Dear Matthieu,
see below for my answer...
On Wed, 9 Nov 2016 10:05:14 +0100, Matthieu Vanhoutte <[log in to unmask]> wrote:
>Dear Christian,
>
>Thanks for the detailed answers !
>
>Please see in-line below.
>
>Best regards,
>Matthieu
>
>
>2016-11-09 9:48 GMT+01:00 Christian Gaser <[log in to unmask]>:
>
>> Dear Matthieu,
>>
>> On Tue, 8 Nov 2016 20:27:27 +0100, Matthieu Vanhoutte <
>> [log in to unmask]> wrote:
>>
>> >Dear experts,
>> >
>> >Does anybody have answer to my questions in last mail ?
>> >
>> >Best regards,
>> >Matthieu
>> >
>> >> Le 3 nov. 2016 à 16:44, Matthieu Vanhoutte <[log in to unmask]>
>> a écrit :
>> >>
>> >> Dear CAT12's experts,
>> >>
>> >> 1) Could you precise me what is the impact of TPM maps onto central
>> surface and surface extracted data ?
>> The TPM is only used for the initial registration and segmentation.
>> Because CAT is using a segmentation approach without use of priors the
>> impact of the TPM on the overall results will be minor, except for children
>> data. For children data I would strongly recommend to create customized
>> TPMs. Please also keep in mind that the surfaces are extracted using
>> segmentations in native space without any spatial registration. However,
>> the initial registration can also bias your final segmentation and
>> skull-stripping...
>>
>> >>
>> >> 2) In case of atrophied patients, wouldn't it better to create own TPM
>> with DARTEL for surface features ? (and own customized TPM with TOM too ? )
>> There are different TPMs or templates used in CAT12:
>> TPM - used for initial registration and segmentation (followed by
>> segmentation without use of priors)
>> DARTEL template - used for high-dimensional spatial registration
>> Freesurfer surface template - used for surface registration
>>
>> Of course, a customized DARTEl template will influence the results of your
>> spatial registration. However, the impact on surfaces will be minor,
>> because here a surface registration is used (where no opportunity exists to
>> create own templates). Thus, a customized DARTEL template will influence
>> mostly your VBM data, but almost nothing your surface data (there might be
>> some minor influence because some internal atlas maps are used to fill
>> ventricles and subcortical areas and to divide the hemispheres).
>> Personally, I only create customized templates for children data, but not
>> for atrophied brains. Of course, there are some advantages in using a
>> customized DARTEL template for atrophied brains, but you will loose some
>> useful features such as ROI analysis (which is DARTEL atlas based) and the
>> ease of processing your data with an existing template and the
>> advantages/weaks of using a customized template might also depend on your
>> data. In case of doubts create a customized DARTEL template and compare the
>> results to the standard template.
>>
>> >>
>> >> 3) Finally for statistical analysis, are preprocessed surface features
>> extracted compatible with Freesurfer's fsaverage allowing possibly to use
>> another framework than SPM for analysis ?
>> The surface features such as thickness, gyrification are saved in
>> freesurfer format and could be also used with other software. However, even
>> if CAT12 uses some freesurfer naming conventions freesurfer is expecting
>> certain files and folders that don't exist in CAT12.
>> The surface features such as thickness are estimated in native space and
>> need to be resampled/transformed according to the fsaverage surface (after
>> surface registration). This is done in CAT12 with the "resample & smooth
>> surfaces" function (where the surface registration is considered) and the
>> resampled data are finally in gifti format and could be used with any other
>> software. This is similar to the step in freesurfer where the command
>> "mris_preproc" is used to resample data into fsaverage space. However, the
>> output format in freesurfer is different and all files are concatenated
>> into a single mgh-file.
>>
>
>I managed to open thickness in native space (without any extension file as
>'gii') as overlay onto the central surface within freeview. However, when I
>tried to open resampled thickness, which has '.gii' extension, I only
>managed to open it as a surface on freeview and have not seen the thickness
>values.
This is the issue with the additional "." in the filename which I mentioned in another answer to your question. We will solve this issue...
>Moreover, once load thickness resampled gifti file and fsaverage space, I
>could see that the the two surfaces weren't identical.
The freesurfer fsaverage surface is used for spatial registration and as common mesh structure (tetrahedral mesh with defined number of vertices and faces) for the resampled files. Of course, the shapes will be different because of the different coordinates. Only the underlying mesh structure is from fsaverage and should be the same. The shape is from the individual data and will be different.
Best,
Christian
>
>Could you give me an explanation ?
>
>
>>
>> Best,
>>
>> Christian
>>
>> >>
>> >> Best regards,
>> >> Matthieu
>> >>
>> >>
>> >
>> >
>>
>>
>>
>
|