Microcalcification detection in mammography image using computer-aided detection based on convolutional neural network

Breast cancer is one of the most common among women and survival rate tends to be low when its stage found high treated. To improve breast survival, early detection critical. There are two ways for cancer: diagnosis screening. make an accurate in cancer, appearance masses micro-calcifications on mammography image important indicators. It time-consuming challenging identify micro-calcification from mammogram images by human eye because size appearance. Several Computer-Aided Detection (CADe) have been developed support radiologists automatic next recommended treatment. Most current CADe systems at this time started using Convolutional Neural Network (CNN) implement mammograms their quantitative results very satisfying, average level accuracy more than 90%. However, methods used fragments a complete which then included program. This research conducts automated approach detect location any with simple way. At first, preprocessing algorithms were applied enhance quality. After that, region was labeled segmentation based Radiologist's expertise. The positive label contains pixels taken train network. A total 354 INbreast dataset study. Finally, trained network utilized area automatically images. process can help as assistant radiologist increase regions. proposed system performance measured according error values Mean Squared Logarithmic Error (MSLE) technique find out difference between predicted model actual values. best MSLE loss value obtained achieved 0.05 0.95.