WIP for standardized pre-processing pipeline for OPM data collected at NatMEG.

Overview

This repository contains scripts for pre-processing OPM data collected at NatMEG. The scripts are designed to be modular and can be adapted for different datasets or analysis needs. The main focus is on reading in the data, performing Independent Component Analysis (ICA) for artifact removal, and preparing the data for source reconstruction.

To do

• Read data from BIDS, loop over subjects from CSV/list
• Read parameters from separate config file
  • Add option to exclude some subjects/sessions
• Recommend SSP, make sure channels are identical between actual and empty room recording.
• Consider level for paradigm/task in BIDS/derivatives/sub/ses/mod/TASK/ to simplify structuring of data processed with different parameters
• Parallelize processing
• Separate MRI and HPI preparation pipeline
• Separate OPM and SQUID analysis
• Don't specify params.manual_ica more than once (now in params, function and when calling function?)
• Save intermediary results and figures
• Create quarto markdown template for report generation

Scripts

opm_analysis_main.m

Main script to run analysis. Specify paths to data and parameters (e.g. filter cutoff, pre- and post-stim time window) here.

ica_MEG.m

Function to run ICA on OPM or SQUID data (depending on params.ch) with option to manually select components to remove from data.

read_osMEG.m

Read in On Scalp MEG data from OPM sensors, including filtering, downsampling, epoching and identification of bad channels (using opm_badchannels.m).

opm_badchannels.m

Function to detect bad channels in OPM data. Return badchs with:

• badchs_flat : Flat channels, i.e. channels with no signal.
• badchs_std: Channels with excessive standard deviation, likely contaminated by high levels of noise or artifacts.
• badchs_neighbors: Channels that have poor correlation with their spatial neighbors.
• badchs_outlier: Channels that show outlier behavior in their frequency spectrum.

read_cvMEG.m

• Read MEG data from conventional SQUID sensors, including filtering and epoching.

timelock.m

Create time-locked epochs for OPM or SQUID data. Baseline/demean data and save butterfly plot.

prepare_mri.m

Prepare MRI for source reconstruction.

fit_hpi.m

?

fit_mne.m

Fit MNE source model for both OPM and SQUID data?

create_preproc_report.m

Not in use

insidePointcloud.m

Not in use