Automatic Classification of Anterior Talofibular Ligament Based on 2D Convolutional Neural Network.

IF 2 4区医学 Q3 MEDICINE, RESEARCH & EXPERIMENTAL

Current Medical Science Pub Date : 2026-08-01 Epub Date: 2026-08-03 DOI:10.1007/s11596-026-00241-5

Feng Li, Xiao-Shan Wang, Ting Li, Gang Wu

{"title":"Automatic Classification of Anterior Talofibular Ligament Based on 2D Convolutional Neural Network.","authors":"Feng Li, Xiao-Shan Wang, Ting Li, Gang Wu","doi":"10.1007/s11596-026-00241-5","DOIUrl":null,"url":null,"abstract":"Objective: To investigate the feasibility of 2D convolutional neural networks (CNNs) in the automatic classification of anterior talofibular ligaments (ATFLs) on MR images.Methods: A total of 560 transverse T2-weighted MR images of the ATFL were collected from Center A, and 96 from Center B; manual segmentation of the ATFL was performed. The ATFL segmentation model was trained on YOLO11 and was validated on images from Center B. The dice similarity coefficient (DSC) between manual and automatic segmentation was calculated. A total of 1,103 T2-weighted MR images of the ATFL were further collected from Center C and divided into three groups: normal, partial, and total tear, and ATFL was automatically segmented for all the images. The 2D ResNet model was then trained for ATFL classification. Finally, the segmentation model and classification model were applied to 420 images from Center D.Results: The median DSC for the YOLO11 segmentation model was 0.95. For Center D data, the automatic workflow achieved an accuracy of 92.6% (389/420). It showed 95.0% (190/200) sensitivity and 93.6% (206/220) specificity for abnormal ATFL detection, slightly below the junior radiologist's 97.0% (194/200) sensitivity and 95.5% (210/220) specificity, but the difference did not reach statistical significance (P = 0.22). Automatic classification of the Center D dataset took 3 minutes, compared with manual 14 minutes for the junior radiologist.Conclusion: Automatic segmentation and classification of ATFL on MR images based on CNNs are feasible for evaluating ATFL.","PeriodicalId":10820,"journal":{"name":"Current Medical Science","volume":" ","pages":"1121-1126"},"PeriodicalIF":2.0000,"publicationDate":"2026-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Current Medical Science","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1007/s11596-026-00241-5","RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"2026/8/3 0:00:00","PubModel":"Epub","JCR":"Q3","JCRName":"MEDICINE, RESEARCH & EXPERIMENTAL","Score":null,"Total":0}

引用次数: 0

Abstract

Objective: To investigate the feasibility of 2D convolutional neural networks (CNNs) in the automatic classification of anterior talofibular ligaments (ATFLs) on MR images.

Methods: A total of 560 transverse T2-weighted MR images of the ATFL were collected from Center A, and 96 from Center B; manual segmentation of the ATFL was performed. The ATFL segmentation model was trained on YOLO11 and was validated on images from Center B. The dice similarity coefficient (DSC) between manual and automatic segmentation was calculated. A total of 1,103 T2-weighted MR images of the ATFL were further collected from Center C and divided into three groups: normal, partial, and total tear, and ATFL was automatically segmented for all the images. The 2D ResNet model was then trained for ATFL classification. Finally, the segmentation model and classification model were applied to 420 images from Center D.

Results: The median DSC for the YOLO11 segmentation model was 0.95. For Center D data, the automatic workflow achieved an accuracy of 92.6% (389/420). It showed 95.0% (190/200) sensitivity and 93.6% (206/220) specificity for abnormal ATFL detection, slightly below the junior radiologist's 97.0% (194/200) sensitivity and 95.5% (210/220) specificity, but the difference did not reach statistical significance (P = 0.22). Automatic classification of the Center D dataset took 3 minutes, compared with manual 14 minutes for the junior radiologist.

Conclusion: Automatic segmentation and classification of ATFL on MR images based on CNNs are feasible for evaluating ATFL.

查看原文本刊更多论文

基于二维卷积神经网络的距腓骨前韧带自动分类。

目的：探讨二维卷积神经网络（cnn）在距腓骨前韧带（atfl） MR图像自动分类中的可行性。方法：从A中心采集ATFL横t2加权MR图像560张，从B中心采集96张；手动分割ATFL。在YOLO11上对ATFL分割模型进行训练，并在来自Center b的图像上进行验证，计算手动和自动分割的骰子相似系数（DSC）。进一步从C中心收集ATFL的t2加权MR图像1103张，分为正常撕裂、部分撕裂和完全撕裂三组，并对所有图像自动分割ATFL。然后训练二维ResNet模型进行ATFL分类。最后，将分割模型和分类模型应用于d中心的420幅图像。结果：YOLO11分割模型的DSC中位数为0.95。对于D中心数据，自动工作流的准确率达到92.6%（389/420）。对异常ATFL检测的敏感性为95.0%(190/200)，特异性为93.6%(206/220)，略低于初级放射科医师97.0%（194/200）的敏感性和95.5%（210/220）的特异性，但差异无统计学意义（P = 0.22）。Center D数据集的自动分类需要3分钟，而初级放射科医生的手动分类需要14分钟。结论：基于cnn的MR图像ATFL自动分割与分类对ATFL评价是可行的。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Current Medical Science Biochemistry, Genetics and Molecular Biology-Genetics

CiteScore

4.70

自引率

0.00%

发文量

126

期刊介绍： Current Medical Science provides a forum for peer-reviewed papers in the medical sciences, to promote academic exchange between Chinese researchers and doctors and their foreign counterparts. The journal covers the subjects of biomedicine such as physiology, biochemistry, molecular biology, pharmacology, pathology and pathophysiology, etc., and clinical research, such as surgery, internal medicine, obstetrics and gynecology, pediatrics and otorhinolaryngology etc. The articles appearing in Current Medical Science are mainly in English, with a very small number of its papers in German, to pay tribute to its German founder. This journal is the only medical periodical in Western languages sponsored by an educational institution located in the central part of China.