Semi-automatic segmentation of renal cortex and medulla based on dynamic magnetic resonance images

摘要

Image-based functional analysis of the kidney plays an increasingly important role in the clinical application. Efficiently and accurately segment the renal cortex and the medulla in MR images will be very helpful for doctor's clinical diagnosis. This paper proposed a semi-automatic segmentation method of renal cortex and medulla based on dynamic magnetic resonance (MR) images of pigs. This segmentation method includes the 3D registration, subtraction and the amendment by using morphology method. The accuracy and precision of the segmentation method on dynamic MR images were evaluated by 11 model pigs. The correlation of renal cortex segmentation result between our method and manual method from doctor in 15 kidneys was 0.9874, while the correlation of signal intensity of the segmented result was 0.9901. Besides, it took less than 1 minute to segment three phases of renal cortex, compared to 20 minutes for manual segmentation. Moreover, by using our semi-automatic segmentation method, we can extract the renal cortex from non-contrast scan images.