dowdlelt / meica_tool

Useful for making plots from MEICA

These images are now produced in tedana, which can be found here: https://github.com/ME-ICA/tedana

This "toolbox" is no longer maintained, updated and may not even work anymore. The figures are produced by default in the much improved multi-echo denoising package tedana.

meica_tool

I've created a handy matlab script that works with meica.py (https://bitbucket.org/prantikk/me-ica) v3 - from the experimental branch.

It creates a series of figures that are useful for visualizing the output in a quick manner, including component timeseries from meica.py, color coded on whether they were:

• BOLD-like - green
• Non-BOLD - red
• r2 weighted - pink
• Ignored - black.

2017/09/22 update - now more 4ier - enjoy a fft plot.

Each plot includes brain slices of the component beta values (from TED/betas_OC.nii)

• motion parameters and framewise displacement
• kappa vs rho scatter plot, where size is proportaional to variance, colors as above
• kappa vs rho line plot
• Bar plot of variance explained
• tSNR figures, with histograms

It then creates a bar plot showing the relative variance of each of those categories.

Its (still) ugly code, but effective...for now.

Current dependencies include:

• spm, for file loading, for now
• load_nii, from Nifti Tools

But these few functions will eventually be packaged together and included.

Usage

• Add to matlab path
• run meica_component_displayer(tr), where tr is the repitition of your EPI timeseries in seconds.
• select the meica.py output folder, ex. meica_nback_e1.label
• wait a bit

Example Figures

Thanks to bramila framewise displacement and detrend code (from https://git.becs.aalto.fi/bml/bramila/tree/master) for dvars calculation

meica_tool's Issues

TSNR plots are deceptive

The plots of E2 vs MEDN are a bit tricky, because the MEDN time series will typically be detrended and to some extent highpass filtered due to MEICA. This may or may not bias these figures to make MEDN look even better.

A simple detrending might be a solution.

TSNR measures in relevant areas

While the plots are nice, it would be more useful to know the TSNR in grey matter, separately from white matter and CSF. Need to see if MEICA segments the tissues. If not, could use the SPM segmentation, assuming the delay is acceptable.

More useful figures

For later, easy QC:
Add T2star figures
Add vein figures

motion plots are interpreted wrong.

Uses the motion.1D file to plot motion, and in turn Framewise Displacement. Unfortunately the motion calculation is produced by the 3dvolreg (https://afni.nimh.nih.gov/pub/dist/doc/program_help/3dvolreg.html)
option -dfile which has the following attributes:
roll = rotation about the I-S axis }
pitch = rotation about the R-L axis } degrees CCW
yaw = rotation about the A-P axis }
dS = displacement in the Superior direction }
dL = displacement in the Left direction } mm
dP = displacement in the Posterior direction }

so, have to correct plots, and supply information to bramila_framewise correctly.