A Comparison of Standalone and Hybrid System in Diagnosing Anterior Cruciate Ligament (ACL) Injury

This paper presents a comparison of standalone and hybrid diagnosing system. The hybrid system applied sorting and learning system on diagnosing ACL knee injury. The development of the system involves processes of intelligent system, sorting injury and classifying injury based on MRI. Sorting is done through support vector machine (SVM), which identifies and isolates potential ACL injuries. The raw data were extracted by the image-processing phase using k-mean clustering system with parameter of C = 2(3) and gamma = 2(-4). The injury classification applied an ANN model using data scheme of training (70%), testing (20%) and validation (10%). The outcome is categorized into three types of injuries; normal, partial and crucial. Results of ensemble sorting and learning system improved the current analysis system performance by an average accuracy of 86.2% and it is proven by ROC curve that shows sensitivity at 81.7%.