Wilson Nicole A, Park Hyung-Soon, Lee Kyoung Soub, Barron Lauren K, Warner Brad W
Division of Pediatric Surgery, St Louis Children's Hospital, St Louis, MO; Department of Surgery, Washington University in St Louis, St Louis, MO.
Department of Mechanical Engineering, Korea Advanced Institute of Science and Technology, Seoul, Korea.
J Am Coll Surg. 2017 Aug;225(2):266-273.e1. doi: 10.1016/j.jamcollsurg.2017.04.009. Epub 2017 Apr 23.
Bowel length is an important factor in the management of patients with intestinal failure and short bowel syndrome. However, it is difficult to accurately measure the length of the small intestine. Our aim was to demonstrate the feasibility of a dedicated software algorithm for calculating small intestine length based on magnetic resonance enterography (MRE) images.
A custom algorithm for image thresholding, wall identification, segmentation, and path extraction was designed and implemented. Algorithm output included the Euclidean distance from the pylorus to the terminal ileum and coordinates of points along that route. For validation, 10 C57BL/6 mice underwent bowel preparation before MRE. After sacrifice, the entire small bowel was removed and the ex vivo small intestine length was measured in a standardized manner. Calculated and ex vivo length measurements were compared.
Mean absolute difference and mean percentage difference between the 2 modalities were 1.8 ± 3.8 cm (p = 0.24) and 9.4% ± 6.0%, respectively.
This study demonstrates the feasibility of a dedicated, computer-vision-based algorithm that can reliably and accurately calculate the length of the small intestine using MRE images. A validated, noninvasive approach to measuring small intestine length that can be implemented using available clinical imaging can have a dramatic clinical impact on management in patients with short bowel syndrome, Crohn's disease, and other intestinal disorders.
肠长度是肠衰竭和短肠综合征患者管理中的一个重要因素。然而，准确测量小肠长度很困难。我们的目的是证明一种基于磁共振肠造影（MRE）图像计算小肠长度的专用软件算法的可行性。
设计并实施了一种用于图像阈值处理、肠壁识别、分割和路径提取的定制算法。算法输出包括从幽门到回肠末端的欧几里得距离以及沿该路径的点的坐标。为了进行验证，10只C57BL/6小鼠在进行MRE之前进行了肠道准备。处死后，取出整个小肠并以标准化方式测量离体小肠长度。比较计算得出的长度和离体长度测量值。
两种测量方式之间的平均绝对差值和平均百分比差值分别为1.8±3.8厘米（p = 0.24）和9.4%±6.0%。
本研究证明了一种基于计算机视觉的专用算法的可行性，该算法可以使用MRE图像可靠且准确地计算小肠长度。一种经过验证的、可使用现有临床成像技术实施的非侵入性测量小肠长度的方法，可能会对短肠综合征、克罗恩病和其他肠道疾病患者的管理产生重大临床影响。
