Supraspinatus Segmentation From Shoulder Ultrasound Images Using a Multilayer Self-Shrinking Snake

Abstract

Over people's lifetimes, the prevalence of shoulder pain exceeds 70%. In particular, 70% of shoulder pain is caused by the rotator cuff lesions that are located in the supraspinatus area. The automatic and quantitative segmentation of the supraspinatus area can provide a more-objective and accurate assessment of the rotator cuff lesions. In this paper, 108 shoulder ultrasound images comprised the image database to evaluate the proposed segmentation method, and a multilayer self-shrinking snake (S3), based on a multilayer segmentation framework, was used to achieve optimal segmentation. Using a rough initial contour that enclosed the supraspinatus area, the modified snake was shrunken with an iteration procedure according to the boundary conditions that included the elasticity, curvature, gradient, and distance. In the performance evaluation, S3 achieved an F-measure of 0.85. The success of S3 could provide more-objective location information to physicians diagnosing rotator cuff lesions.

Keywords

• Ultrasound
• shoulder pain
• rotator cuff
• segmentation
• supraspinatus segmentation
• snake