New draft describing quality control (QC) tools in AFNI

Hello-

Here is a new draft describing several new quality control (QC) tools for FMRI in AFNI:

A Set of FMRI Quality Control Tools in AFNI: Systematic, in-depth and interactive QC with afni_proc.py and more, by Paul A Taylor, Daniel Glen, Gang Chen, Robert W Cox, Taylor Hanayik, Chris Rorden, Dylan M Nielson, Justin K Rajendra, Richard C Reynolds. bioRxiv 2024.03.27.586976

It describes both quantitative and qualitative tools. There are a lot of descriptions about the APQC HTML functionality that is created by afni_proc.py. These include interactive features for saving QC ratings and comments, making the QC more easily discussable and sharable.

The QC contains systematic images and text to help quickly sort through many features of the data, and the new interactive buttons also facilitate opening up the AFNI GUI to investigate the data with image panels, graph views and InstaCorr running, as well as opening up NiiVue instances directly within the webpage.

You can even open up multiple APQC HTMLs (like across a data collection), and simply by double clicking the gold text of any image or section, you can make all pages jump to the same location, and then navigate with Ctrl+Tab and Ctrl+Shift+Tab through the different subjects quickly. It is very nice to be able to check quickly for multiple properties across one subject as well as one property across multiple subjects!

We have also added new "localized" QC, because it is vital for you to know about the quality of your data in your particular regions of interest. These include potential warnings about ROI shape (empty voxels, low resolution, shapes that might be very susceptible to partial voluming) and TSNR (low values, dropoff of values within the ROI). We have automatically loaded ROIs for a growing number of templates, and you can specify your own:

There are examples of using the classic gen_ss_review_table.py to do quantitative, scriptable QC across subjects. We have added a new program very recently called gtkyd_check to help you Get To Know Your Data (GTKYD) quickly even before processing starts.

These tools greatly streamline the QC process in important ways. Please check out the draft above for details and examples.

--pt

2 Likes

There is now also a live demo of some of the QC rating+comment and embedded NiiVue functionality, here:
https://afni.github.io/qc-demo-repo/
Many thanks to Taylor Hanayik for setting that up.

--pt

2 Likes