Abstract

The prevention and diagnosis of chronic diseases have been the top priority of the biomedical sector with advancements in technology, and computer-aided diagnostics (CADs) have aided a lot in this regard. Still, with the advancements in CADs, accurate diagnosis with the least amount of human error is always a challenging task for doctors, surgeons and medical staff for long period.

With the recent advancements in computer science, and especially in the areas of artificial intelligence (AI) and Machine Learning (ML), the biomedical sector has truly aided and powered effective disease diagnosis. Knee osteoarthritis is a chronic disease that happens in adults and it needs to be diagnosed at early stages to prevent harm to patients and their lives. With the advancements in AI, this disease can be diagnosed with the least amount of error and will aid in controlling the factor of human error that happens in manual diagnosis.

This research aims to analyse, evaluate and develop an intelligent model using artificial intelligence that will be capable of identifying the severity of knee osteoarthritis in five possible classes using a dataset that contains X-ray scanned images of patients. The classes indicate the severity level of osteoarthritis from normal to severe. Deep Learning models like convolutional neural networks (CNNs) will be employed for the task since the dataset that is to be used is image related.

The dataset for knee osteoarthritis will be obtained from Kaggle, which is an online platform for AI and machine learning researchers. For achieving higher accuracy to put our research work competitive, pre-trained CNNs will be employed for image classification via transfer learning (TL). This research is initiated with a target not only to draw useful insights but also to highlight challenges for future practitioners and researchers working in this field.

Through a variety of observations and experiments, the system has been assessed. The accuracy performance of the experimental data is encouraging. The model has made predictions that are 86% accurate.

Intelligent Identification of Knee Osteoarthritis Severity Using Deep Learning