Enhancing Local Public Health Services Through AI-Driven Knee Osteoarthritis Severity Assessment Using An Optimized Deep Vision Transformer

Povzetek

Knee Osteoarthritis (Knee OA) is a prevalent condition that affects elderly individuals. Knee OA diagnosis includes knee X-ray images. Proposed Tetrachoric correlated AlexNet based Golden Jackal Optimized Deep Vision Transformer (TA-GJODVT) model is developed for efficient classification of knee osteoarthritis’ severity levels. Numerous X-ray knee images gathered from the dataset in acquisition process. Then, the collected images are preprocessed using Adaptive Wiener Filter for eliminating the noise and enhancing the visibility without blurring. After that, segmentation process is performed by applying using Tetrachoric Correlated AlexNet model for segmenting joint space and extracting Region of Interest (RoI).  Followed by this, Gradient Tuned Golden Jackal Optimized Deep Vision Transformer is employed to perform feature extraction, classification and fine-tuning. In addition, Gradient tuned Golden Jackal Optimization is applied to execute hyperparameter tuning of transformer model for improving the accuracy of severity grading with lesser error rate. Experimental evaluation is carried out on several factors. The quantitative analysis of proposed TA-GJODVT Model achieves superior performance in classifying knee OA classification with minimum time and error rate compared to existing techniques.