Severity Classification of Knee Osteoarthritis from XRay Images using Deep Learning

Prakash, Pranav (2024) Severity Classification of Knee Osteoarthritis from XRay Images using Deep Learning. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (2MB) | Preview
Preview	PDF (Configuration Manual) Download (1MB) | Preview

Abstract

Osteoarthritis of the knee (KOA) is the most common cause of disability caused by cartilage degeneration and consequent joint failure. Getting diagnosed early is key to avoiding long-term pain and degeneration, mental health, and mobility issues. While traditional manual analysis of X-ray images can be subjective, time consuming and inconsistent, automated solutions are desired. Advances in deep learning have recently demonstrated effective use of CNN architectures including DenseNet121, EfficientNetB0 and MobileNet to perform superior to traditional machine learning methods on medical imaging tasks. This study explores the effectiveness of these models for classifying KOA severity at three levels of granularity: A three-class, five-class, and binary classification, on a dataset graded using the Kellgren–Lawrence system. Accuracy, F1-score and confusion matrices were used in evaluation of the models. Results were found to indicate that binary classification with DL models consistently outperformed conventional ML methods, with DenseNet121 being the most accurate (78.27%). As classification granularity decreased, performance improved, confirming the contribution of simplified tasks to ameliorating class imbalance and improve generalization. ML models such as Random Forest had a moderate outcome, but they exhibited a failure to deal with high dimensional data. Finally, this research has shown the basis for DL as an automation method of KOA diagnostics, and potential areas are suggested for future research exploiting hybrid models and clinical metadata integration.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Rustam, Furqan UNSPECIFIED
Uncontrolled Keywords:	CNN (Convolutional Neural Network); KOA (Knee Osteoarthritis); Classification; Accuracy; F1-Score; Machine Learning
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science R Medicine > Diseases > Disabilities R Medicine > Healthcare Industry Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	04 Sep 2025 10:32
Last Modified:	04 Sep 2025 10:32
URI:	https://norma.ncirl.ie/id/eprint/8776

Actions (login required)

View Item