Dear Andy,
If the registration is only slightly wrong, then it is likely that all the
problems are due to geometric distortion of the images. This
distortion can be corrected by using fieldmaps and prelude/fugue.
However, without them I'm afraid that our existing software cannot
perform any more corrections as it does not know what the
distortions are and the image information in EPI is insufficient to
drive standard non-linear registration methods.
If you are currently de-weighting areas of know signal loss then
that is about the best linear registration you are likely to get.
What you could do is also try to upweight areas near you region
of interest in order to emphasize any mis-registrations in these
areas and hence get better registration there at the expense of
poorer registration somewhere else. I'm afraid that is the best
I can offer you without fieldmap data.
All the best,
Mark
Silver, Andrew wrote:
>Hi Mark,
>The registration does not completely fail. In fact, it's only off by a
>small amount. For example, we are seeing activation in the white matter
>just below the superior temporal gyrus. Our task is designed to activate
>the STG, so I'm assuming that our alignment is simply off by a slight
>amount. I do BET the anatomical image, and it's pretty good.
>Unfortunately, I do not have fieldmaps and am not applying distortion
>correction. Like I mentioned in my previous email, I am hand drawing a
>binary mask that I load into FLIRT's input weighting option. I'm just not
>sure what else to do. Thanks again.
>-Andy
>
>-----Original Message-----
>From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf
>Of Mark Jenkinson
>Sent: Tuesday, May 30, 2006 1:50 PM
>To: [log in to unmask]
>Subject: Re: [FSL] alignment help
>
>Hi,
>
>It is hard to know what to do without more information.
>
>What happens when you register?
>Does it completely fail (brain out by 90 or 180 degrees)?
>Or is it just out by a small amount (ventricle mismatch)?
>Did you use BET on your anatomical image?
>Is the result from BET good?
>Have you got fieldmaps and are you applying distortion correction?
>
>EPI and anatomical registration is not easy, but we can normally get a
>pretty good alignment.
>
>Let me know a few more details and we'll see what we can do.
>All the best,
> Mark
>
>
>Andrew Silver wrote:
>
>
>
>>FSL Community,
>>I've been having trouble co-registering my EPI images to my high-res
>>anatomical image for some time now. I've tried two methods for getting
>>better alignment:
>>1. change degrees of freedom from 12 to 6 2. make a binary mask using
>>fslview to weight out the signal loss areas in the functional images.
>>
>>Is there anything else I can try? I feel as though the co-registration
>>shouldn't be this difficult. Thanks.
>>-Andy
>>
>>
>>
>>
|