Hi
There are numerous issues when combining data of variable length - see e.g. some of the FBIRN papers. The release version of melodic will only allow
you to use data with the same number of vols in the same analysis and while you can 'hack' it to work (e.g. by doing what you suggest and running separate ICAs) I suggest you simply extract 180 vols from the 260 data and feed all that 180 vol data into a single analysis.
hth
Christian
On 27 Jul 2011, at 07:43, Krista Wisner wrote:
> Hello,
>
> I recently acquired resting-state data from a longitudinal study and was planing on running melodic temp-concat followed by dual-regression. In this data set, there seems to be a mix of scan lengths although the TR is constant across all scans. Some subjects have 180 vol for both scans, some have 260 vols for both scans, and others have 180 vol for one scan and 260 for the other.
> I would of course like to maximize my N, but not at the cost of my methodology. I see in the Biswal et al. 2010 mutli-site paper that scans of different lengths can indeed be included in the same melodic. That being said, I was curious what considerations, alterations, or additional effects must be added to the analysis so that the difference in length across scans is minimized, e.g. a confound variable during dual regression to account for scan length differences. But I presume there is more involved than that, thus I ask.
>
> Thank you in advance,
> Krista Wisner
|