InstaCorr not using selected dataset, strange behavior

AFNI Message Board
1

Hey folks,

I am seeing some strange behavior in InstaCorr. Specifically it appears that it is refusing to use the selected dataset. For background, I am doing some testing with AFNI output, specifically, the pb06* blurred data. I convert it to .nii format (3dAFNItoNIFTI), process it through a toolbox (rapidtide), which produces a nifti output. Interestingly the identifier codes (from 3dinfo) are identical between the processed data and the original dataset - perhaps this is the important detail…

The nifti output can be opened in AFNI, the timesteps are correct, and the data still resembles a brain. However, if I attempt to perform instacorr on this dataset (filtered.nii) AFNI will do one of two things. If the original pb06blurnii file remains in the folder, InstaCorr will quietly use that data, despite it not being selected. This can be seen in the terminal window, which states that it is processing the pb06*blur data, even while the InstaCorr window shows that the filtered data is selected.

The other outcome, if I have isolated the filtered.nii data into a separate folder, is InstaCorr pops up an error window, stating that the ‘Start’ value is too large. I did not enter a start value. The error is the same even if I enter a start value of 0.

Is it possible that the identical identifiers are confusing AFNI?

2

A follow-up, it definitely appears to be an identifier code issue. I used 3dCalc to ‘convert’ the filtered_data to AFNI BRIK format (-a data -expr ‘a’) and this altered the identifier code. I was then able to run InstaCorr with no issues.

I’m still relatively new to AFNI, is there somewhere it would give a warning that multiple ‘identical’ datasets were detected? It did provide warnings that the files had “dimensions altered since AFNI extension was added” but I wasn’t certain what to make of that.

3

This issue comes into play mostly in the graphical interfaces,
such as the afni GUI. The reason datasets have unique
identifiers is because the same dataset name can reside in a
different directory. The afni GUI has to handle such cases, so
datasets get the identifier.

Most programs will not care about this.

When modifying a NIFTI dataset from AFNI, it might be
safer to just throw away the AFNI extension. There are
other issues you could run into this way.

  • rick