Neural Network Based Automatic Detection of Lesion Diagnosis in Mammogram Using Image Fusion

Breast cancer is the second-most common and leading cause of cancer death among women. A Computer Aided Diagnosis (CAD) system for the classification of microcalcification (tumor) lesions in mammogram images is presented in this work. The process uses tumor characteristics in mammogram images, such as shapes (circle or elliptical), sizes (2mm to 5mm), locations (distance from center) and intensities (pixel values) for the isolation of the tumor which depends on manual tracing by experts. This project proposes automatic mammogram tumor detection and isolation of tumor cells from mammogram images using image fusion and neural networks. The performance of the proposed algorithm is also compared to the existing methods. The system classifies the mammogram images as normal or abnormal, and abnormal severity as benign or malignant. The experiments results demonstrate that this approach can provide better classification rate. The evaluation of the system is carried on Mammography Image Analysis society (MIAS) database. Keywords— CAD, Neural Network classifier, Mammography, SVM classifier.