Hi,
It seems to me that infants' brain is not well mylinated and the contrasts
of their FA maps look different from that of adults. That may explain the
registration issues that you have got there. You may want to make a
population-specific template to register these subjects on.
Best
Longchuan
----- Original Message -----
From: "Gwenaëlle DOUAUD" <[log in to unmask]>
To: <[log in to unmask]>
Sent: Tuesday, August 26, 2008 1:57 PM
Subject: [FSL] Re : [FSL] TBSS with neonatals
> Hi Mathias,
>
> I need to know first which version of tbss you are using.
>
> It could only be that the "best target" hasn't been properly affinely
> registered to the MNI152 template...
>
> What does your target_to_MNI152 image look like?
>
> Cheers,
> Gwenaelle
>
>
> --- En date de : Mar 26.8.08, Mathias Engstrom
> <[log in to unmask]> a écrit :
>
>> De: Mathias Engstrom <[log in to unmask]>
>> Objet: [FSL] TBSS with neonatals
>> À: [log in to unmask]
>> Date: Mardi 26 Août 2008, 13h05
>> Hi Everybody,
>>
>> In a neonatal study at our hospital there has been interest
>> in using TBSS on
>> the diffusion data (15 dir overplus, b= 750, voxel
>> size=1.4/1.4/2mm). We have
>> done some preliminary attemps, but been running into
>> registration problems.
>>
>> All input data is in the same orientation (double checked),
>> but all outputs from
>> tbss_3_postreg (e.g. mean_FA ) gets a heavy rotation around
>> the R-L axis
>> and a chopping of the frontal lobe (due to the rotation).
>> From just looking at
>> the mean_FA the bigger tracts seems to have registered
>> fairly, but it is hard
>> to tell due to the strange view it gets presented in.
>>
>> We have used the -n flag when doing the registration, since
>> we have not
>> found/known about any good neonatal target space. The
>> registration has also
>> been tried without only the controls (suggesting that the
>> rotation would be
>> due to patient morphology), but with the same result. The
>> registration has
>> been running nicely on adults, when using the same
>> parameters.
>>
>> Has anyone else experienced similar problems?
>>
>> Any views on if there is something that can be done to
>> improve the results
>> would be very much appreciated, or what this is due to
>> (e.g. to high noise
>> level in the data, to big morphologic differences, etc. )
>>
>> Regards,
>> Mathias Engström, Karolinska University Hospital
>
>
>
> _____________________________________________________________________________
> Envoyez avec Yahoo! Mail. Une boite mail plus intelligente
> http://mail.yahoo.fr
>
>
>
|