Hello SPM Users,
in SPM 8 , defining an explicit mask does skip the threhsold-masking,or do we have to set the variable to -inf in order to tell spm not to use threhsold-masking (and use the explicit mask instead).
2- How can we define HPF Cutoff, rather than using the default value 128
Thank you
Marouane Nassim,B. Eng, Msc
Centre de recherche du CHU Ste-Justine
3175 Côte-Sainte-Catherine
Bloc 8, 2ième étage, bureau 2806
Montréal, Québec, Canada, H3T 1C5
Tél./phone: 514-345-4931, ext. 4046 ou 5702
Fax: 514-345-4801
email: [log in to unmask]
________________________________________
From: SPM (Statistical Parametric Mapping) [[log in to unmask]] on behalf of MCLAREN, Donald [[log in to unmask]]
Sent: October 28, 2011 11:17 AM
To: [log in to unmask]
Subject: Re: [SPM] 回复:Re: [SPM] ANOVA contrast
See inline responses.
Best Regards, Donald McLaren
=================
D.G. McLaren, Ph.D.
Postdoctoral Research Fellow, GRECC, Bedford VA
Research Fellow, Department of Neurology, Massachusetts General Hospital and
Harvard Medical School
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.
2011/10/28 <[log in to unmask]>:
> As for the conjunction Dr Mclaren mentioned in his response, I wonder
> whether one can perform conjunction on two separate one-sample t test models
> instead of a single 2nd level ANOVA or factorial model because one might
> prefer partitioned errors to pooled errors when performing a multiway ANOVA
> analysis.
If there are no violations of spherecity, then pooling is acceptable.
If you can correct for violations of spherecity, then pooling is also
acceptable. In the case of between subject effects -- there is no
spherecity violations and thus pooling is very common. The more
subjects, the better the variance estimates will be. This works across
subjects. In imaging, the data is also pooled across voxels, but the
same statements hold. The idea behind pooling across voxels is that
you get a better measure of the variance-covariance between levels of
different factors. However, the ability to correct for violations of
non-spherecity in a NxM within-subject design is dependent on the
specified covariance stucture and the actual covariance structure.
Differences between these would lead to uncorrected violations in
spherecity. The effect these violations is something that I am
currently testing.
>
>
>
> For example, let's say that we have a 2 by 2 within-subject design with
> factors and corresponding levels A1 A2 and B1 B2. And I am interested in the
> overlapping regions activated in A1-A2 and B1-B2, i.e. the conjunction
> of both main effects.
>
I would personally use two models to test this conjunction (or GLM
Flex -- http://www.nmr.mgh.harvard.edu/harvardagingbrain/People/AaronSchultz/GLM_Flex.html).
>
>
> So in 1st level GLM with regressors A1B1 A1B2 A2B1 A2B2, uni-dimension
> F-contrasts or T-contrasts ( [1 1 -1 -1] and [1 -1 1 -1]) can be drawn. On
> second level, I plan to build two seperate GLMs, each of which models the
> main effects of A and B respectively ( thus the error is partitioned over
> different effects of interest). Since I only have 1 contrast of interest
> for each main effect per subject, one-sample t test might seem the most
> straightforward choice.
>
>
>
> Then the problem arises because I don't know how to perform conjunction
> analysis of both main effects. Any ideas or comments?
Step1: Threshold the image and save it (I'd suggest peak_nii as it
creates several nice images for viewing)
Step2: In imcalc, use the two thresholded images (or *clusters.nii
from peak_nii) with the following equation: (i1>0)+2*(i2>0)
The areas of the new image that have a value of 3 will be in both
effects, the areas that have a value of 1 will be from the contrast of
the first imade, the areas with a value of 2 will be for the contrast
for the second image. The only reason for multiplying by 2 is to parse
out effects isolated in contrast of image 1 from those isolated in
contrast of image 2.
Hope this helps.
>
>
>
> Many thanks
>
>
>
> Ce
>
>
>
> ----- 原始邮件 -----
> 发件人:MCLAREN, Donald <[log in to unmask]>
> 收件人:[log in to unmask]
> 主题:Re: [SPM] ANOVA contrast
> 日期:2011-10-27 12:05:04
>
> As far as I know there is not a single contrast that will test that
> relationship. The reason being is as follows:
> G1>G2 --> 1 -1
> G2>G3 --> 1 -1
> Now combining these two would produce 1 0 -1 (G1>G3). G2 has no
> effect. Because you are looking for difference, the mean of the
> contrast has to be 0.
>
> It seems like the better way to ask the question is: where is G1>G2 AND
> G2>G3.
> Use 2 contrasts and then either:
> (1) form a conjunction in SPM; or
> (2) create 2 thresholded maps, make them into binary images, and then
> sum the images. Regions where the value is 2, then G1>G2 AND G2>G3.
>
> Best Regards, Donald McLaren
> =================
> D.G. McLaren, Ph.D.
> Postdoctoral Research Fellow, GRECC, Bedford VA
> Research Fellow, Department of Neurology, Massachusetts General Hospital and
> Harvard Medical School
> 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 Wed, Oct 26, 2011 at 3:28 PM, Shashwath Meda <[log in to unmask]>
> wrote:
>> Hi All - What would be the contrast to identify voxels that show a
>> group1>group2>group3 directional effect in a one-way ANOVA design?
>>
>> --
>> with regards,
>>
>> Shashwath Meda
>>
>> ***********************************************
>> Imaging Genetics Analyst III
>> 515-F Light Hall
>> Vanderbilt University Medical Center
>> Cell: (313) 505-6847
>> Off: (615) 875-3086
>>
>> ***********************************************
>>
>
|