DPARSF outputs/results for second level analyses

Submitted by tsheng75 on Sat, 01/05/2013 - 08:02

Hi guys,

I am a relatively new user of DPARSF/A and I am wondering which output/results files are the most appropriate to feed into second level group analyses? For example, does one use the 'ALFFmap', 'mALFFmap', or 'zALFFmap' files from each individual subject for a group-level t-test or regression? Some of the online documentation (albeit for previous versions of DPARSF) seem to suggest using the 'm' or 'z' files, but is that appropriate? In task-based fMRI, the con* images are fed into second level (and they are neither mean-normed nor z-scored). Please advise, thanks!

Best,
Tong

Log in or register to post comments
3956 reads

Submitted by YAN Chao-Gan on Sun, 01/13/2013 - 11:58 Permalink

Re: DPARSF outputs/results for second level analyses

Hi,

ALFFmap is the one neither mean-normed nor z-scored.

However, given there are big baseline variation across subjects, some standardization procedures seems necessary for this analysis (we are doing a study on this issue, but a half way yet). Actually, the results of m (divide by mean) or z (subtract the mean and then divide by the STD) files are pretty similar. Zuo et al., 2010 Neuroimage reported the z-score transformation could improve normality across subjects, thus "z" files are not bad choices for group analyses.