Machine Learning Characterization of Ictal and Interictal States in EEG Aimed at Automated Seizure Detection

Abstract

Background: The development of automated seizure detection methods using EEG signals could be of great importance for the diagnosis and the monitoring of patients with epilepsy. These methods are often patient-specific and require high accuracy in detecting seizures but also very low false-positive rates. The aim of this study is to evaluate the performance of a seizure detection method using EEG signals by investigating its performance in correctly identifying seizures and in minimizing false alarms and to determine if it is generalizable to different patients. Methods: We tested the method on about two hours of preictal/ictal and about ten hours of interictal EEG recordings of one patient from the Freiburg Seizure Prediction EEG database using machine learning techniques for data mining. Then, we tested the obtained model on six other patients of the same database. Results: The method achieved very high performance in detecting seizures (close to 100% of correctly classified positive elements) with a very low false-positive rate when tested on one patient. Furthermore, the model portability or transfer analysis revealed that the method achieved good performance in one out of six patients from the same dataset. Conclusions: This result suggests a strategy to discover clusters of similar patients, for which it would be possible to train a general-purpose model for seizure detection.

Training Datasets for Epilepsy Analysis: Preprocessing and Feature Extraction from Electroencephalography Time Series

“…Table 2 lists the related research efforts, which primarily focused on the detection of epileptic seizures (for further details, see Section 4). Specifically, the work cited in [4] provides a comprehensive account of the seizure detection analyses, efforts to reduce false alarms, and the portability of models, conducted using the training datasets we make openly available. Table 2 also presents the performance metrics of models trained with ML algorithms (k-NN, MLP, SVM, BayesNet) using the datasets described in this paper, highlighting the effectiveness of our methodology in creating them.…”

“…The datasets discussed in this study, in whole or in part, served as the training sets for developing the detection models described in [4]. We would like to highlight the adaptability of our proposed TrB tool, which allows for the extraction of more detailed data by adjusting its parameters, like the windowing temporal parameters L and S. Should users need more specialized data, we are open to providing them upon request, underlining our commitment to supporting collaborative research in the field of Data Science for epilepsy analysis.…”

“…methods) = 1080 (1) All the datasets that we make available are obtained considering length L = 2 and sliding S = 1 as temporal parameters of the sliding window. This choice follows the methodology described in [4]. A selection of windowing time parameters L and S is reported in [6].…”

“…• Bj is the j-th band number, with j = 8, 13, 21, 30, 40, 70. Each of these band numbers corresponds, respectively, to: α (8-13 Hz), β 1 (13-21 Hz), β 2 (21-30 Hz), low γ (30-40 Hz), medium γ (40-70 Hz) and high γ (70-120 Hz) bands[4]. •…”

EEG seizure detection: concepts, techniques, challenges, and future trends

“…This approach included dense convolutional blocks, feature attention modules, residual blocks, and the hypercolumn technique [49]. Gaetano Zazzaro and Luigi Pavone evaluate the performance of a seizure detection system by studying its performance in correctly identifying seizures and in minimizing false alarms and to decide if it is generalizable to several patients [146]. In [74] explore the possibilities of wearable multimodal monitoring in epilepsy and identify effective strategies for seizure-detection.…”