Accuracy Verification of Automatic Musculoskeletal Segmentation for Upright CT

Abstract

Musculoskeletal models that are faithful to medical images are important for clarifying the causes of movement disorders. Since the kinematic properties of the musculoskeletal system change depending on the body position, it is desirable to create a musculoskeletal model of standing movements based on medical images scanned in the standing position. However, medical images scanned in the upright position are less common than those in the supine position, and data are limited. Therefore, the final goal of our project is to quantify musculoskeletal deformation in the upright position and predict it from the images scanned in the supine position in order to effectively use images scanned in the supine position. In this study, as a first step, we automatically segmented supine and upright CT images scanned for the same subject, and determined the segmentation accuracy with respect to the body position.