The Lossless Pipeline
Overview
Teaching: 20 min
Exercises: 10 minQuestions
What are the inputs and outputs of the Lossless pipeline?
What kinds of decisions does the Lossless pipeline make?
Objectives
Understand the flow of decisions in the Lossless pipeline.
This section is designed to outline the key procedures of each of the Lossless pipeline scripts.
Scalpart (s01)
- Load an initialized BIDS file.
- Initialize marks
- Optional staging
- Co-register to MNI surface
- Special case average re-reference
- Compute ch_sd for time
- Compute ch_sd for channels
- Re-reference data
- High pass filter
- Low pass filter
- Low R channel annotation
- Bridge channel annotation
- Flag rank channel
- Re-reference data
- Low R time annotation
- Mark gaps
- Save state
- Purge data for Amica
- Create Amica parameter files
Initial Amica (s02)
The Amica script is run once initially.
Compart (s03)
- Load a
*_sa.set
file. - Load initial Amica model.
- Recreate data via ICs
- Compute log likelihood
- Compute ic_sd1 annotation
- Mark gaps
- Save state
- Purge data for Amica
- Create Amica parameter files
Amica (s04a, s04b, s04c)
The Amica script is run three more times in parallel after teh compnent artefact rejection to ensure that the Amica procedure is replicable if run multiple times on the same dataset.
Concat and Dipfit (s05)
- Load a
*_compart_data.set
file. - Load AMICA model A
- Recreate data via model A ICs
- Compute log likelihood for model A
- Load models B and C
- Compute log likelihood for models B and C
- ISCtest
- Compute ic_sd2 annotation
- Theta/delta power annotation
- Alpha power annotation
- Beta power annotation
- Low gamma annotation
- High gamma annotation
- Mark gaps
- Dipole fitting
- Rotate montage back to EEGLAB
- Save state
- Purge data
- ICLabel
- Load previous state, add back ICLabel info
- Marks updates
- BIDS export
Key Points
Decisions are made via criteria functions.
Parameters are changed via batch configuration files.