Date of Award

December 2014

Degree Type

Thesis

Degree Name

Master of Science

Department

Engineering

First Advisor

Brian S.R. Armstrong

Committee Members

Kristian M. O'Connor, Yi Hu, Jun Zhang

Keywords

Functional Method, Hip, Joint Parameters, Kinematic Model, Knee, Optimization

Abstract

Biomechanics, generally speaking, concerns the application of engineeringprinciples to the study of living things. This work is concerned withhuman movement analysis, a subfield of biomechanics, where the methodsof classical mechanics are applied to human movement. This field hascontributed to the general understanding of human movement, and itstechniques are used in the diagnosis and treatment of disease. Centralto the field is the process of measuring human movement. Since classicalmechanics deals with the motion of rigid bodies, and ideal measurementsystem would be able to accurately record the exact pose --- combinedposition and orientation --- of the bones. The techniques that reachthis ideal require exposure to radiation or the insertion of metalpins into bones. Non-invasive methods are far more commonly used,and these involve the optical tracking of special markers placed overthe skin on each segment of the body being studied. Motion capturesystems are able to accurately record the pose of the markers, butthey bear no repeatable relationship to the pose of the underlyingbone. Many techniques are employed to bridge the gap between the two.The most direct technique finds three or more points on each bonenear the surface of the skin, called Anatomical Landmarks (ALs), anduses them to define the bone's pose relative to the motion trackingmarkers. There are concerns about the reliability of this method;the same experimenter performing this procedure multiple times onthe same subject will choose slightly different points on the bone,leading to variation in its orientation. The problem is exacerbatedwhen multiple experimenters are involved. This affects the abilityto compare data across time or between working groups; it may alsolead to erroneous interpretations of data. Furthermore, this techniquecannot be used directly to locate the hip joint center; instead, ALsat the pelvis are used as independent variables in a regression equationwhich statistically predicts the hip joint center location. Such techniqueshave begun to show reasonable reliability only recently.

An alternative approach is to orient the bones based on a mathematicalanalysis of the motion of the tracking markers while the subject moves.This is the domain of functional and optimization methods. Functionalmethods are commonly used to find two joint parameters in particular:the center of the hip joint and the axis of rotation of the knee.Once found, these parameters are used to determine the orientationof the bones relative to the tracking markers. Functional methodsare subject specific and operator independent but may be biased dueto the presence of Soft Tissue Artifact (STA), which is the measurementerror caused by the movement of tissue in between the tracking markersand the underlying bone. Optimization methods estimate joint parametersby fitting a kinematic model of the joints under study to motion datawhich records a subject exercising those joints. Unlike functionalmethods, which estimate parameters for a single joint, optimizationmethods may estimate the parameters of multiple joints in some circumstances.The parameters of a kinematic model incorporating multiple jointsmay be estimated as long as the relative pose of the end segmentsof the model is measured with more Degrees of Freedom (DoF) than themodel itself possesses. The key insight of this work is thata kinematic model which contains a spherical hip joint and a 2 DoFcompound hinge knee joint may be fitted to motion data from the pelvisand lower leg. There are two benefits to this procedure. First, thethigh is known to be affected by a high degree of STA; by removingdependence on data from the thigh, this method gains the potentialfor more accurate joint parameter estimates. Second, once fitted tomovement data, the model provides an estimate of the pose of the femur.One may investigate STA at the thigh by comparing the pose of thethigh markers to the model's estimate of the pose of the femur. Typically,medical imaging or invasive methods are required to investigate STA;this procedure is accessible and safe.

In summary, this work presents a technique which has the potentialto make the non-invasive measurement of human movement more reliable.This technique also provides the possibility of estimating soft tissueartifact at the thigh in a safe and convenient manner.

Share

COinS