Newsroom
Headlines
Calendar
Seminar Series
Publications
Media Information
An Investigation and Expansion of Musculoskeletal Modeling and Analysis Techniques
John Wade Kelly
Date: 2008-07-07
Degree: MS - Electrical Engineering
Advisory Committee
Carol Giuliani - Committee MemberEdward Grant - Committee Chair
H. Troy Nagle - Committee Member
Abstract
A better understanding of human movement and its underlying dynamics is essential in developing more effective methods of rehabilitation and in assisting the diagnosis of physical ailments. Currently, the methods for analyzing and modeling human motion lag behind the methods available for capturing motion data. Many software packages have been developed to close this gap; one of the most recent and promising packages is OpenSim. However, a major problem exists with OpenSim: it can only handle a small range of formats used for capturing motion. Here, a new software package was developed that allowed motion capture files in the C3D format to be used in OpenSim. There were still minor problems associated with OpenSim's ability to analyze this data, however, ultimately, the results obtained from this new software package proved to be as accurate as that obtained from an analysis conducted using the same data and a proprietary software package for musculoskeletal analysis. Also, like other modeling software, OpenSim can only calculate the parameters associated with motion; it does not reduce the data to relevant statistics or determine patterns related to motion. To address these limitations two large sets of motion data were analyzed: first a motion data set from recovering stroke patients, and second, a data set from healthy subjects. Features of interest were extracted from the data sets and used to create a pattern classifier that recognized the distinct motion patterns exhibited by recovering stroke victims. Experiments with this new proof-of-concept system proved that C3D motion capture data could be successfully imported into OpenSim and analyzed, and then important motion patterns could be extracted and classified to show abnormalities in human movement.
Download the complete document from the NCSU Libraries: 
