This paper presents a fully developed computer aided diagnosis (CAD) system for early knee Osteoarthritis (OA) detection using knee medical imaging and machine learning algorithms. The medical images are first preprocessed in the Fourier domain using a circular Fourier filter. Osteoarthritis (OA) is a chronic disease prevalent worldwide. There are two primary forms of osteoarthritis, affecting mainly the fingers, thumbs, spine, hips, knees, and big-toes, while secondary occurs with pre-existing joint anomalies. OA most commonly occurs in older individuals, but it can occur in adults of any age. OA is also known as degenerative joint disease, degenerative arthritis, and arthritis with wear and tear. Diagnosis of such disease is normally rendered by examining the joint scan of the Image with X-ray of knee. MRI analyzes are conducted by well-trained radiologists and orthopedists. The other side of this study is that it requires time and can be subject to loss of precision. The manual segmentation of images from a great number of scan images is a tedious and time-consuming procedure. Automatic segmentation and interpretation of joint MRI scans is therefore needed to increase the precision of clinical outcomes and bone calculation. In recent years, the advent of deep-learning technologies within medical systems is causing a transition. This can process a large volume of data to have greater precision. Deep learning methods can thus be used effectively for the automatic segmentation of MRI scans. This paper offers an overview of the different models and their output using deep learning and image processing techniques. Diagnosis of the illness is conferred from the image data. The state-of-the-art analysis is then discussed on Convolutional Neural Network (CNN) and Image Recognition. Finally, a comparative overview of the proposed model with other state-of-the-art techniques are presented based on the performance metrics. In the proposed detection algorithm, we found that all the suggested models achieved a higher level of predictive accuracy, greater than 90%, in detecting osteoarthritis. However, the best-performing model was the pretrained VGG-16 model with edge detection, which achieved a training accuracy of 99% and a testing accuracy of 92%.