Multi-Center Validation Study of Automated Classification of Pathological Slowing in Adult Scalp Electroencephalograms Via Frequency Features

Abstract

Pathological slowing in the electroencephalogram (EEG) is widely investigated for the diagnosis of neurological disorders. Currently, the gold standard for slowing detection is the visual inspection of the EEG by experts, which is time-consuming and subjective. To address those issues, we propose three automated approaches to detect slowing in EEG: Threshold-based Detection System (TDS), Shallow Learning-based Detection System (SLDS), and Deep Learning-based Detection System (DLDS). These systems are evaluated on channel-, segment-, and EEG-level. The three systems perform prediction via detecting slowing at individual channels, and those detections are arranged in histograms for detection of slowing at the segment- and EEG-level. We evaluate the systems through Leave-One-Subject-Out (LOSO) cross-validation (CV) and Leave-One-Institution-Out (LOIO) CV on four datasets from the US, Singapore, and India. The DLDS achieved the best overall results: LOIO CV mean balanced accuracy (BAC) of 71.9%, 75.5%, and 82.0% at channel-, segment- and EEG-level, and LOSO CV mean BAC of 73.6%, 77.2%, and 81.8% at channel-, segment-, and EEG-level. The channel- and segment-level performance is comparable to the intra-rater agreement (IRA) of an expert of 72.4% and 82%. The DLDS can process a 30 min EEG in 4 s and can be deployed to assist clinicians in interpreting EEGs.

Time–Frequency Decomposition of Scalp Electroencephalograms Improves Deep Learning-Based Epilepsy Diagnosis

“…We also achieved LOIO BAC of 76.1% and LOSO BAC of 69.3% with an ensemble of three components: ConvNet for detecting IEDs, Template Matching for detecting IEDs, and Spectral features for classifying EEGs. 10 In another multi-center study, Peh et al 25 proposed an EEG classifier based on slowing and reported an LOIO BAC 82.0% and LOSO BAC of 81.8%. Marleen et al 31 achieved an AUC of 0.94 with 2D ConvNet for IED detection.…”

“…Marleen et al 31 achieved an AUC of 0.94 with 2D ConvNet for IED detection. In these five studies, 6,8,10,25,30 the ConvNet detector is investigated with only the preprocessed EEG signal. In our preliminary study, we proposed a 1D ConvNet that exploited frequency sub-bands as features for the detection of IEDs, and we tested it on a dataset containing 554 EEGs.…”

Multi-center Assessment of CNN-Transformer with Belief Matching Loss for Patient-independent Seizure Detection in Scalp and Intracranial EEG

“…We perform seizure detection őrst at individual channels (channel-level detection), next at multi-channel segments (segment-level detection), and at last, we detect the start and end points of the seizures in the entire multi-channel EEG (EEG-level detection) [65][66][67] (see Supplementary Figure 1). The pipeline of the proposed seizure detector is displayed in Figure 3.…”

“…We assess the channel-and segment-level seizure classiőers through the following metrics: accuracy (ACC), balanced accuracy (BAC), sensitivity (SEN), speciőcity (SPE), F1 score (F1), and expected calibration error (ECE) 70 . As the seizure and non-seizure classes are imbalanced, we evaluate the results mainly in terms of BAC 67 .…”

“…Also, we implemented class weights that are inversely proportional to the class frequency in the training data during training. This allows us to optimize the loss function on an imbalanced dataset without overőtting67 . Finally, we optimized parameters within the CNN via nested CV on the training data, with an 80:20% split for training and validation.…”