From MEG Core
Revision as of 14:18, 26 November 2019 by Jstout (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search


Epilepsy analysis at the NIH is comprised of several components

  1. Copy data from MEG server and MRI folder
  2. Coregistration
  3. Beamformer Analysis
  4. Kurtosis Analysis of virtual sensors
  5. Ranking of virtual sensors
  6. Review of sensor data and virtual sensors
  7. Coregistration of MRI data to the orthohull data

Analysis Code and Configuration

The current analysis pipeline is performed through several Jupyter notebooks that are found on the internal repository

  1. Configure the computer MEG code dependencies
  1. Configure the epilepsy code


  1. Copy data from server (MEG_do_01.sh)
    1. Enter information into the MEG file (####FIX NAME)
    2. Run MEG_do_01.sh $DATE to pull the MEG data from the MEG repository and process the DICOM data
  2. Coregistration
    1. Add LPA, RPA, Nasion fiducials to the dataset
    2. MEG_do_02.sh
      1. Opens Afni for the marking of the tags
      2. Runs orthohull to realign the MRI to be aligned along the LPA/RPA/Nasion axis
  3. SAM analysis
    1. Constrain orientation (beta power analysis)
    2. Search for excess kurtosis
    3. Mapping voxels with highest excess kurtosis
    4. Creating 'virtual sensors' at the locations with excessive kurtosis
  4. Reviewing the SAM virtual sensors along with the sensor timeseries
    1. Open the CTF dataset using the CTF DataEditor (may need to be run through the singularity container if dependencies have not been met on the os)
    2. Add SAM output (this will apply the sensor weights to create a virtual sensor time series)
    3. Determine if the regions with high kurtosis contain epileptiform activity
  1. SAMepi parameters
    1. Covariance 10-70 (20-70 or 30-100Hz)

Software Requirements SAMscrV* MEG_do_0*.sh - These commands must be launched from the scripts dir as the commands use the pwd command to structure the processing paths and determine the location of the patient data

Issues with installation At the top level of SAM software, the make command must be run. The links to the compiled SAM binaries must

SAMsrcV3 - requires python 2.7 SAMsrcV5 - compiled for python 3.7+