A model-based validation scheme for organ segmentation in CT scan volumes.

In this study, we propose a novel approach for accurate 3-D organ segmentation in the CT scan volumes. Instead of using the organ's prior information directly in the segmentation process, here we utilize the knowledge of the organ to validate a large number of potential segmentation outcomes that are generated by a generic segmentation process. For this, an organ space is generated based on the principal component analysis approach using which the fidelity of each segment to the organ is measured. We detail applications of the proposed method for the 3-D segmentation of human kidney and liver in computed tomography scan volumes. For evaluation, the public database of the MICCAI's 2007 grand challenge workshop has been incorporated. Implementation results show an average Dice similarity measure of 0.90 for the segmentation of the kidney. For the liver segmentation, the proposed algorithm achieves an average volume overlap error of 8.7 % and an average surface distance of 1.51 mm.