WorldCIST'15 - 3rd World Conference on Information Systems and Technologies

Full Program »

Using data mining techniques to support breast cancer diagnosis

More than ever, in breast cancer research, many computer aided diagnostic systems have been developed in order to reduce false-positives diagnosis. In this work, we present a data mining based approach which might support oncologists in the process of breast cancer classification and diagnose. A reliable database with 410 images was used containing microcalcifications, masses and also normal tissue findings. We applied two feature extraction techniques, specifically the gray level co-occurrence matrix and the gray level run length matrix, and for classification purposes several data mining classifiers were also used. The results revealed great percentages of positive predicted value (approximately 70%) and very good accuracy values in terms of distinction of mammographic findings (>65%) and classification of BI-RADSĀ® scale (>75%). The best predictive method and the best performance on the distinction of microcalcifications found was the Random Forest classifier.


Joana Diz    
Institute of Engineering-Polytechnic of Porto

Goreti Marreiros    
GECAD - Knowledge Engineering and Decision Support Group, Institute of Engineering-Polytechnic of Porto

Alberto Freitas    
Center for Health Technology and Services Research (CINTESIS) and Department of Health Information and Decision Sciences (CIDES)


Powered by OpenConf®
Copyright ©2002-2013 Zakon Group LLC