Method of Kidney Image Segmentation Based on Improved C-V Model

Abstract

Kidney medical image segmentation is the key step of medical image analysis and non-invasive computer aided diagnosis system in related kidney diseases. Based on the traditional Chan-Vese model, according to the continuity and redundancy of the kidney tissues between slices in the CT sequence images, combined with local statistical information for improving the curve evolution, combined with the initial contour based on a narrowband evolution curve and the termination conditions by using the biological continuity of adjacent slices, a kidney tissue segmentation model based on energy minimization was proposed. The model was used to process the 24 sets of standard segmentation test data sets. The segmentation results showed that the average PRA and DSC indices have improved over traditional models, reached 0.961 and 94.68%, respectively, the kidney tissue could be located and segmented efficiently and accurately.