Random Forest-based Algorithm for Sleep Spindle Detection in Infant EEG

Abstract

Sleep spindles are associated with normal brain development, memory consolidation and infant sleep-dependent brain plasticity and can be used by clinicians in the assessment of brain development in infants. Sleep spindles can be detected in EEG, however, identifying sleep spindles in EEG recordings manually is very time-consuming and typically requires highly trained experts. Research on the automatic detection of sleep spindles in infant EEGs has been limited to-date. In this study, we present a novel supervised machine learning-based algorithm to detect sleep spindles in infant EEG recordings. EEGs collected from 141 ex-term born infants and 6 ex-preterm born infants, recorded at 4 months of age (adjusted), were used to train and test the algorithm. Sleep spindles were annotated by experienced clinical physiologists as the gold standard. The dataset was split into training (81 ex-term), validation (30 exterm), and testing (30 ex-term + 6 ex-preterm) set. 15 features were selected for input into a random forest algorithm. Sleep spindles were detected in the ex-term infant EEG test set with 92.1% sensitivity and 95.2% specificity. For ex-preterm born infants, the sensitivity and specificity were 80.3% and 91.8% respectively. The proposed algorithm has the potential to assist researchers and clinicians in the automated analysis of sleep spindles in infant EEG.

Spontaneous Activity Predicts Survival of Developing Cortical Neurons

“…This likely reflects differences in neuronal subtypes or maturational stages across individual neurons and indicates the multiplicity of factors regulating cell death. To identify these and reveal predictive features of survival or cell death of individual neurons, we applied machine learning, which use as diagnostic and prognostic tool is recently emerging to assess brain development based on EEG recordings of preterm infants ( Wei L et al, 2020 ; for review, see Tataranno et al, 2021 ) and for early prediction of spontaneous activity in cortical networks in vitro ( Cabrera-Garcia et al, 2021 ). The overall good performance of the applied classifiers demonstrated how the survival fate of immature neurons is predictable based on activity features from single neuron, cluster and network level.…”

Hierarchical fusion detection algorithm for sleep spindle detection

“…This method could be popularized for clinical disease diagnosis instead of artificial spindle detection as it improves the speed of disease diagnosis and enables patients to receive rapid treatment ( Imtiaz and Rodriguez-Villegas, 2014 ). At the same time, according to this test, the study of spindles on human intelligence and memory can save substantial experimental time ( Wei et al, 2020 ). Therefore, effective and rapid spindle detection method is a common research direction.…”

Spindle-AI: Sleep Spindle Number and Duration Estimation in Infant EEG

“…The EEG data were filtered using a 4 th order Butterworth filters (IIR) within the frequency bands of interest: delta (0-4 Hz), theta (4-8 Hz), alpha (8-12 Hz), sigma (12.5-15 Hz), sleep spindle (10.5-16 Hz) [40] and beta (12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30). The mean absolute amplitude, signal envelope (estimated using the Hilbert transform), relative power in each band and absolute power in each band was estimated for each epoch.…”

“…The approach combines several randomized decision trees and aggregates their predictions by averaging. The study builds on our previous preliminary work to develop a method to detect sleep spindles in infant EEGs [24]. Spindle-AI has been implemented as a web server freely available for academic use at http://lisda.ucd.ie/Spindle-AI/.…”