Hi
Together with Andre and Nikos, we have collected some potentially very interesting results via an fMRI study on task switching.
The issue is this. When we do group analyses with cluster correction, 3 very large blobs are found in contrasting cond 1 with cond 2.
However, in playing with fslview it is clear that within these blobs there are more interesting localised islands of activity.
My dilemma is this if we simply report the cluster-corrected blobs that is okay but what I really want to discuss are these other ROIs.
So how do I get this published and satisfy the statisitcal reviewer?
I cannot do a simple a priori ROI analysis because some of the conditions are relatively novel and we cant predict where the ROIs might be.
We can do this via email or I can talk about this at YNiC next week perhaps?
Philip. ******************************************************************** Philip Quinlan E-Mail: ptq1@york.ac.uk Department of Psychology FAX: (01904) 323181 The University of York Tel: (01904) 320000 Ext. 3135 Heslington Direct : (01904) 323135 York YO10 5DD U.K. ********************************************************************
This is a problem that others have struggled with as well.
I agree that choosing to repeat the same statistical analysis in the clustered blobs to see if there are subdivisions might be hard to argue.
Have you tried using MELODIC rather than the standard FSL method?
It may be that the subdivisions explain different aspects of the variance and therefore a principal components approach may tease out the subcomponents quite naturally.
It may also be the case that we should try out the method in A Multimodal Approach to Representational Similarity Analysis by Li Su et al. http://www.mrc-cbu.cam.ac.uk/people/nikolaus.kriegeskorte/rsa.html
Gary
Gary Green
York Neuroimaging Centre The Biocentre York Science Park Innovation Way Heslington York YO10 5DG
http://www.ynic.york.ac.uk https://www.ynic.york.ac.uk/about-us/people/ggrg
tel. +44 (0) 1904 435349 PA - Claire Fox : +44 (0) 1904 435329 or Claire.Fox@ynic.york.ac.uk
fax +44 (0) 1904 435356 mobile +44 (0) 788 191 3004
philip quinlan wrote:
Hi
Together with Andre and Nikos, we have collected some potentially very interesting results via an fMRI study on task switching.
The issue is this. When we do group analyses with cluster correction, 3 very large blobs are found in contrasting cond 1 with cond 2.
However, in playing with fslview it is clear that within these blobs there are more interesting localised islands of activity.
My dilemma is this if we simply report the cluster-corrected blobs that is okay but what I really want to discuss are these other ROIs.
So how do I get this published and satisfy the statisitcal reviewer?
I cannot do a simple a priori ROI analysis because some of the conditions are relatively novel and we cant predict where the ROIs might be.
We can do this via email or I can talk about this at YNiC next week perhaps?
Philip.
Philip Quinlan E-Mail: ptq1@york.ac.uk Department of Psychology FAX: (01904) 323181 The University of York Tel: (01904) 320000 Ext. 3135 Heslington Direct : (01904) 323135 York YO10 5DD U.K.
-- ynic-users mailing list ynic-users@ynic.york.ac.uk https://www.ynic.york.ac.uk/mailman/listinfo/ynic-users
Given that my MRI experience is limited, I haven't really got a clue what you're talking about, but it sounds interesting and potentially crucial, so I think it would be good for you to speak at YNiC, and there may be some good suggestions that come out of it.
If i'm in the minority in this thinking, you could always just talk me through it to satisfy my own curious mind.
Garreth
Hi
Together with Andre and Nikos, we have collected some potentially very interesting results via an fMRI study on task switching.
The issue is this. When we do group analyses with cluster correction, 3 very large blobs are found in contrasting cond 1 with cond 2.
However, in playing with fslview it is clear that within these blobs there are more interesting localised islands of activity.
My dilemma is this if we simply report the cluster-corrected blobs that is okay but what I really want to discuss are these other ROIs.
So how do I get this published and satisfy the statisitcal reviewer?
I cannot do a simple a priori ROI analysis because some of the conditions are relatively novel and we cant predict where the ROIs might be.
We can do this via email or I can talk about this at YNiC next week perhaps?
Philip.
Philip Quinlan E-Mail: ptq1@york.ac.uk Department of Psychology FAX: (01904) 323181 The University of York Tel: (01904) 320000 Ext. 3135 Heslington Direct : (01904) 323135 York YO10 5DD U.K.
-- ynic-users mailing list ynic-users@ynic.york.ac.uk https://www.ynic.york.ac.uk/mailman/listinfo/ynic-users
-
Garreth Prendergast -
Gary Green -
philip quinlan