Hi Stephane,
I have had some preprocessing troubles with one particular individual. I used your pre-processing 2 approach. The latter approach, if I am not mistaken, is based on the "unified segmentation approach" which had been shown to be superior to the classical approach of your pre-processing 1.
I realized that my subject had a pretty significant signal drop out of parts of the temporal lobes which may have cause issues with the segmentation or the coregistration.
My suggestion would be to manually re-orient the images to approximately the right angle using the "display image" and the "re-orient" functions and then proceed.
Michael
________________________________________
From: SPM (Statistical Parametric Mapping) [[log in to unmask]] On Behalf Of Stephane De Brito [[log in to unmask]]
Sent: Monday, December 20, 2010 10:39 AM
To: [log in to unmask]
Subject: [SPM] SPM8: Normalisation problem - steep angle
Dear SPM experts,
I am sorry for the long email, but you are my last resort option after asking colleagues, an internal email list, and perusing the SPM list.
I am facing a problem in the pre-processing of my fMRI data with SPM8 and wonder if someone already had this problem and could provide me with some clues as to what the issue might be. Essentially, after the normalisation of my data, I noticed that, for 5 subjects, the normalised images were distorted with quite a steep angle as seen on the sagital view. On word document attached, the first two scan are from two different subjects for whom I get what I would expect, but the other 5 scans have a much steeper angle despite being accurately registered to the EPI template.
Here is how I did the pre-processing of the scans:
1. Realign and unwrap (with field maps)
2. Normalise to EPI template
3. Smooth
On the pdf attachment, I put 4 scans of one of my 5 problematic cases (#14). The 1st scan has been pre-processed using a script for each of the 3 steps described above. The 2nd scan is the same subject, but all the steps have been done with the GUI – the 1st and 2nd scans look identical suggesting that there is nothing wrong with my pre-processing scripts. The 3rd and 4th have been done via the GUI and for these I have re-centered the origin to the ac-pc line for all the EPI, the structural and the FM prior to doing the pre-processing proper. The 4th scan has been preprocessed like the 3rd one, but unwarped using the FM.
For both the 3rd and the 4th scans I did the following:
1. Realign and unwrap (with FM for the 4th scans)
2. Co-register the structural to the mean EPI (meanuf.img). Coregister: reference = meanuf and source = subject’s T1.
3. Segmented the T1 of the subject
4. Normalise (write only) the uf files using the warps (sn.mat file) obtained during the segmentation.
At the end of the day, the angle is still there, albeit less steep. The main differences between pre-processing 1 (realign+unwrap, normalise, smooth) and pre-processing 2 (realign+unwarp,co-register, segment, normalise) is that in the 2nd the angle is less pronounced, but the scans seem less stretched…not sure which of the two is better. For the 2 processing pipelines, the registration is good and, most importantly, we have the coverage of our key ROIs (amygdala, temporal lobes and most of PFC and ACC), but not for the most superior-posterior part of the brain.
My main concern is that these 5 subjects will influence the final mask for the analyses and lead to a significant reduction in the portion of the brain that will be covered. Do you have any suggestions? Is this angle due to the fact that the subject might have had their head positioned at a different angle from the others? However, by looking at the original scans I cannot see a big difference between the good ones and the problematic ones…Should I try my group analyses with and without those 5 participants to see if it makes a difference in the results?
Thanks in advance for your help.
Best wishes,
Stephane
-------------------------------------------------------------------
Stephane De Brito, PhD
Postdoctoral Research Associate
Developmental Risk & Resilience Unit
Division of Psychology and Language Sciences
University College London
26 Bedford Way
London WC1H 0AP
Tel: +44(0)20 7679 5394 (internal: 25394)
http://www.ucl.ac.uk/clinical-psychology/Research-Groups/DVR/index.htm
|
