Mammograms Segmentation and extraction for breast cancer regions based on region growing

Abstract

Medical imaging is an essential part of modern healthcare, where it’s technologists take X-rays, mammograms, ultrasounds and computed tomography images to help diagnose patients' injuries and diseases. In this work, the application of region growing technique was explored to the problem of image segmentation, extracting and finding the boundary of different breast tissue regions cancer in mammograms images. Our search focus on two parts, the first is detecting the cancer area. The detection algorithm used on 117 dataset images, the result is detection cancer in 115 image and two images are shifted. Then the second step started with origin these images using segmentation method based on region growing. The goal of the segmentation algorithm here is to see if the region-growing algorithm could separate different intensities for the different breast patterns. This algorithm is applied with selecting a seed point to provide the hard constraint, whereas the seed point are selected based on user-defined. Region growing has been explored on images of various imaging modalities but not on mammograms just yet. Therefore, this article is mainly focused on using region-growing algorithm to perform segmentation to increase the visibility of different breast densities in mammography images. Our proposed methodology for the segmentation of mammograms has been tested on Mini-MIAS database mammogram images. The results show that the proposed algorithms are fast for image segmentation into regions with edges detection, region extraction, and region features calculations such as region area, mean, max, minimum intensity values, and no. of pixels etc