Coarse-to-fine Kidney Segmentation Incorporating Abnormality Detection and Correction

Abstract

In this paper, we propose and validate a coarse-to-fine kidney segmentation method from Computed Tomography (CT) images, i.e., predicting a coarse label based on the entire image and a fine label based on the coarse segmentation and cropped image patches. A key difference between the two stages lies in how input images were preprocessed. For the coarse segmentation, each 2D CT slice was normalized to be of the same image size (but possible different pixel size), and for the fine segmentation, each 2D CT slice was first resampled to be of the same pixel size and then cropped to be of the same image size. In other words, the image inputs to the coarse segmentation were 2D CT slices of the same image size whereas those to the fine segmentation were 2D CT patches of the same image size as well as the same pixel size. In addition, we designed an abnormality detection method based on component analysis between two stages and used another 2D convolutional neural network to correct the abnormality regions. A total of 168 CT images were used to train the proposed framework and evaluations were conducted qualitatively on another 42 testing images. The proposed method showed promising results and achieved an average DSC of 94.53 % on the testing data.