Document Type

Thesis

Date of Degree

Fall 2014

Degree Name

MS (Master of Science)

Degree In

Electrical and Computer Engineering

First Advisor

Punam K. Saha

Abstract

The finite element method (FEM) has been widely applied to various medical imaging applications over the past two decades. The remarkable progress in high-resolution imaging techniques has allowed FEM to draw great research interests in computing trabecular bone (TB) stiffness from three-dimensional volumetric imaging. However, only a few results are available in literature on applying FEM to multi-row detector CT (MDCT) imaging due to the challenges posed by limited spatial resolution. The research presented here develops new methods to preserve TB structure connectivity and to generate high-quality mesh representation for FEM from relatively low resolution images available at MDCT imaging. Specifically, it introduced a space-variant hysteresis algorithm to threshold local trabecular structure that preserves structure connectivity. Also, mesh generation algorithms was applied to represent TB micro-architecture and mesh quality was compared with that generated by traditional methods. TB stiffness was computed using FEM simulation on micro-CT (µ-CT) and MDCT images of twenty two cadaveric specimens of distal tibia. Actual stiffness of those specimens were experimentally determined by mechanical testing and its correlation with computed stiffness was analyzed. The observed values of linear correlation (r2) between actual bone stiffness and computed stiffness from µ-CT and MDCT imaging were 0.95 and 0.88, respectively. Also, reproducibility of the FEM-based computed bone stiffness was determined from repeat MDCT scans of cadaveric specimens and the observed intra-class correlation coefficient was a high value of 0.98. Experimental results demonstrate the feasibility of application of FEM with high sensitivity and reproducibility on MDCT imaging of TB at distal tibia under in vivo condition.

Public Abstract

Over the past two decades, trabecular bone (TB) study has drawn research interests to determine bone quality, understand causes of diseases, and assess risks of fractures. The finite element method (FEM) is an advanced computer-aided method that computes stiffness and simulates behaviors of target object. With the remarkable progress of high-resolution imaging techniques, FEM has been widely applied in TB study. However, only a few research focuses on FEM application on multi-row detector CT (MDCT) imaging because of limited spatial resolution. Since MDCT has advantages of fast acquisition, widely availability and capability of visualizing large bone structures, it has a great potential for clinical use. In this work, a framework of FEM modeling, which possesses several desired properties meeting requirements of FEM simulation, is presented and applied on MDCT and micro-CT (μ-CT). The framework introduces a space-variant hysteresis image processing method and applies state-of-the-art mesh generation algorithms to form an automatic system. Bone stiffness of TB is computed from twenty two cadaveric subjects using mechanical testing and FEM simulation on μ-CT and MDCT. Reproducibility of repeat MDCT scans was found to be 0.98. Linear correlations between mechanical testing and μ-CT, MDCT were found to be R2=0.95 and R2=0.88, respectively. The result demonstrates the feasibility and high sensitivity of the proposed framework. It is a fundamental tool for future studies of TB using MDCT imaging.

Keywords

publicabstract, finite element, mechanical testing, mesh generatioin, multi-row detector CT, space-variant hysteresis, Young's modulus

Pages

vi, 30 pages

Bibliography

Includes bibliographical references (pages 27-30).

Copyright

Copyright 2014 Cheng Chen

Share

COinS