Roundup of denoising and unringing algorithms from ISMRM

nipreps / dmriprep

dMRIPrep is a robust and easy-to-use pipeline for preprocessing of diverse dMRI data. The transparent workflow dispenses of manual intervention, thereby ensuring the reproducibility of the results.

https://www.nipreps.org/dmriprep

Apache License 2.0

63 stars 24 forks source link

Roundup of denoising and unringing algorithms from ISMRM #132

Open josephmje opened 3 years ago

josephmje commented 3 years ago

Patch2Self was brought to my attention by a colleague and was also announced on Neurostars yesterday: https://neurostars.org/t/patch2self-self-supervised-denoiser-for-diffusion-mri-data-let-your-old-data-live-a-new-life/17671

The paper explaining more can also be found here: https://arxiv.org/pdf/2011.01355.pdf They include comparisons with MP-PCA.

The method is also implemented in dipy.

EDIT: I'm editing this issue to make note of some of the new methods and comparisons introduced at ISMRM.

oesteban commented 3 years ago

Looks incredible. Although, I have to say that something presented as "magic" makes me nervous.

josephmje commented 3 years ago

Denoising

MP-PCA: implemented in dipy, MRtrix and DESIGNER
MP-PCA non-local patch:
Patch2Self: implemented in dipy
using NORDIC in combination with Patch2Self?

Unringing

subvoxel shifts
RPG correction: implemented in DESIGNER

ShreyasFadnavis commented 2 years ago

Hi @josephmje and @oesteban ! I only saw this now :) I am happy to help with Patch2Self if this is still of interest! Also nothing magical in Patch2Self really...

Let me know what I can help with! I am also working on NORDIC + Patch2Self --> Do you guys have a Pythonic Implementation of NORDIC? Here is an example of the excellent potential it has: