Deep learning framework for the detection and classification of knee osteoarthritis using multi modal images

Abstract

Osteoarthritis being a most prevalent degenerative musculoskeletal newlinedisease is affecting almost 5% of the global population. The human knees are newlinethe most common joints affected by osteoarthritis and is characterized by newlineirreversible degeneration of the articular cartilage at the ends of the bones such newlineas femoral, tibial, and patella cartilages. Knee osteoarthritis is a progressive newlinedisease that affects the entire knee joint and is a condition driven by mechanical newlinewear and tear and biochemical changes. Known risk factors for Osteoarthritis newline(OA) include aging, obesity, and previous knee injuries. The presence of newlineknee osteoarthritis in humans makes them lack in their day-to-day life and newlinedrastically affects their lifestyle. Due to this, it is always essential to identify the newlineoccurrence of osteoarthritis at the earliest and enable the patients suffering from newlinethe disease to initiate treatment so that they to lead regular activities in a newlinepain-free manner. Many techniques are available to diagnose the early newlineoccurrence of OA disease and still, there has been a requirement to develop newlineeffective techniques. newlineAt this juncture, this thesis is intended for the development of novel newlinedeep learning neural network models, as these Deep Learning (DL) models newlineare newlinehighly effective for feature extraction of the image datasets. newlineNovel deep-learning neural network models based on the concept of newlineconvolutional neural models and recurrent learning models have been newlinedeveloped in this thesis for early identification and grading of the severity level newlineof osteoarthritis patients. newline