Breast Cancer, one of the cancer types, is the cause of the second highest mortality rate in the world. Breast cancer normally happens in women and early detection is the key to increase the survival rate. A digital mammogram is a type of medical imaging which is used by a radiologist to predict the presence of cancer tumor in the breast region. However, there were reports on 10-30 % of breast cancer cases being undetected due to either the incorrect diagnosis or misinterpretation of the digital mammogram by radiologists. Hence, this study aims to use a machine learning technique for the interpretation of digital mammograms to detect the early presence of breast cancer to reduce mortality in breast cancer. A Computer-aided System was developed for the diagnosis by first classifying a mammogram into the normal and abnormal categories before the identification of benign or malignant breast tumors. Feature extraction and artificial neural network were used in the classification process via supervised learning. The outcomes of this study showed that the net produced for classification between normal and abnormal, as well as between benign and malignant, can provide a considerably high accuracy of 89% and 96%, respectively, in their diagnosis. In this study, we also took a further step in utilizing the nets produced by the CAD program and tested them with another 10 data samples from the mini-MIAS dataset, the result revealed 100% accuracy in their classification results
Volume 12 | 02-Special Issue
Pages: 642-649
DOI: 10.5373/JARDCS/V12SP2/SP20201116