Residual Conv-Transformer with Hybrid Attention-Based Knee Osteoarthritis Classification and Optimized Echo State Network for Severity Assessment

Currently, the huge population worldwide are affected by Osteoarthritis (OA), a common type of arthritis. The impact of OA is often seen on the joints of the body, including the knee, spine, and hands. OA is common in the elderly population, which commonly occurs in the knees. Knee OA develops due to damage to cartilage in the knee joint, which is diagnosed by severe knee pain. However, detecting and classifying Knee OA at the primary stage is hard. Ensuring effective treatments at the initial phase of OA may assist in mitigating the complications associated with it. Therefore, an automatic and robust Knee OA detection framework is introduced in the work. Here, the essential images are accumulated from the benchmark datasets. Then, the Residual Convolutional Transformer with Hybrid Attentional Module (ResConvT-HAM) is proposed for classifying the Knee OA. If the person is affected with Knee OA, then the severity of the disease is analyzed using the Adaptive Echo State Network (Adap-ESN), where the parameters are tuned using the Modified Decision Variable Update-based Spider-Tailed Viper and Bird Optimizer (MDVU-STVBO) for improving the performance of the system in the severity assessment. The developed framework is employed by the medical practitioner to provide accurate results in the Knee OA classification and severity assessment. At last, the efficiency of the introduced technique is compared with the previous findings to prove the efficiency of the designed approach.