Shortest Path Algorithm for Obtaining Muscular Anthropometric Data from MRI

Abstract

In the process of generating image based subject specific musculoskeletal models and the simulation of rescaled generic musculoskeletal models, the accurate segmentation of the anatomical structures of interest from medical images determines the efficiency of the musculoskeletal system models. Efficiency is highly influenced by the image segmentation technique used. This paper presents a semi-automatic segmentation algorithm based on the Dijkstra's shortest path algorithm for obtaining the origin and insertion points, and muscle paths from a magnetic resonance image. This algorithm significantly reduces the processing time while retaining high levels of sensitivity and specificity for the structures to be segmented. Anthropometric parameters calculated from the results obtained with the proposed algorithm are comparable to the results published by other groups of researchers. These results could be used to create an anthropometric parameters database from healthy population and with gait abnormalities that can be used in the development and simulation of rescaled generic models. In addition, the shortest path algorithm proposed in this paper could be used by the medical experts as a training algorithm of model based segmentation algorithm that may reduce processing time even more.