NEAR: An artifact removal pipeline for human newborn EEG data

Abstract

Electroencephalography (EEG) is arising as a valuable method to investigate neurocognitive functions shortly after birth. However, obtaining high-quality EEG data from human newborn recordings is challenging. Compared to adults and older infants, datasets are typically much shorter due to newborns’ limited attentional span and much noisier due to non-stereotyped artifacts mainly caused by uncontrollable movements. We propose Newborn EEG Artifact Removal (NEAR), a pipeline for EEG artifact removal designed explicitly for human newborns. NEAR is based on two key steps: 1) A novel bad channel detection tool based on the Local Outlier Factor (LOF), a robust outlier detection algorithm; 2) A parameter calibration procedure for adapting to newborn EEG data the algorithm Artifacts Subspace Reconstruction (ASR), developed for artifact removal in mobile adult EEG. Tests on simulated data showed that NEAR outperforms existing methods in removing representative newborn non-stereotypical artifacts. NEAR was validated on two developmental populations (newborns and 9-month-old infants) recorded with two different experimental designs (frequency-tagging and ERP). Results show that NEAR artifact removal successfully reproduces established EEG responses from noisy datasets, with a higher statistical significance than the one obtained by existing artifact removal methods. The EEGLAB-based NEAR pipeline is freely available at https://github.com/vpKumaravel/NEAR .

Introducing RELAX (the Reduction of Electroencephalographic Artifacts): A fully automated pre-processing pipeline for cleaning EEG data - Part 2: Application to Event-Related Potentials

“…Future research using simulated data or hybrid data with simulated artifacts added to real data that has already been cleaned might help address the unknown ground truth issue, but simulated data comes with its own complications. For example, it is difficult to determine whether the simulated artifacts are analogous enough to real artifacts to provide an effective test (Kumaravel et al, 2022; Mumtaz et al, 2021; Rošťáková & Rosipal, 2021). We chose not to use simulations for this reason, and instead preferred to comprehensively assess real data using previously validated cleaning efficacy metrics and measures of the amount of variance explained by well-established experimental manipulations.…”

EEG artifact removal using sub-space decomposition, nonlinear dynamics, stationary wavelet transform and machine learning algorithms

“…This problem could be tackled by considering training and testing errors together and also performing some validation methods such as k-fold cross-validation. In addition, the proposed artifact detection method is based on SOBI, which is mainly effective for stereo-typed artifacts like the ones we had in most research experiments, while in other practical ones, such as newborn’s EEG preprocessing, we mostly face non-stereotyped artifacts ( Kumaravel et al, 2022 ). This point should also be taken into consideration while employing our proposed methods.…”

“…Then BSS methods can be applied to decomposed signals. Previous studies have shown that for EEG artifact removal, the combination of EEG subspace decomposition methods such as ICA-family methods and wavelet transforms could lead to acceptable results ( Zikov et al, 2002 ; Brychta et al, 2007b ; Delorme et al, 2007 ; Kumaravel et al, 2022 ). We used this to propose our method, which is mainly based on SOBI and SWT.…”

Adaptable and Robust EEG Bad Channel Detection Using Local Outlier Factor (LOF)

“…For example, in adult EEG, the artifacts have well-defined temporal and spatial features such as eye blinks (here, ICA is a good solution). Instead, developmental EEG collected from newborns, infants, or young children present more challenges in cleaning as the artifacts are primarily due to uncontrolled motion (here, ASR processing before ICA is recommended [ 13 ]). As such, artifact removal tools developed for adult EEG might not be optimal for newborn EEG.…”

“…In previous work, we introduced LOF for the first time on EEG data as the first step of a pipeline for artifact removal in developmental studies [ 13 ]. Here, we provide a complete characterization of LOF, presenting further development and validation of the method in the following three directions: (1) We present a novel, robust, and fully automatic method for computing LOF key parameters from a single dataset with annotated bad channels; (2) To test LOF adaptability to any kind of data, we validate LOF on newborn, infant, and adult datasets.…”