Hi, I am conducting VBM analysis with CAT12 toolbox in Matlab. I got mwp1 and mwp2 files after the segmentation. I know that mwp1 is modulated and normalized gray matter while mwp2 is modulated and normalized white matter. I just wonder how to interpret the values of each voxel in these files? I guess they are probability maps and the value 0.6, for example, may mean that this voxel has a 60% change to be identified as gray matter (in mwp1) or white matter (in mwp2). However, I found that the values change a lot after modulation/normalization because I found a lot of value of voxels larger than 0.5 in the unmodulated images (i.e. p1, p2), while for the the modulated images (i.e. mwp1, mwp2) most of the values are even smaller than 0.5. Is that normal? It should not the case that the voxel is absolutely white matter but the percentage of being white matter is lower than 0.2! And this also leads to another problem when I applied the second level calculation with a minimum threshold masking larger than 0.5-- I would get an error in matlab saying that no voxel was significant. And with a lower minimum threshold masking (0.2 for example), I would have an edge effect which I don't want to see.
Thank you.
I hope this can help you:
http://www.neuro.uni-jena.de/cat12/CAT12-Manual.pdf
“Modulated normalized” creates tissue class images in alignment with the template, but multiplies (“modulates”) the voxel values with the Jacobian determinant (i.e., linear and non-linear components) derived from the spatial normalization. This is useful for VBM analyses and allows the comparison of the absolute amount of tissue (e.g. “volume” of gray matter; Good et al., 2001; Neuroimage). Please note that in this type of modulation you must use TIV as covariate to correct different brain sizes. An alternative approach is to use global scaling with TIV if TIV is too strongly correlated with your covariates of interest. For more information, see the section “Checking for Design Orthogonality”. If you use the expert mode, you can optionally choose whether you want to modulate your data for non-linear terms only, which was the default for previous VBM-toolbox versions. This produces tissue class images in alignment with the template, but the voxel values are only multiplied by the non-linear components. This is useful for VBM analyses and allows comparison of the absolute amount of tissue corrected for individual brain sizes. This option is similar to using “Affine+non-linear” (see above) in combination with “global normalization” (when building the statistical model with the traditional PET design late one). Although this type of modulation was used in as default previous VBM-toolbox versions it is no longer recommended as the use of standard modulation in combination with TIV as covariate provides more reliable results (Malone et al., 2015; Neuroimage).
As the manual states, the modulated images contain in each voxel the volume prior to spatial normalization. If you calculate the sum of all voxels of such a spatially normalized modulated tissue segment and compare it to the tissue segment volume in native (subject, non-normalized) space, it will be nearly the same (rounding errors etc.). I would recommend studying the above CAT12 manual mentioned by Ricardo Faustino and here the statistics section.
And, if there is no significant result, perhaps it is because of no significant result…
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis