A Visual Validation Method Based on Point Mappings for Medical Image Segmentation

Validation for image segmentation plays an important role in medical image processing. Point-mappings based validations process segmentation results by matching corresponding points on surfaces of the segmentation model and the standard model, then evaluate the matches locally and globally. This paper proposes four algorithms of building point-mappings; and creatively applies perfect matching algorithm of weighed bipartite graph on it. Furthermore, detailed local and global evaluation methods are also presented. Finally, experiments demonstrate that such validation method is competent to evaluate 2D or 3D segmentation scientifically, expeditiously and accurately