Module 7 CAT12 VBM

What is a VBM analysis?

Voxel-based morphometry (VBM) allows you to assess group level behavioral relationships or group difference in gray matter structure (white matter can be done as well but typically isn't). It is based on estimations of brain matter density. Density is a proportion (think of it like the volume of a structure divided by head size).

As a side note: There’s some debate whether or not VBM measures volume or density. What it does depends on the tool and how the data are processed. Some processing of data for a VBM does what is called a modulation, which effectively transforms the density value into a volume (in effect, removes the adjustment for head size [this isn’t what’s going on but is close enough for our purposes]). However, there’s not clear consensus if the value is a volume or a density so I usually default to calling what you are comparing density, just to be safe.

In practice we are measuring gray matter volume differences or correlations with a VBM analysis; however, because there are additional factors to consider (e.g., whether or not to also include head size as a covariate: https://link.springer.com/article/10.1007/s11682-016-9656-9Links to an external site.), I’ll use density and volume interchangeably when talking about an VBM analysis and usually default to density.

Watch this video comparing FreeSurfer with VBM: https://www.youtube.com/watch?v=TT8FJUq-QKo

If you’re interested in learning more about VBM, this is an important paper to read: https://www.fil.ion.ucl.ac.uk/spm/doc/papers/am_vbmreview.pdf

“VBM can be thought of as comparing the relative concentration of gray or white matter structures in the spatially normalized images (i.e. the proportion of gray or white matter to all tissue types within a region). There are cases, however, when the objective of the study is to identify regional differences in the volume of a particular tissue (gray or white matter), which requires the information about absolute volumes to be preserved. Here a further processing step, which is usually referred to as “modulation”, can be incorporated to compensate for the effect of spatial normalisation…. In short, the multiplication of the spatially normalised gray matter (or other tissue class) by its relative volume before and after warping has critical implications for the interpretation of what VBM is actually testing for. Without this adjustment, VBM can be thought of as comparing the relative concentration of gray or white matter structures in the spatially normalized images. With the adjustment, VBM can be thought of as comparing the absolute volume of gray or white matter structures. The two approaches are known as “non-modulated” and “modulated” VBM, respectively.”

So what this means is just think of this as a **volume ** and don’t worry about the density aspect.

Note that this will require about 4-8 hours of processing (you don't have to do anything during that time), but factor that into your schedule for working on this. Once that is done, the rest is relatively quick to do.

Once data are set up, running a VBM analysis is straightforward. The most challenging parts are quality assessment and setting up the GLM (General Linear Model). You do not need to worry about quality assessment for the class assignment. T1-weighted images are provided.

For the VBM analysis, organization of files depends on what type of analysis you are doing. If you are running a between-group analysis, the convention is to group or name all the "control" MRIs so they group together and all the "patient" MRIs so they group together. For example, prepend con_ for control and pat_ for patient. In CAT12 it's easiest to have all group MRI in one directory and another group in another directory. For the homework, this has already been done for you.

Data are stored in a directory called vbm, which is in /blue/psy4930/share/data/Module7

Copy that to your directory using the Terminal (or can use the file system in the browser in ood.ufl.edu): cp -fr /blue/psy4930/share/data/Module7/vbm /blue/psy4930/share/students/{gatorlink}/ Use a Matlab session for this (I've had good experiences with Matlab R2020b but you can try a newer version). Request 8 cores, about 32 GB of RAM, and to be safe, request at least 24 hours. You will need to use psy4930-b under QoS (refer to the video) and make sure Cluster partition has "default" selected.

My tutorial is mostly based on one created by Dr. Jahn. You can refer to these. We are using different data but his guide is a great resource.

Step 1: https://andysbrainbook.readthedocs.io/en/latest/CAT12/CAT12_03_Preprocessing.html

Step 2: https://andysbrainbook.readthedocs.io/en/latest/CAT12/CAT12_04_Analysis.html

Here is the walk-through video: https://youtu.be/x6tm8ewq8zk?si=DKYJ9U_sCBBe0hfv

I recommend changing the quality to 1440p60 if you have fast enough internet. That will make everything on the screen as clear as possible.

Some of the video covers what's on the previous page. Most of it covers what's after that. Again, the preprocessing can take 6+ hours, so plan accordingly. Check the next page to see what you submit (it will be a screenshot).

What You Turn In

For this assignment, please upload a screenshot of the Matlab session showing some results.