EFRNet: A Deep Hybrid Model for End-to-End Knee Osteoarthritis Detection and Progression Assessment

Knee osteoarthritis (KOA), a common joint disorder affecting mobility, is traditionally assessed through digital X-ray images. These method dependents on expert skill and prone to errors due to subjectivity and noise. To overcome these issues, an enhanced Faster Regional Convolutional Neural Network (EFRCNN) has been proposed for KOA detection and classification. This method incorporates a hybrid pretrained module combining ResNet101, VGG19, and MobileNetV2 to improve feature extraction from input images. Additionally, a modified Anscombe transform is utilized to address the noise and low-resolution issues. To evaluate the effectiveness of the proposed method, publicly available KOA image data from the Osteoarthritis Initiative has been collected. The efficacy of the proposed model is comprehensively evaluated utilizing various metrics. The outcomes indicate that the proposed method outperforms existing methodologies, with an accuracy of 98.70%.