Print

Print
Stephen is absolutely right that you want to bandpass the data (Fox et al suggest .008 to .09, I think). My comment was more directed at the issue of using an AR(1) correlation structure to further filter the data.
 
The Fischer Z transform is not a Z-statistic, but rather the normalized form of the correlation coefficent (r). As such, it is independent of scaling and passing it to the second level is the same as passing the normalized betas. Thus, its still a random-effects analysis or mixed-effects depending on the terminology. It is not a fixed effects though.

On Thu, Jun 26, 2008 at 12:21 AM, Stephen J. Fromm <[log in to unmask]> wrote:
On Wed, 25 Jun 2008 06:59:54 -0500, MCLAREN, Donald
<[log in to unmask]> wrote:

>(1) If you are doing resting connectivity, I am not aware of any papers
that
>actually remove the "autocorrelation" or prewhiten the data. Although, the
>papers by Mike Fox all remove the confounds of the ventricles and white
>matter that might contain some information about the autocorrelation
issue.
>It is probably worth looking into the effects of adding an AR1 filter to
the
>data.

The papers I've read about connectivity analysis (which mostly aren't
resting state papers), including one by Fox, all do temporal filtering
(usually bandpass) on the data.  Of course, maybe the original poster has
scientific reasons for not wanting to filter out drift.

>(2) In the case of resting connectivity, you probably don't need to
>normalize the data. The reason for this is simply that you are inputting
the
>fisher Z-transform (derived from the r). The intensity difference between
>subject has no effect since the correlations are independent of any
>multiplicative scaling that would be applied in intensity normalization.

Yes, but in that case you might be looking at a fixed-effects measure of
connectivity, not a random effects measure.  (I myself haven't worked out
the details on that, and it depends on the precise definition of
connectivity you use.  But taking Z stats to the second level would lead
to a fixed effects measure, it seems to me.)  Not that there's necessarily
anything wrong with that, but it should be understood as such.


>On Wed, Jun 25, 2008 at 6:22 AM, Stephen J. Fromm <[log in to unmask]>
>wrote:
>
>> On Tue, 24 Jun 2008 14:11:03 +1000, Nicholas Bradfield
>> <[log in to unmask]> wrote:
>>
>> >dear all,
>> >does anyone know where i can obtain an extension/toolbox for matlab
that
>> will allow me to correlate an roi time series with every other voxel and
>> output a summary image that will contain the correlation coefficients at
>> each voxel so that i can enter it into a second level analysis?
>> >any help would be greatly appreciated.
>> >nick
>>
>> In addition to what Volkmar said, I think you need to consider the
>> following points:
>>
>> (1) Filtering.  There's a lot of low-frequency noise in EPI time series,
>> and there's a good chance that that noise will introduce a large amount
of
>> spurious correlation.
>> (2) Intensity normalization.  For one subject, you don't have to worry
>> about anything.  But for many subjects combined in a random effects
>> analysis, you have to decide how and whether to normalize the "seed"
time
>> series.
>>
>> >--
>> >Nicholas Bradfield BBNSc (Hons)
>> >Provisional Psychologist
>> >Candidate for Doctorate of Clinical Neuropsychology
>> >Department of Medicine (Neurosciences)
>> >Monash Medical Centre
>> >5th floor, E Block
>> >246 Clayton Road, Clayton 3168
>> >Victoria, Australia
>> >telephone:+61 3 9594 5543
>> >mobile: 0419 386 255
>> >facsimile: +61 3 9594 6495
>>
>=========================================================================
>>
>
>
>
>--
>Best Regards, Donald McLaren
>=====================
>D.G. McLaren
>University of Wisconsin - Madison
>Neuroscience Training Program
>Tel: (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.
>



--
Best Regards, Donald McLaren
=====================
D.G. McLaren
University of Wisconsin - Madison
Neuroscience Training Program
Tel: (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.