Accurate 3d Kidney Segmentation Using Unsupervised Domain Translation And Adversarial Networks

Computed tomography urography imaging is routinely performed to evaluate the kidneys. Kidney 3D segmentation and reconstruction from urographic images provides physicians with an intuitive visualization way to accurately diagnose and treat kidney diseases, particularly used in surgical planning and outcome analysis before and after kidney surgery. While 3D fully convolution networks have achieved a big success in medical image segmentation, they get trapped in clinical unseen data and cannot be adapted in deferent modalities with one training procedure. This study proposes an unsupervised domain adaptation or translation method with 2D networks to deeply learn urographic images for accurate kidney segmentation. We tested our proposed method on clinical urography data. The experimental results demonstrate our proposed method can resolve the domain shift problem of kidney segmentation and achieve the comparable or better results than supervised learning based segmentation methods.

1. Introduction

Retrograde intrarenal surgery (RIRS) is a minimally invasive surgical procedure to remove renal stones within the kidneys using a flexible ureteroscope. While RIRS has some advantages of less trauma, faster recovery times and fewer complications, its stone-free rate depends on surgical planning of several anatomical 3D kidney shape parameters mainly including the infundibulopelvic angle, infundibular length, and infundibular width [1]. Surgical planning for RIRS critically requires 3D kidney segmentation to accurately determine these parameters. Various kidney segmentation methods are reported in the scientific community. Particularly, deep learning based kidney extraction methods are widely discussed. Fig. 1:

An example of kidney distribution on CT and CTU