Segmentation of knee cartilage: A comprehensive review

Abstract

The main aim of this study is a complex critical review of the methods applicable for the knee cartilage segmentation. Segmentation of the knee cartilage is intended for a quantitative and qualitative analysis of cartilage morphological structure. This task is important in a field of the clinical practice in the context of an early diagnosis of the pathological changes, such as the osteoarthritis. The cartilage segmentation methods are divided into manual, semiautomatic and automatic approaches, each of this group is associated with a certain level of the user interaction. Generally, the knee cartilage segmentation and extraction can be performed by various approaches including the edge tracking, intensity based methods, supervised learning, energy minimization, statistical methods and multiregional segmentation methods. Using of particular segmentation method is depended on a compromise which user is willing to accept with a respect to the robustness, segmentation purpose, computational time, accuracy and level of user interaction. Some of the presented methods are intended for a detection of cartilage shape, moreover other methods are able to identify the pathological changes badly recognizable from the native image records.