Date of Award
12-2009
Degree Name
Master of Science in Engineering
Department
Electrical and Computer Engineering
First Advisor
Dr. Ikhlas Abdel-Qader
Second Advisor
Dr. Massood Z. Atashbar
Third Advisor
Dr. Liang Dong
Access Setting
Masters Thesis-Campus Only
Abstract
Early detection of Breast Cancer has been proven to save lives. Today, mammography is the best method for breast cancer detection. Studies have shown that 10% - 30% of women who undergo mammography have negative mammograms. Out of these, two-thirds are false negative. One of the root causes is identified to be the inability of the reader to detect these abnormalities due to various reasons such as poor image quality, image noise, or eye fatigue. This study focuses on the development of an algorithm for classifying mammograms as normal or abnormal. The algorithm integrates the Principal Component Analysis (PCA) and Fisher Linear Discriminant (FLD) methods, proving to be a successful combination for detecting abnormalities in mammograms. The algorithm is tested on the Mammographic Image Analysis Society (MIAS) Digital Mammogram Database which has normal and abnormal mammograms with a 93.75% classification success rate. Results show PCA-FLD outperforms PCA and FLD alone. The algorithm is highly dependent on its parameters and suggestions for their selection and future work are provided.
Recommended Citation
Sarfraz, Memuna, "An Integrated PCA-FLD Classification System for Mammographic Images" (2009). Masters Theses. 297.
https://scholarworks.wmich.edu/masters_theses/297