Date of Award

Spring 1986

Document Type

Thesis - Restricted

Degree Name

Master of Science (MS)

Department

Biomedical Engineering

First Advisor

Doerr, Thomas A.

Second Advisor

Hock, Jeffrey

Abstract

Computerized Tomography (CT) provides a means for obtaining cross-sectional images of an object under examination. Three-dimensional properties of an object, such as its shape, volume and surface area, can be determined from a series of images of contiguous cross-sections. Traditionally, the object is viewed by displaying each of the cross-sectional images separately. Three-dimensional properties of the object can then be inferred by mentally connecting the series of cross-sectional images. Unfortunately, some objects are difficult to visualize through inference. For applications requiring the visualization of an object's shape, an alternative viewing mode is available. Surface display of an object allows the object's shape to be viewed in a single image, as opposed to requiring the observer to connect a series of cross-sections. Surface display of an object is accomplished by extracting a mathematical representation of the object surface from the CT data. A synthesized image, resembling a photograph of the object, is then constructed from the defined surface. The algorithms used to extract a representation of an object surface from CT data, and render synthetic images of an object, are computationally intensive. The implementation bf surface display algorithms in dedicated hardware and software has been investigated. Typically, the amount of processing power available for software-only implementations forces a trade-off to be made between image quality and the time required for image rendering. Two existing algorithms for surface display illustrate this trade-off. One algorithm provides high quality images, but requires a time-consuming preprocessing step to derive the mathematical surface representation. The other algorithm eliminates the pre-processing step, but the resulting images are of much lower quality than the images produced with the former algorithm. The goal of the research was to modify the latter algorithm to produce higher quality images, without adding a pre-processing step. The result of the research is an enhanced algorithm, providing higher quality images than those of the algorithm employing pre-processing.

Share

COinS

Restricted Access Item

Having trouble?