Breast Cancer Detection using Image Segmentation

Abstract

Breast cancer is the most prevalent kind of malignancy in women, and diagnostic systems that apply artificial intelligence algorithms for breast imaging have shown positive results. Two methods that increase the precision of detecting breast tu-mors from mammography images are a multiclass support vector machine model and a deep convolutional neural network (DCNN) using K-means clustering. To accurately diagnose breast tumors, the pectoral muscle (PM) border must still be dis-tinguished from the rest of the breast tissue. By merging the transfer learning model with a number of pre-trained CNN structures, this research offers an Ensemble-Net model for distinguishing the PM boundary from the rest breast region in mammograms. The segmentation procedure consists of 2 steps. According to the input, various regions of interest are formed in the initial phase and include the object.