help > QA histogram "bug"
Dec 6, 2018  10:12 AM | Jon Dudley
QA histogram "bug"
Hi all,

This does not apply to most projects but...

Starting in version 18.a (or perhaps 17.f), it appears that the before and after denoising histograms of voxelwise connectivity are generated based off 1,000 (random?) voxels from the set of voxels in the MNI-aligned file ~/conn/utils/surf/referenceGM.nii with values of 0.25 or higher.

As a result, if your CONN project is dealing with data normalized to a different standard space (e.g. a neonate atlas) or if you are using a restricted analysis mask then the QA histograms will not be accurately reflecting the distribution of voxelwise correlations in your data before or after denoising. ETA: this also affects carpet plots.

Cheers!
Jon

Threaded View

TitleAuthorDate
QA histogram "bug"
Jon Dudley Dec 6, 2018
Alfonso Nieto-Castanon Dec 9, 2018
Jon Dudley Dec 10, 2018