When I lower the threshold I get more activation but the pattern is the same, where everything is really scattered.
From: Chris Watson [[log in to unmask]]
Sent: Wednesday, March 30, 2011 1:28 PM
To: Meera Paleja
Cc: [log in to unmask]
Subject: Re: [SPM] "dalmation" pattern in data
What does it look like when you lower the threshold, e.g. to p<0.01?
Since your data are unsmoothed, there are a lot more voxels/resels to
deal with and thus a bigger multiple comparisons issue.
Meera Paleja wrote:
> It is normalized. I haven't checked it without normalization, but I'm not sure how normalization would contribute to this random scattered pattern of activity...?
> The other contrasts show similar activation patterns but they're not identical. I did try putting a +1 in front of my conditions and I got activity everywhere as expected.
> From: Chris Watson [[log in to unmask]]
> Sent: Tuesday, March 29, 2011 6:23 PM
> To: Meera Paleja
> Cc: [log in to unmask]
> Subject: Re: [SPM] "dalmation" pattern in data
> Is your data normalized? If so, have you checked to see what it looks
> like without normalization?
> What about other contrasts? Do they look similar? Have you tried, say, a
> contrast of just a +1 over each key press? Then you should at least see
> something in motor cortex.
> Meera Paleja wrote:
>> Attached is a figure for one subject, but this is a fairly consistent pattern across all my subjects for all contrasts.
>> This was a high-resolution fMRI experiment that used a medial temporal lobe shim. The task involved seeing an image of an object on the floor of a computerized room. This was followed by the recognition phase, where subjects saw an image with two objects, one in the same location as the original and another in a different location and were asked to indicate which is in the same location using a key press.
>> The image file attached here is a contrast comparing the recognition phase with a baseline letter task, where participants are asked to indicate whether a displayed letter is a vowel or consonant using a key press.
>> From: Chris Watson [[log in to unmask]]
>> Sent: Tuesday, March 29, 2011 5:03 PM
>> To: Meera Paleja
>> Cc: [log in to unmask]
>> Subject: Re: [SPM] "dalmation" pattern in data
>> Could you attach a screenshot? And explain the experiment?
>> Meera Paleja wrote:
>>> Hi SPM-ers,
>>> I am seeing odd activation patterns in my data, specifically "dalmation-like" activation, where one or two voxel "clusters" are activated throughout the brain, with very few clusters larger than that. Does anyone know what might account for this? I haven't smoothed the data as I am interested in very small regions (hippocampal subregions)- could the lack of smoothing be a reason for these strange patterns?
>>> Meera Paleja, B.Sc. (Hons.), M.A.
>>> PhD. Student, Psychological Science
>>> Brain Imaging and Memory Lab
>>> Department of Psychology
>>> Ryerson University
>>> Phone: (416) 979-5000 ex. 2192
>>> email: [log in to unmask]
>>> “Perhaps the most fascinating and mysterious universe of all is the one within us.” -- Carl Sagan