Abstract
Anterior talofibular ligament (ATFL) injuries are common ankle injuries that require accurate grading for effective treatment planning. However, conventional diagnostic methods, including manual MRI interpretation, often lack objectivity and reproducibility. Radiomics, a technique that extracts quantitative features from medical images, offers a promising solution for enhancing diagnostic precision. This study developed a radiomics model based on MRI fat-suppressed proton density-weighted turbo spin-echo images to grade ATFL injuries. A dataset of 467 arthroscopically confirmed cases (276 partial tears, 191 complete tears) was analyzed, and 28 key features were selected for model construction using machine learning classifiers. The support vector machine (SVM) model achieved the best performance, with an AUC of 0.955 (95% CI: 0.931-0.980) on the training set and 0.844 (95% CI: 0.781-0.906) on the validation set. Decision curve analysis and confusion matrix results demonstrated the model's strong predictive accuracy and clinical utility. This SVM-based radiomics model offers a reliable, non-invasive approach for precise ATFL injury diagnosis and grading, with significant potential for improving clinical decision-making and personalized treatment.