Segmentation and volumetric measurement of renal cysts and parenchyma from MR images of polycystic kidneys using multi-spectral analysis method

Abstract

For segmentation and volume measurement of renal cysts and parenchyma from kidney MR images in subjects with autosomal dominant polycystic kidney disease (ADPKD), a semi-automated, multi-spectral anaylsis (MSA) method was developed and applied to T1- and T2-weighted MR images. In this method, renal cysts and parenchyma were characterized and segmented for their characteristic T1 and T2 signal intensity differences. The performance of the MSA segmentation method was tested on ADPKD phantoms and patients. Segmented renal cysts and parenchyma volumes were measured and compared with reference standard measurements by fluid displacement method in the phantoms and stereology and region-based thresholding methods in patients, respectively. As results, renal cysts and parenchyma were segmented successfully with the MSA method. The volume measurements obtained with MSA were in good agreement with the measurements by other segmentation methods for both phantoms and subjects. The MSA method, however, was more time-consuming than the other segmentation methods because it required pre-segmentation, image registration and tissue classification- determination steps.