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

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.

Keywords

Convolutional neural network / Anterior talofibular ligament / YOLO11 / Artificial intelligence / Deep learning / Magnetic resonance imaging / Ligament injury classification

References

[1] van Putte-Katier N, van Ochten JM, van Middelkoop M, et al.. Magnetic resonance imaging abnormalities after lateral ankle trauma in injured and contralateral ankles. Eur J Radiol., 2015, 84(12): 2586-2592

[2] Koris J, Calder JDF, Dalmau-Pastor M, et al.. Deltoid ligament injuries: a review of the anatomy, diagnosis and treatments. Knee Surg Sports Traumatol Arthrosc., 2024, 32(12): 3052-3064

[3] Lynch SA, Renström PAFH. Treatment of acute lateral ankle ligament rupture in the athlete. Phys Med., 1999, 27(1): 61-71

[4] Delahunt E, Bleakley CM, Bossard DS, et al.. Clinical assessment of acute lateral ankle sprain injuries (ROAST): 2019 consensus statement and recommendations of the International Ankle Consortium. Br J Sports Med., 2018, 52(20): 1304-1310

[5] Fong DT, Hong Y, Chan LK, et al.. A systematic review on ankle injury and ankle sprain in sports. Sports Med., 2007, 37(1): 73-94

[6] Kunde AMH, Vosseller JT, Dahne M, et al.. Combining radiographic and CT measurements to rival MRI for the diagnosis of acute isolated syndesmotic injury. Arch Orthop Trauma Surg., 2023, 143(11): 6631-6639

[7] Pi Y, Guo Q, Jiang D, et al.. Acute distal rupture of the calcaneus-fibular ligament near the calcaneus insertion: magnetic resonance imaging for diagnostic value and comparison to surgical findings—a retrospective case series study. J Foot Ankle Surg., 2023, 62(4): 712-718

[8] Baltes TPA, Arnáiz J, Geertsema L, et al.. Diagnostic value of ultrasonography in acute lateral and syndesmotic ligamentous ankle injuries. Eur Radiol., 2021, 31(4): 2610-2620

[9] Choufani C, Bilichtin E, Demory D, et al.. Value of arthroscopy in the management of acute lesions of the distal tibiofibular joint. Int Orthop., 2024, 48(9): 2439-2443

[10] Großterlinden LG, Hartel M, Yamamura J, et al.. Isolated syndesmotic injuries in acute ankle sprains: diagnostic significance of clinical examination and MRI. Knee Surg Phys Traumatol Arthrosc., 2016, 24(4): 1180-1186

[11] Chen CM, Huang YS, Fang PW, et al.. A computer-aided diagnosis system for differentiation and delineation of malignant regions on whole-slide prostate histopathology image using spatial statistics and multidimensional DenseNet. Med Phys., 2020, 47(3): 1021-1033

[12] Wu SH, Li MD, Tong WJ, et al.. Adaptive dual-task deep learning for automated thyroid cancer triaging at screening US. Radiol Artif Intell., 2025, 7(3 ArticleID: e240271

[13] Liu X, Wu D, Xie H, et al.. Clinical evaluation of AI software for rib fracture detection and its impact on junior radiologist performance. Acta Radiol., 2022, 63(11): 1535-1545

[14] Sahoo PK, Gupta P, Lai YC, et al.. Localization of colorectal cancer lesions in contrast-computed tomography images via a deep learning approach. Bioengineering (Basel)., 2023, 10(8): 972

[15] Ni M, Chen W, Zhao Q, et al.. Deep learning approach for MRI in the classification of anterior talofibular ligament injuries. J Magn Reson Imaging., 2023, 58(5): 1544-1556

[16] Nunes GA, Martinez LM, Cordier G, et al.. The ATFL inferior fascicle, the CFL and the PTFL have a continuous footprint at the medial side of the fibula. Knee Surg Phys Traumatol Arthrosc., 2023, 31(11): 5207-5213

[17] Nakasa T, Ikuta Y, Ota Y, et al.. Safe angles of ATFL and CFL anchor insertion into anatomical attachment of fibula in a lateral ankle ligament repair. J Orthop Sci., 2021, 26(1): 156-161

[18] Cardone BW, Erickson SJ, Den Hartog BD, et al.. MRI of injury to the lateral collateral ligamentous complex of the ankle. J Comput Assist Tomogr., 1993, 17(1): 102-107

[19] Schneck CD, Mesgarzadeh M, Bonakdarpour A. MR imaging of the most commonly injured ankle ligaments. Part II. Ligament Injuries. Radiol., 1992, 184(2): 507-512

Funding

National Key R&D Program of China (No. 2024YFC2417700)

Hubei Provincial Health Science and Technology Project (No. WJ2025M136)

Hainan Provincial Key Laboratory of Biomedical Engineering Open Project(No. BME20240004)

RIGHTS & PERMISSIONS

The Author(s), under exclusive licence to the Huazhong University of Science and Technology

Just Accepted

This article has successfully passed peer review and final editorial review, and will soon enter typesetting, proofreading and other publishing processes. The currently displayed version is the accepted final manuscript. The officially published version will be updated with format, DOI and citation information upon launch. We recommend that you pay attention to subsequent journal notifications and preferentially cite the officially published version. Thank you for your support and cooperation.