Mass segmentation in Automated 3-D Breast Ultrasound Using Deep learning

Automated 3-D breast ultrasound(ABUS) is a novel system for breast screening‎. ‎It has been proposed as a supplementary modality to mammography for detection and diagnosis of breast cancers‎. ‎Although ABUS has better performance for dense breasts‎, ‎reading ABUS images is time-consuming and exhausting‎. ‎A computer-aided detection (CAD) system can be helpful for interpretation of ABUS images‎. ‎Mass Segmentation in CADe and CADx systems play the leading role because it affects the performance of succeeding stages‎. ‎Besides‎, ‎it is a very challenging task because of the vast variety in size‎, ‎shape, and texture of masses‎. ‎Moreover, imbalanced datasets make segmentation harder‎. ‎A novel mass segmentation approach based on deep learning is introduced in this paper‎. ‎The deep network that is used in this study for image segmentation is inspired by U-net which has been used broadly for dense segmentation in recent years‎‎. ‎Performance was determined using a dataset of 50 masses including 38 malignant and 12 benign masses‎.