Dear Jonathan, cc List,
I did something similar with even smaller (midbrain) FOV, and had
registration issues as well.
My solution: get 1 whole brain EPI volume (eg, same sequence as your 1s
timeseries data, only with more slices and a correspondingly adapted
flip angle) right after functional scanning, and take graet care you
angulate this image exactly the same as your 14 slice functional time
series scans (eg only translate your FOV, do NOT change the angles). By
doing so your EPI distortion is nearly identical to your time series
data.
This whole brain EPI will likely register well with your time series
data, as it is exactly the same contrast, and has the same spatial EPI
distortions.
You could now do 2 things for registration with anatomy:
1. Coregister this whole brain EPI to your T1, choose time series as
'other' (after you registered your whole brain EPI to your timeseries).
This requires that your EPI has minimal spatial distortions, which is
usually an issue. Especially as you are focussing on OFC. Consider
acquiring both phase and magnitude images for phase unwarping your EPI
images.
2. Forget about registering to T1, and use whole brain EPI directly for
normalizing your time series to an EPI template. Depending on your image
contrast and the required spatial accuracy of the results, you could
even use 'unified segmentation' for this, but be sure you can reasonably
segment this whole brain EPI. For that, you might want to record 10
whole brain EPIs or so, and average them, for better image SNR.
Good luck,
Bas
--------------------------------------------------
Dr. S.F.W. Neggers
Division of Brain Research
Rudolf Magnus Institute for Neuroscience
Utrecht University Medical Center
Visiting : Heidelberglaan 100, 3584 CX Utrecht
Room B.01.1.03
Mail : Huispost B.01.206, P.O. Box 85500
3508 GA Utrecht, the Netherlands
Tel : +31 (0)88 7559609
Fax : +31 (0)88 7555443
E-mail : [log in to unmask]
--------------------------------------------------
> -----Oorspronkelijk bericht-----
> Van: SPM (Statistical Parametric Mapping)
> [mailto:[log in to unmask]] Namens Jonathan Malmaud
> Verzonden: woensdag 15 juli 2009 9:07
> Aan: [log in to unmask]
> Onderwerp: [SPM] Registration trouble when I don't have
> full-brain coverage
>
> Hi,
> I am having trouble registering and normalizing fMRI data
> that was gathered using a sequence with an RT of 1s
> consisting of only 14 slices centered on the OFC in four subjects.
> Normally I use 44 slices to cover the whole brain and
> normalize my data by taking the mean of a subject's
> functional scans and warping it to a EPI template. However,
> this procedure is failing badly now that I only have coverage
> of a small area of the brain.
> Registration between a subject's functional scans and
> T1-weighted anatomical scan is also giving poor results. Does
> anyone have any advice on how to achieve usable registration results?
>
> Much appreciated,
> Jon Malmaud
> California Institute of Technology
>
|