Automated Detection and Classification of Leg Bone Cancer Using MRI and Deep Learning Techniques

Abstract

Leg bone cancer is a severe musculoskeletal disorder that requires early and accurate diagnosis to improve patient survival and optimize treatment outcomes. This study presents an automated detection framework utilizing deep learning algorithms, specifically Convolutional Neural Networks (CNN) and Deep Neural Networks (DNN), for precise detection and classification of cancerous regions in MRI leg bone images. To enhance image quality, a median filter is applied during preprocessing to remove noise while preserving essential structural details. A Region of Interest (ROI) based approach focuses analysis on affected bone areas and eliminates irrelevant regions, improving detection accuracy and computational efficiency. The CNN and DNN models are comparatively evaluated using image based metrics to identify the most effective algorithm. Additionally, the study emphasizes patient awareness, exploring the role of lifestyle and dietary habits on disease occurrence and post-treatment recovery. The proposed framework facilitates early diagnosis, clinical decision support, preventive care, and treatment monitoring, contributing to improved patient outcomes.