[Mristudio-users] test-retest

[Miguel Burgaleta]

Thanks a lot, Susumu. It is very clear now.

Miguel


On Apr 2, 2011, at 8:54 PM, susumu mori wrote:

> Please see below;
>
> On Fri, Apr 1, 2011 at 4:39 AM, Miguel Burgaleta
> <miguel.burgaleta at uam.es> wrote:
>>
>> Hi Susumu,
>>
>> Thanks for your reply.
>>
>> I find your suggestion very helpful! However, with that method I  
>> don't
>> see how to separate the reproducibility of the tract from the rater
>> reliability. If the tract in time 1 is different from that in time 2,
>> we don't really know to what extent it is caused by small errors in
>> the ROI placement, or by problems of reproducibility...
>>
>> Another possibility would be to do tractography in time 1 twice, and
>> once in time 2, so that we can look at the time1-time2  
>> reproducibility
>> relative to the time1-time1 reliability.
>
> Yes, for intra/inter-operator reproducibility, you have to pick one
> data (e.g. Time 1) and perform ROI drawing multiple times by the same
> and different operators.
>
> For inter-measurement reproducibility, you have to have Time 1 and
> Time 2 data. In this case, if you draw ROI independently on Time 1 and
> 2, you have both inter-measurement and intra-operator reproducibility
> mixed.
>
>>
>> Also, can ROIs be transformed with Diffeomap as well (just like I
>> suggested in my previous email)? That would help ensure that
>> differences are due to reproducibility only.
>>
>
> To remove the intra-operator factor from the inter-measurement
> reproducibility, you can transform the ROI used for Time 1 to Time 2.
> In this case, I suggest you to save the ROIs as the "Binary Map"
> format. For example, suppose you drew 3 ROIs for tractography on Time
> 1. If you save the ROIs as the Binary Map, you'll get 4 files. Three
> files are 1/0 binary mask files for 3 ROIs. The final one is a header
> file in a text format. You can read and study the header. There is an
> explanation section at the bottom of the header file.
>
> Then you can apply the transformation matrix to the 3 ROI files in
> DiffeoMap. As mentioned in the earlier email, you can use, for
> example, b0 (Time1) to b0 (Time2) linear transformation and save the
> transformation matrix. This matrix can be applied to the 3 ROI files.
> Save these 3 new transformed ROI files. Then, rewrite the text header
> files and specify the filenames of the 3 new ROI files.
>
> You can load all necessary files (vector and FA files) for Time 2 for
> tractography in DtiStudio. Then, you can load the revised Binary Map,
> which will load the 3 new ROI files and apply the ROIs.
>
> Hope this will do.
>
> Susumu
>
>
>> Thanks,
>> Miguel
>>
>>
>> On Apr 1, 2011, at 12:15 AM, susumu mori wrote:
>>
>>> Hi Miguel,
>>>
>>> You can do it using Diffeomap (or FSL or SPM). First do tractography
>>> in image 1 and 2 independently. Save the results as a 3D 1/0 mask in
>>> DtiStudio.
>>> Then run a rigid transformation between the image 1 and 2. You can  
>>> do
>>> FA vs FA but b0 vs b0 could be easier because FA maps could have  
>>> some
>>> random intense pixels outside the brain that may interfere with the
>>> rigid alignment. Save the transformation matrix and apply it to your
>>> 1/0 binary fiber mask image. Then you can superimpose the aligned #1
>>> and original #2 see the reproducibility. You can do all these very
>>> easily by DiffeoMap.
>>>
>>> Susumu
>>>
>>> On Wed, Mar 30, 2011 at 12:56 PM, Miguel Burgaleta
>>> <miguel.burgaleta at uam.es> wrote:
>>>>
>>>> Hi,
>>>>
>>>> We have a sample of DWI acquired twice for each subject, 1 month in
>>>> between, and would like to analyze the test-retest  
>>>> reproducibility of
>>>> a selection of tracts with DTIStudio. We would also like to see the
>>>> effect of outlier pixel rejection on the reproducibility.  
>>>> However, we
>>>> are not sure of how to proceed so that we control for the intra- 
>>>> rater
>>>> reliability.
>>>>
>>>> One option we thought of is to perform tractography at time 1, save
>>>> both binary tracts and ROIs, and then register each FA image in
>>>> time 1
>>>> to its corresponding FA image in time 2, apply that  
>>>> transformation to
>>>> the ROIs and get the tract for time 2. Is this feasible with
>>>> DTIStudio?
>>>>
>>>> If you know a better way to do this, please let me know.
>>>>
>>>> Best,
>>>> Miguel
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> mristudio-users mailing list
>>>> mristudio-users at mristudio.org
>>>> http://lists.mristudio.org/mailman/listinfo/
>>>> Unsubscribe, send a blank email to: mristudio-users-unsubscribe at mristudio.org
>>>>
>>> _______________________________________________
>>> mristudio-users mailing list
>>> mristudio-users at mristudio.org
>>> http://lists.mristudio.org/mailman/listinfo/
>>> Unsubscribe, send a blank email to: mristudio-users-unsubscribe at mristudio.org
>>
>> _______________________________________________
>> mristudio-users mailing list
>> mristudio-users at mristudio.org
>> http://lists.mristudio.org/mailman/listinfo/
>> Unsubscribe, send a blank email to: mristudio-users-unsubscribe at mristudio.org
>>
> _______________________________________________
> mristudio-users mailing list
> mristudio-users at mristudio.org
> http://lists.mristudio.org/mailman/listinfo/
> Unsubscribe, send a blank email to: mristudio-users-unsubscribe at mristudio.org