ACL Injury Detection and Classification utilizing Magnetic Resonance Imaging Scans and Deep Learning Techniques

Abstract

One of the most common injuries incurred through physical activity is the anterior cruciate ligament tear or ACL tear. ACL injuries tend to have a low self-recovery rate and as such, they usually require surgery in order to reconstruct or repair the torn ligament. The study aims to improve upon the already established methods when trying to diagnose and classify potential ACL injuries through the use of convolutional neural networks (CNNs) and transfer learning. Different parameters are tested to find the optimum and best-performing model based on specific performance metrics. A web-based decision support tool for assessing and diagnosing ACL injury utilizing knee MRIs integrating the best performing CNN model is developed serving as a valuable decision support tool in different healthcare applications.