Sorry for the late response to this thread.
One quite important thing to consider is the difference in b-values between
both groups. If the differences in b-values were achieved by changing the
diffusion-encoding gradient timing, then your ADC along each direction is
going to be probing different spatial scales (the longer you allow the water
to diffuse, the longer the spatial scale). For example, even in water along
very oriented fibers, if you don't allow the water to diffuse far enough to
reach the fiber barriers along the perpendicular direction, the diffusion
(at that scale) would look isotropic.
The diffusion timing is normally controlled by TE; so a different TE would
normally indicate different diffusion timing. Do you have the same TE in
both groups? Do you know the timing?
Even in the event that the diffusion timing was exactly the same for both
groups, diffusion is linear only in a first approximation. That means that
if you were to measure the DWI for the same subject at two different
b-values along the same direction, the computed ADC won't be exactly the
same (even if the noise were negligible). Therefore, your values for MD and
FA would be compromised.
In principle, it is a very BAD IDEA to mix data collected at two different
b-values when using a linear model analysis.
In your case, I would follow David Gutman's suggestion and include some
parameter in your analysis across groups. (However, given that you only
have 3 scans with the second set of parameters, I don't know if you have
sufficient statistical power to control for the different protocols).
HTH,
-Pablo
On Fri, 17 Apr 2009 22:23:23 -0700, liang wang <[log in to unmask]> wrote:
>Hi Matt and David,
>
>Thanks for your prompt answer and suggestion.
>
>Liang
>
>2009/4/17 David Gutman <[log in to unmask]>
>
>> One bigger question is what's the difference between the first group of
>> patients and the second group of patients/subjects you are trying to group
>> together... couldn't you simply include some sort of parameter in your model
>> (e.g. scanner 1/scanner 2) and correct for that in whatever the model is?
>> Even if data is collected on the same scanner but at different locations,
>> it's still likely there's some sort of systematic "difference" that could in
>> theory bias the results..
>>
>>
>> dg
>>
>> On Fri, Apr 17, 2009 at 1:58 PM, Matt Glasser <[log in to unmask]> wrote:
>>
>>> I am not sure what the acquisition number is referring to. It takes
>>> values of 5, 1006, 6, and 5. All the scans say the number of ave is 2,
>>> however. Perhaps someone more familiar with Philips scanners knows better.
>>> Regarding the TR, that should not make any significant difference, unless
>>> your scanner produces artifacts at low TR (a problem I am not aware of
>>> effecting Philips scanners).
>>>
>>>
>>>
>>> Peace,
>>>
>>>
>>>
>>> Matt.
>>>
>>> gg
>>> ------------------------------
>>>
>>> *From:* FSL - FMRIB's Software Library [mailto:[log in to unmask]] *On
>>> Behalf Of *liang wang
>>> *Sent:* Friday, April 17, 2009 12:39 PM
>>> *To:* [log in to unmask]
>>> *Subject:* Re: [FSL] Effect of different DTI protocol on results
>>>
>>>
>>>
>>> Hi Matt,
>>>
>>> Thanks for your great answer. All data were collected with 3T scanner.
>>> As you know, later 3 sequences indeed resulted in some same parameters
>>> (2*2*2 isotropic, the same bvalue 800). However, these sequence had
>>> different TR and acquisition number. I am not sure if two different measure
>>> have an effect on DTI results. Based on your comments, can I merge these
>>> data collected in later 3 sequence.
>>>
>>> Thanks again,
>>> Liang
>>>
>>> 2009/4/17 Matt Glasser <[log in to unmask]>
>>>
>>> Looks like one protocol has voxels of 2.3x2.3x2.0mm with 0.5mm gap while
>>> the others are 2x2x2mm isotropic. All of the sequences have 16 diffusion
>>> directions and 1 b0 with 2 averages, however the later 3 sequences have a
>>> bvalue of 800 instead of 1000. It is unclear from the PAR files if you were
>>> using a 1.5T or a 3T. I would be concerned about the following things: 1)
>>> You may get more partial volume artifact in the 2.3x2.3x2.0 w/ 0.5mm gap,
>>> which would tend to lower FA values and raise MD. The change in bvalue
>>> might affect FA and MD, but I am not 100% sure on this.
>>>
>>>
>>>
>>> Peace,
>>>
>>>
>>>
>>> Matt.
>>> ------------------------------
>>>
>>> *From:* liang wang [mailto:[log in to unmask]]
>>> *Sent:* Thursday, April 16, 2009 9:44 PM
>>> *To:* [log in to unmask]
>>> *Subject:* Effect of different DTI protocol on results
>>>
>>>
>>>
>>> Hi Matt,
>>>
>>> Thanks for your reply on FSL maillist. However, I don't attached slightly
>>> bigger files to you. Just email you privately. These subjects' PAR-header
>>> are different in # slice, TR, or acquisition number. Thanks your attention.
>>>
>>> Liang
>>>
>>> --
>>>
>>> Liang Wang, PhD
>>> Postdoctoral Fellow
>>> Woodward Lab
>>> Department of Psychiatry
>>> University of British Columbia
>>> BC Mental Health & Addiction Services
>>> 938 West 28th Avenue
>>> Vancouver BC V5Z 4H4
>>> Telephone: 1-604-875-2000 (ext. 4735)
>>> Fax: 1-604-875-3871
>>> Email: [log in to unmask]
>>>
>>>
>>>
>>>
>>> ------------------------------
>>>
>>> *From:* FSL - FMRIB's Software Library [mailto:[log in to unmask]] *On
>>> Behalf Of *liang wang
>>> *Sent:* Thursday, April 16, 2009 9:13 PM
>>> *To:* [log in to unmask]
>>> *Subject:* [FSL] Effect of different DTI protocol on results
>>>
>>>
>>>
>>> Hi FSL group,
>>>
>>>
>>>
>>> I have a set of DTI data. However these data involve different scan
>>> parameters, such as TR, number of slices, FOV, acquisition number... Can you
>>> tell me which parameters have serious effect on DTI results (eg. FA and MD)
>>> because I want to merge these data into a single study together.
>>>
>>> Liang Wang
>>>
>>>
>>> --
>>> Liang Wang, PhD
>>> Postdoctoral Fellow
>>> Woodward Lab
>>> Department of Psychiatry
>>> University of British Columbia
>>> BC Mental Health & Addiction Services
>>> 938 West 28th Avenue
>>> Vancouver BC V5Z 4H4
>>> Telephone: 1-604-875-2000 (ext. 4735)
>>> Fax: 1-604-875-3871
>>> Email: [log in to unmask]
>>>
>>>
>>>
>>>
>>> --
>>> Liang Wang, PhD
>>> Postdoctoral Fellow
>>> Woodward Lab
>>> Department of Psychiatry
>>> University of British Columbia
>>> BC Mental Health & Addiction Services
>>> 938 West 28th Avenue
>>> Vancouver BC V5Z 4H4
>>> Telephone: 1-604-875-2000 (ext. 4735)
>>> Fax: 1-604-875-3871
>>> Email: [log in to unmask]
>>>
>>
>>
>>
>> --
>> David A Gutman, M.D. Ph.D.
>> Department of Psychiatry & Behavioral Sciences
>> Emory University School of Medicine
>>
>
>
>
>--
>Liang Wang, PhD
>Postdoctoral Fellow
>Woodward Lab
>Department of Psychiatry
>University of British Columbia
>BC Mental Health & Addiction Services
>938 West 28th Avenue
>Vancouver BC V5Z 4H4
>Telephone: 1-604-875-2000 (ext. 4735)
>Fax: 1-604-875-3871
>Email: [log in to unmask]
>
|