Construction and clinical application of the cochlear automatic segmentation and measurement model based on U-HRCT

Construction and clinical application of the cochlear automatic segmentation and measurement model based on U-HRCT | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Construction and clinical application of the cochlear automatic segmentation and measurement model based on U-HRCT Dewu Yang, Mengshi Zhang, Yili Feng, Shuguang Han, Ruowei Tang, and 6 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5972058/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Objectives In order to improve the accuracy and efficiency of data acquisition for cochlear structure, this study built an intelligent segmentation and automatic measurement model of cochlear based on U-HRCT to solve the difficulty of performing an accurate preoperative evaluation of the cochlea. Methods The data of 86 patients (172 ears in total) who received U-HRCT and were diagnosed with normal ears were retrospectively collected. In this study, TransUnet neural network was used to construct automatic segmentation model of cochlear, 7 direct measurement indexes were selected and 3 indirect measurement indexes were defined. The effect of automatic segmentation was evaluated by subjective method, compared with manual measured data, and the influencing factors of cochlear structure were analyzed by multiple linear regression method. Finally, the Violin Plot method is used to calculate the normal value range of cochlear structural parameters. Results The index of automatic segmentation DSC reached 98%, and the results of automatic measurement were consistent with those of manual measurement (P > 0.05). Except for regularity parameters CCR and HRD, cochlear structural parameters had high correlation with gender (P 0.05). Except VRD, the correlation between the cochlear structure parameter and the location was low (P > 0.05). Conclusions This study built an automatic segmentation and measurement model of cochlear structure, obtained the reference value range of cochlear structure parameters, and provided reference data for clinical diagnosis and treatment. Clinical relevance statement The above indexes have important reference value for the automatic identification of cochlear structural abnormalities in clinic, and in-depth data mining of case samples will be carried out in subsequent studies. Cochlea Base of cochlea Cochlear duct Regularity Intelligent segmentation Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Key Points Manual measurement of cochlear CT image is mostly used, and the accuracy and efficiency of manual measurement are poor. Automatic segmentation of cochlea reached good level, and the automatic measurement results were in good agreement with the manual measurement results. The reference range of cochlear structural parameters of different genders can provide possibilities for improving the accuracy of disease diagnosis and treatment and reducing the risk of surgery. 1. Introduction The cochlea is an important part of the bone labyrinth of the internal ear, which contains important structures responsible for converting acoustic vibrations into neural signals. The overall structure of the cochlea is a spiral bone duct, which is formed by the cochlear duct coiled 2.5–2.75 cycles around the cochlear axis[1]. When the cochlear structure is abnormal due to congenital developmental defects or acquired injuries, which can lead to hearing loss or even deafness, cochlear implantation can restore partial hearing for patients, and is suitable for patients with extremely severe hearing loss or total deafness[2]. Therefore, the measurement of cochlear structure can help diagnose the deformity of the internal ear, relieve the pressure of clinical diagnosis, provide a basis for later surgical treatment, and provide design references for cochlear implant manufacturers. High resolution CT images can show the structure of the cochlea, which is the first choice in the diagnosis of inner ear diseases. However, its ability to detect the spatial edge of the fine structure of the inner layer is limited, and the accuracy of some test data is poor. With the advent of ultra-high resolution CT (U-HRCT) dedicated to the ear department, the spatial resolution of the internal ear structure can reach 50 ~ 100µm, which provides high-quality CT images (as shown in Fig. 1) for the identification of the cochlea structure before surgery, and can clearly display the minute structure and abnormal lesions of the cochlea, which is an important data basis for structural segmentation. It has become a key technical means for cochlear imaging and evaluation before cochlear implantation [3]. In the domestic and foreign researches on CT image segmentation of cochlear, in addition to the traditional threshold classification, regional growth and atlas construction methods, the neural network machine learning method is mostly used to automatically identify the cochlear structure. Floris Heutink et al. developed and validated the automatic cochlear segmentation and measurement system based on deep learning. The results of automatic segmentation show that the Dice coefficient is 0.90 ± 0.03[4]. Zhenhua Li et al. investigate the feasibility of a deep learning method based on a UNETR model for fully automatic segmentation of the cochlea in temporal bone CT images, and the Dice coefficient of the normal cochlear test set was 0.92[5]. Yi Lv et al. proposed a lightweight three-dimensional convolutional neural network (W-Net) for segmenting multiple targets including cochlea, ossicular chain, and facial nerve from conventional CT images of temporal bone and achieves human-level accuracy[6]. However, the segmentation accuracy needs to rely on the accuracy of data annotation and image quality, so the generalization ability and robustness of the model are poor. By extracting the cochlear Central Line from the cochlear segmentation results, the length and curvature of the cochlear Central Line, the volume and area of the cochlea and other quantitative indexes were calculated, and the physiological function of the cochlea was further evaluated. Ibraheem Al-Dhamari et al. proposed an automated fast cochlear segmentation, length, and volume estimation method from clinical 3D multimodal images, and evaluated using 3D landmarks located by two experts[7]. Tawfiq Khurayzi et al. developed ontological planning software (OTOPLAN) allows surgeons to directly measure parameters such as the diameter and width of the cochlea and then automatically calculate the CDL[8]. However, the selection of cochlear measurement indexes has not yet formed a normative standard, the results produced by different measurement methods are somewhat different, and there is less research experience in automatic measurement. Intelligent measurement and anomaly judgment put forward higher requirements on the data capacity and segmentation accuracy of training samples. In this study, based on the automatic segmentation research of deep learning algorithm, the cochlear automatic segmentation model was built through TransUnet neural network algorithm, and automatic fitting measurement was used to realize the automatic measurement of cochlear structural parameters. The problems of inaccuracy of traditional manual measurement and the limitation of two-dimensional image measurement were solved. Automatic segmentation and measurement models provide important diagnostic information for personalized treatment of cochlear surgery, reducing surgical risk and improving surgical success. 2. Methods 2.1 Automatic segmentation model of cochlear based on TransUnet neural network The TransUnet neural network model combines the advantages of U-net and Transformer to become an important improvement scheme for medical image segmentation [9-10]. The TransUnet model is a U-shaped Encoder-Decoder structure, and the network architecture is shown in Figure 2. Among them, Encoder input the original image into the self-convolutional neural network (CNN) for feature extraction, and after linear projection, the toenized image is serialized and position encoded, while Decoder up-samples the encoded features and connects them through the resolution feature map of the encoder CNN. The combination achieves accurate positioning and segmentation [11-12]. Based on the overall segmentation of the labyrinth of the internal ear [13], this study made full use of the prior information of anatomical structure as guidance, designed the pseudo-label proxy task, combined with the correlation between the overall and substructure segmentation, designed the teacher-student framework, and finally realized the segmentation of the substructure of the internal ear, including the three semicircular canals, vestibule and cochlea [14]. The DSC index of cochlea is 98%, which provides important support for the quantitative analysis of cochlear structure. 2.2 Automatic cochlear measurement based on precise segmentation (1)Measuring indexes. Based on the actual needs of clinical surgical evaluation and related research progress, the cochlear measurement indexes were defined, including direct and indirect measurement indexes. A. Direct indexes include: ① Basal long diameter (BLD), defined as the longest distance between the boundary points of the projection of the cochlear base; ② Basal short diameter (BSD), defined as the distance from the midpoint of BLD to the boundary of the projection of the cochlear base, which is on the vertical line passing through the midpoint of BLD; ③ Cochlear height (HGT), defined as the vertical distance between the cupula cochlea and the base of the cochlea; ④ Cochlear duct length (CDL), defined as the length of the fitted centerline of the cochlear duct; ⑤ Cochlear volume (VOL), defined as the total number of pixel points enclosed by the cochlear shell; ⑥ Two-turn length (2TL), defined as the length from the entrance of the cochlea to the second turn (i.e., the cochlear axis); ⑦ Basal turn length (BTL), defined as the length from the entrance of the cochlea to the basal turn (the innermost layer of the cochlea). B. Indirect indexes include: ① Cochlear Curvature Ratio (CCR), defined as the ratio of the cochlear duct length to the mean of the long and short diameters; ② Height Regularity Degree (HRD), defined as the rate of the cochlear height to the mean of the long and short diameters; ③ Volume Regularity Degree (VRD), defined as the rate of the cochlear volume tothe mean of the long and short diameters. (2) Measurement Method. The measurement of cochlear structural parameters is achieved by designing computerized automatic measurement algorithms for different indexes based on the precise segmentation of the cochlear structure. The automatic measurement of cochlear structure data mainly includes processes such as detecting cochlear voxel connected domain , calibrating spatial position, and measuring key data, etc. as shown in Figure 3. A. Denoising of segmentation results based on connected region analysis: The connected domains composed of adjacent pixel coordinates with the same segmentation category in the segmentation results are identified and marked, and the largest array of cochlear voxel group in the connected domains is retrieved through pixel traversal and label distribution. The other smaller noise points are removed[15]. B. Posture calibration of cochlea based on principal component analysis: Posture calibration of cochlea structural adopts the principal component analysis to analyze the coordinates of point cloud data and find the three perpendicular principal vectors with the largest spatial distribution. As the new coordinates of calibration, it rotate the target to realize the calibration of the spatial position of the cochlea [16-17]. C. Measurements of key structural data include five direct indexes: BLD, BSD, HGT, CDL, and VOL. Among these, the length of the cochlear duct is estimated using the formula (CDL = 4.16BLD - 4mm). The lengths from the entrance of the cochlea to the base turn and the second turn of the cochlear duct are set as the base-turn length (BTL) and two-turn length (2TL) , respectively. The estimation formulas are 2TL = 3.65 × (BLD - 1mm) and BTL = 2.43 × (BLD - 1mm)[18-19]. 2. Materials 2.1 Basic data A total of 86 patients (172 ears) underwent U-HRCT scanning in the Department of Radiology, Beijing Friendship Hospital Affiliated to Capital Medical University from July 2021 to September 2023. This study was reviewed by the Hospital Ethics Committee (IRB: 2024-P2-241) and approved. 2.2 Image acquisition All patients underwent U-HRCT examination using the U-HRCT scanner (Ultra3D, LargeV, Beijing) with a voltage of 100-110 kVp, a current of 120-180 mAs, and a view field of 65 mm. The thickness and spacing of slice are both set to 0.1 mm. Isotropic axial images are obtained after scanning for 20s. 2.3 Data Analysis SPSS25.0 was used for statistical analysis of basic data, segmentation effect, and measured data. Metrical data with normal distribution were expressed as 𝑥̅±𝑠, and paired sample t test was used for comparison of cases; when metrical data did not meet normal distribution, Wilcoxon rank sum test was used for comparison of cases in the same group. The difference was statistically significant with P < 0.05. (1) Analyse basic data. The basic data of 86 patients with normal cochlea were collected and demographic analysis was conducted on data of 172 ears in terms of age, gender, left/right ear, etc. (2) Analyse automatic segmentation effect. An evaluation team composed of 3 staff members with 5 years or more related work experience evaluated the effect of cochlear auto segmentation by subjective evaluation. The evaluation dimensions included segmentation accuracy, image clarity and diagnostic applicability. The scoring criteria are shown in Table 1. Table 1 Subjective evaluation criteria for the effect of cochlear auto-segmentation Scoring Criteria segmentation accuracy image clarity diagnostic applicability Very good (5points) Complete structure, accurate Smooth edges, clearly visible Fully meet diagnostic requirements Good(4points) Complete structure, slight deviation Smooth edges, visible Basically meet Average(3points) Basically complete Complete structure, obvious deviation Rough edges, visible Not meet Poor(2points) Basically complete Complete structure, obvious distortion Rough edges, vaguely visible May lead to missed diagnosis Very poor (1point) Incomplete Invisible May lead to misdiagnosis (3) Measurement data analysis. Firstly, the accuracy of measurement data is analyzed. One observer used Mimics software (19.0.0.347, Materialise, Belgium) to manually measure BLD, BSD, HGT, CDL, 2TL, BTL, as shown in Figure 4. After repeated measure for 3 times, the mean value was taken to compare the difference between manual and automatic measurement data. Secondly, by comparing the difference of indexdata among different age, gender or location, multiple linear regression was used to analyze the influencing factors of cochlear structure. Finally, the clinical value of the measurement data was analyzed and the Violin Plot method was used to calculate the normal value range. 3. Results 3.1 Analysis results of basic data A total of 86 patients were included in this study, including 50 males and 36 females, aged from 19 to 80 years old, with an average age of (38.36 ± 16.56) years old, with 86 left ears and 86 right ears, totaling 172, as shown in Table 2 . Table 2 Demographic information Characteristic Value Number of ears 172 Mean age (mean ± SD) 38.36 ± 16.56 Sex (n, %) Female 100(58.14) Male 72(41.86) Laterality (n, %) Right 86(50) Left 86(50) 3.2 Analysis results of cochlear auto-segmentation effect Compared with the original two-dimensional CT cochlear image and the automatic segmentation 3D image, MeshLeb software was used to perform the subjective evaluation of the segmentation effect, and the segmentation accuracy, image clarity and diagnostic applicability were analyzed. The evaluation results were (4.38 ± 0.75), (4.49 ± 0.66) and (4.58 ± 0.67) points, all of which reached a good level. The total number of poor and very poor were 4.75% (8/168), 1.78%(3/168) and 2.98%(5/168), respectively, which were controlled at a low level. 3.3 Analysis results of automatic measurement data (1) Accuracy of measurement data. The direct indexes of cochlear structure, namely, BLD, BSD, HGT, CDL, 2TL and BTL, and the differences among different measurement methods were not statistically significant (P > 0.05), as shown in Table 3 . Table 3 Analysis results of the accuracy of automatic measurement data Measurement data manual measurement Automatic measurement statistic P BLD(mm) 9.55 ± 0.44 9.60 ± 0.549 1.713 0.090 BSD(mm) 6.78 ± 0.37 6.82 ± 0.32 1.912 0.059 HGT(mm) 4.28 ± 0.38 4.24 ± 0.31 1.515 0.133 CDL(mm) 35.97 ± 2.09 35.94 ± 2.28 0.195 0.846 2TL(mm) 31.52 ± 1.77 31.40 ± 2.00 1.202 0.232 BTL(mm) 20.75 ± 1.90 20.90 ± 1.33 1.579 0.117 (2) Influence factors of cochlear structure. Multiple linear regression was used to analyze the influence degree of factors such as age, gender and location on the cochlear structure, and the results were shown in Table 4 . The direct measurement indexes of cochlear structure had very high correlation with gender (M/F), and the positive correlation coefficient were significant (P < 0.01). The indirect measurement indexes of CCR and HRD had poor correlation with gender, while the VRD had a high correlation with gender. Except for cochlear volume and volume regularity, other parameters had low correlation with age, and the negative correlation coefficient was not statistically significant(P > 0.05). The direct indexes and the indirect indexes of CCR and HRD were not significantly correlated with the left/right ear (R/L) (P > 0.05), while the VRD was significantly correlated with the left/right ear (R/L) (P < 0.05). The constant terms of multiple linear regression were statistically significant (P < 0.01). Table 4 Regression coefficients of influencing factors of cochlear structure Measurement Data Age(Y) Gender(M/F) Location(R/L) Constant term(d) BLD(mm) -0.004 0.332 ** 0.162 8.975 ** BSD(mm) -0.002 0.298 ** 0.056 6.367 ** HGT(mm) -0.001 0.184 ** -0.040 4.090 ** CDL(mm) -0.015 1.381 ** 0.673 33.335 ** VOL(mm) -0.099 * 7.955 ** -0.625 69.323 ** 2TL(mm) -0.013 1.212 ** 0.589 29.108 ** BTL(mm) -0.009 0.806 ** 0.393 19.380 ** CCR 0.000 -0.002 0.017 4.350 ** HRD 0.000 0.003 -0.010 0.532 ** VRD(mm 2 ) -0.009 ** 0.603 ** -0.195 * 9.048 ** * mean statistically significant difference (P < 0.05); ** mean statistically significant difference (P < 0.01). (3) Reference value of cochlear structure According to the results of multiple linear regression analysis of the influencing factors of cochlear structure, the reference values of cochlear structure parameters were calculated according to gender, that is, the Violin Plot method was used to analyze the distribution of each parameter data set. As shown in Fig. 5. The values in male group were as follows: BLD=[8.30,11.26], BSD=[6.21,7.74], HGT=[3.57,5.08], CDL=[30.51,42.83], VOL=[52.16,107.56], 2TL=[26.63,37.43], BTL=[17.72,24.92], CCR=[4.12,4.65], HRD=[0.42,0.62], VRD=[7.75,11.51]; The values in female group were : BLD=[8.21,10.72], BSD=[5.95,7.38], HGT=[3.57,4.70], CDL=[30.15,40.59], VOL=[56.53,89.21], 2TL=[26.31,35.47], BTL=[17.51,23.63], CCR=[4.09,4.66], HRD=[0.42,0.61], VRD=[7.68,10.46]. 4 Discussion Based on the data acquisition and analysis of ear U-HRCT images, this study proposed the cochlear automatic segmentation and measurement algorithm, which can be used to analyze clinical cases and assist precision diagnosis and treatment, and effectively solve the problem that traditional manual measurement methods are difficult to promote. We took a relatively large sample of Chinese normal adults as the research objects, and verified the accuracy of automatic segmentation and measurement algorithm by comparing clinical evaluation and manual measurement. The results showed that all automatic segmentation of cochlea reached good level, and the automatic measurement results were in good agreement with the manual measurement results. We proposed the reference range of cochlear structural parameters of different genders by using the quartile method to provide individual guidance for the diagnosis of clinical cochlear malformations and the formulation of preoperative protocols, which can provide possibilities for improving the accuracy of disease diagnosis and treatment and reducing the risk of surgery. Compared with domestic and foreign research results, the reliability of cochlear automatic measurement data in this study is higher. Masahiro Takahashi et al. used MRI to measure cochlear volume and used CT to measure cochlear duct length[20].Anandhan Dhanasingh et al. measured the structural parameters of the cochlea base with micro-computed tomography (µCT) images, and investigated the correlation between the diameter (A value), the width (B value) or the straight portion of the cochlea basal turn (S value) with the scala tympani (ST)[21]. Floris Heutink et al. proposed a method to automatically segment and measure the human cochlea in clinical ultra-high-resolution (UHR) CT images, and investigate differences in cochlea size for personalized implant planning[22]. Ferhat Geneci et al. investigated the inner ear anatomy accurately in detail by micro-computed tomography (micro-CT) to contribute to the data related to the inner ear anatomy[23].There is little difference between relevant research results and the results of this study, which further confirms the accuracy of cochlear automatic segmentation and measurement model. In addition, Jie Tang et al. used multiple axial micro-computed tomography and high-resolution CT 3D digital reconstruction model of the skull to analyze the temporal bones of 15 cadavers[24]. Raabid Hussain et al. analyzed 1000 + clinical temporal bone CT images using a web-based image analysis tool. Cochlear size and shape parameters were obtained to determine population statistics and perform regression and correlation analysis[25].The above researches provide important supplement for the accurate analysis of cochlear structure and adjacent relationship, and also provide references for the subsequent quantitative research of cochlea. In addition to conventional measurement indicators, this study innovatively proposed three measurement indexes representing the regularity of cochlear structure, including CCR, HRD and VRD, to analyze the degree of correlation between cochlear tube length, cochlear height and volume with the size of the cochlear base. The results showed that the correlation between CCR or HRD with age, sex or location was poor, indicating that the structural stability and similarity of normal adults were high, and the influencing factors were few, while the correlation between VRD and sex was high, which could be caused by non-uniform units and dimensions of volume and long/short diameter. The above indexes have important reference value for the automatic identification of cochlear structural abnormalities in clinic, and in-depth data mining of case samples will be carried out in subsequent studies. 5. Limitations Compared with domestic and foreign research results there are still limitations in this study. Considering the complexity of the cochlear structure and volume size, the measure of the cochlear duct length is calculated by the size of the cochlear base, and the direct measure is more difficult and less accurate. In the future, the algorithm will be improved to enhance the possibility of direct measure. The dependence of measurement results on automatic segmentation accuracy will be considered. It will also optimize the automatic segmentation of the minute structure of the cochlea duct, and build a more scientific and comprehensive evaluation index system of cochlear structure. 6. Conclusions In this study, an automatic segmentation and measurement model of cochlear structure was constructed based on U-HRCT image collection of clinical samples. The research results showed that the cochlear structure was segmented completely, the edges were smooth, and the details were basically visible, which met the needs of clinical diagnosis. The difference between the automatic measurement results and the manual measurement results was not obvious, and the measurement accuracy was good. In terms of the clinical application of the model, cochlear structure is closely related to patients’ gender, but it is less affected by age and location in the adult. Statistical analysis of cochlear structure parameters based on gender differences provides important references for clinical diagnosis of cochlea. Abbreviations U-HRCT Ultra-high resolution CT BLD Basilar length diameter BSD Basilar short diameter HGT height CDL Cochlear duct length VOL volume 2DL Two-turn length BTL Basilar turn length CCR Cochlear Curvature Ratio HRD Height Regularity Degree VRD Volume Regularity Degree Declarations Funding and a cknowledgements This study has received funding by the National Natural Science Foundation of China (Grant Nos. 62371316, 82302282, 62276012), Beijing Science and Technology Plan Project (Grant No. Z241100009024020), Beijing Scholar 2015 (Grant No. [2015]160), and Capital’s Funds for Health Improvement and Research (Grant No. 2022-1-1111). Guarantor The scientific guarantor of this publication is Hongxia Li. Conflict of Interest: The authors of this manuscript declare no relationships with any companies, whose products or services may be related to the subject matter of the article. Statistics and Biometry: One of the authors has significant statistical expertise. Informed Consent: Written informed consent was waived by the Institutional Review Board. Ethical Approval: Institutional Review Board approval was obtained. Methodology Retrospective Experimental multicenter study Author Contribution Dewu Yang and Mengshi Zhang wrote the main manuscript text, Yili Feng, Shuguang Han, and Ruowei Tang were responsible for data collection and analysis, Li Zhuo and Ye Zhang prepared figures, Pengfei Zhao, Xiaoguang Li, Zhenchang Wang and Hongxia Yin were responsible for research design and guidance. All authors reviewed the manuscript. References Defourny J, Lallemend F, Malgrange B. Structure and development of cochlear afferent innervation in mammals. Am J Physiol Cell Physiol. 2011 Oct;301(4):C750-61. doi: 10.1152/ajpcell.00516.2010. Epub 2011 Jul 13. PMID: 21753183. Lu S, Wei X, Chen B, Chen J, Zhang L, Yang M, Sun Z, Shi Y, Kong Y, Liu S, Li Y. A new phenomenon of cochlear otosclerosis: an acquired or congenital disease? - A clinical report of cochlear otosclerosis. Acta Otolaryngol. 2021 Jun;141(6):551-556. doi: 10.1080/00016489.2021.1906947. Epub 2021 Apr 5. PMID: 33819124. Geerardyn A, Zhu M, Klabbers T, Huinck W, Mylanus E, Nadol JB Jr, Verhaert N, Quesnel AM. Human Histology after Structure Preservation Cochlear Implantation via Round Window Insertion. Laryngoscope. 2024 Feb;134(2):945-953. doi: 10.1002/lary.30900. Epub 2023 Jul 26. PMID: 37493203. Heutink F, Koch V, Verbist B, van der Woude WJ, Mylanus E, Huinck W, Sechopoulos I, Caballo M. Multi-Scale deep learning framework for cochlea localization, segmentation and analysis on clinical ultra-high-resolution CT images. Comput Methods Programs Biomed. 2020 Jul;191:105387. doi: 10.1016/j.cmpb.2020.105387. Epub 2020 Feb 15. PMID: 32109685. Li Z, Zhou L, Tan S, Tang A. Application of UNETR for automatic cochlear segmentation in temporal bone CTs. Auris Nasus Larynx. 2023 Apr;50(2):212-217. doi: 10.1016/j.anl.2022.06.008. Epub 2022 Aug 13. PMID: 35970625. YI LV, JIA KE, YING XU, et al. Automatic segmentation of temporal bone structures from clinical conventional CT using a CNN approach[J]. The international journal of medical robotics + computer assisted surgery :,2021,17(2):e2229. DOI:10.1002/rcs.2229. Al-Dhamari I, Helal R, Abdelaziz T, Waldeck S, Paulus D. Automatic cochlear multimodal 3D image segmentation and analysis using atlas-model-based method. Cochlear Implants Int. 2024 Jan;25(1):46-58. doi: 10.1080/14670100.2023.2274199. Epub 2023 Nov 3. PMID: 37922404. Khurayzi T, Almuhawas F, Sanosi A. Direct measurement of cochlear parameters for automatic calculation of the cochlear duct length. Ann Saudi Med. 2020 May-Jun;40(3):212-218. doi: 10.5144/0256-4947.2020.218. Epub 2020 Jun 4. PMID: 32493102; PMCID: PMC7270618. Li L, Ma H. RDCTrans U-Net: A Hybrid Variable Architecture for Liver CT Image Segmentation. Sensors (Basel). 2022 Mar 23;22(7):2452. doi: 10.3390/s22072452. PMID: 35408067; PMCID: PMC9003011. Wang H, Chi J, Wu C, Yu X, Wu H. Degradation Adaption Local-to-Global Transformer for Low-Dose CT Image Denoising. J Digit Imaging. 2023 Aug;36(4):1894-1909. doi: 10.1007/s10278-023-00831-y. Epub 2023 Apr 28. PMID: 37118101; PMCID: PMC10407009. Yang B, Chang Y, Liang Y, Wang Z, Pei X, Xu XG, Qiu J. A Comparison Study Between CNN-Based Deformed Planning CT and CycleGAN-Based Synthetic CT Methods for Improving iCBCT Image Quality. Front Oncol. 2022 May 30;12:896795. doi: 10.3389/fonc.2022.896795. Erratum in: Front Oncol. 2023 Mar 07;13:1159770. doi: 10.3389/fonc.2023.1159770. PMID: 35707352; PMCID: PMC9189355. Huang L, Jiang H, Li S, Bai Z, Zhang J. Two stage residual CNN for texture denoising and structure enhancement on low dose CT image. Comput Methods Programs Biomed. 2020 Feb;184:105115. doi: 10.1016/j.cmpb.2019.105115. Epub 2019 Oct 8. PMID: 31627148. Xiaoguang Li, Ziyao Zhu, Hongxia Yin, Zhenchang Wang, Li Zhuo, and Yichao Zhou, Labyrinth net: a robust segmentation method for inner ear labyrinth in CT images. Computers in Biology and Medicine, 2022, vol. 146, 105630, https://doi.org/10.1016/j.compbiomed.2022.105630. Yating Qin. U-HRCT ORIENTED LABYRINTH SUB-STRUCTURE SEGMENTATION AND MALFORMATION DETECTION. Master's degree thesis of Beijing University of Technology (China).2024. Cohen AD, Yang B, Fernandez B, Banerjee S, Wang Y. Improved resting state functional connectivity sensitivity and reproducibility using a multiband multi-echo acquisition. Neuroimage. 2021 Jan 15;225:117461. doi: 10.1016/j.neuroimage.2020.117461. Epub 2020 Oct 16. PMID: 33069864; PMCID: PMC10015256. Presley BM, Sklar JC, Hazelwood SJ, Berg-Johansen B, Klisch SM. Balance Assessment Using a Smartwatch Inertial Measurement Unit with Principal Component Analysis for Anatomical Calibration. Sensors (Basel). 2023 May 9;23(10):4585. doi: 10.3390/s23104585. PMID: 37430500; PMCID: PMC10222509. Anthony EC, Kam OK, Klisch SM, Hazelwood SJ, Berg-Johansen B. Balance Assessment Using a Handheld Smartphone with Principal Component Analysis for Anatomical Calibration. Sensors (Basel). 2024 Aug 23;24(17):5467. doi: 10.3390/s24175467. PMID: 39275378; PMCID: PMC11397924. Alexiades G, Dhanasingh A, Jolly C. Method to estimate the complete and two-turn cochlear duct length. Otol Neurotol. 2015 Jun;36(5):904-7. doi: 10.1097/MAO.0000000000000620. PMID: 25299827. Schendzielorz P, Ilgen L, Mueller-Graff T, Noyalet L, Völker J, Taeger J, Hagen R, Neun T, Zabler S, Althoff D, Rak K. Precise Evaluation of the Cochlear Duct Length by Flat-panel Volume Computed Tomography (fpVCT)-Implication of Secondary Reconstructions. Otol Neurotol. 2021 Mar 1;42(3):e294-e303. doi: 10.1097/MAO.0000000000002951. PMID: 33555750. Takahashi M, Arai Y, Sakuma N, Yabuki K, Sano D, Nishimura G, Oridate N, Usami SI. Cochlear volume as a predictive factor for residual-hearing preservation after conventional cochlear implantation. Acta Otolaryngol. 2018 Apr;138(4):345-350. doi: 10.1080/00016489.2017.1393840. Epub 2017 Nov 17. PMID: 29148288. Dhanasingh A, Swords C, Bance M, Van Rompaey V, Van de Heyning P. Cochlear Size Assessment Predicts Scala Tympani Volume and Electrode Insertion Force- Implications in Robotic Assisted Cochlear Implant Surgery. Front Surg. 2021 Sep 30;8:723897. doi: 10.3389/fsurg.2021.723897. Erratum in: Front Surg. 2021 Oct 27;8:789184. doi: 10.3389/fsurg.2021.789184. PMID: 34660676; PMCID: PMC8514755. Heutink F, Koch V, Verbist B, van der Woude WJ, Mylanus E, Huinck W, Sechopoulos I, Caballo M. Multi-Scale deep learning framework for cochlea localization, segmentation and analysis on clinical ultra-high-resolution CT images. Comput Methods Programs Biomed. 2020 Jul;191:105387. doi: 10.1016/j.cmpb.2020.105387. Epub 2020 Feb 15. PMID: 32109685. Geneci F, Uzuner MB, Bilecenoğlu B, Torun Bİ, Orhan K, Ocak M. Examination of inner ear structures: a micro-CT study. Acta Otolaryngol. 2022 Jan;142(1):1-5. doi: 10.1080/00016489.2021.2015078. Epub 2022 Jan 5. PMID: 34985378. Tang J, Tang X, Li Z, Liu Y, Tan S, Li H, Ke R, Wang Z, Gong L, Tang A. Anatomical Variations of the Human Cochlea Determined from Micro-CT and High-Resolution CT Imaging and Reconstruction. Anat Rec (Hoboken). 2018 Jun;301(6):1086-1095. doi: 10.1002/ar.23730. Epub 2018 Feb 21. PMID: 29160929. Hussain R, Frater A, Calixto R, Karoui C, Margeta J, Wang Z, Hoen M, Delingette H, Patou F, Raffaelli C, Vandersteen C, Guevara N. Anatomical Variations of the Human Cochlea Using an Image Analysis Tool. J Clin Med. 2023 Jan 8;12(2):509. doi: 10.3390/jcm12020509. PMID: 36675438; PMCID: PMC9867191. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-5972058","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":415829996,"identity":"d00a6c72-4ead-400b-8be0-5b6b3e11d16e","order_by":0,"name":"Dewu Yang","email":"","orcid":"","institution":"Beijing Health Vocational College","correspondingAuthor":false,"prefix":"","firstName":"Dewu","middleName":"","lastName":"Yang","suffix":""},{"id":415829998,"identity":"97d9d9e9-f994-457f-813c-741d97fd9911","order_by":1,"name":"Mengshi Zhang","email":"","orcid":"","institution":"Capital Medical University","correspondingAuthor":false,"prefix":"","firstName":"Mengshi","middleName":"","lastName":"Zhang","suffix":""},{"id":415829999,"identity":"aeaac7e7-5a96-43bf-9588-306e775b1f54","order_by":2,"name":"Yili Feng","email":"","orcid":"","institution":"Capital Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yili","middleName":"","lastName":"Feng","suffix":""},{"id":415830000,"identity":"bfe3d381-46bb-455a-b28c-bb05aaf88da6","order_by":3,"name":"Shuguang Han","email":"","orcid":"","institution":"Capital Medical University","correspondingAuthor":false,"prefix":"","firstName":"Shuguang","middleName":"","lastName":"Han","suffix":""},{"id":415830001,"identity":"a8464606-2958-45aa-88cc-aa069e5e9c0a","order_by":4,"name":"Ruowei Tang","email":"","orcid":"","institution":"Capital Medical University","correspondingAuthor":false,"prefix":"","firstName":"Ruowei","middleName":"","lastName":"Tang","suffix":""},{"id":415830003,"identity":"336bd57f-cff7-4af6-bfa8-361bfdd5e158","order_by":5,"name":"Li Zhuo","email":"","orcid":"","institution":"Beijing University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Li","middleName":"","lastName":"Zhuo","suffix":""},{"id":415830004,"identity":"a93d1a6f-d01a-4ff2-8fe6-5d6efcaf8320","order_by":6,"name":"Ye Zhang","email":"","orcid":"","institution":"Beijing Health Vocational College","correspondingAuthor":false,"prefix":"","firstName":"Ye","middleName":"","lastName":"Zhang","suffix":""},{"id":415830006,"identity":"92f3c630-1ddd-487b-9992-7908ff28c8b5","order_by":7,"name":"Pengfei Zhao","email":"","orcid":"","institution":"Capital Medical University","correspondingAuthor":false,"prefix":"","firstName":"Pengfei","middleName":"","lastName":"Zhao","suffix":""},{"id":415830007,"identity":"09e43239-d867-44e1-87e7-1d53668bd3e1","order_by":8,"name":"Xiaoguang Li","email":"","orcid":"","institution":"Beijing University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Xiaoguang","middleName":"","lastName":"Li","suffix":""},{"id":415830008,"identity":"91b4b66b-041c-4146-ac9e-b9d00128bc53","order_by":9,"name":"Zhenchang Wang","email":"","orcid":"","institution":"Capital Medical University","correspondingAuthor":false,"prefix":"","firstName":"Zhenchang","middleName":"","lastName":"Wang","suffix":""},{"id":415830009,"identity":"d0bcdadf-a9ec-45b7-9227-91a041fcddb5","order_by":10,"name":"Hongxia Yin","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAz0lEQVRIiWNgGAWjYBACfvbmAwYfKv7XN7Y3EKlFsudYQuGMM8yMzT0HiNRicCPH4DNvCzNj+4wEYl125oDhZt4GNmbemY833mCosYkmqIOxvSHZcO4OHjbJ2WnFFgzH0nIbCGlh5jlwzODtGQkew9k5ZhKMDYcJa2GTSGz/wdtmIGF/8wyRWngkkhkMedsSDBhn8BCpRYLnGIPhjDMHEhh7gH5JIMYv9sf7PwCjEqil/fDGGx9qbAhrQQYGEgmkKIdoIVXHKBgFo2AUjAwAAPRlRQJZpY7kAAAAAElFTkSuQmCC","orcid":"","institution":"Capital Medical University","correspondingAuthor":true,"prefix":"","firstName":"Hongxia","middleName":"","lastName":"Yin","suffix":""}],"badges":[],"createdAt":"2025-02-06 09:38:17","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5972058/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5972058/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":76687034,"identity":"b9284f01-83b3-450f-a29c-bcb5d3a0f5f9","added_by":"auto","created_at":"2025-02-19 16:21:05","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":223840,"visible":true,"origin":"","legend":"\u003cp\u003eU-HRCT images of cochlea at different standard planes\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-5972058/v1/ab0f4277b38bd74aac1cff1b.png"},{"id":76687880,"identity":"6f601593-7013-4101-8561-c3e13e4b6822","added_by":"auto","created_at":"2025-02-19 16:29:05","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":414384,"visible":true,"origin":"","legend":"\u003cp\u003eThe network architecture of TransUnet model\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-5972058/v1/6a4b37afb0a9fe2ace82e47a.png"},{"id":76687878,"identity":"63afb176-202e-4d05-ba87-92fdb45fecc9","added_by":"auto","created_at":"2025-02-19 16:29:05","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":447502,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSchematic diagram of cochlear automatic measurement process\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-5972058/v1/0e103c99339c18a6fad87504.png"},{"id":76687040,"identity":"248baf1b-5bbe-44f5-9803-848b30ddd76d","added_by":"auto","created_at":"2025-02-19 16:21:05","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":246418,"visible":true,"origin":"","legend":"\u003cp\u003eSchematic diagram of manual measurement of cochlea\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-5972058/v1/7b6a01d7b305d5d5f35eb51b.png"},{"id":76687051,"identity":"70b2d85d-a9ea-460e-9a97-cc7c3fa0a4c9","added_by":"auto","created_at":"2025-02-19 16:21:06","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":128570,"visible":true,"origin":"","legend":"\u003cp\u003eViolin Plot of cochlear structural parameters\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-5972058/v1/68945b47630b100411976b1d.png"},{"id":79815404,"identity":"d6cf709c-7a95-4041-90ee-5620936a3246","added_by":"auto","created_at":"2025-04-03 07:38:23","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2337528,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5972058/v1/5fc4f6b8-c332-4bae-ba5a-b048c126c01c.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Construction and clinical application of the cochlear automatic segmentation and measurement model based on U-HRCT","fulltext":[{"header":"Key Points","content":"\u003cul\u003e\n \u003cli\u003eManual measurement of cochlear CT image is mostly used, and the accuracy and efficiency of manual measurement are poor.\u003c/li\u003e\n \u003cli\u003eAutomatic segmentation of cochlea reached good level, and the automatic measurement results were in good agreement with the manual measurement results.\u003c/li\u003e\n \u003cli\u003eThe reference range of cochlear structural parameters of different genders can provide possibilities for improving the accuracy of disease diagnosis and treatment and reducing the risk of surgery.\u003c/li\u003e\n\u003c/ul\u003e"},{"header":"1. Introduction","content":"\u003cp\u003eThe cochlea is an important part of the bone labyrinth of the internal ear, which contains important structures responsible for converting acoustic vibrations into neural signals. The overall structure of the cochlea is a spiral bone duct, which is formed by the cochlear duct coiled 2.5–2.75 cycles around the cochlear axis[1]. When the cochlear structure is abnormal due to congenital developmental defects or acquired injuries, which can lead to hearing loss or even deafness, cochlear implantation can restore partial hearing for patients, and is suitable for patients with extremely severe hearing loss or total deafness[2]. Therefore, the measurement of cochlear structure can help diagnose the deformity of the internal ear, relieve the pressure of clinical diagnosis, provide a basis for later surgical treatment, and provide design references for cochlear implant manufacturers.\u003c/p\u003e\n\u003cp\u003eHigh resolution CT images can show the structure of the cochlea, which is the first choice in the diagnosis of inner ear diseases. However, its ability to detect the spatial edge of the fine structure of the inner layer is limited, and the accuracy of some test data is poor. With the advent of ultra-high resolution CT (U-HRCT) dedicated to the ear department, the spatial resolution of the internal ear structure can reach 50 ~ 100µm, which provides high-quality CT images (as shown in Fig. 1) for the identification of the cochlea structure before surgery, and can clearly display the minute structure and abnormal lesions of the cochlea, which is an important data basis for structural segmentation. It has become a key technical means for cochlear imaging and evaluation before cochlear implantation [3].\u003c/p\u003e\n\u003cp\u003eIn the domestic and foreign researches on CT image segmentation of cochlear, in addition to the traditional threshold classification, regional growth and atlas construction methods, the neural network machine learning method is mostly used to automatically identify the cochlear structure. Floris Heutink et al. developed and validated the automatic cochlear segmentation and measurement system based on deep learning. The results of automatic segmentation show that the Dice coefficient is 0.90 ± 0.03[4]. Zhenhua Li et al. investigate the feasibility of a deep learning method based on a UNETR model for fully automatic segmentation of the cochlea in temporal bone CT images, and the Dice coefficient of the normal cochlear test set was 0.92[5]. Yi Lv et al. proposed a lightweight three-dimensional convolutional neural network (W-Net) for segmenting multiple targets including cochlea, ossicular chain, and facial nerve from conventional CT images of temporal bone and achieves human-level accuracy[6]. However, the segmentation accuracy needs to rely on the accuracy of data annotation and image quality, so the generalization ability and robustness of the model are poor. By extracting the cochlear Central Line from the cochlear segmentation results, the length and curvature of the cochlear Central Line, the volume and area of the cochlea and other quantitative indexes were calculated, and the physiological function of the cochlea was further evaluated. Ibraheem Al-Dhamari et al. proposed an automated fast cochlear segmentation, length, and volume estimation method from clinical 3D multimodal images, and evaluated using 3D landmarks located by two experts[7]. Tawfiq Khurayzi et al. developed ontological planning software (OTOPLAN) allows surgeons to directly measure parameters such as the diameter and width of the cochlea and then automatically calculate the CDL[8]. However, the selection of cochlear measurement indexes has not yet formed a normative standard, the results produced by different measurement methods are somewhat different, and there is less research experience in automatic measurement. Intelligent measurement and anomaly judgment put forward higher requirements on the data capacity and segmentation accuracy of training samples.\u003c/p\u003e\n\u003cp\u003eIn this study, based on the automatic segmentation research of deep learning algorithm, the cochlear automatic segmentation model was built through TransUnet neural network algorithm, and automatic fitting measurement was used to realize the automatic measurement of cochlear structural parameters. The problems of inaccuracy of traditional manual measurement and the limitation of two-dimensional image measurement were solved. Automatic segmentation and measurement models provide important diagnostic information for personalized treatment of cochlear surgery, reducing surgical risk and improving surgical success.\u003c/p\u003e"},{"header":"2. Methods","content":"\u003cp\u003e\u003cstrong\u003e2.1 Automatic segmentation model of cochlear based on TransUnet neural network\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe TransUnet neural network model combines the advantages of U-net and Transformer to become an important improvement scheme for medical image segmentation [9-10]. The TransUnet model is a U-shaped Encoder-Decoder structure, and the network architecture is shown in Figure 2. Among them, Encoder input the original image into the self-convolutional neural network (CNN) for feature extraction, and after linear projection, the toenized image is serialized and position encoded, while Decoder up-samples the encoded features and connects them through the resolution feature map of the encoder CNN. The combination achieves accurate positioning and segmentation [11-12].\u003c/p\u003e\n\u003cp\u003eBased on the overall segmentation of the labyrinth of the internal ear [13], this study made full use of the prior information of anatomical structure as guidance, designed the pseudo-label proxy task, combined with the correlation between the overall and substructure segmentation, designed the teacher-student framework, and finally realized the segmentation of the substructure of the internal ear, including the three semicircular canals, vestibule and cochlea [14]. The DSC index of cochlea is 98%, which provides important support for the quantitative analysis of cochlear structure.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.2 Automatic cochlear measurement based on precise segmentation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(1)Measuring indexes. Based on the actual needs of clinical surgical evaluation and related research progress, the cochlear measurement indexes were defined, including direct and indirect measurement indexes.\u003c/p\u003e\n\u003cp\u003eA. Direct indexes include:\u0026nbsp;① Basal long diameter (BLD), defined as the longest distance between the boundary points of the projection of the cochlear base; ② Basal short diameter (BSD), defined as the distance from the midpoint of BLD to the boundary of the projection of the cochlear base, which is on the vertical line passing through the midpoint of BLD; ③ Cochlear height (HGT), defined as the vertical distance between the cupula cochlea and the base of the cochlea; ④ Cochlear duct length (CDL), defined as the length of the fitted centerline of the cochlear duct; ⑤ Cochlear volume (VOL), defined as the total number of pixel points enclosed by the cochlear shell; ⑥ Two-turn length (2TL), defined as the length from the entrance of the cochlea to the second turn (i.e., the cochlear axis); ⑦ Basal turn length (BTL), defined as the length from the entrance of the cochlea to the basal turn (the innermost layer of the cochlea).\u003c/p\u003e\n\u003cp\u003eB. Indirect indexes include:\u0026nbsp;① Cochlear Curvature Ratio (CCR), defined as the ratio of the cochlear duct length to the mean of the long and short diameters; ② Height Regularity Degree (HRD), defined as the rate of the cochlear height to the mean of the long and short diameters; ③ Volume Regularity Degree (VRD), defined as the rate of the cochlear volume tothe mean of the long and short diameters.\u003c/p\u003e\n\u003cp\u003e(2) Measurement Method. The measurement of cochlear structural parameters is achieved by designing computerized automatic measurement algorithms for different indexes based on the precise segmentation of the cochlear structure. The automatic measurement of cochlear structure data mainly includes processes such as detecting cochlear voxel connected domain , calibrating spatial position, and measuring key data, etc. as shown in Figure 3.\u003c/p\u003e\n\u003cp\u003eA. Denoising of segmentation results based on connected region analysis: The connected domains composed of adjacent pixel coordinates with the same segmentation category in the segmentation results are identified and marked, and the largest array of cochlear voxel group in the connected domains is retrieved through pixel traversal and label distribution. The other smaller noise points are removed[15].\u003c/p\u003e\n\u003cp\u003eB. Posture calibration of cochlea based on principal component analysis: Posture calibration of cochlea structural adopts the principal component analysis to analyze the coordinates of point cloud data and find the three perpendicular principal vectors with the largest spatial distribution. As the new coordinates of calibration, it rotate the target to realize the calibration of the spatial position of the cochlea [16-17].\u003c/p\u003e\n\u003cp\u003eC. Measurements of key structural data include five direct indexes: BLD, BSD, HGT, CDL, and VOL. Among these, the length of the cochlear duct is estimated using the formula (CDL = 4.16BLD - 4mm). The lengths from the entrance of the cochlea to the base turn and the second turn of the cochlear duct are set as the base-turn length (BTL) and two-turn length (2TL) , respectively. The estimation formulas are 2TL = 3.65 \u0026times; (BLD - 1mm) and BTL = 2.43 \u0026times; (BLD - 1mm)[18-19].\u003c/p\u003e\n\u003ch3\u003e2. Materials\u003c/h3\u003e\n\u003cp\u003e\u003cstrong\u003e2.1 Basic data\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA total of 86 patients (172 ears) underwent U-HRCT scanning in the Department of Radiology, Beijing Friendship Hospital Affiliated to Capital Medical University from July 2021 to September 2023. This study was reviewed by the Hospital Ethics Committee (IRB: 2024-P2-241) and approved.\u003c/p\u003e\n\u003cp\u003e2.2 Image acquisition\u003c/p\u003e\n\u003cp\u003eAll patients underwent U-HRCT examination using the U-HRCT scanner (Ultra3D, LargeV, Beijing) with a voltage of 100-110 kVp, a current of 120-180 mAs, and a view field of 65 mm. The thickness and spacing of slice are both set to 0.1 mm. Isotropic axial images are obtained after scanning for 20s.\u003c/p\u003e\n\u003cp\u003e2.3 Data Analysis\u003c/p\u003e\n\u003cp\u003eSPSS25.0 was used for statistical analysis of basic data, segmentation effect, and measured data. Metrical data with normal distribution were expressed as\u0026nbsp;𝑥̅\u0026plusmn;𝑠, and paired sample t test was used for comparison of cases; when metrical data did not meet normal distribution, Wilcoxon rank sum test was used for comparison of cases in the same group. The difference was statistically significant with P \u0026lt; 0.05.\u003c/p\u003e\n\u003cp\u003e(1) Analyse basic data. The basic data of 86 patients with normal cochlea were collected and demographic analysis was conducted on data of 172 ears in terms of age, gender, left/right ear, etc.\u003c/p\u003e\n\u003cp\u003e(2) Analyse automatic segmentation effect. An evaluation team composed of 3 staff members with 5 years or more related work experience evaluated the effect of cochlear auto segmentation by subjective evaluation. The evaluation dimensions included segmentation accuracy, image clarity and diagnostic applicability. The scoring criteria are shown in Table 1.\u003c/p\u003e\n\u003cp\u003eTable 1 Subjective evaluation criteria for the effect of cochlear auto-segmentation\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"551\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eScoring Criteria\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003esegmentation accuracy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eimage clarity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003ediagnostic applicability\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 103px;\"\u003e\n \u003cp\u003eVery good\u003c/p\u003e\n \u003cp\u003e(5points)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 173px;\"\u003e\n \u003cp\u003eComplete structure, accurate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 137px;\"\u003e\n \u003cp\u003eSmooth edges, clearly visible\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 137px;\"\u003e\n \u003cp\u003eFully meet diagnostic requirements\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 103px;\"\u003e\n \u003cp\u003eGood(4points)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 173px;\"\u003e\n \u003cp\u003eComplete structure, slight deviation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 137px;\"\u003e\n \u003cp\u003eSmooth edges, visible\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 137px;\"\u003e\n \u003cp\u003eBasically meet\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 103px;\"\u003e\n \u003cp\u003eAverage(3points)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 173px;\"\u003e\n \u003cp\u003eBasically complete Complete structure, obvious deviation\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 137px;\"\u003e\n \u003cp\u003eRough edges, visible\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 137px;\"\u003e\n \u003cp\u003eNot meet \u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 103px;\"\u003e\n \u003cp\u003ePoor(2points)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 173px;\"\u003e\n \u003cp\u003eBasically complete Complete structure, obvious distortion\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 137px;\"\u003e\n \u003cp\u003eRough edges, vaguely visible\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 137px;\"\u003e\n \u003cp\u003eMay lead to missed diagnosis\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 103px;\"\u003e\n \u003cp\u003eVery poor\u003c/p\u003e\n \u003cp\u003e(1point)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 173px;\"\u003e\n \u003cp\u003eIncomplete\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 137px;\"\u003e\n \u003cp\u003eInvisible\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 137px;\"\u003e\n \u003cp\u003eMay lead to misdiagnosis\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e(3) Measurement data analysis. Firstly, the accuracy of measurement data is analyzed. One observer used Mimics software (19.0.0.347, Materialise, Belgium) to manually measure BLD, BSD, HGT, CDL, 2TL, BTL, as shown in Figure 4. After repeated measure for 3 times, the mean value was taken to compare the difference between manual and automatic measurement data. Secondly, by comparing the difference of indexdata among different age, gender or location, multiple linear regression was used to analyze the influencing factors of cochlear structure. Finally, the clinical value of the measurement data was analyzed and the Violin Plot method was used to calculate the normal value range.\u003c/p\u003e"},{"header":"3. Results","content":"\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003e3.1 Analysis results of basic data\u003c/h2\u003e \u003cp\u003eA total of 86 patients were included in this study, including 50 males and 36 females, aged from 19 to 80 years old, with an average age of (38.36\u0026thinsp;\u0026plusmn;\u0026thinsp;16.56) years old, with 86 left ears and 86 right ears, totaling 172, as shown in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\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\u003eDemographic information\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"2\"\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 \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCharacteristic\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eValue\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNumber of ears\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e172\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMean age (mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e38.36\u0026thinsp;\u0026plusmn;\u0026thinsp;16.56\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSex (n, %)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\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\u003e100(58.14)\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\u003e72(41.86)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLaterality (n, %)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRight\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e86(50)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLeft\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e86(50)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e3.2 Analysis results of cochlear auto-segmentation effect\u003c/h2\u003e \u003cp\u003eCompared with the original two-dimensional CT cochlear image and the automatic segmentation 3D image, MeshLeb software was used to perform the subjective evaluation of the segmentation effect, and the segmentation accuracy, image clarity and diagnostic applicability were analyzed. The evaluation results were (4.38\u0026thinsp;\u0026plusmn;\u0026thinsp;0.75), (4.49\u0026thinsp;\u0026plusmn;\u0026thinsp;0.66) and (4.58\u0026thinsp;\u0026plusmn;\u0026thinsp;0.67) points, all of which reached a good level. The total number of poor and very poor were 4.75% (8/168), 1.78%(3/168) and 2.98%(5/168), respectively, which were controlled at a low level.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003e3.3 Analysis results of automatic measurement data\u003c/h2\u003e \u003cp\u003e(1) Accuracy of measurement data. The direct indexes of cochlear structure, namely, BLD, BSD, HGT, CDL, 2TL and BTL, and the differences among different measurement methods were not statistically significant (P\u0026thinsp;\u0026gt;\u0026thinsp;0.05), as shown in Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e.\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\u003eAnalysis results of the accuracy of automatic measurement data\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=\"char\" char=\"\u0026plusmn;\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\"\u0026plusmn;\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMeasurement data\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003emanual measurement\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAutomatic measurement\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003estatistic\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eP\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBLD(mm)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e \u003cp\u003e9.55\u0026thinsp;\u0026plusmn;\u0026thinsp;0.44\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e \u003cp\u003e9.60\u0026thinsp;\u0026plusmn;\u0026thinsp;0.549\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.713\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.090\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBSD(mm)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e \u003cp\u003e6.78\u0026thinsp;\u0026plusmn;\u0026thinsp;0.37\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e \u003cp\u003e6.82\u0026thinsp;\u0026plusmn;\u0026thinsp;0.32\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.912\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.059\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHGT(mm)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e \u003cp\u003e4.28\u0026thinsp;\u0026plusmn;\u0026thinsp;0.38\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e \u003cp\u003e4.24\u0026thinsp;\u0026plusmn;\u0026thinsp;0.31\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.515\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.133\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCDL(mm)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e \u003cp\u003e35.97\u0026thinsp;\u0026plusmn;\u0026thinsp;2.09\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e \u003cp\u003e35.94\u0026thinsp;\u0026plusmn;\u0026thinsp;2.28\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.195\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.846\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2TL(mm)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e \u003cp\u003e31.52\u0026thinsp;\u0026plusmn;\u0026thinsp;1.77\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e \u003cp\u003e31.40\u0026thinsp;\u0026plusmn;\u0026thinsp;2.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.202\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.232\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBTL(mm)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e \u003cp\u003e20.75\u0026thinsp;\u0026plusmn;\u0026thinsp;1.90\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e \u003cp\u003e20.90\u0026thinsp;\u0026plusmn;\u0026thinsp;1.33\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.579\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.117\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(2) Influence factors of cochlear structure.\u003c/p\u003e \u003cp\u003eMultiple linear regression was used to analyze the influence degree of factors such as age, gender and location on the cochlear structure, and the results were shown in Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e. The direct measurement indexes of cochlear structure had very high correlation with gender (M/F), and the positive correlation coefficient were significant (P\u0026thinsp;\u0026lt;\u0026thinsp;0.01). The indirect measurement indexes of CCR and HRD had poor correlation with gender, while the VRD had a high correlation with gender. Except for cochlear volume and volume regularity, other parameters had low correlation with age, and the negative correlation coefficient was not statistically significant(P\u0026thinsp;\u0026gt;\u0026thinsp;0.05). The direct indexes and the indirect indexes of CCR and HRD were not significantly correlated with the left/right ear (R/L) (P\u0026thinsp;\u0026gt;\u0026thinsp;0.05), while the VRD was significantly correlated with the left/right ear (R/L) (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05). The constant terms of multiple linear regression were statistically significant (P\u0026thinsp;\u0026lt;\u0026thinsp;0.01).\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\u003eRegression coefficients of influencing factors of cochlear structure\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=\"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=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMeasurement Data\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAge(Y)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGender(M/F)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eLocation(R/L)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eConstant term(d)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBLD(mm)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-0.004\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.332\u003csup\u003e**\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.162\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e8.975\u003csup\u003e**\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBSD(mm)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-0.002\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.298\u003csup\u003e**\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.056\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e6.367\u003csup\u003e**\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHGT(mm)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.184\u003csup\u003e**\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-0.040\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e4.090\u003csup\u003e**\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCDL(mm)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-0.015\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.381\u003csup\u003e**\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.673\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e33.335\u003csup\u003e**\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVOL(mm)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-0.099\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e7.955\u003csup\u003e**\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-0.625\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e69.323\u003csup\u003e**\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2TL(mm)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-0.013\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.212\u003csup\u003e**\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.589\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e29.108\u003csup\u003e**\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBTL(mm)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-0.009\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.806\u003csup\u003e**\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.393\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e19.380\u003csup\u003e**\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCCR\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-0.002\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.017\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e4.350\u003csup\u003e**\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHRD\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.003\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-0.010\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.532\u003csup\u003e**\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVRD(mm\u003csup\u003e2\u003c/sup\u003e)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-0.009\u003csup\u003e**\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.603\u003csup\u003e**\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-0.195\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e9.048\u003csup\u003e**\u003c/sup\u003e\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* mean statistically significant difference (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05); ** mean statistically significant difference (P\u0026thinsp;\u0026lt;\u0026thinsp;0.01).\u003c/p\u003e \u003cp\u003e(3) Reference value of cochlear structure\u003c/p\u003e \u003cp\u003eAccording to the results of multiple linear regression analysis of the influencing factors of cochlear structure, the reference values of cochlear structure parameters were calculated according to gender, that is, the Violin Plot method was used to analyze the distribution of each parameter data set. As shown in Fig.\u0026nbsp;5. The values in male group were as follows: BLD=[8.30,11.26], BSD=[6.21,7.74], HGT=[3.57,5.08], CDL=[30.51,42.83], VOL=[52.16,107.56], 2TL=[26.63,37.43], BTL=[17.72,24.92], CCR=[4.12,4.65], HRD=[0.42,0.62], VRD=[7.75,11.51]; The values in female group were : BLD=[8.21,10.72], BSD=[5.95,7.38], HGT=[3.57,4.70], CDL=[30.15,40.59], VOL=[56.53,89.21], 2TL=[26.31,35.47], BTL=[17.51,23.63], CCR=[4.09,4.66], HRD=[0.42,0.61], VRD=[7.68,10.46].\u003c/p\u003e \u003c/div\u003e"},{"header":"4 Discussion","content":"\u003cp\u003eBased on the data acquisition and analysis of ear U-HRCT images, this study proposed the cochlear automatic segmentation and measurement algorithm, which can be used to analyze clinical cases and assist precision diagnosis and treatment, and effectively solve the problem that traditional manual measurement methods are difficult to promote. We took a relatively large sample of Chinese normal adults as the research objects, and verified the accuracy of automatic segmentation and measurement algorithm by comparing clinical evaluation and manual measurement. The results showed that all automatic segmentation of cochlea reached good level, and the automatic measurement results were in good agreement with the manual measurement results. We proposed the reference range of cochlear structural parameters of different genders by using the quartile method to provide individual guidance for the diagnosis of clinical cochlear malformations and the formulation of preoperative protocols, which can provide possibilities for improving the accuracy of disease diagnosis and treatment and reducing the risk of surgery.\u003c/p\u003e \u003cp\u003eCompared with domestic and foreign research results, the reliability of cochlear automatic measurement data in this study is higher. Masahiro Takahashi et al. used MRI to measure cochlear volume and used CT to measure cochlear duct length[20].Anandhan Dhanasingh et al. measured the structural parameters of the cochlea base with micro-computed tomography (\u0026micro;CT) images, and investigated the correlation between the diameter (A value), the width (B value) or the straight portion of the cochlea basal turn (S value) with the scala tympani (ST)[21]. Floris Heutink et al. proposed a method to automatically segment and measure the human cochlea in clinical ultra-high-resolution (UHR) CT images, and investigate differences in cochlea size for personalized implant planning[22]. Ferhat Geneci et al. investigated the inner ear anatomy accurately in detail by micro-computed tomography (micro-CT) to contribute to the data related to the inner ear anatomy[23].There is little difference between relevant research results and the results of this study, which further confirms the accuracy of cochlear automatic segmentation and measurement model.\u003c/p\u003e \u003cp\u003eIn addition, Jie Tang et al. used multiple axial micro-computed tomography and high-resolution CT 3D digital reconstruction model of the skull to analyze the temporal bones of 15 cadavers[24]. Raabid Hussain et al. analyzed 1000\u0026thinsp;+\u0026thinsp;clinical temporal bone CT images using a web-based image analysis tool. Cochlear size and shape parameters were obtained to determine population statistics and perform regression and correlation analysis[25].The above researches provide important supplement for the accurate analysis of cochlear structure and adjacent relationship, and also provide references for the subsequent quantitative research of cochlea.\u003c/p\u003e \u003cp\u003eIn addition to conventional measurement indicators, this study innovatively proposed three measurement indexes representing the regularity of cochlear structure, including CCR, HRD and VRD, to analyze the degree of correlation between cochlear tube length, cochlear height and volume with the size of the cochlear base. The results showed that the correlation between CCR or HRD with age, sex or location was poor, indicating that the structural stability and similarity of normal adults were high, and the influencing factors were few, while the correlation between VRD and sex was high, which could be caused by non-uniform units and dimensions of volume and long/short diameter. The above indexes have important reference value for the automatic identification of cochlear structural abnormalities in clinic, and in-depth data mining of case samples will be carried out in subsequent studies.\u003c/p\u003e"},{"header":"5. Limitations","content":"\u003cp\u003eCompared with domestic and foreign research results there are still limitations in this study. Considering the complexity of the cochlear structure and volume size, the measure of the cochlear duct length is calculated by the size of the cochlear base, and the direct measure is more difficult and less accurate. In the future, the algorithm will be improved to enhance the possibility of direct measure. The dependence of measurement results on automatic segmentation accuracy will be considered. It will also optimize the automatic segmentation of the minute structure of the cochlea duct, and build a more scientific and comprehensive evaluation index system of cochlear structure.\u003c/p\u003e"},{"header":"6. Conclusions","content":"\u003cp\u003eIn this study, an automatic segmentation and measurement model of cochlear structure was constructed based on U-HRCT image collection of clinical samples. The research results showed that the cochlear structure was segmented completely, the edges were smooth, and the details were basically visible, which met the needs of clinical diagnosis. The difference between the automatic measurement results and the manual measurement results was not obvious, and the measurement accuracy was good. In terms of the clinical application of the model, cochlear structure is closely related to patients\u0026rsquo; gender, but it is less affected by age and location in the adult. Statistical analysis of cochlear structure parameters based on gender differences provides important references for clinical diagnosis of cochlea.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eU-HRCT Ultra-high resolution CT\u003c/p\u003e\n\u003cp\u003eBLD Basilar length diameter\u003c/p\u003e\n\u003cp\u003eBSD Basilar short diameter\u003c/p\u003e\n\u003cp\u003eHGT height\u003c/p\u003e\n\u003cp\u003eCDL Cochlear duct length\u003c/p\u003e\n\u003cp\u003eVOL volume\u003c/p\u003e\n\u003cp\u003e2DL Two-turn length\u003c/p\u003e\n\u003cp\u003eBTL Basilar turn length\u003c/p\u003e\n\u003cp\u003eCCR Cochlear Curvature Ratio\u003c/p\u003e\n\u003cp\u003eHRD Height Regularity Degree\u003c/p\u003e\n\u003cp\u003eVRD Volume Regularity Degree\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;and a\u003c/strong\u003e\u003cstrong\u003ecknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study has received funding by the National Natural Science Foundation of China (Grant Nos. 62371316, 82302282, 62276012), Beijing Science and Technology Plan Project (Grant No. Z241100009024020), Beijing Scholar 2015 (Grant No. [2015]160), and Capital’s Funds for Health Improvement and Research (Grant No. 2022-1-1111).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eGuarantor\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe scientific guarantor of this publication is Hongxia Li.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflict of Interest:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors of this manuscript declare no relationships with any companies, whose products or services may be related to the subject matter of the article.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStatistics and Biometry:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eOne of the authors has significant statistical expertise.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eInformed Consent:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWritten informed consent was waived by the Institutional Review Board.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthical Approval:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eInstitutional Review Board approval was obtained.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethodology\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003eRetrospective\u003c/li\u003e\n \u003cli\u003eExperimental\u003c/li\u003e\n \u003cli\u003emulticenter study\u003c/li\u003e\n\u003c/ul\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eDewu Yang and Mengshi Zhang wrote the main manuscript text, Yili Feng, Shuguang Han, and Ruowei Tang were responsible for data collection and analysis, Li Zhuo and Ye Zhang prepared figures, Pengfei Zhao, Xiaoguang Li, Zhenchang Wang and Hongxia Yin were responsible for research design and guidance. All authors reviewed the manuscript.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eDefourny J, Lallemend F, Malgrange B. Structure and development of cochlear afferent innervation in mammals. Am J Physiol Cell Physiol. 2011 Oct;301(4):C750-61. doi: 10.1152/ajpcell.00516.2010. Epub 2011 Jul 13. PMID: 21753183.\u003c/li\u003e\n\u003cli\u003eLu S, Wei X, Chen B, Chen J, Zhang L, Yang M, Sun Z, Shi Y, Kong Y, Liu S, Li Y. A new phenomenon of cochlear otosclerosis: an acquired or congenital disease? - A clinical report of cochlear otosclerosis. Acta Otolaryngol. 2021 Jun;141(6):551-556. doi: 10.1080/00016489.2021.1906947. Epub 2021 Apr 5. PMID: 33819124.\u003c/li\u003e\n\u003cli\u003eGeerardyn A, Zhu M, Klabbers T, Huinck W, Mylanus E, Nadol JB Jr, Verhaert N, Quesnel AM. Human Histology after Structure Preservation Cochlear Implantation via Round Window Insertion. Laryngoscope. 2024 Feb;134(2):945-953. doi: 10.1002/lary.30900. Epub 2023 Jul 26. PMID: 37493203.\u003c/li\u003e\n\u003cli\u003eHeutink F, Koch V, Verbist B, van der Woude WJ, Mylanus E, Huinck W, Sechopoulos I, Caballo M. Multi-Scale deep learning framework for cochlea localization, segmentation and analysis on clinical ultra-high-resolution CT images. Comput Methods Programs Biomed. 2020 Jul;191:105387. doi: 10.1016/j.cmpb.2020.105387. Epub 2020 Feb 15. PMID: 32109685.\u003c/li\u003e\n\u003cli\u003eLi Z, Zhou L, Tan S, Tang A. Application of UNETR for automatic cochlear segmentation in temporal bone CTs. Auris Nasus Larynx. 2023 Apr;50(2):212-217. doi: 10.1016/j.anl.2022.06.008. Epub 2022 Aug 13. PMID: 35970625.\u003c/li\u003e\n\u003cli\u003eYI LV, JIA KE, YING XU, et al. Automatic segmentation of temporal bone structures from clinical conventional CT using a CNN approach[J]. The international journal of medical robotics + computer assisted surgery :,2021,17(2):e2229. DOI:10.1002/rcs.2229.\u003c/li\u003e\n\u003cli\u003eAl-Dhamari I, Helal R, Abdelaziz T, Waldeck S, Paulus D. Automatic cochlear multimodal 3D image segmentation and analysis using atlas-model-based method. Cochlear Implants Int. 2024 Jan;25(1):46-58. doi: 10.1080/14670100.2023.2274199. Epub 2023 Nov 3. PMID: 37922404.\u003c/li\u003e\n\u003cli\u003eKhurayzi T, Almuhawas F, Sanosi A. Direct measurement of cochlear parameters for automatic calculation of the cochlear duct length. Ann Saudi Med. 2020 May-Jun;40(3):212-218. doi: 10.5144/0256-4947.2020.218. Epub 2020 Jun 4. PMID: 32493102; PMCID: PMC7270618.\u003c/li\u003e\n\u003cli\u003eLi L, Ma H. RDCTrans U-Net: A Hybrid Variable Architecture for Liver CT Image Segmentation. Sensors (Basel). 2022 Mar 23;22(7):2452. doi: 10.3390/s22072452. PMID: 35408067; PMCID: PMC9003011.\u003c/li\u003e\n\u003cli\u003eWang H, Chi J, Wu C, Yu X, Wu H. Degradation Adaption Local-to-Global Transformer for Low-Dose CT Image Denoising. J Digit Imaging. 2023 Aug;36(4):1894-1909. doi: 10.1007/s10278-023-00831-y. Epub 2023 Apr 28. PMID: 37118101; PMCID: PMC10407009.\u003c/li\u003e\n\u003cli\u003eYang B, Chang Y, Liang Y, Wang Z, Pei X, Xu XG, Qiu J. A Comparison Study Between CNN-Based Deformed Planning CT and CycleGAN-Based Synthetic CT Methods for Improving iCBCT Image Quality. Front Oncol. 2022 May 30;12:896795. doi: 10.3389/fonc.2022.896795. Erratum in: Front Oncol. 2023 Mar 07;13:1159770. doi: 10.3389/fonc.2023.1159770. PMID: 35707352; PMCID: PMC9189355.\u003c/li\u003e\n\u003cli\u003eHuang L, Jiang H, Li S, Bai Z, Zhang J. Two stage residual CNN for texture denoising and structure enhancement on low dose CT image. Comput Methods Programs Biomed. 2020 Feb;184:105115. doi: 10.1016/j.cmpb.2019.105115. Epub 2019 Oct 8. PMID: 31627148.\u003c/li\u003e\n\u003cli\u003eXiaoguang Li, Ziyao Zhu, Hongxia Yin, Zhenchang Wang, Li Zhuo, and Yichao Zhou, Labyrinth net: a robust segmentation method for inner ear labyrinth in CT images. Computers in Biology and Medicine, 2022, vol. 146, 105630, https://doi.org/10.1016/j.compbiomed.2022.105630.\u003c/li\u003e\n\u003cli\u003eYating Qin. U-HRCT ORIENTED LABYRINTH SUB-STRUCTURE SEGMENTATION AND MALFORMATION DETECTION. Master\u0026apos;s degree thesis of Beijing University of Technology (China).2024.\u003c/li\u003e\n\u003cli\u003eCohen AD, Yang B, Fernandez B, Banerjee S, Wang Y. Improved resting state functional connectivity sensitivity and reproducibility using a multiband multi-echo acquisition. Neuroimage. 2021 Jan 15;225:117461. doi: 10.1016/j.neuroimage.2020.117461. Epub 2020 Oct 16. PMID: 33069864; PMCID: PMC10015256.\u003c/li\u003e\n\u003cli\u003ePresley BM, Sklar JC, Hazelwood SJ, Berg-Johansen B, Klisch SM. Balance Assessment Using a Smartwatch Inertial Measurement Unit with Principal Component Analysis for Anatomical Calibration. Sensors (Basel). 2023 May 9;23(10):4585. doi: 10.3390/s23104585. PMID: 37430500; PMCID: PMC10222509.\u003c/li\u003e\n\u003cli\u003eAnthony EC, Kam OK, Klisch SM, Hazelwood SJ, Berg-Johansen B. Balance Assessment Using a Handheld Smartphone with Principal Component Analysis for Anatomical Calibration. Sensors (Basel). 2024 Aug 23;24(17):5467. doi: 10.3390/s24175467. PMID: 39275378; PMCID: PMC11397924.\u003c/li\u003e\n\u003cli\u003eAlexiades G, Dhanasingh A, Jolly C. Method to estimate the complete and two-turn cochlear duct length. Otol Neurotol. 2015 Jun;36(5):904-7. doi: 10.1097/MAO.0000000000000620. PMID: 25299827.\u003c/li\u003e\n\u003cli\u003eSchendzielorz P, Ilgen L, Mueller-Graff T, Noyalet L, V\u0026ouml;lker J, Taeger J, Hagen R, Neun T, Zabler S, Althoff D, Rak K. Precise Evaluation of the Cochlear Duct Length by Flat-panel Volume Computed Tomography (fpVCT)-Implication of Secondary Reconstructions. Otol Neurotol. 2021 Mar 1;42(3):e294-e303. doi: 10.1097/MAO.0000000000002951. PMID: 33555750.\u003c/li\u003e\n\u003cli\u003eTakahashi M, Arai Y, Sakuma N, Yabuki K, Sano D, Nishimura G, Oridate N, Usami SI. Cochlear volume as a predictive factor for residual-hearing preservation after conventional cochlear implantation. Acta Otolaryngol. 2018 Apr;138(4):345-350. doi: 10.1080/00016489.2017.1393840. Epub 2017 Nov 17. PMID: 29148288.\u003c/li\u003e\n\u003cli\u003eDhanasingh A, Swords C, Bance M, Van Rompaey V, Van de Heyning P. Cochlear Size Assessment Predicts Scala Tympani Volume and Electrode Insertion Force- Implications in Robotic Assisted Cochlear Implant Surgery. Front Surg. 2021 Sep 30;8:723897. doi: 10.3389/fsurg.2021.723897. Erratum in: Front Surg. 2021 Oct 27;8:789184. doi: 10.3389/fsurg.2021.789184. PMID: 34660676; PMCID: PMC8514755.\u003c/li\u003e\n\u003cli\u003eHeutink F, Koch V, Verbist B, van der Woude WJ, Mylanus E, Huinck W, Sechopoulos I, Caballo M. Multi-Scale deep learning framework for cochlea localization, segmentation and analysis on clinical ultra-high-resolution CT images. Comput Methods Programs Biomed. 2020 Jul;191:105387. doi: 10.1016/j.cmpb.2020.105387. Epub 2020 Feb 15. PMID: 32109685.\u003c/li\u003e\n\u003cli\u003eGeneci F, Uzuner MB, Bilecenoğlu B, Torun Bİ, Orhan K, Ocak M. Examination of inner ear structures: a micro-CT study. Acta Otolaryngol. 2022 Jan;142(1):1-5. doi: 10.1080/00016489.2021.2015078. Epub 2022 Jan 5. PMID: 34985378.\u003c/li\u003e\n\u003cli\u003eTang J, Tang X, Li Z, Liu Y, Tan S, Li H, Ke R, Wang Z, Gong L, Tang A. Anatomical Variations of the Human Cochlea Determined from Micro-CT and High-Resolution CT Imaging and Reconstruction. Anat Rec (Hoboken). 2018 Jun;301(6):1086-1095. doi: 10.1002/ar.23730. Epub 2018 Feb 21. PMID: 29160929.\u003c/li\u003e\n\u003cli\u003eHussain R, Frater A, Calixto R, Karoui C, Margeta J, Wang Z, Hoen M, Delingette H, Patou F, Raffaelli C, Vandersteen C, Guevara N. Anatomical Variations of the Human Cochlea Using an Image Analysis Tool. J Clin Med. 2023 Jan 8;12(2):509. doi: 10.3390/jcm12020509. PMID: 36675438; PMCID: PMC9867191.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"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":"Cochlea, Base of cochlea, Cochlear duct, Regularity, Intelligent segmentation","lastPublishedDoi":"10.21203/rs.3.rs-5972058/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5972058/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eObjectives\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn order to improve the accuracy and efficiency of data acquisition for cochlear structure, this study built an intelligent segmentation and automatic measurement model of cochlear based on U-HRCT to solve the difficulty of performing an accurate preoperative evaluation of the cochlea.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe data of 86 patients (172 ears in total) who received U-HRCT and were diagnosed with normal ears were retrospectively collected. In this study, TransUnet neural network was used to construct automatic segmentation model of cochlear, 7 direct measurement indexes were selected and 3 indirect measurement indexes were defined. The effect of automatic segmentation was evaluated by subjective method, compared with manual measured data, and the influencing factors of cochlear structure were analyzed by multiple linear regression method. Finally, the Violin Plot method is used to calculate the normal value range of cochlear structural parameters.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe index of automatic segmentation DSC reached 98%, and the results of automatic measurement were consistent with those of manual measurement (P \u0026gt; 0.05). Except for regularity parameters CCR and HRD, cochlear structural parameters had high correlation with gender (P \u0026lt; 0.05), and except volume related parameters VOL and VRD, cochlear structural parameters had low correlation with age (P \u0026gt; 0.05). Except VRD, the correlation between the cochlear structure parameter and the location was low (P \u0026gt; 0.05).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study built an automatic segmentation and measurement model of cochlear structure, obtained the reference value range of cochlear structure parameters, and provided reference data for clinical diagnosis and treatment.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eClinical relevance statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe above indexes have important reference value for the automatic identification of cochlear structural abnormalities in clinic, and in-depth data mining of case samples will be carried out in subsequent studies.\u003c/p\u003e","manuscriptTitle":"Construction and clinical application of the cochlear automatic segmentation and measurement model based on U-HRCT","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-02-19 16:21:00","doi":"10.21203/rs.3.rs-5972058/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":"a14bf813-9048-4980-8619-61746374112e","owner":[],"postedDate":"February 19th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-07-21T19:38:17+00:00","versionOfRecord":[],"versionCreatedAt":"2025-02-19 16:21:00","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-5972058","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5972058","identity":"rs-5972058","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}