Segmentation of tumour from mammogram images using U-SegNet: a hybrid approach
M. Ravikumar, P. G. Rachana, B. J. Shivaprasad · Computer Methods in Biomechanics and Biomedical Engineering Imaging & Visualization · 2022
Breast cancer is the most common cancer type around the world which majorly affects women. Early detection of breast cancer helps in increasing survival rate. Segmentation helps in easy identification of abnormalities in image and draw conclusion whether the image is normal or not. So, a hybrid approach U-SegNet is proposed, which is fully convolutional neural network. It extends U-Net by integrating SegNet for improved mass detection. The proposed method is evaluated for its accuracy on publicly available Digital Database for Screening Mammography (DDSM) dataset. A comparison analysis is done on the proposed method compared with three other models such as Watershed, Fuzzy c-means and U-Net model; it is found that the proposed method gives good results.