Cyril,
Thanks for your reply, but I don't fully understand. I have 3 PET images
for each subject, each a different condition. (I'm using SPM2)
I think this is what you're saying:
Set up my analysis as usual (ie "multi-subject: conditions &
covariates") and then create 3 Con images for contrasts [-1 0 1], [-1 1
0] & [0 -1 1].
I guess there is no single contrast in this model to find the linear
trend(?) so I use the Con images to set up a new model using the Basic
Model: Simple Regression with correlates [2 0 -2].
(this confuses me as why would you use "2" instead of "1")
Regards,
Joel
cyril pernet wrote:
> Hi Joel
>
>> Hi,
>> How can I look for a linear trend between three (or an odd number of)
>> conditions?
>> EG Conditions A, B, C. Contrast [-1 0 1] finds regions where C>A
>> independ
>> ent of B (I think) but I'd like to find C>B>A.
>> (I'm not sure that masked contrasts or conjunctions will work but I
>> can't
>> quite get my head around that...)
>>
>>
> try a simple correlation analysis where you enter your 3 types of con
> images
> the contrast [2 0 -2] for A, B, C should reveal, if any, a linear
> increase from C to A :-)
>
> best
> cyril
>
>
--
-----------------------------------------------
Joel Dunn
Research Assistant
PET Imaging Centre
Guy's, King's and St Thomas' School of Medicine
Lower Ground Floor, Lambeth Wing,
St Thomas' Hospital, Lambeth Palace Road,
London SE1 7EH UK Tel: 020 7188 7445
Email: [log in to unmask] Fax: 020 7620 0790
-----------------------------------------------
|