How Much Data is Enough? Optimization of Data Collection for Artifact Detection in EEG Recordings

TL;DR

This work tackles the cost and efficiency of collecting EEG/EMG data for artifact detection by adopting a data-oriented design and optimizing artifact-task selection. Using mel-spectrogram features derived from EEG-based EMG channels and three CNN-based architectures, the authors evaluate subject-specific and generalized models across three analyses, reducing artifact tasks from twelve to three and repeating tasks from ten to as few as one. Key findings show high recall for certain artifact classes and substantially improved specificity when excluding certain occipitalis contractions, though generalization across subjects remains limited due to small sample size. The study provides a practical framework for economical data collection in EEG/EMG artifact detection and highlights avenues for future work, including larger datasets and integration of spatial connectivity information to further boost performance.

Abstract

Objective. Electroencephalography (EEG) is a widely used neuroimaging technique known for its cost-effectiveness and user-friendliness. However, various artifacts, particularly biological artifacts like Electromyography (EMG) signals, lead to a poor signal-to-noise ratio, limiting the precision of analyses and applications. The currently reported EEG data cleaning performance largely depends on the data used for validation, and in the case of machine learning approaches, also on the data used for training. The data are typically gathered either by recruiting subjects to perform specific artifact tasks or by integrating existing datasets. Prevailing approaches, however, tend to rely on intuitive, concept-oriented data collection with minimal justification for the selection of artifacts and their quantities. Given the substantial costs associated with biological data collection and the pressing need for effective data utilization, we propose an optimization procedure for data-oriented data collection design using deep learning-based artifact detection. Approach. We apply a binary classification between artifact epochs (time intervals containing artifacts) and non-artifact epochs (time intervals containing no artifact) using three different neural architectures. Our aim is to minimize data collection efforts while preserving the cleaning efficiency. Main results. We were able to reduce the number of artifact tasks from twelve to three and decrease repetitions of isometric contraction tasks from ten to three or sometimes even just one. Significance. Our work addresses the need for effective data utilization in biological data collection, offering a systematic and dynamic quantitative approach. By providing clear justifications for the choices of artifacts and their quantity, we aim to guide future studies toward more effective and economical data collection in EEG and EMG research.

Figures (14)

• Figure 1: EEG channel positions on the head. The marked channels are used to derive the according EMG signals.
• Figure 1: Misclassification of analysis #1 isometric contractions: One person, one model using Vision Transformer Small. The x-axis shows the subject ID and the number of task repetitions. data_size_val_total includes both EO and artifact epochs in the validation data. Artifacts are labeled with their IDs (refers to \ref{['tab_EMG_artifact_ID']}).
• Figure 2: (a) Illustration of a mel-spectrogram depicting an artifact epoch, particularly, the jaw tensing task. The EMG channels are concatenated along the x-axis, representing time within each channel area. (b) Zoomed-in view of channel F9-F7 from (a). (c) Exemplary mel-spectrogram showcasing a non-artifact epoch from EO recordings.
• Figure 2: Misclassification of analysis #1 isometric contractions: Subject cross-validation using Vision Transformer Small. The x-axis shows the number of total subjects and the number of task repetitions. data_size_val_total includes both EO and artifact epochs in the validation data. Artifacts are labeled with their IDs (refers to \ref{['tab_EMG_artifact_ID']}).
• Figure 3: Overall results from all three analyses with models using all repetitions and all subjects in the form of the average of all three DL architectures (also refer to Tables \ref{['tab_result_sum_full']} and \ref{['tab_result_sum_selected']} in Appendix \ref{['Appendix_result_sum_tables']}).