I was curious what sort of automated dwi QC pipelines are available? I’m aware that there are several tool-dependent (eddyQC) and preproc-pipeline-dependent (dmriprep-viewer) tools, but was curious if there has been development on either ones that focus on raw data or ones that are more pipeline-flexible.
You can use the classifier described here: An analysis-ready and quality controlled resource for pediatric brain white-matter research | Scientific Data
It predicts whether or not a DWI image will pass expert quality control ratings based on automated image quality metrics. The most informative metrics was called average neighbor correlation (described here: Differential Tractography as a Track-Based Biomarker for Neuronal Injury - PMC).
Ah yes, I was wondering if there was an implementation of Yeh et al. 2019’s QC metrics outside qsiprep. Thank you for bringing it to my attention!
No problem! I think that the classifier uses both raw and preprocessed image quality, so it might require QSIPrep, I would have to confirm that.
The classifier in that paper used only outputs of qsiprep (including Yeh’s NDC) for QC, and not raw images.
We have also begun some work to incorporate some DWI-specific metrics into mriqc - for raw data - but that’s at a rather preliminary stage.
Yes, I should have clarified: All of the metrics are prepared by QSIPrep, but the metrics relate to both the raw and preprocessed images (see the table here: https://www.nature.com/articles/s41597-022-01695-7/tables/2)
There is also DTIPrep which operates on raw data, but I would think that is outdated and no longer maintaned.
Hi @ajschadler ,
Thank you so much!
Would you also happen to know if there are any python implementations of the QC metrics mentioned in Yeh 2019?
I am aware that qsiprep has an implementation using DSI studio.
nevermind, found your implementation
Don’t hesitate, if you have any issues.
There are a couple of new features ready in the coming weeks.
Would you happen to have a link to your dmriqc ISMRM poster?
Here it is theaud-etal-ismrm22.pdf
Have a good day