Breast mass segmentation based on ultrasonic entropy maps and attention\n gated U-Net

We propose a novel deep learning based approach to breast mass segmentation\nin ultrasound (US) imaging. In comparison to commonly applied segmentation\nmethods, which use US images, our approach is based on quantitative entropy\nparametric maps. To segment the breast masses we utilized an attention gated\nU-Net convolutional neural network. US images and entropy maps were generated\nbased on raw US signals collected from 269 breast masses. The segmentation\nnetworks were developed separately using US image and entropy maps, and\nevaluated on a test set of 81 breast masses. The attention U-Net trained based\non entropy maps achieved average Dice score of 0.60 (median 0.71), while for\nthe model trained using US images we obtained average Dice score of 0.53\n(median 0.59). Our work presents the feasibility of using quantitative US\nparametric maps for the breast mass segmentation. The obtained results suggest\nthat US parametric maps, which provide the information about local tissue\nscattering properties, might be more suitable for the development of breast\nmass segmentation methods than regular US images.\n