Print

Print
For small clusters, eig and mean should be very similar. I haven't seen a reviewer have an issue with using the mean or eig for PPI analyses - and as long as the region is small (see comment below), I wouldn't accept/reject the paper on this choice.

For larger clusters, they can diverge. One issue with eig for larger clusters is the potential that part of the cluster that contains the first eigenvariate is different in different groups leading to potentially comparing different areas across subjects.

Best Regards, Donald McLaren
=================
D.G. McLaren, Ph.D.
Research Fellow, Department of Neurology, Massachusetts General Hospital and
Harvard Medical School
Postdoctoral Research Fellow, GRECC, Bedford VA
Website: http://www.martinos.org/~mclaren
Office: (773) 406-2464
=====================
This e-mail contains CONFIDENTIAL INFORMATION which may contain PROTECTED
HEALTHCARE INFORMATION and may also be LEGALLY PRIVILEGED and which is
intended only for the use of the individual or entity named above. If the
reader of the e-mail is not the intended recipient or the employee or agent
responsible for delivering it to the intended recipient, you are hereby
notified that you are in possession of confidential and privileged
information. Any unauthorized use, disclosure, copying or the taking of any
action in reliance on the contents of this information is strictly
prohibited and may be unlawful. If you have received this e-mail
unintentionally, please immediately notify the sender via telephone at (773)
406-2464 or email.

On Tue, Jan 20, 2015 at 4:13 AM, Dennis van der Meer <[log in to unmask]> wrote:
Hi Mark,
Thanks! I figured there probably wouldn't be a single correct answer to this, but your comments are helpful nonetheless.
I got a tip yesterday that using the first eigenvariate (of the cluster mean) may be a good alternative (i.e. simply adding --eig when applying fslmeants), do you also have any thoughts on this?
Kind regards, Dennis

2015-01-08 10:22 GMT+01:00 Mark Jenkinson <[log in to unmask]>:
Hi,

I donšt think there is a simple, single answer to this and opinions might
vary.  My personal opinion is that using an average means that you will
have a stronger signal related to your task but could reduce other signals
that might be of interest for correlations with other areas, particularly
if your region spans different resting-state networks (which would not be
obvious from the task results alone). Also, averaging across slices can be
an issue with respect to slice timing changes not being accounted for.  So
I suspect that the peak voxel approach is more common for these reasons,
and you might have less trouble with reviewers if you stuck to that, but
other people might have different opinions.

All the best,
        Mark

On 08/01/2015 07:50, "Dennis van der Meer" <[log in to unmask]> wrote:

>Hi there,
>I'm hoping to get feedback on my method for seed region selection for a
>PPI analysis. I first ran a straightforward fMRI analysis for a working
>memory task, and found two significant clusters (thresholded at Z=2.6,
>p=.001). As I wanted to get one homogeneous cluster with strong signal, I
>thresholded the data again at Z=3.7, leaving me with one cluster of 61
>voxels for which I extracted the mean timeseries (in every subjects
>native space) and used this timeseries for the PPI analysis. I chose this
>approach because I thought this might better capture signal/ be more
>robust than basing my seed on a peak voxel, even though the latter is
>suggested in several online tutorials. Can someone explain the pro's and
>cons of either approach?
>Thanks! Kind regards, Dennis