Building an Open Source Classifier for the Neonatal EEG Background: A Systematic Feature-Based Approach From Expert Scoring to Clinical Visualization

Abstract

Neonatal brain monitoring in the neonatal intensive care units (NICU) requires a continuous review of the spontaneous cortical activity, i.e., the electroencephalograph (EEG) background activity. This needs development of bedside methods for an automated assessment of the EEG background activity. In this paper, we present development of the key components of a neonatal EEG background classifier, starting from the visual background scoring to classifier design, and finally to possible bedside visualization of the classifier results. A dataset with 13,200 5-minute EEG epochs (8–16 channels) from 27 infants with birth asphyxia was used for classifier training after scoring by two independent experts. We tested three classifier designs based on 98 computational features, and their performance was assessed with respect to scoring system, pre- and post-processing of labels and outputs, choice of channels, and visualization in monitor displays. The optimal solution achieved an overall classification accuracy of 97% with a range across subjects of 81–100%. We identified a set of 23 features that make the classifier highly robust to the choice of channels and missing data due to artefact rejection. Our results showed that an automated bedside classifier of EEG background is achievable, and we publish the full classifier algorithm to allow further clinical replication and validation studies.

Neonatal EEG graded for severity of background abnormalities in hypoxic-ischaemic encephalopathy

“…These automated methods could produce a continuous objective measure of EEG activity that could be easily scaled to monitor a high-volume of neonates, far beyond what would be humanly possible. Many methods have been developed to generate background grading systems [8][9][10][11][12][13][14][15][16][17] . This existing body of work highlights the potential of signal processing and machine learning methods to construct accurate classifiers of background EEG.…”

“…Thus far, progress has been confined to individual research groups pursuing different approaches. Comparing methods is difficult for many reasons 17 , including the lack of an accepted standard grading scheme 3 and freely-available EEG data. Aiming to address some of these limitations-and inspired by the success of an open-access neonatal EEG data set with EEG.…”

“…The EEG data could also be used to develop an automated EEG grading algorithm. These classification algorithms use signal processing and machine-learning methods to extract information from the EEG that is characteristic of the particular grade of EEG [8][9][10][11][12][13][14][15][16][17] . The first stage in algorithm development is to preprocess the data.…”

Neurophysiology of the Developing Cerebral Cortex: What We Have Learned and What We Need to Know

“…It is not surprising that these activity patterns also occur in the cerebral cortex of preterm human babies, when the cortex resembles in its structure and function the cortex of a newborn rodent (Colonnese et al, 2010 ; Luhmann and Fukuda, 2020 ). The activity patterns recorded in different cortical areas of preterm human babies are termed spindle bursts, delta brushes, tracé discontinue, tracé alternant , and spontaneous activity transients (SAT) (Vanhatalo and Kaila, 2006 ; Milh et al, 2007 ; Colonnese et al, 2010 ; Chipaux et al, 2013 ; Koolen et al, 2016 ) and serve as biomarkers for the cognitive and motor development of the child (Iyer et al, 2015 ; Tokariev et al, 2019 ; Moghadam et al, 2021 ).…”

Modeling Neonatal EEG Using Multi-Output Gaussian Processes

“…Features are evaluated using a NN. Critically, this method caters from multivariate features indirectly by using ICA and not modelling cross-correlations using GPs [34] A neonatal EEG background classifier is developed, it application ranges from visual background scoring to classifier design. Methods in the classifier include SVM, NN, RNN, which are fed with 98 features including amplitude, complexity and oscillatory behaviour of EEG.…”