PhD Proposal

A Dynamic-Image Computational Framework for Modeling Anatomical Joints

Md. Abedul Haque

Thursday May 17, 2012
10:00 am - Eli Lilly Room Sennott Square 6106

Abstract

We propose a dynamic-image driven, computational framework for the modeling and simulation of multi-articular anatomical joints. The framework uses anatomic knowledge, computational methods, static and dynamic medical images to accuraely measure bone kinematics and to model the subject-specific geometry of anatomical joints. The resulting models have application in biomechanical simulation, computer animation, and orthopaedic surgery.

The first component of the framework is an automated tracking method for measuring with sub-millimeter accuracy multi-articular 3D motion. The automation of the method enables the study of subject-specific joint kinematics over large groups of population. The accuracy of the method enables the modeling and analysis of small anatomical features that are difficult to capture in-vivo using existing imaging techniques. The second component of the framework is a set of computational tools to model joint soft-tissue structures. We propose to build hybrid, dynamic-image driven geometric models that will combine the complementary strengths of the accurate but static models used in orthopaedics and the dynamic but low level-of-detail multibody simulations used in humanoid computer animation. Leveraging dynamic images and reconstructed motion, this component will allow the modeling and simulation of small anatomical features and of their dynamic behavior. The third and last component of the framework will enable the generation of predictive, subject-specific models and simulations of healthy and symptomatic joints. The predictive models will help to identify, understand and validate hypotheses about joint disorders.

Dissertation Adviser

Dr. Dr. G. Elisabeta Marai, Department of Computer Science

Committee Members

Dr. Milos Hauskrecht, Department of Computer Science
Dr. Jingtao Wang, Department of Computer Science
Dr. Nancy Pollard, Department of Computer Science, Carnegie Mellon University