Visualizing intestines for diagnostic assistance of ileus based on intestinal region segmentation from 3D CT images

Abstract

This paper presents a visualization method of intestine (the small and large intestines) regions and their stenosed parts caused by ileus from CT volumes. Since it is difficult for non-expert clinicians to find stenosed parts, the intestine and its stenosed parts should be visualized intuitively. Furthermore, the intestine regions of ileus cases are quite hard to be segmented. The proposed method segments intestine regions by 3D FCN (3D U-Net).Intestine regions are quite difficult to be segmented in ileus cases since the inside the intestine is filled with fluids. These fluids have similar intensities with intestinal wall on 3D CT volumes. We segment the intestine regions by using 3D U-Net trained by a weak annotation approach. Weak-annotation makes possible to train the 3D U-Net with small manually-traced label images of the intestine. This avoids us to prepare many annotation labels of the intestine that has long and winding shape. Each intestine segment is volume-rendered and colored based on the distance from its endpoint in volume rendering. Stenosed parts (disjoint points of an intestine segment) can be easily identified on such visualization. In the experiments, we showed that stenosed parts were intuitively visualized as endpoints of segmented regions, which are colored by red or blue.

Small bowel radiology

“…Since the inner cylinder was free of contact with sub-adjacent anatomy, a cylindrical shape constraint applied on this augmented branch guided the network to generate a topologically correct segmentation on both qualitative and quantitative evaluation. Similarly, it has been demonstrated that a 3D U-Net could be trained for fully automated small bowel segmentation in patients with ileus even with sparsely annotated CT datasets [79].…”

Intestine segmentation from CT volume based on bidirectional teaching

“…Up to now, researchers have presented some methods for intestine segmentation task. [4][5][6][7] U-Net and its various optimization methods get good results in organ segmentation tasks. However, those methods just use pixel-level labels for training, which is time-consuming for medical images.…”

A Graph-theoretic Algorithm for Small Bowel Path Tracking in CT Scans

“…There have been research efforts to develop automatic methods for small bowel segmentation for the last decade. [3][4][5][6] Considering the high difficulty of labeling the small bowel, the recent works based on deep learning presented data-efficient methods each by training with sparsely annotated CT volumes, 4 by incorporating a cylindrical shape prior, 5 or by developing an unsupervised domain adaptation technique for the small bowel. 6 While the segmentation is useful for detecting lesions, blockages, and for distinguishing the bowels from adjacent lesions in the mesentery, it may be insufficient to identify the whole structure of the small bowel especially for the parts with lumpy segmentation due to the aforementioned touching issue (Fig.…”