Dear,
Thank you for the reply!
We looked into the Cbeta.img and found voxels with positive values and
voxels with label 'Not a Number'.
Can someone explain what these values actually mean, are these already the
posterior probabilities?
And what does it mean if voxels are labeled as 'Not a Number'? Can this be
the reason why we find the message ' no suprathreshold clusters ' when
computing the PPMs for some ROI's?
We thought that, by lowering the effect size and posterior probability
threshold to zero, you should be able to compute PPMs for all voxels (or
ROI's) even when the probabilities are very small , or is this a wrong
reasoning?
kind regards,
Freja
----- Original Message -----
From: "Will Penny" <[log in to unmask]>
To: "Freja Gheysen" <[log in to unmask]>
Cc: <[log in to unmask]>
Sent: Tuesday, April 06, 2010 7:03 PM
Subject: Re: [SPM] Bayesian 2nd level analysis
> Dear Freja,
>
> Freja Gheysen wrote:
>> Dear all,
>>
>> When using Bayesian 2nd level analyses with SPM5 and computing PPM’s,
>> I sometimes get the message "no suprathreshold clusters" even when
>> lowering the prior to 0
>
> I think you must mean 'lowering the effect size to zero'.
>
> and lowering the posterior probability
>> threshold to 0 and even though this analysis works perfectly in other
>> contrasts for the same ROI.
>
> Yes - this is odd.
>
> Have a look at the estimated beta image for the contrast. If your contrast
> has a single 1 in eg the third position then its cbeta03.img
> that you need to look at.
>
> If there are positive values in this image they should show up in the
> PPM.
>
> If you still have a problem send me your SPM.mat file & con images (if its
> only a few).
>
> Best, Will.
>
>>
>> Does anyone ever encountered a similar problem or has any idea what
>> could go wrong here? I would greatly appreciate any help on this
>> issue!
>>
>> kind regards, Freja Gheysen
>>
>>
>
> --
> William D. Penny
> Wellcome Trust Centre for Neuroimaging
> University College London
> 12 Queen Square
> London WC1N 3BG
>
> Tel: 020 7833 7475
> FAX: 020 7813 1420
> Email: [log in to unmask]
> URL: http://www.fil.ion.ucl.ac.uk/~wpenny/
>
>
|