5b+t9t//oBoaf5sCdaS/plxxj+gFvk32LXMyN3GwAPR0iDNuK0GZLgxw8c2kC08OLS83SzNYwBxmGTvtgOJbRI5xQwf+xKM23jSii2waDHsz9348UfFYXt+9vbHH35uq0vsl0jfzJDwLUG2n/3wxhvYtDSASET4H2ZggzHZMFWDgTwavw6HulEwCkbBKBjJAADhWmG6SGmJfAAAAABJRU5ErkJggg==","orcid":"","institution":"Massachusetts General Hospital","correspondingAuthor":true,"prefix":"","firstName":"James","middleName":"Michael","lastName":"Hillis","suffix":""}],"badges":[],"createdAt":"2024-11-20 03:40:17","currentVersionCode":2,"declarations":"","doi":"10.21203/rs.3.rs-5487343/v2","doiUrl":"https://doi.org/10.21203/rs.3.rs-5487343/v2","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":70792913,"identity":"f4bfc534-380b-4455-b46b-ad7051121234","added_by":"auto","created_at":"2024-12-06 19:33:00","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":567727,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCohort selection flowchart\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Hillsfig1.png","url":"https://assets-eu.researchsquare.com/files/rs-5487343/v2/8ccec52f7a55078e618509cf.png"},{"id":70792749,"identity":"febbb827-572f-43da-8cf3-3ca8118d1004","added_by":"auto","created_at":"2024-12-06 19:25:00","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":505051,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eModel performance. \u003c/strong\u003eA and B, Receiver operating characteristic curves\u003cstrong\u003e \u003c/strong\u003efor the thin series (A) and thick series (B). The shaded region reflects the bootstrapped 95% CIs. The selected point on each graph reflects the performance at the operating point.\u003c/p\u003e","description":"","filename":"Hillsfig2.png","url":"https://assets-eu.researchsquare.com/files/rs-5487343/v2/a191e5a35f3c81c996f0af8f.png"},{"id":70792747,"identity":"62f41daf-4961-4dd6-b7d7-c84604a00f59","added_by":"auto","created_at":"2024-12-06 19:25:00","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":896020,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eExample cases. \u003c/strong\u003eA and B, Examples images taken from true positive (A) and false negative (B) cases. The model score is the classification score between 0 and 1 provided by the model.\u003c/p\u003e","description":"","filename":"Hillsfig3.png","url":"https://assets-eu.researchsquare.com/files/rs-5487343/v2/75567b2bb91786f1128ccf4b.png"},{"id":77292017,"identity":"baa02970-97e7-42f1-add8-95dbf86bf7c5","added_by":"auto","created_at":"2025-02-27 06:46:54","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2321720,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5487343/v2/cc43af2b-5549-453f-ab09-4ef6aad9e796.pdf"},{"id":70792745,"identity":"1474d0d5-2cf9-4dd0-87bc-97a1dec9cdef","added_by":"auto","created_at":"2024-12-06 19:25:00","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":35235,"visible":true,"origin":"","legend":"","description":"","filename":"STARD2015Checklist.docx","url":"https://assets-eu.researchsquare.com/files/rs-5487343/v2/e63eb1936271492336df3887.docx"}],"financialInterests":"Competing interest reported. This study was funded by Annalise-AI. Annalise-AI were involved in the design and conduct of the study; preparation, review, and approval of the manuscript; and decision to submit the manuscript for publication. Annalise-AI were not involved in the collection, management, analysis, and interpretation of the data. AG, INC, SFM, JKC, MAH, EL, ALM, ASS, KB, JC, WM, SP, SR, BCB, JMH are employees of Mass General Brigham and/or Massachusetts General Hospital, which had received institutional funding from Annalise-AI for the study.","formattedTitle":"Evaluation of an artificial intelligence model for identification of obstructive hydrocephalus on computed tomography of the head","fulltext":[{"header":"Introduction","content":"\u003cp\u003eHydrocephalus reflects the dilatation of all or part of the cerebrospinal-fluid-containing spaces within the brain.\u003csup\u003e1\u003c/sup\u003e It is commonly categorized into obstructive and non-obstructive mechanisms, with the former potentially reflecting a neurologic emergency given the presence of increased intracranial pressure. Common causes include intracranial hemorrhage, especially subarachnoid hemorrhage and intraventricular hemorrhage, and masses, particularly intraventricular, cerebellar or pineal tumors.\u003csup\u003e2\u003c/sup\u003e Its presence may necessitate the emergent placement of a ventricular drain to reduce intracranial pressure in addition to the management of the underlying cause.\u003c/p\u003e \u003cp\u003eArtificial intelligence (AI) has already been applied to many emergent neuroradiologic findings especially large vessel occlusion and intracranial hemorrhage identification.\u003csup\u003e3\u003c/sup\u003e Its key application has been to triage suspected abnormal radiology cases so that they can be interpreted sooner by radiologists and other clinicians, and so that time-critical interventions can commence.\u003csup\u003e4\u003c/sup\u003e Such software-as-medical-devices are often referred to as computer assisted triage devices (CADt).\u003c/p\u003e \u003cp\u003eIt has been proposed that AI could similarly be used to identify obstructive hydrocephalus on head CT and this study assessed the performance of an AI model by comparing its outputs to consensus neuroradiologist interpretations. The model was provided separately with thin (\u0026le;\u0026thinsp;1.5mm) and thick (\u0026gt;\u0026thinsp;1.5 and \u0026le;\u0026thinsp;5mm) axial series from each case so that the performance on different slice thicknesses could be calculated. Its performance was also evaluated across multiple subgroups to assess its robustness including in the presence or absence of common co-occurring findings.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eStudy design\u003c/h2\u003e \u003cp\u003eThis retrospective standalone model performance study was conducted using radiology cases from five hospitals within the Mass General Brigham (MGB) network between 2016 and 2022 using similar methods to previously published studies about intracranial hemorrhage, mass effect and vasogenic edema identification.\u003csup\u003e5,6\u003c/sup\u003e It was approved by the MGB Institutional Review Board with waiver of informed consent. It was conducted in accordance with relevant guidelines and regulations including the Health Insurance Portability and Accountability Act (HIPAA). This manuscript follows the Standards for Reporting Diagnostic Accuracy (STARD 2015) reporting guideline.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eCase selection\u003c/h3\u003e\n\u003cp\u003e The cohort was selected in a consecutive manner based on review of original radiology reports. The cohort size for each of the positive and negative cases was based on powering calculations as described in the \u003cspan refid=\"Sec8\" class=\"InternalRef\"\u003estatistical analysis\u003c/span\u003e section below. The positive cases were identified through a natural language processing search engine (Nuance mPower Clinical Analytics) followed by manual report review by an American Board of Radiology (ABR)-certified neuroradiologist. The negative cases were identified by taking the next negative case acquired on the same CT scanner after each positive case to avoid temporal and technical bias.\u003c/p\u003e \u003cp\u003eThe cohort considered all CT head cases performed at a hospital including inpatient and outpatient settings; there were no limitations on the original CT head clinical indication. The CT head cases were obtained from patients at least 18 years of age. The CT head cases were taken from unique patients; only the first CT head from a given patient was included.\u003c/p\u003e \u003cp\u003eAll cases were deidentified and underwent an image quality review by an ABR-certified neuroradiologist. The relevant series for the model interpretations were selected at the same time as described under the \u003cspan refid=\"Sec5\" class=\"InternalRef\"\u003eseries selection\u003c/span\u003e section below. The review was performed using the FDA-cleared eUnity image visualization software (Version 6 or higher) and an internal web-based annotation system that utilized the REDCap electronic data capture tools hosted at MGB.\u003csup\u003e7,8\u003c/sup\u003e\u003c/p\u003e\n\u003ch3\u003eSeries selection\u003c/h3\u003e\n\u003cp\u003eThe model was provided with a single selected series at the time of model inference. These series were non-contrast thin (\u0026le;\u0026thinsp;1.5mm) and thick (\u0026gt;\u0026thinsp;1.5 and \u0026le;\u0026thinsp;5mm) axial series for each CT head case. The series were selected such that the thin series was the thinnest available series\u0026thinsp;\u0026le;\u0026thinsp;1.5mm; the thick series was randomized between the thinnest and thickest available series\u0026thinsp;\u0026gt;\u0026thinsp;1.5 and \u0026le;\u0026thinsp;5mm to ensure representation of slice thicknesses across the entire range. After series selection, a DICOM metadata review was additionally performed to ensure that the slice thickness was within the appropriate range and that there was a consistent slice interval (with tolerance of \u0026plusmn;\u0026thinsp;0.2mm).\u003c/p\u003e\n\u003ch3\u003eGround truth interpretations\u003c/h3\u003e\n\u003cp\u003eGround truth interpretations were performed by up to three ABR-certified neuroradiologists. They answered whether obstructive hydrocephalus was \u0026ldquo;Present\u0026rdquo; or \u0026ldquo;Absent\u0026rdquo;. The definition of obstructive hydrocephalus was \u0026ldquo;Enlargement of one of more ventricles due to obstruction\u0026rdquo;. They provided their interpretations independently, without access to the original radiology reports and in different worklist orders. They used the same image visualization software and annotation system as was used in the image quality review. They had access to the entire CT head case (i.e., were not restricted to the series selected for model inference). For determining consensus interpretations, a \u0026ldquo;2\u0026thinsp;+\u0026thinsp;1\u0026rdquo; strategy was used: the first two neuroradiologists interpreted every case and a third neuroradiologist then interpreted cases with discrepant interpretations.\u003c/p\u003e \u003cp\u003eThe first two neuroradiologists also identified the presence of the co-existing findings or abnormalities of intracranial hemorrhage; parenchymal abnormality including ischemia, mass, cyst, encephalomalacia; and ventricular drain including extraventricular drain, ventriculoperitoneal shunt. These findings were considered present for a case if either of the neuroradiologists interpreted them as being present.\u003c/p\u003e\n\u003ch3\u003eModel inference\u003c/h3\u003e\n\u003cp\u003eThe evaluated AI model was version 3.1.0 of the Annalise Enterprise CTB Triage Trauma device. It is the same AI model used by the Annalise Enterprise (CTB module) device, which is commercially available in some non-US markets and whose development has been previously described.\u003csup\u003e9\u003c/sup\u003e In brief, it consists of an ensemble of five neural networks with three heads: one for classification, one for left-right localization and one for segmentation. It can identify 130 different radiological findings and was trained on over 200,000 CT head cases, which were each labelled by at least three radiologists.\u003c/p\u003e \u003cp\u003eThe Annalise Enterprise CTB Triage Trauma device only provides binary classification outputs about the identification of findings, which is consistent with US Food and Drug Administration (FDA) regulations for CADt devices. The model was installed at MGB for use in this study and received only the DICOM-formatted CT head cases. It outputted a classification score between 0 and 1 for obstructive hydrocephalus. A binary output could be derived using a prespecified default operating point (0.185900).\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eStatistical analysis\u003c/h2\u003e \u003cp\u003eThe statistical analysis was performed in R (version 4.0.2) on the full analysis set. The predefined endpoints included the area under the receiver operating characteristic curve (AUC) for the identification of obstructive hydrocephalus for each of thin and thick series. The AUCs were calculated using the consensus annotations and the classification scores from the AI model. The predefined endpoints also included the sensitivity and specificity at the operating point. They were calculated by comparing the binary model output at the operating point with the consensus annotations (i.e., by calculating the number of true positive, false negative, true negative and false positive cases).\u003c/p\u003e \u003cp\u003eThe AUCs, sensitivities and specificities were calculated as exploratory analyses for the subgroups of sex, age groups, ethnicity, race and CT scanner manufacturer. They were also calculated as exploratory analyses for subgroups based on the presence or absence of the co-existing findings or abnormalities. The positive predictive values (PPVs) and negative predictive values (NPVs) were calculated as exploratory analyses at different levels of assumed prevalence. The demographic parameters and scanner manufacturer were derived from clinical databases or DICOM fields for each radiology case; any missing data were treated as \u0026ldquo;Unavailable\u0026rdquo; and no data were imputed.\u003c/p\u003e \u003cp\u003eAll CIs were calculated using bootstrapped intervals with 2,000 resamples. The sample size was powered based on preliminary model results at a balanced operating point to ensure the lower bound of the 95% CI for sensitivity was \u0026gt;\u0026thinsp;80% and for specificity was \u0026gt;\u0026thinsp;80%.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cp\u003eThe cohort of CT head cases included 181 thin series and 200 thick series (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). The model successfully performed inference on 177 (97.8%) thin series and 194 (97.0%) thick series. The final analysis set of thin series cases included 77 (43.5%) positive cases and 100 (56.5%) negative cases; there were 98 (55.4%) women and 79 (44.6%) men; the mean (\u0026plusmn;\u0026thinsp;SD) age was 60.8 (\u0026plusmn;\u0026thinsp;18.8) years (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). The final analysis set of thick series cases included 87 (44.8%) positive cases and 107 (55.2%) negative cases; there were 106 (54.6%) women and 88 (45.4%) men; the mean (\u0026plusmn;\u0026thinsp;SD) age was 61.0 (\u0026plusmn;\u0026thinsp;18.8) years.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eDemographic and technical breakdown of CT head cases.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cp\u003eThin series\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003eThick series\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePositive\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNegative\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePositive\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eNegative\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eTotal\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e77\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e87\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e107\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eSex\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFemale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e44 (57.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e54 (54.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e51 (58.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e55 (51.4%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e33 (42.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e46 (46.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e36 (41.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e52 (48.6%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eAge\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026le;\u0026thinsp;65 years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e45 (58.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e49 (49.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e49 (56.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e52 (48.6%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026gt;\u0026thinsp;65 years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e32 (41.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e51 (51.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e38 (43.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e55 (51.4%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD (years)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e57.9\u0026thinsp;\u0026plusmn;\u0026thinsp;19.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e63.0\u0026thinsp;\u0026plusmn;\u0026thinsp;17.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e58.5\u0026thinsp;\u0026plusmn;\u0026thinsp;19.8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e63.1\u0026thinsp;\u0026plusmn;\u0026thinsp;17.8\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eEthnicity\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHispanic\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7 (9.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e10 (10.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e7 (8.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e10 (9.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNot Hispanic\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e63 (81.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e81 (81.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e73 (83.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e87 (81.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUnavailable\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7 (9.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e9 (9.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e7 (8.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e10 (9.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eRace\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAsian\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5 (6.5%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1 (1.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e7 (8.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1 (0.9%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBlack or African American\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4 (5.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e9 (9.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4 (4.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e11 (10.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNative Hawaiian or Other Pacific Islander\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0 (0.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1 (1.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0 (0.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1 (0.9%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWhite\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e58 (75.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e77 (77.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e66 (75.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e82 (76.6%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOther\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7 (9.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e4 (4.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e7 (8.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e4 (3.7%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTwo or more\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0 (0.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2 (2.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0 (0.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2 (1.9%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDeclined\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1 (1.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1 (1.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1 (1.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1 (0.9%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUnavailable\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2 (2.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e5 (5.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2 (2.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e5 (4.7%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eManufacturer\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGE Healthcare\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e8 (10.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e19 (19.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e16 (18.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e24 (22.4%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNeuroLogica\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1 (1.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1 (1.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2 (2.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1 (0.9%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSiemens\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e66 (85.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e78 (78.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e66 (75.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e79 (73.8%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eToshiba\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2 (2.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2 (2.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3 (3.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3 (2.8%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThe model performed with an AUC of 0.988 (95% CI: 0.971 to 0.998) on thin series and 0.986 (95% CI: 0.969 to 0.997) on thick series (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e and Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). These results corresponded to a sensitivity of 93.5% (95% CI: 88.3 to 98.7%) and specificity of 95.0% (95% CI: 90.0 to 99.0%) on thin series, and a sensitivity of 93.1% (95% CI: 87.4 to 97.7%) and specificity of 95.3% (95% CI: 90.7 to 99.1%) on thick series. All demographic and scanner manufacturer subgroups had an AUC of at least 0.95 for both thin series (Supplementary Table\u0026nbsp;1) and thick series (Supplementary Table\u0026nbsp;2). The sensitivities and specificities of subgroups were at least 80% whenever there were at least 5 positive or negative cases respectively for the subgroup.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eModel performance summary.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMetric\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eThin series\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eThick series\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePositive N\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e77\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e87\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNegative N\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e107\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAUC (95% CI)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.988 (0.971 to 0.998)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.986 (0.969 to 0.997)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSensitivity (95% CI)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e93.5% (88.3 to 98.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e93.1% (87.4 to 97.7)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSpecificity (95% CI)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e95.0% (90.0 to 99.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e95.3% (90.7 to 99.1)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe PPVs and NPVs were calculated at different prevalences with recognition that the clinical implications of positive and negative model outputs are also dependent on the prevalence of a finding (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). For an assumed prevalence of 0.10, the PPV was 67.5% (95% CI: 51.6 to 91.3%) and the NPV was 99.2% (95% CI: 98.6 to 99.8%) on thin series, and the PPV was 68.9% (95% CI: 53.1 to 91.4%) and the NPV was 99.2% (95% CI: 98.5 to 99.7%) on thick series.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePPVs and NPVs at different levels of assumed prevalence.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cp\u003eThin series\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003eThick series\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAssumed prevalence\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePPV (95% CI)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNPV (95% CI)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePPV (95% CI)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eNPV (95% CI)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e0.05\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e49.6%\u003c/p\u003e \u003cp\u003e(33.6 to 83.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e99.6%\u003c/p\u003e \u003cp\u003e(99.3 to 99.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e51.2%\u003c/p\u003e \u003cp\u003e(35.0 to 83.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e99.6%\u003c/p\u003e \u003cp\u003e(99.3 to 99.9)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e0.10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e67.5%\u003c/p\u003e \u003cp\u003e(51.6 to 91.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e99.2%\u003c/p\u003e \u003cp\u003e(98.6 to 99.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e68.9%\u003c/p\u003e \u003cp\u003e(53.1 to 91.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e99.2%\u003c/p\u003e \u003cp\u003e(98.5 to 99.7)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e0.15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e76.7%\u003c/p\u003e \u003cp\u003e(62.9 to 94.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e98.8%\u003c/p\u003e \u003cp\u003e(97.8 to 99.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e77.9%\u003c/p\u003e \u003cp\u003e(64.3 to 94.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e98.7%\u003c/p\u003e \u003cp\u003e(97.7 to 99.6)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e0.20\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e82.4%\u003c/p\u003e \u003cp\u003e(70.6 to 95.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e98.3%\u003c/p\u003e \u003cp\u003e(96.8 to 99.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e83.3%\u003c/p\u003e \u003cp\u003e(71.8 to 96.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e98.2%\u003c/p\u003e \u003cp\u003e(96.8 to 99.4)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eA potential mimic finding for hydrocephalus is ventriculomegaly in the setting of age-related atrophy. The age subgroups were therefore separated further into 18 to 49 years, 50 to 64 years, 65 to 79 years and \u0026ge;\u0026thinsp;80 years (Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). The model did not appear to be impacted by age-related atrophy, with the model correctly interpreting all cases from patients aged\u0026thinsp;\u0026ge;\u0026thinsp;80 years. The model did show decreased sensitivities for the 50\u0026ndash;64 years age group for both the thin and thick series (82.4% (95% CI: 64.7 to 100.0%) and 84.2% (95% CI: 68.4 to 100.0%) respectively); in both situations, the decrease in these sensitivities reflected only 3 false negative cases.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eModel performance for age subgroups.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePositive N\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNegative N\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAUC \u003c/p\u003e \u003cp\u003e(95% CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSensitivity \u003c/p\u003e \u003cp\u003e(95% CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eSpecificity (95% CI)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eThin Series\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e18\u0026ndash;49 years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e24\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e23\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.996\u003c/p\u003e \u003cp\u003e(0.978 to 1.000)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e95.8%\u003c/p\u003e \u003cp\u003e(87.5 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e87.0%\u003c/p\u003e \u003cp\u003e(73.9 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e50\u0026ndash;64 years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e17\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e25\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.979\u003c/p\u003e \u003cp\u003e(0.922 to 1.000)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e82.4%\u003c/p\u003e \u003cp\u003e(64.7 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e96.0%\u003c/p\u003e \u003cp\u003e(88.0 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e65\u0026ndash;79 years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e26\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e37\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.986\u003c/p\u003e \u003cp\u003e(0.953 to 1.000)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e96.2%\u003c/p\u003e \u003cp\u003e(88.5 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e97.3%\u003c/p\u003e \u003cp\u003e(91.9 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026ge;\u0026thinsp;80 years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003cp\u003e(1.000 to 1.000)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e100.0%\u003c/p\u003e \u003cp\u003e(100.0 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e100.0%\u003c/p\u003e \u003cp\u003e(100.0 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eThick Series\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e18\u0026ndash;49 years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e26\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e25\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.995\u003c/p\u003e \u003cp\u003e(0.978 to 1.000)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e96.2%\u003c/p\u003e \u003cp\u003e(88.5 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e88.0%\u003c/p\u003e \u003cp\u003e(72.0 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e50\u0026ndash;64 years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e19\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e26\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.978\u003c/p\u003e \u003cp\u003e(0.919 to 1.000)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e84.2%\u003c/p\u003e \u003cp\u003e(68.4 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e96.2%\u003c/p\u003e \u003cp\u003e(88.5 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e65\u0026ndash;79 years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.988\u003c/p\u003e \u003cp\u003e(0.955 to 1.000)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e93.3%\u003c/p\u003e \u003cp\u003e(83.3 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e97.5%\u003c/p\u003e \u003cp\u003e(92.5 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026ge;\u0026thinsp;80 years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e12\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e16\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003cp\u003e(1.000 to 1.000)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e100.0%\u003c/p\u003e \u003cp\u003e(100.0 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e100.0%\u003c/p\u003e \u003cp\u003e(100.0 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThe model was also assessed in the presence or absence of intracranial hemorrhage, parenchymal abnormality and ventricular drain to ensure it was identifying obstructive hydrocephalus rather than these co-occurring findings (Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e5\u003c/span\u003e). Intracranial hemorrhage is one of many causes of obstructive hydrocephalus; the model achieved sensitivity of at least 88.5% both in its presence and absence, suggesting the model identified obstructive hydrocephalus regardless of whether the cause was intracranial hemorrhage. A parenchymal abnormality was present on almost all positive cases and the subgroup analysis was most relevant in ensuring that the model did not produce false positive interpretations when one was present; the model maintained a specificity of at least 90.0% both in the presence and absence of a parenchymal abnormality. A ventricular drain is a treatment for obstructive hydrocephalus and a hypothetic non-specific model could learn to identify it rather than obstructive hydrocephalus during training; this model demonstrated sensitivity of at least 82.4% when it was present and at least 95.2% when it was absent.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eModel performance for co-existing findings or abnormalities subgroups.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePositive N\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNegative N\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAUC \u003c/p\u003e \u003cp\u003e(95% CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSensitivity \u003c/p\u003e \u003cp\u003e(95% CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eSpecificity (95% CI)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eThin Series\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eIntracranial hemorrhage\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePresent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e51\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e32\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.978\u003c/p\u003e \u003cp\u003e(0.939 to 0.999)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e96.1%\u003c/p\u003e \u003cp\u003e(90.2 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e90.6%\u003c/p\u003e \u003cp\u003e(81.2 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAbsent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e26\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e68\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.995\u003c/p\u003e \u003cp\u003e(0.983 to 1.000)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e88.5%\u003c/p\u003e \u003cp\u003e(73.1 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e97.1%\u003c/p\u003e \u003cp\u003e(92.6 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eParenchymal abnormality\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePresent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e72\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.980\u003c/p\u003e \u003cp\u003e(0.947 to 0.997)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e93.1%\u003c/p\u003e \u003cp\u003e(86.1 to 98.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e90.0%\u003c/p\u003e \u003cp\u003e(80.0 to 97.5)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAbsent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e60\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.997\u003c/p\u003e \u003cp\u003e(0.970 to 1.000)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e100.0%\u003c/p\u003e \u003cp\u003e(100.0 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e98.3%\u003c/p\u003e \u003cp\u003e(95.0 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eVentricular drain\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePresent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003cp\u003e(1.000 to 1.000)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e86.7%\u003c/p\u003e \u003cp\u003e(66.7 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e100.0%\u003c/p\u003e \u003cp\u003e(100.0 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAbsent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e62\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e96\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.987\u003c/p\u003e \u003cp\u003e(0.970 to 0.998)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e95.2%\u003c/p\u003e \u003cp\u003e(88.7 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e94.8%\u003c/p\u003e \u003cp\u003e(89.6 to 99.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eThick Series\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eIntracranial hemorrhage\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePresent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e59\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e33\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.975\u003c/p\u003e \u003cp\u003e(0.932 to 0.998)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e94.9%\u003c/p\u003e \u003cp\u003e(88.1 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e90.9%\u003c/p\u003e \u003cp\u003e(81.8 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAbsent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e28\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e74\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.993\u003c/p\u003e \u003cp\u003e(0.977 to 1.000)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e89.3%\u003c/p\u003e \u003cp\u003e(75.0 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e97.3%\u003c/p\u003e \u003cp\u003e(93.2 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eParenchymal abnormality\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePresent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e81\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.976\u003c/p\u003e \u003cp\u003e(0.937 to 0.997)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e92.6%\u003c/p\u003e \u003cp\u003e(86.4 to 97.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e90.0%\u003c/p\u003e \u003cp\u003e(80.0 to 97.5)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAbsent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e67\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.998\u003c/p\u003e \u003cp\u003e(0.978 to 1.000)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e100.0%\u003c/p\u003e \u003cp\u003e(100.0 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e98.5\u003c/p\u003e \u003cp\u003e(95.5 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eVentricular drain\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePresent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e17\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003cp\u003e(1.000 to 1.000)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e82.4%\u003c/p\u003e \u003cp\u003e(64.7 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e100.0%\u003c/p\u003e \u003cp\u003e(100.0 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAbsent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e70\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e103\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.986\u003c/p\u003e \u003cp\u003e(0.969 to 0.998)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e95.7%\u003c/p\u003e \u003cp\u003e(90.0 to 100.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e95.1%\u003c/p\u003e \u003cp\u003e(91.2 to 99.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis retrospective diagnostic study assessed the performance of an AI model in identifying obstructive hydrocephalus on head CT. The model achieved an AUC of 0.988 on thin series and 0.986 on thick series, and sensitivities and specificities of at least 93.1% at the pre-defined operating point for both series. These results were mostly maintained across demographic subgroups including sex, age, ethnicity and race, and scanner manufacturer subgroups. These results were also consistent with the expected performance of CADt devices through the US FDA and this model has subsequently received 510(k) clearance through the US FDA as the first CADt device that identifies obstructive hydrocephalus.\u003csup\u003e10\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eA key consideration with radiologic AI models is ensuring that they are identifying the intended finding and not a mimic or co-occurring finding. While hopefully the radiology and data science communities are more aware of this potential pitfall now, previous studies have shown that AI models may learn incorrect \u0026ldquo;shortcuts\u0026rdquo; for the detection of findings including non-imaging features like radiologist annotations on abnormal cases.\u003csup\u003e11,12\u003c/sup\u003e Obstructive hydrocephalus is a finding that has many mimics and common co-occurring findings. Here we showed that the model performance was not impacted in older age subgroups, suggesting age-related atrophy was not leading to false positive case identification.\u003c/p\u003e \u003cp\u003eWe similarly showed that the model identified obstructive hydrocephalus regardless of the presence or absence of the co-occurring findings of intracranial hemorrhage, parenchymal abnormality and ventricular drain. Intracranial hemorrhage is a common cause of obstructive hydrocephalus; we considered the maintenance of sensitivity in the presence or absence of intracranial hemorrhage to show that the model could identify obstructive hydrocephalus from both hemorrhagic and non-hemorrhagic causes. A parenchymal abnormality can likewise reflect a cause of obstructive hydrocephalus; however, almost all cases that were positive for obstructive hydrocephalus had a parenchymal abnormality, which limited the conclusions from this subgroup analysis. At the same time, there were many cases with a parenchymal abnormality that were negative for obstructive hydrocephalus; we considered the maintenance of a specificity of 90% among these cases to suggest that the model was identifying obstructive hydrocephalus rather than an underlying parenchymal abnormality causing obstructive hydrocephalus. The model performed similarly in the presence or absence of a ventricular drain; we were reassured that it identified obstructive hydrocephalus in the absence of a ventricular drain especially as this clinical scenario is more likely to reflect patients who have not yet received treatment for obstructive hydrocephalus.\u003c/p\u003e \u003cp\u003eA takeaway of having a robust model is that it could play a key role in the clinical environment. CADt devices aim to aid in the prioritization and triage of cases with suspected time sensitive findings with an ultimate goal of enabling management to commence sooner.\u003csup\u003e13\u003c/sup\u003e Obstructive hydrocephalus is a critical finding with management steps including the possible placement of a ventricular drain. This model therefore has the potential to positively impact patient care and it will be important to formally evaluate its effectiveness moving forward.\u003c/p\u003e \u003cp\u003eThe main limitation of this study is that it was a retrospective study outside of the clinical workflow. The effectiveness of the model will ultimately be best assessed in a prospective manner within the clinical environment. An initial assessment could consider key metrics before and after the implementation of this model. From a radiologic endpoint perspective, these metrics could include changes in the time to interpretation and time to notification of the responding clinician for cases with obstructive hydrocephalus. From a broader clinical endpoint perspective, these metrics could consider time to intervention for patients with obstructive hydrocephalus, percentage of patients who receive an intervention, and morbidity or mortality rates.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThis diagnostic study assessed an AI model that accurately detected obstructive hydrocephalus. It maintained this performance across many subgroup analyses suggesting its robustness.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e \u003ch2\u003eAccess to data and data analysis\u003c/h2\u003e \u003cp\u003eJMH had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. AG and SFM performed the statistical analyses.\u003c/p\u003e \u003c/p\u003e\u003cp\u003e\u003cstrong\u003eCompeting Interests\u003c/strong\u003e\u003cp\u003eThis study was funded by Annalise-AI. Annalise-AI were involved in the design and conduct of the study; preparation, review, and approval of the manuscript; and decision to submit the manuscript for publication. Annalise-AI were not involved in the collection, management, analysis, and interpretation of the data. AG, INC, SFM, JKC, MAH, EL, ALM, ASS, KB, JC, WM, SP, SR, BCB, JMH are employees of Mass General Brigham and/or Massachusetts General Hospital, which had received institutional funding from Annalise-AI for the study.\u003c/p\u003e\u003c/p\u003e\u003ch2\u003eFunding\u003c/h2\u003e \u003cp\u003eThis study was funded by Annalise-AI. Annalise-AI were involved in the design and conduct of the study; preparation, review, and approval of the manuscript; and decision to submit the manuscript for publication. Annalise-AI were not involved in the collection, management, analysis, and interpretation of the data. AG, INC, SFM, JKC, MAH, EL, ALM, ASS, KB, JC, WM, SP, SR, BCB, JMH are employees of Mass General Brigham and/or Massachusetts General Hospital, which had received institutional funding from Annalise-AI for the study.\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eThe authors contributed to the paper as follows: study conception and design: INC, SFM, JKC, WAM, BCB, JMH; data collection: AG, INC, SFM, JKC, MAH, EL, ALM, ASS, BCB, JMH; ground truth radiologist interpretations: KB, JC, WAM, SP, SR; statistical analysis: AG, SFM; draft manuscript preparation: AG, INC, SFM, JMH. All authors reviewed and approved the manuscript.\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003eThe authors thank the broader Mass General Brigham Data Science Office and Annalise teams for their assistance with this project.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eThe data from this study include protected health information and are not openly available. Components of the data may be available from the corresponding author upon reasonable request.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003e1. Matson, D. D. Hydrocephalus. \u003cem\u003eN Engl J Med\u003c/em\u003e \u003cb\u003e271\u003c/b\u003e, 1360\u0026ndash;1362, doi:10.1056/NEJM196412242712608 (1964).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e2. Wijdicks, E. F. M. \u003cem\u003eThe practice of emergency and critical care neurology\u003c/em\u003e. Second edition. edn, (Oxford University Press, 2016).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e3. Chandrabhatla, A. S. \u003cem\u003eet al.\u003c/em\u003e Artificial Intelligence and Machine Learning in the Diagnosis and Management of Stroke: A Narrative Review of United States Food and Drug Administration-Approved Technologies. \u003cem\u003eJ Clin Med\u003c/em\u003e \u003cb\u003e12\u003c/b\u003e, doi:10.3390/jcm12113755 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e4. Rajpurkar, P. \u0026amp; Lungren, M. P. The Current and Future State of AI Interpretation of Medical Images. \u003cem\u003eN Engl J Med\u003c/em\u003e \u003cb\u003e388\u003c/b\u003e, 1981\u0026ndash;1990, doi:10.1056/NEJMra2301725 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e5. Hillis, J. M. \u003cem\u003eet al.\u003c/em\u003e Evaluation of an Artificial Intelligence Model for Identification of Intracranial Hemorrhage Subtypes on Computed Tomography of the Head. \u003cem\u003eStroke: Vascular and Interventional Neurology\u003c/em\u003e \u003cb\u003e4\u003c/b\u003e, e001223, doi:10.1161/SVIN.123.001223 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e6. Newbury-Chaet, I. \u003cem\u003eet al.\u003c/em\u003e Evaluation of an Artificial Intelligence Model for Identification of Mass Effect and Vasogenic Edema on CT of the Head. \u003cem\u003eAJNR Am J Neuroradiol\u003c/em\u003e \u003cb\u003e45\u003c/b\u003e, 1528\u0026ndash;1535, doi:10.3174/ajnr.A8358 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e7. Harris, P. A. \u003cem\u003eet al.\u003c/em\u003e The REDCap consortium: Building an international community of software platform partners. \u003cem\u003eJ Biomed Inform\u003c/em\u003e \u003cb\u003e95\u003c/b\u003e, 103208, doi:10.1016/j.jbi.2019.103208 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e8. Harris, P. A. \u003cem\u003eet al.\u003c/em\u003e Research electronic data capture (REDCap)--a metadata-driven methodology and workflow process for providing translational research informatics support. \u003cem\u003eJ Biomed Inform\u003c/em\u003e \u003cb\u003e42\u003c/b\u003e, 377\u0026ndash;381, doi:10.1016/j.jbi.2008.08.010 (2009).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e9. Buchlak, Q. \u003cem\u003eet al.\u003c/em\u003e Effects of a comprehensive brain computed tomography deep-learning model on radiologist detection accuracy: a multireader, multicase study. (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e10. US Food and Drug Administration. \u003cem\u003eK231094 (Annalise Enterprise CTB Triage - OH)\u003c/em\u003e, \u0026lt;https://www.accessdata.fda.gov/cdrh_docs/pdf23/K231094.pdf\u0026gt; (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e11. DeGrave, A. J., Janizek, J. D. \u0026amp; Lee, S. I. AI for radiographic COVID-19 detection selects shortcuts over signal. \u003cem\u003emedRxiv\u003c/em\u003e, doi:10.1101/2020.09.13.20193565 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e12. Zech, J. R. \u003cem\u003eet al.\u003c/em\u003e Variable generalization performance of a deep learning model to detect pneumonia in chest radiographs: A cross-sectional study. \u003cem\u003ePLoS Med\u003c/em\u003e \u003cb\u003e15\u003c/b\u003e, e1002683, doi:10.1371/journal.pmed.1002683 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e13. U.S. Department of Health and Human Services. \u003cem\u003e21CFR 892.2080 Radiological computer aided triage and notification software\u003c/em\u003e, \u0026lt;https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/cfrsearch.cfm?fr=892.2080\u0026gt; (2020).\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"},{"header":"Supplemental Tables","content":" \u003cp\u003e\u003cstrong\u003eSupplemental Table 1\u003c/strong\u003e: Demographic and scanner manufacturer subgroup performance for identifying obstructive hydrocephalus in thin series.\u003c/p\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Taba\" border=\"1\"\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colspan=\"5\" nameend=\"c6\" namest=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003eThin Series (Operating Point 0. 0.1859)\u003c/div\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003ePositive N\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003eNegative N\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003eAUC (95% CI)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003eSensitivity (95% CI)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003eSpecificity (95% CI)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003eOverall\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e77\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e100\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.988\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(0.971 to 0.998)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e93.5%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(88.3 to 98.7)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e95.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(90.0 to 99.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003eSex\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eFemale\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e44\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e54\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.978\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(0.950 to 0.997)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e90.9%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(81.8 to 97.7)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e94.4%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(88.9 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eMale\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e33\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e46\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.999\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(0.994 to 1.000)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e97.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(90.9 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e95.7%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(89.1 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003eAge\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u0026le;\u0026thinsp;65 years\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e45\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e49\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.986\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(0.962 to 0.998)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e91.1%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(82.2 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e91.8%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(83.7 to 98.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u0026gt;\u0026thinsp;65 years\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e32\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e51\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.991\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(0.967 to 1.000)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e96.9%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(90.6 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e98.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(94.1 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003eEthnicity\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eHispanic\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e7\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e10\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e1.000\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(1.000 to 1.000)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eNot Hispanic\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e63\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e81\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.984\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(0.962 to 0.997)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e92.1%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(84.1 to 98.4)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e95.1%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(90.1 to 98.8)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eUnavailable\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e7\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e9\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e1.000\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(1.000 to 1.000)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e88.9%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(66.7 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003eRace\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eAsian\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e5\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e1\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e1.000\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(1.000 to 1.000)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eBlack or African American\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e4\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e9\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e1.000\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(0.833 to 1.000)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e75.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(25.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e88.9%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(66.7 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eNative Hawaiian or Other Pacific Islander\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e0\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e1\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e-\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e-\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eWhite\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e58\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e77\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.986\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(0.966 to 0.999)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e93.1%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(86.2 to 98.3)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e96.1%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(90.9 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eOther\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e7\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e4\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e1.000\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(1.000 to 1.000)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eTwo or more\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e0\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e2\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e-\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e-\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eDeclined\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e1\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e1\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e1.000\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(1.000 to 1.000)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eUnavailable\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e2\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e5\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e1.000\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(1.000 to 1.000)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e80.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(40.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003eManufacturer\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eGE Healthcare\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e8\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e19\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e1.000\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(0.961 to 1.000)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e84.2%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(68.4 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eNeuroLogica\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e1\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e1\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e1.000\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(1.000 to 1.000)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(0.0. to 0.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eSiemens\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e66\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e78\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.991\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(0.973 to 1.000)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e93.9%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(87.9 to 98.5)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e97.4%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(93.6 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eToshiba\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e2\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e2\u003c/div\u003e \u003c/td\u003e \n\u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e1.000\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(1.000 to 1.000)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003cbr/\u003e \n\u003cp\u003e\u003cstrong\u003eSupplemental Table 2\u003c/strong\u003e: Demographic and scanner manufacturer subgroup performance for identifying obstructive hydrocephalus in thick series.\u003c/p\u003e\n\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Tabb\" border=\"1\"\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colspan=\"5\" nameend=\"c6\" namest=\"c2\"\u003e \n\u003cdiv class=\"SimplePara\"\u003eThick Series (Operating Point 0. 0.1859)\u003c/div\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003ePositive N\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003eNegative N\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003eAUC (95% CI)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003eSensitivity (95% CI)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003eSpecificity (95% CI)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003eOverall\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e87\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e107\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.986\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(0.969 to 0.997)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e93.1%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(87.4 to 97.7)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e95.3%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(90.7 to 99.1)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003eSex\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eFemale\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e51\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e55\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.979\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(0.949 to 0.998)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e90.2%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(80.4 to 98.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e94.5%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(87.3 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eMale\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e36\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e52\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.997\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(0.987 to 1.000)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e97.2%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(91.7 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e96.2%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(90.4 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003eAge\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u0026le;\u0026thinsp;65 years\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e49\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e52\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.984\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(0.958 to 0.998)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e91.8%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(83.7 to 98.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e92.3%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(84.6 to 98.1)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u0026gt;\u0026thinsp;65 years\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e38\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e55\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.989\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(0.961 to 1.000)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e94.7%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(86.8 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e98.2%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(94.5 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003eEthnicity\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eHispanic\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e7\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e10\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e1.000\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(1.000 to 1.000)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eNot Hispanic\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e73\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e87\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.984\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(0.962 to 0.998)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e91.8%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(84.9 to 97.3)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e95.4%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(90.8 to 98.9)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eUnavailable\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e7\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e10\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e1.000\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(1.000 to 1.000)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e90.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(70.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003eRace\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eAsian\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e7\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e1\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e1.000\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(1.000 to 1.000)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eBlack or African American\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e4\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e11\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.955\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(0.795 to 1.000)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e75.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(25.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e90.9%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(72.7 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eNative Hawaiian or Other Pacific Islander\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e0\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e1\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e-\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e-\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eWhite\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e66\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e82\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.986\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(0.965 to 0.999)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e92.4%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(84.8 to 98.5)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e96.3%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(91.5 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eOther\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e7\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e4\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e1.000\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(1.000 to 1.000)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eTwo or more\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e0\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e2\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e-\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e-\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eDeclined\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e1\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e1\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e1.000\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(1.000 to 1.000)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eUnavailable\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e2\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e5\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e1.000\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(1.000 to 1.000)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e80.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(40.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003eManufacturer\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eGE Healthcare\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e16\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e24\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.982\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(0.927 to 1.000)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e87.5%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(75.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eNeuroLogica\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e2\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e1\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e1.000\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(1.000 to 1.000)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(0.0 to 0.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eSiemens\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e66\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e79\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.991\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(0.972 to 1.000)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e93.9%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(87.9 to 98.5)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e97.5%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(93.7 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eToshiba\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e3\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e3\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e1.000\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(1.000 to 1.000)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e100.0%\u003c/div\u003e \u003cdiv class=\"SimplePara\"\u003e(100.0 to 100.0)\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003cbr/\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"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":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-5487343/v2","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5487343/v2","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eIntroduction:\u003c/h2\u003e \u003cp\u003eObstructive hydrocephalus is a critical radiographic finding requiring emergent treatment. Its identification on head computed tomography (CT) by an artificial intelligence (AI) model could facilitate sooner life-saving interventions, although there are common co-occurring findings including intracranial hemorrhage that can confound this interpretation. This study assessed the accuracy of an AI model (Annalise Enterprise CTB) at identifying obstructive hydrocephalus including in the presence or absence of other findings.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eThis retrospective cohort included 177 thin (\u0026le;\u0026thinsp;1.5mm) series and 194 thick (\u0026gt;\u0026thinsp;1.5 and \u0026le;\u0026thinsp;5mm) series from 200 non-contrast head CT cases. These cases were obtained from patients aged\u0026thinsp;\u0026ge;\u0026thinsp;18 years at 5 hospitals in the United States. Each case was interpreted independently by up to three neuroradiologists. Each series was then interpreted by the AI model.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eThe AI model performed with area under the curve 0.988 (95% confidence interval (CI): 0.971 to 0.998) on thin series and 0.986 (95% CI: 0.969 to 0.997) on thick series. These results were broadly maintained in subgroups for the presence or absence of intracranial hemorrhage, parenchymal abnormality and ventricular drain, and across demographic and scanner manufacturer subgroups.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eThe AI model accurately identified obstructive hydrocephalus in this dataset. Its performance in subgroup analyses reflected its robustness.\u003c/p\u003e","manuscriptTitle":"Evaluation of an artificial intelligence model for identification of obstructive hydrocephalus on computed tomography of the head","msid":"","msnumber":"","nonDraftVersions":[{"code":2,"date":"2024-12-06 19:24:55","doi":"10.21203/rs.3.rs-5487343/v2","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}},{"code":1,"date":"2024-11-28 13:44:24","doi":"10.21203/rs.3.rs-5487343/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"491ab7de-76b1-4897-a6b4-1619f8bcd9d9","owner":[],"postedDate":"December 6th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":41268363,"name":"Health sciences/Neurology/Neurological disorders/Hydrocephalus"},{"id":41268364,"name":"Health sciences/Medical research/Translational research"},{"id":41268365,"name":"Health sciences/Health care/Medical imaging/Brain imaging"},{"id":41268366,"name":"Health sciences/Health care/Medical imaging/Tomography/Computed tomography"}],"tags":[],"updatedAt":"2025-02-27T06:38:46+00:00","versionOfRecord":[],"versionCreatedAt":"2024-12-06 19:24:55","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v2","identity":"rs-5487343","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5487343","identity":"rs-5487343","version":["v2"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}