Date Approved

12-31-2003

Embargo Period

5-3-2016

Document Type

Thesis

Degree Name

M.S. in Engineering

Department

Electrical & Computer Engineering

College

Henry M. Rowan College of Engineering

First Advisor

Mandayam, Shreekanth

Subject(s)

Breast--Examination; Breast--Radiography

Disciplines

Electrical and Computer Engineering

Abstract

The percentage of radiodense (bright) tissue in a mammogram has been correlated to an increased risk of breast cancer. This thesis presents an automated method to quantify the amount of radiodense tissue found in a digitized mammogram. The algorithm employs a radial basis function neural network in order to segment the breast tissue region from the remainder of the X-ray. A spatially varying Neyman-Pearson threshold is used to calculate the percentage of radiodense tissue and compensate for the effects of tissue compression that occurs during a mammography procedure. Results demonstrating the efficacy of the technique are demonstrated by exercising the algorithm on two separate sets of mammograms - one obtained from Brigham Women's Hospital, Harvard Medical School and the other set obtained from Fox Chase Cancer Center and digitized at Rowan University. The results of the algorithm compare favorably with a previously established manual segmentation technique.

Share

COinS