Deformable object segmentation in ultra-sound images

摘要: This thesis analyses the current strategies to segment breast lesions in Ultra-Sound (US) data and proposes a fully automatic methodology for generating accurate segmentations of US with low false positive rates. The proposed approach targets segmentation as minimization procedure multi-label probabilistic framework that takes advantage min-cut/max- flow Graph-Cut (GC) inferring appropriate label from set tissue labels all pixels within target image. image is divided into contiguous regions so belonging particular region would share same by end process. From training dataset stochastic models are built order infer each main it splits problem segmenting tissues present images subtasks can be taken care individually