AFNI Message Board

Dear AFNI users-

We are very pleased to announce that the new AFNI Message Board framework is up! Please join us at:

https://discuss.afni.nimh.nih.gov

Existing user accounts have been migrated, so returning users can login by requesting a password reset. New users can create accounts, as well, through a standard account creation process. Please note that these setup emails might initially go to spam folders (esp. for NIH users!), so please check those locations in the beginning.

The current Message Board discussion threads have been migrated to the new framework. The current Message Board will remain visible, but read-only, for a little while.

Sincerely, AFNI HQ

History of AFNI updates  

Previous Message Next Message
Message List New Topic Print View
|
nic
July 23, 2014 06:42PM
Hi Rick,

Thank you for your answer and resulting aha moments! Some follow-up questions:

“... applying a maximum censor number or fraction limit in a similar manner to BLOCK ... The key is just having enough (relatively independent) samples for each TENT beta.”

My current thoughts on exclusion decisions based on max censored TRs in a block design (using censor limit of 0.4 mm in afni_proc.py; the prior TR is not censored; voxel size 3x3x3):
1) total censor frac (all TRs) does not exceed 30% (total/task TRs are 115/71 for breath-hold). Is this sensible?
2) task/rest TRs censored: at least 30 TRs need to be left for statistical power while addressing motion artifacts (suggested on the AFNI message board)
3) can apply similar measures to TENT but this increases the # participants to be dropped and don't have 30 TRs in a given tent (except instruction TRs) ... As an example, the breath-hold paradigm has a total of 115 TRs and 71/task when using BLOCK, and 20/21/21/20/20/21/43 TRs in the various tents.

As such, what are "enough (relatively independent) samples for each TENT beta"? With independent, do you mean temporally independent, i.e. checking the pattern/# of affected TRs across vs within blocks for a given tent? How to determine enough?

The participants are older adults with/without cardiovascular disease and their data require a bit more thinking on exclusion criteria while ensuring valid data in the analyses and maximizing the sample.

Following this paper more or less, I'm using normalized breath-hold betas to scale betas of a checker board task and I wasn't sure which breath-hold beta to choose for this analysis, i.e., if the average of one set of betas is better than the other. Shouldn't the betas estimating the shape (iresp) as a measure of cerebrovascular/hemodynamic components somehow make more sense to use than the betas estimating the response model fit (REML)? I'm going back and forth on this as I continue learning about TENT modeling.

The rest of the questions are answered; for some reason I overlooked the breath F-stat.

Indeed, for 'TENTzero(12.5,42.5,8)', the REML output gives 6 betas for the 8 intervals and the iresp output has 13 betas being on the TR grid for 30 sec.

Thank you very much!

Nic
Reply Quote
RSS
Subject Author Posted

Re: Question about TENT/SPLIN - number of parameters

nic July 21, 2014 05:40PM

Re: Question about TENT/SPLIN - number of parameters

rick reynolds July 23, 2014 09:54AM

Re: Question about TENT/SPLIN - number of parameters

nic July 23, 2014 06:42PM

Re: Question about TENT/SPLIN - number of parameters

rick reynolds July 24, 2014 04:36PM

Re: Question about TENT/SPLIN - number of parameters

nic July 30, 2014 07:31PM

Re: Question about TENT/SPLIN - number of parameters

nic July 30, 2014 08:58PM

Re: Question about TENT/SPLIN - number of parameters

rick reynolds July 31, 2014 12:38PM

Re: Question about TENT/SPLIN - number of parameters

nic August 10, 2014 07:06PM