Fast Acquisition for Diffusion Tensor Tractography

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

*Corresponding author for this work

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
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

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

Fingerprint

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

Cite this