TSNR extreme values

AFNI Message Board
,
1

AFNI version info (afni -ver): AFNI_25.0.00

Hi AFNI team,
I have a 7T dataset with many sessions for the same participant.
For some of the runs I got extremely high TSNR values (average of 280 with a max of 623).
This run happened to be a part of the session that all other scans were aligned to (I don't know if it has any influence, I hope not).
What can cause these values?

Thank you!
Maya

2

Look at the original time series for clues. One cause, for example, can be values are pegged at a ceiling value of 4095 on some Siemens acquisitions. In that case, the data doesn't change, and noise can drop to 0 for part of the run. Alternatively, low maximum values or even 0 values can result in little change over the course of a run with high TSNR.

3

Hi,
Thanks!
The original data looks like the other scans without any extreme values.
What else should I check?

4

Hi Maya,

Please feel free to send the QC report as a pdf to the gmail address afni dot bootcamp.

But it is common for us to see (final) averages over 200, so that does not immediately seem tooooo extreme. Note that the final TSNR averages are higher than those computed initially or after volreg. What are typical averages for your other subjects?

-rick

5

Howdy-

In addition to Daniel and Rick's comments, I would note it is helpful to look where these values are occuring: GM, WM, CSF, other?

Is this resting state or task-based, and is the TSNR in question before or after regression modeling?

As another step to check for potential weirdness in the data, you can view the time series themselves with the Graph Viewer (in the APQC HTML, the GV button above many images), and if there are any suspicions of oddities, then one of my first things to do would be to run InstaCorr and surf around the brain (in the APQC HTML, the IC button above many images).

--pt