Temporal EEG Imaging for Drowsy Driving Prediction

Cheng, Eric Juwei; Young, Ku-Young; Lin, Chin-Teng · 2019 · Crossref

DOI: 10.3390/app9235078

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study addresses the challenge of accurately predicting drowsy driving, a major cause of vehicle accidents. Drowsiness is an ambiguous state that does not occur instantaneously, making it difficult to identify using single-point electroencephalography (EEG) measurements. While EEG is a robust tool for monitoring brain dynamics, raw signals are often obscured by artifacts, and traditional feature extraction methods (like Fast Fourier Transform) fail to capture spatial information. Furthermore, manual artifact removal techniques like Independent Component Analysis are unsuitable for real-time applications. To overcome these limitations, the authors propose a temporal EEG imaging method that leverages deep learning to predict driver drowsiness without manual artifact removal. The experimental design involved 38 healthy young adults performing a 90-minute driving task in a virtual reality environment simulating night-time highway driving. Participants engaged in an event-related lane-keeping task where random lane departures occurred every 5–10 seconds. Drowsiness was quantified by response time (RT), defined as the time taken to steer the vehicle back to the center lane after a deviation. EEG data were recorded using 32 electrodes. The raw signals were filtered and down-sampled, then transformed into the frequency domain using FFT. The mean power of theta, alpha, and beta bands was extracted and mapped to RGB pixel values. These values were interpolated using the Clough–Tocher scheme to generate 32×32 color images representing spatial brain activity. Crucially, the proposed method created "temporal EEG images" by linearly combining a sequence of five consecutive EEG images, weighting recent frames more heavily than older ones, to capture temporal dynamics. These images were fed into a Convolutional Neural Network (CNN) for classification. The results demonstrated that incorporating temporal information significantly improved prediction performance. By analyzing a sequence of EEG images rather than a single time point, the model could better distinguish between alert and drowsy states, even when instantaneous EEG patterns were similar. The temporal approach accounted for the gradual decline in vigilance and the potential for sudden recovery, which single-frame analysis often missed. The study confirmed that combining spatial EEG features with temporal behavior via CNNs provides a more robust and accurate method for drowsiness detection compared to traditional single-point image-based methods. The significance of this work lies in its contribution to real-time brain-computer interface applications for safety-critical systems. By eliminating the need for manual artifact removal and leveraging the temporal nature of drowsiness, the proposed method offers a viable solution for automatic, real-time drowsy driving prediction. This approach enhances the reliability of EEG-based monitoring systems, potentially reducing traffic accidents caused by driver fatigue.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-nvidia on 2026-08-10; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• drowsiness detection algorithms
• drowsiness
• microsleep

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Empirical Findings: physiological data
• Methodological Resource: tool software
• Theoretical Contribution: computational model