Deep Learning Solutions for Knee Osteoarthritis Prediction with Optimized Convolutional Neural Networks

A generative in nature joint disorder that is predominantly affects middle-aged and older adults is knee osteoarthritis (KOA). However, there are significant obstacles to an objective and effective early diagnosis due to technical bottlenecks including noise, artifacts, and modality. In this study, a comprehensive Deep Learning (DL) approach is presented that Adaptive Wiener Filter (AWF) for preprocessing, Segmentation using K Means Clustering, feature selection using Gray Level Co-Occurrence Matrix (GLCM), and Sea Gull Optimization Convolutional Neural Network (SGO-CNN) for Knee KOA prediction. As a result, it will support KOA research as well as draw attention to shortcomings and possible issues with use in clinical practice. The first phase involves Adaptive Wiener Filter. The proposed approach is validated using Python Software with KOA prediction dataset. Performance image, including accuracy, precision, recall and f-score, demonstrates superior’s results in predicting KOA prediction utilizing the proposed approach.