A Microcalcification Detection Using Multi-Layer Support Vector Machine in Korean Digital Mammogram

A computer-aided diagnosis (CAD) system has been examined to reduce the effort of radiologist. In the mammogram, it is helpful to improve the diagnostic accuracy of malignancy microcalcifications in early stage of detecting breast cancer. In this paper, we propose a microcalcification detection method using multi-layer support vector machine (SVM) classifiers to determine whether microcalcifications are malignant or benign tumors. The proposed microcalcification detection is divided into two steps, each of which uses a SVM classifier. First, potential ROIs (Region of interest) those are suspicious as malignant tumors are detected as a coarse detection level. And then, each ROI is classified whether it is malignant or not. The proposed algorithm is applied to the Korean digital mammogram. Experimental result showed that the proposed method would outperform conventional method using ANN (artificial neural networks).