Dear SPMers,
I have been dealing (and struggling) with VBM analysis for a while and I
have some major issues to be addressed. Even though I have been using John
Ashburner's scripts to create my customized templates and to segment my
raw images, I would really like to run the whole procedure in GUI mode
since its very beginning in order to get further insight into this process
and to perform a quality control of my analysis (the CSF partition is
pretty poor in a few individuals). As far as I understood by reading
several papers, the steps for the image pre-processing are the following:
First-pass normalization/segmentation
a. normalization of the raw images to the default SPM2 T1-MRI template
b. segmentation of the normalized images into GM, WM, and CSF partitions
c. smoothing of the normalized segmented images using a 8-mm FWHM kernel
d. creation of study-specific GM, WM, and CSF templates by averaging all
the respective individual GM, WM, and CSF normalized, segmented, smoothed
images
Second-pass normalization/segmentation
1. segmentation of the raw anatomical images into GM, WM, and CSF
partitions in native space
2. normalization of the individual GM, WM, and CSF partitions to their
respective study-specific template
3. normalization of the original MRI images in native space by using one
of the three sets of optimal normalization parameters (GM, WM, or CSF)
obtained in the previous step, according to the type of VBM analysis
4. segmentation of the optimally-normalized MRI images using study-
specific templates created in the first-pass pre-processing
Is this sequence correct? More importantly, if the first-pass
normalization/segmentation is rather straightforward and feasible in GUI
mode, I cannot say the same about the second-pass pre-processing. These
are my doubts:
Step 1) When I click the "Segment" button, I am asked whether the images
are already normalized or not. Am I supposed to choose "No" (because they
are raw images and, therefore, not actually normalized), and select "T1"
as modality, or "Yes" (because otherwise they will be normalized before
being segmented), to skip this step and go on to the segmentation, or
something else I cannot think of?
Step 2) When I click the "Normalize" button, I am prompted for an "Image
to determine parameters" and "Images to write normalized" for each
subjects. I cannot understand which is which since I have only one image
for each segment in each subject. As to the templates, I guess they are my
customized GM template for *_seg1.img, my customized WM template for
*_seg2.img, and my customized CSF template for *_seg3.img.
Step 3) This step is completely obscure to me. How can I perform it in GUI
mode? If the raw MRI images are the ones to be normalized, which is the
template? Am I supposed to create also a whole brain template, besides the
GM template, the WM template, and the CSF template, in the first-pass pre-
processing to be used as template in this normalization step? If so, steps
1 & 2 of this second-pass pre-processing are useless. Please, enlighten me.
Step 4) When I click the "Segment" button, I am asked whether the images
are already normalized or not. I think this time I need to choose "Yes",
because the images are actually normalized. But, I am not allowed to
choose a template in this step. What am I supposed to do? Should I replace
the default templates with my customized templates in the "SPM2 apriori"
directory for them to be used by default in this step?
I apologize for the huge deal of questions I asked in my first email ever
in this forum. I am really looking forward to receiving suggestions and
directions which allow me to move on in my GUI analysis at your earliest
convenience. Thank you very much.
Best regards,
Nicola Pannacciulli
Obesity and Diabetes Clinical Research Branch
Phoenix Epidemiology and Clinical Research Branch
National Institute of Diabetes & Digestive & Kidney Diseases
National Institutes of Health
Phoenix, AZ, USA
|