A supervised DCE-MR images classification strategy is proposed in this study. First, the training set was obtained by an automated seeds extraction procedure. Subsequently, support vector machine (SVM) and random walk algorithms were employed as two separate classification approaches to achieve image segmentations, respectively. The automated segmentations and a repeated manual segmentation were compared quantitatively with a reference manual segmentation. The average similarity indexes for SVM, random walker and repeated manual segmentation were 0.78, 0.76 and 0.72, respectively. The results indicate that the proposed strategy yield a satisfied similarity with manual segmentation and is more stable than the manual segmentation.
This abstract and the presentation materials are available to members only; a login is required.