A Systematic Survey on Segmentation Algorithms for Musculoskeletal Tissues in Ultrasound Imaging

Abstract

Ultrasound imaging is widely used for the clinical assessment and study of musculoskeletal tissues because of its capacity for real-time imaging, low cost, high availability and portability. Objectively identifying and segmenting these tissues in ultrasound images can enhance disease diagnosis and biomechanical research. Manual segmentation is tedious, time-consuming and examiner-dependent. At the same time, ultrasound images suffer from poor image quality and low contrast between different regions in the image, making visual interpretation difficult. Hence, there is a need for reliable algorithms for computerised segmentation. This paper reviews the techniques developed for automated and semi-automated segmentation of vital musculoskeletal tissues (i.e. tendon, ligament, bone, muscle, plantar fascia and cartilage) from ultrasound images. This paper comprehensively explains each methodology and discusses distinguishing features, advantages and limitations to help the reader decide the most appropriate method on an application-specific basis.