eduzhai > Applied Sciences > Engineering >

Enhancing Fiber Orientation Distributions using convolutional Neural Networks

  • king
  • (0) Download
  • 20210506
  • Save

... pages left unread,continue reading

Document pages: 29 pages

Abstract: Accurate local fiber orientation distribution (FOD) modeling based ondiffusion magnetic resonance imaging (dMRI) capable of resolving complex fiberconfigurations benefits from specific acquisition protocols that sample a highnumber of gradient directions (b-vecs), a high maximum b-value(b-vals), andmultiple b-values (multi-shell). However, acquisition time is limited in aclinical setting and commercial scanners may not provide such dMRI sequences.Therefore, dMRI is often acquired as single-shell (single b-value). In thiswork, we learn improved FODs for commercially acquired MRI. We evaluatepatch-based 3D convolutional neural networks (CNNs)on their ability to regressmulti-shell FOD representations from single-shell representations, where therepresentation is a spherical harmonics obtained from constrained sphericaldeconvolution (CSD) to model FODs. We evaluate U-Net and HighResNet 3D CNNarchitectures on data from the Human Connectome Project and an in-housedataset. We evaluate how well each CNN model can resolve local fiberorientation 1) when training and testing on datasets with the same dMRIacquisition protocol; 2) when testing on a dataset with a different dMRIacquisition protocol than used to train the CNN models; and 3) when testing ona dataset with a fewer number of gradient directions than used to train the CNNmodels. Our approach may enable robust CSD model estimation on single-shelldMRI acquisition protocols with few gradient directions, reducing acquisitiontimes, facilitating translation of improved FOD estimation to time-limitedclinical environments.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×