@inproceedings{af00aa3cab0445c49ed401b857c7e952,
title = "Mass classification in digitized mammograms using texture features and artificial neural network",
abstract = "A technique is proposed to classify regions of interests (ROIs) of digitized mammograms into mass and non-mass regions using texture features and artificial neural network (ANN). Fifty ROIs were extracted from the MIAS MiniMammographic Database, with 25 ROIs containing masses and 25 ROIs containing normal breast tissue only. Twelve texture features were derived from the gray level co-occurrence matrix (GLCM) of each region. The sequential forward selection technique was used to select four significant features from the twelve features. These significant features were used in the ANN to classify the ROI into either mass or non-mass region. By using leave-one-out method on the 50 images using the four significant features, classification accuracy of 86% was achieved for ANN. The test result using the four significant features is better than the full set of twelve features. The proposed method is compared with some existing works and promising results are obtained.",
keywords = "Artificial neural network, Mammogram, Mass classification",
author = "Wong, {Man To} and Xiangjian He and Hung Nguyen and Yeh, {Wei Chang}",
year = "2012",
doi = "10.1007/978-3-642-34500-5_19",
language = "English",
isbn = "9783642344992",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
number = "PART 5",
pages = "151--158",
booktitle = "Neural Information Processing - 19th International Conference, ICONIP 2012, Proceedings",
edition = "PART 5",
note = "19th International Conference on Neural Information Processing, ICONIP 2012 ; Conference date: 12-11-2012 Through 15-11-2012",
}