Question Posted 06/10/16:
Dear Experts,

I see from prior helplist questions that the .nii header can be unreliable for determining TR (many I have tried say TR = 0). In my IDA data collection, when I click on each subject's image description, I see that some participants have TR = 3000 and others have TR = 2250. Is that information reliable? Or should I always download the dicom as well as the .nii files to verify? I'm also wondering if there is a way to validly combine subjects with differing TR in one analysis. Thank you for your assistance.
Response posted 06/10/16 by ADNI MRI Core:
Thank you for your question to the ADNI MRI Core.

We always recommend using the raw DICOM image headers to determine actual protocol values. For the most part those values are very reliable and is what is checked for protocol compliance by the MRI core.

What type of analysis are you going that would be combining images with different TRs. We have verified that the 3D T1 scans are comparable across vendors, but have not combined other scans as those are analyzed individually.


Response posted 06/10/16 by Jessica:
Thank you for your response. My question was regarding the resting state fMRI data. I realize the TR comes in during individual subject preprocessing, but in the past I have always had participant samples within which all individuals have the same TR value. In the ADNI dataset, at least based on the information by subject that is available by clicking on "resting state fMRI" under "image description" in a searched image collection, it appears that TR may be 3000ms or 2250ms (that I have noted so far). I will definitely check to see if those differences are borne out with the dicom header information check. If there are differences, will that impact group-level analyses in any way? Thank you again.