Untangling the Small Intestine in 3D cine-MRI using Deep Stochastic TrackingDownload PDF

Feb 10, 2021 (edited Feb 22, 2021)MIDL 2021 Conference SubmissionReaders: Everyone
  • Keywords: Deep learning, cine-MRI, centerline extraction, small intestine, motility
  • TL;DR: We extract small intestine centerlines in MR images by stochastically sampling a CNN-based directional classifier
  • Abstract: Motility of the small intestine is a valuable metric in the evaluation of gastrointestinal disorders. Cine-MRI of the abdomen is a non-invasive imaging technique allowing evaluation of this motility. While 2D cine-MR imaging is increasingly used for this purpose in both clinical practice and in research settings, the potential of 3D cine-MR imaging has been largely underexplored. In the absence of image analysis tools enabling investigation of the intestines as 3D structures, the assessment of motility in 3D cine-images is generally limited to the evaluation of movement in separate 2D slices. Hence, to obtain an untangled representation of the small intestine in 3D cine-MRI, we propose a method to extract a centerline of the intestine, thereby allowing easier (visual) assessment by human observers, as well as providing a possible starting point for automatic analysis methods quantifying peristaltic bowel movement along intestinal segments. The proposed method automatically tracks individual sections of the small intestine in 3D space, using a stochastic tracker built on top of a CNN-based orientation classifier. We show that the proposed method outperforms a non-stochastic iterative tracking approach.
  • Registration: I acknowledge that publication of this at MIDL and in the proceedings requires at least one of the authors to register and present the work during the conference.
  • Authorship: I confirm that I am the author of this work and that it has not been submitted to another publication before.
  • Paper Type: methodological development
  • Source Latex: zip
  • Primary Subject Area: Application: Radiology
  • Secondary Subject Area: Learning with Noisy Labels and Limited Data
11 Replies