Related Experiment Video
Updated: Dec 26, 2025
Co-analysis of Brain Structure and Function using fMRI and Diffusion-weighted Imaging
Published on: November 8, 2012
Multifold Acceleration of Diffusion MRI via Deep Learning Reconstruction from Slice-Undersampled Data
Yoonmi Hong1, Geng Chen1, Pew-Thian Yap1
1Department of Radiology and BRIC, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA.
Abstract:
Diffusion MRI (dMRI), while powerful for characterization of tissue microstructure, suffers from long acquisition time. In this paper, we present a method for effective diffusion MRI reconstruction from slice-undersampled data. Instead of full diffusion-weighted (DW) image volumes, only a subsample of equally-spaced slices need to be acquired. We show that complementary information from DW volumes corresponding to different diffusion wavevectors can be harnessed using graph convolutional neural networks for reconstruction of the full DW volumes. The experimental results indicate a high acceleration factor of up to 5 can be achieved with minimal information loss.
Insights
This study introduces a novel method for faster diffusion MRI scans by reconstructing full images from undersampled slices. Graph convolutional neural networks enable high acceleration factors with minimal data loss.
Area of Science:
- Medical Imaging
- Neuroscience
- Biophysics
Background:
- Diffusion MRI (dMRI) is crucial for analyzing tissue microstructure.
- However, dMRI's long acquisition times limit its clinical applicability.
- Current methods often require lengthy scanning protocols.
Purpose of the Study:
- To develop an effective method for reconstructing diffusion-weighted (DW) MRI volumes from slice-undersampled data.
- To enable significant acceleration of dMRI acquisition without compromising image quality.
- To leverage complementary information across diffusion wavevectors for improved reconstruction.
Main Methods:
- Acquisition of only a subsample of equally-spaced slices instead of full DW image volumes.
- Utilizing graph convolutional neural networks (GCNNs) to process undersampled data.
- Harnessing complementary information from DW volumes with different diffusion wavevectors.
Main Results:
- Successful reconstruction of full DW MRI volumes from slice-undersampled data.
- Demonstration of a high acceleration factor, up to 5x, achievable with the proposed method.
- Minimal information loss observed during the reconstruction process, preserving key microstructural details.
Conclusions:
- The presented method offers a viable solution for accelerating dMRI acquisition.
- GCNNs effectively reconstruct dMRI data from undersampled slices by utilizing multi-vector diffusion information.
- This approach has the potential to significantly reduce scan times in clinical settings.
