A Fully Automatic Fine-Tuned Deep Learning Model for Knee Osteoarthritis Detection and Progression Anal

This study presents a fully automated deep learning framework tailored for the detection and progression analysis of knee osteoarthritis (OA). Leveraging state-of-the-art algorithms in classification and object detection, the framework integrates an ensemble of deep convolutional neural networks (CNNs), including Improved CNN, Xception, Inception V3, ResNet50, Inception ResNetV2, DenseNet121, and DenseNet169 with fine-tuned layers. These models are optimized to accurately classify knee OA from medical imaging data while providing detailed analysis of disease progression. For object detection, the framework employs variants of the YOLO (You Only Look Once) architecture, namely YOLOv5-GhostNet, YOLOv5, YOLOv6, and YOLOv7, ensuring robust detection of pathological features in knee images. The proposed approach aims to streamline diagnostic workflows by automating the detection of OA indicators and offering insights into disease development over time. Experimental results demonstrate the effectiveness of the ensemble approach in achieving high accuracy and reliability in both classification and detection tasks, positioning the framework as a promising tool for enhancing clinical decision-making and patient care in knee OA management.