Artefact detection in neonatal EEG

A multi-task and multi-channel convolutional neural network for semi-supervised neonatal artefact detection

Objective. Automated artefact detection in the neonatal electroencephalogram (EEG) is crucial for reliable automated EEG analysis, but limited availability of expert artefact annotations challenges the development of deep learning models for artefact detection. This paper proposes a semi-supervised deep learning approach for artefact detection in neonatal EEG that requires few labelled data by training a multi-task convolutional neural network (CNN).Approach. An unsupervised and a supervised objective were jointly optimised by combining an autoencoder and an artefact classifier in one multi-output model that processes multi-channel EEG inputs. The proposed semi-supervised multi-task training strategy was compared to a classical supervised strategy and other existing state-of-the-art models. The models were trained and tested separately on two different datasets, which contained partially annotated multi-channel neonatal EEG. Models were evaluated using the F1-statistic and the relevance of the method was investigated in the context of a functional brain age (FBA) prediction model.Main results. The proposed multi-task and multi-channel CNN methods outperformed state-of-the-art methods, reaching F1 scores of 86.2% and 95.7% on two separate datasets. The proposed semi-supervised multi-task training strategy was shown to be superior to a classical supervised training strategy when the amount of labels in the dataset was artificially reduced. Finally, we found that the error of a brain age prediction model correlated with the amount of automatically detected artefacts in the EEG segment.Significance. Our results show that the proposed semi-supervised multi-task training strategy can train CNNs successfully even when the amount of labels in the dataset is limited. Therefore, this method is a promising semi-supervised technique for developing deep learning models with scarcely labelled data. Moreover, a correlation between the error of FBA estimates and the amount of detected artefacts in the corresponding EEG segments indicates the relevance of artefact detection for robust automated EEG analysis.

Automated Detection and Removal of Cardiac and Pulse Interferences from Neonatal EEG Signals

Electrical cardiac and pulsatile interference is very difficult to remove from electroencephalographic (EEG) signals, especially if recorded in neonates, for which a small number of EEG channels is used. Several methods were proposed, including Blind Source Separation (BSS) methods that required the use of artificial cardiac-related signals to improve the separation of artefactual components. To optimize the separation of cardiac-related artefactual components, we propose a method based on Independent Component Analysis (ICA) that exploits specific features of the real electrocardiographic (ECG) signals that were simultaneously recorded with the neonatal EEG. A total of forty EEG segments from 19-channel neonatal EEG recordings with and without seizures were used to test and validate the performance of our method. We observed a significant reduction in the number of independent components (ICs) containing cardiac-related interferences, with a consequent improvement in the automated classification of the separated ICs. The comparison with the expert labeling of the ICs separately containing electrical cardiac and pulsatile interference led to an accuracy = 0.99, a false omission rate = 0.01 and a sensitivity = 0.93, outperforming existing methods. Furthermore, we verified that true brain activity was preserved in neonatal EEG signals reconstructed after the removal of artefactual ICs, demonstrating the effectiveness of our method and its safe applicability in a clinical context.

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

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.

Relationship Between Early Functional and Structural Brain Developments and Brain Injury in Preterm Infants

Background

Recent studies explored the relationship between early brain function and brain morphology, based on the hypothesis that increased brain activity can positively affect structural brain development and that excitatory neuronal activity stimulates myelination.

Objective

To investigate the relationship between maturational features from early and serial aEEGs after premature birth and MRI metrics characterizing structural brain development and injury, measured around 30weeks postmenstrual age (PMA) and at term. Moreover, we aimed to verify whether previously developed maturational EEG features are related with PMA.

Design/Methods

One hundred six extremely preterm infants received bedside aEEGs during the first 72h and weekly until week 5. 3T-MRIs were performed at 30weeks PMA and at term. Specific features were extracted to assess EEG maturation: (1) the spectral content, (2) the continuity [percentage of spontaneous activity transients (SAT%) and the interburst interval (IBI)], and (3) the complexity. Automatic MRI segmentation to assess volumes and MRI score was performed. The relationship between the maturational EEG features and MRI measures was investigated.

Results

Both SAT% and EEG complexity were correlated with PMA. IBI was inversely associated with PMA. Complexity features had a positive correlation with the cerebellar size at 30weeks, while event-based measures were related to the cerebellar size at term. Cerebellar width, cortical grey matter, and total brain volume at term were inversely correlated with the relative power in the higher frequency bands.

Conclusions

The continuity and complexity of the EEG steadily increase with increasing postnatal age. Increasing complexity and event-based features are associated with cerebellar size, a structure with enormous development during preterm life. Brain activity is important for later structural brain development.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12311-021-01232-z.

Quantitative Preterm EEG Analysis: The Need for Caution in Using Modern Data Science Techniques

Hemodynamic changes during neonatal transition increase the vulnerability of the preterm brain to injury. Real-time monitoring of brain function during this period would help identify the immediate impact of these changes on the brain. Neonatal EEG provides detailed real-time information about newborn brain function but can be difficult to interpret for non-experts; preterm neonatal EEG poses even greater challenges. An objective quantitative measure of preterm brain health would be invaluable during neonatal transition to help guide supportive care and ultimately protect the brain. Appropriate quantitative measures of preterm EEG must be calculated and care needs to be taken when applying the many techniques available for this task in the era of modern data science. This review provides valuable information about the factors that influence quantitative EEG analysis and describes the common pitfalls. Careful feature selection is required and attention must be paid to behavioral state given the variations encountered in newborn EEG during different states. Finally, the detrimental influence of artifacts on quantitative EEG analysis is illustrated.

Detecting bursts in the EEG of very and extremely premature infants using a multi-feature approach

•

Machine learning approach enables accurate detection of bursts in preterm EEG.

•

Features of amplitude and spectral shape capture discriminating information.

•

Improves reliability of estimates of inter-burst intervals.

Aim: To develop a method that segments preterm EEG into bursts and inter-bursts by extracting and combining multiple EEG features. Methods: Two EEG experts annotated bursts in individual EEG channels for 36 preterm infants with gestational age < 30 weeks. The feature set included spectral, amplitude, and frequency-weighted energy features. Using a consensus annotation, feature selection removed redundant features and a support vector machine combined features. Area under the receiver operator characteristic (AUC) and Cohen’s kappa (κ) evaluated performance within a cross-validation procedure. Results: The proposed channel-independent method improves AUC by 4–5% over existing methods (p < 0.001, n=36), with median (95% confidence interval) AUC of 0.989 (0.973–0.997) and sensitivity–specificity of 95.8–94.4%. Agreement rates between the detector and experts’ annotations, κ=0.72 (0.36–0.83) and κ=0.65 (0.32–0.81), are comparable to inter-rater agreement, κ=0.60 (0.21–0.74). Conclusions: Automating the visual identification of bursts in preterm EEG is achievable with a high level of accuracy. Multiple features, combined using a data-driven approach, improves on existing single-feature methods.