Dear all,
Thanks for all the answers. I eventually used midtrans.
However, I am wondering: should I apply bias field correction on the two raw T1 images, or rather on the averaged T1?
Thanks,
Vincent
Dear Vincent,
I would use the tool midtrans for this task.
Please see recent postings to this list regarding how to use midtrans.
All the best,
Mark
On 1 Feb 2013, at 16:20, Paul Robinson <[log in to unmask]> wrote:
Hi, Vincent.
avscale in FSL can give you the half-way transforms.
Cheers,
Paul
On Fri, Feb 1, 2013 at 10:58 AM, Rolf A. Heckemann <[log in to unmask]> wrote:
Hi Vincent
I think A is the way to go. B won't work if there's more than a tiny amount of translation.
If you're worried about moving one image too much and the other not at all, it should be possible to bisect the transformation. In IRTK, there's a tool called bisect_dof which does just that. You can then apply the bisected transformation to T1_B and its inversion to T1_A to have the images "meet in the middle". I'm not aware whether an equivalent exists in FSL, but it would probably be straightforward to implement.
Best wishes
Rolf
On 28 January 2013 17:47, Vincent Koppelmans <[log in to unmask]> wrote:
Dear Mark,
Thank you for your answer. I was wondering what the best way would be to average these images. Would you choose option A, B, or something else?
A) Rigid (6 DOF) registration of image B to image A. Subsequently add (fslmaths -add) and average the two images (fslmaths -div 2).
B) Average raw T1_A and T1_B. Rigid (6 DOF) registration of image T1_A and T1_B to the raw mean image. Average the registered images.
To me, option B seems like the most fair one, as it rotates and translates both scans a little bit, and not only one scan the full amount.
What do you suggest?
Thanks for your help,
Vincent
> Yes!
>
> All the best,
> Mark
>
> On 24 Jan 2013, at 15:45, Vincent Koppelmans <[log in to unmask]> wrote:
>> Hi all,
>>
>> I have a large dataset of T1-weighted images. Each subject has been scanned twice with not much more than 45 min in between two scan sessions.
>>
>> I was wondering if I it would make sense to co-register and average these images to increase the SNR (and eventually get a better segmentation).
>>
>> Thanks,
>>
>> Vincent
--
Rolf A Heckemann, MD PhD
Médecin chercheur, Fondation Neurodis
CERMEP - Imagerie du Vivant
Hôpital Neurologique Pierre Wertheimer
59 Boulevard Pinel
69003 Lyon
France
2nd affiliation: Honorary Fellow, Imperial College London
|