From classical to deep learning: review on cartilage and bone segmentation techniques in knee osteoarthritis research

Abstract

Knee osteoarthritis is a major diarthrodial joint disorder with profound global socioeconomic impact. Diagnostic imaging using magnetic resonance image can produce morphometric biomarkers to investigate the epidemiology of knee osteoarthritis in clinical trials, which is critical to attain early detection and develop effective regenerative treatment/therapy. With tremendous increase in image data size, manual segmentation as the standard practice becomes largely unsuitable. This review aims to provide an in-depth insight about a broad collection of classical and deep learning segmentation techniques used in knee osteoarthritis research. Specifically, this is the first review that covers both bone and cartilage segmentation models in recognition that knee osteoarthritis is a “whole joint” disease, as well as highlights on diagnostic values of deep learning in emerging knee osteoarthritis research. Besides, we have collected useful deep learning reviews to serve as source of reference to ease future development of deep learning models in this field. Lastly, we highlight on the diagnostic value of deep learning as key future computer-aided diagnosis applications to conclude this review.