Fast Acquisition for Diffusion Tensor Tractography

Omri Leshem, Nahum Kiryati, Michael Green, Ilya Nelkenbaum, Dani Roizen, Arnaldo Mayer

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Diffusion tensor tractography is a powerful method for in-vivo white matter mapping. Its implementation involves long scanning sessions to capture local diffusion orientations, followed by tedious post-processing to generate accurate tracts. While some initial research was conducted to reduce the number of required gradient directions, the current state-of-the-art still considers acquisition protocol acceleration and automatic tract segmentation as two separate tasks. We aim at optimizing the whole workflow, from acquisition to tract segmentation. We propose a collaborative neural framework for diffusion-encoding color map denoising and white matter tract segmentation. It generates high-quality white matter tracts using DWI acquired for a small number of diffusion-encoding gradient directions (GDs), thus minimizing acquisition and post-processing time. The proposed method is first validated on the high-angular resolution (270 GDs) HCP dataset using a novel spherical k-means method to select a subset of 16 quasi-uniformly distributed GDs. Further validation is provided for a prospective clinical dataset of 10 cases acquired at both 16 and 64 GDs. Encouraging experimental results are obtained using several state-of-the-art neural architectures and training loss functions.

Original languageEnglish
Title of host publicationComputational Diffusion MRI - 14th International Workshop, CDMRI 2023, Held in Conjunction with MICCAI 2023, Proceedings
EditorsMuge Karaman, Remika Mito, Elizabeth Powell, Francois Rheault, Stefan Winzeck
PublisherSpringer Science and Business Media Deutschland GmbH
Pages118-128
Number of pages11
ISBN (Print)9783031472916
DOIs
StatePublished - 2023
Event14th International Workshop on Computational Diffusion MRI, CDMRI 2023 held in conjunction with 26th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2023 - Vancouver, Canada
Duration: 8 Oct 20238 Oct 2023

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume14328 LNCS

Conference

Conference14th International Workshop on Computational Diffusion MRI, CDMRI 2023 held in conjunction with 26th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2023
Country/TerritoryCanada
CityVancouver
Period8/10/238/10/23

Keywords

  • Deep Learning
  • Denoising
  • Diffusion Tensor MRI
  • Segmentation
  • Tractography

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • General Computer Science

Fingerprint

Dive into the research topics of 'Fast Acquisition for Diffusion Tensor Tractography'. Together they form a unique fingerprint.

Cite this