Date of Award

Fall 1994

Document Type

Thesis - Restricted

Degree Name

Master of Science (MS)

Department

Biomedical Engineering

First Advisor

Hock, Jeff

Second Advisor

Rhyne, Ted

Third Advisor

Belfore, Lee

Abstract

The use of medical ultrasound imaging instruments to characterize heart tissue is an attractive procedure. To date clinical and laboratory results have indicated that the viability of a region of heart tissue can be assessed using noninvasive ultrasound imaging procedures which measure the tissues tendency to scatter ultrasound radiation. In order to facilitate the assessment of tissue in a beating heart, one must isolate a portion of the tissue and follow, or track, that region of tissue as it moves during the cardiac cycle. A reliable and automatic means to accomplish this tracking is needed in order to produce accurate tissue assessment. This thesis presents the development and testing of an algorithm to accomplish this region tracking function. The problem of region or feature tracking and detection is not a new topic in the field of image processing. The task of feature region tracking in a medical ultrasound image presents some unique difficulties in this case. A very low signal to noise ratio, poor boundary definition, and changing feature shape all make large demands on this automatic tracking algorithm. In order to address these challenges, the region of interest tracking algorithm developed in this thesis departs from typical detection algorithms and borrows from the random signal analysis techniques more common to the communications field.

Share

COinS

Restricted Access Item

Having trouble?