Identifier

etd-11172006-110852

Degree

Master of Science in Computer Science (MSCS)

Department

Computer Science

Document Type

Thesis

Abstract

A computer-aided diagnostic (CAD) system that uses a unique shape-based classification scheme, the Ellipse-Closed Curve Fitting (ECCF) algorithm, is developed for digital mammogram image analysis. The system is developed to work as a post-processing extension to a previously developed CAD system that locates and segments mass lesions, or tumors, found in digital mammograms into separate images. The ECCF system is implemented in the MATLAB mathematical scripting language and is thus capable of running on multiple platforms. The ECCF algorithm detects edges in tumor images and casts them into closed curve functions. Parameters for an ellipse of best fit for a closed curve function are computed in a way analogous to that in linear regression, where a line of best fit is determined to fit a set of data points. In addition to the shape-fitting algorithm, the ECCF system comprises several other independently functioning components, including auxiliary algorithms and techniques that perform image cropping and edge detection, employed initially to prepare the images for efficient processing, and self-test tools that calculate R2, area matching ratios, and a "shape conformity value" to determine the "goodness of fit". Output generated by the ECCF system for sufficiently large image sets may contain correlations between malignant tumors and their shape that may be captured with data mining techniques, the implementation of which may result in an improved integrated CAD system.

Date

2006

Document Availability at the Time of Submission

Release the entire work immediately for access worldwide.

Committee Chair

John Tyler

DOI

10.31390/gradschool_theses.336

Share

COinS