MEGAP: A Comprehensive Pipeline for Automatic Preprocessing of Large-Scale Magnetoencephalography Data.

Area of Science:

• Neuroimaging
• Biomedical Engineering
• Signal Processing

Background:

• Magnetoencephalography (MEG) data require extensive preprocessing to remove noise and artifacts.
• Existing preprocessing methods lack comprehensive artifact handling and automation, hindering large-scale data analysis and reproducibility.
• There is a need for standardized, automated pipelines for efficient MEG data preprocessing.

Purpose of the Study:

• To develop and validate the first automatic preprocessing pipeline (MEGAP) for large-scale resting-state MEG datasets.
• To address the limitations of manual inspection and lack of standardization in MEG data preprocessing.
• To improve the efficiency, consistency, and reproducibility of MEG data analysis.

Main Methods:

• Developed MEGAP by integrating and sequencing recent algorithms for automated MEG data preprocessing.
• Implemented key features including head movement correction, notch-filter-free line noise removal, muscle artifact annotation, sensor/environmental noise reduction, and automated Independent Component Analysis (ICA) artifact detection.
• Validated MEGAP using simulated and experimental data from the Cambridge Centre for Aging and Neuroscience (Cam-CAN) dataset.

Main Results:

• MEGAP successfully automated noise and artifact reduction in MEG data.
• The pipeline demonstrated substantial improvements in data quality based on Normalized Mean Square Error (NMSE), correlation, and Signal to Noise Ratio (SNR).
• Validated efficacy on both simulated and real-world MEG datasets.

Conclusions:

• MEGAP offers a robust, automated framework for MEG data preprocessing, significantly reducing manual effort.
• The pipeline promotes standardization and reproducibility in neuroimaging research.
• MEGAP facilitates the analysis of large-scale resting-state MEG datasets, enabling greater generalizability.