Magnetic resonance imaging (MRI) has proved to be an invaluable component of pathogenesis research in osteoarthritis. Nevertheless, the detection of a meniscal lesion from magnetic resonance (MR) images is always challenging for both clinicians and researchers, because the surrounding tissues lead to similar signals within MR measurements, thus being difficult to discriminate. Moreover, the size and shape of osteoarthritic and non-osteoarthritic menisci vary to a large extent between individuals of same features, e.g. height, weight, age, etc. An effective way to visualize the entire volume of knee menisci is to segment the menisci voxels from the MR images, which is also useful to evaluate particular properties quantitatively. However, segmentation is a tedious and time-consuming task, and requires adequate training for being done properly. With the advancement of both MRI technology and computer methods, researchers have developed several algorithms to automate the task of meniscus segmentation of the individual knee during the last two decades. The objective of this systematic review was to present available fully automatic and semi-automatic segmentation methods of the knee meniscus published in different scientific articles according to the PRISMA statement. This review should provide a vivid description of the scientific advancements to clinicians and researchers in this field to help developing novel automated methods for clinical applications.
Keywords: Automatic segmentation; Knee; Magnetic resonance imaging (MRI); Meniscus; Review; Semi-automatic segmentation; Soft tissue.
Copyright © 2020 The Authors. Published by Elsevier B.V. All rights reserved.