TY - GEN
T1 - A novel feature reduction framework for digital mammogram image classification
AU - Alharbi, Hajar
AU - Falzon, Gregory
AU - Kwan, Paul
PY - 2016/6/9
Y1 - 2016/6/9
N2 - The visual similarity between normal breast tissues and abnormal lesions in digital mammogram images makes computer-aided diagnosis of breast cancer using automatically detected features a highly error-prone task. Our contribution in this paper is a novel feature reduction framework for selecting the most discriminative features that achieves both efficiency and classification accuracy. Our approach applies five individual feature-ranking methods including Fisher score, minimum redundancy-maximum relevance, relief-f, sequential forward feature selection, and genetic algorithm for sorting the extracted features and selecting the features with highest ranking to setup a classifier. Our method achieves an accuracy of 94.27% and a sensitivity of 98.36% with a specificity of 99.27% on a set of 1,100 mammogram patches taken from image retrieval in medical applications database using a neural network classifier, which competes with state-of-the-art classification accuracy 93.11%. Furthermore, we demonstrate that only 49 out of the 119 extracted features are sufficient to achieve the reported accuracy of normal vs. abnormal classification.
AB - The visual similarity between normal breast tissues and abnormal lesions in digital mammogram images makes computer-aided diagnosis of breast cancer using automatically detected features a highly error-prone task. Our contribution in this paper is a novel feature reduction framework for selecting the most discriminative features that achieves both efficiency and classification accuracy. Our approach applies five individual feature-ranking methods including Fisher score, minimum redundancy-maximum relevance, relief-f, sequential forward feature selection, and genetic algorithm for sorting the extracted features and selecting the features with highest ranking to setup a classifier. Our method achieves an accuracy of 94.27% and a sensitivity of 98.36% with a specificity of 99.27% on a set of 1,100 mammogram patches taken from image retrieval in medical applications database using a neural network classifier, which competes with state-of-the-art classification accuracy 93.11%. Furthermore, we demonstrate that only 49 out of the 119 extracted features are sufficient to achieve the reported accuracy of normal vs. abnormal classification.
KW - Digital mammography
KW - Computer-aided detection/diagnosis (CAD)
KW - breast cancer
KW - Error rates
UR - http://www.scopus.com/inward/record.url?scp=84978786119&partnerID=8YFLogxK
U2 - 10.1109/ACPR.2015.7486498
DO - 10.1109/ACPR.2015.7486498
M3 - Conference contribution
AN - SCOPUS:84978786119
T3 - Proceedings - 3rd IAPR Asian Conference on Pattern Recognition, ACPR 2015
SP - 221
EP - 225
BT - Proceedings - 3rd IAPR Asian Conference on Pattern Recognition, ACPR 2015
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 3rd IAPR Asian Conference on Pattern Recognition, ACPR 2015
Y2 - 3 November 2015 through 6 November 2015
ER -