search

cwatson / brainGraph

Graph theory analysis of brain MRI data
166 stars 48 forks source link

CIVET-output compatibility? #6

Closed sdrakulich closed 5 years ago

sdrakulich commented 5 years ago

Greetings,

I'm hoping to understand if your tool is compatible, or could be made compatible with McGill's CIVET tool. http://www.bic.mni.mcgill.ca/ServicesSoftware/CIVET-2-1-0-Table-of-Contents

Thank you!

cwatson commented 5 years ago

Certainly. It depends on the data being output from CIVET, and therefore what will be input to brainGraph. For example, the Freesurfer data are text files in which each row is a subject and each column is a region of a brain atlas/parcellation (Desikan-Killiany, DKT, or Destrieux). So if you could get me some examples of the data you would like to input to brainGraph, and/or send me the parcellation info (see Chapter 1.4 of the User Guide for format), I can help.

sdrakulich commented 5 years ago

Data that I'd need to use is in the structure

I think it might be best to show you another approach, by Keith Worsley for SurfStat, a "Matlab toolbox for the statistical analysis of univariate and multivariate surface and volumetric data using linear mixed effects models and random field theory".

Masks are created by having those values set to zero in the appropriate places.

There is some vague implementation of such analyses within the institution, but they are not fleshed out it seems...specifically with respect to viewing a mapped output.

The reason I ask is because I'd infinitely prefer (at this stage of my learning) staying in the R ecosystem.

Thank you so much for the help!

cwatson commented 5 years ago

Are you planning on analyzing structural covariance networks? If there are ~40K vertices per hemisphere, then the correlation matrix for a single subject would be 8e4 x 8e4 in size, which would be ~51 GB of memory!

I would have to change most aspects of the package in order to accommodate such data, as it is designed to work with networks with 1-2K vertices (maybe more). However, if you are looking to just do surface-based analyses, then I won't be able to help. You can see other MRI-related R packages here.

cwatson commented 5 years ago

I'm going to close this issue as it seems unrelated to what the package does.