NeuroDETR improves detection of brain signals in continuous EEG recordings
NeuroDETR: Multiscale Discrete Spatiotemporal Tokenization for Asynchronous EEG Event Detection
Machine Learning
Summary
Detecting specific brain activity in ongoing EEG recordings is hard because the signals are mixed with background noise and happen at unpredictable times. The authors designed NeuroDETR, a system that breaks down EEG data into meaningful chunks across time and space to better spot these brain events. It uses a clever attention method to focus on likely event times while still considering the whole recording. Tests across many datasets showed NeuroDETR can find brain events more accurately than previous methods.
What this means in practice
- •For brain-computer interface developers: Enable accurate real-time detection of brain events without needing predefined time windows.$Commercial implications: Improves BCI products by allowing continuous, precise recognition of neural activity events, enhancing user experience and reliability.
- •For neurological monitoring teams: Improve automated detection of meaningful neural events during long EEG recordings for patient monitoring.
Authors
Yaozhi Wen, Kun Zhao, Wei Wei, Haoyang Chen, Kangning Wang, Yi Sun, Shaowei Gu, Dingkun Liu, Shuang Qiu, Huiguang He
Abstract
Asynchronous EEG event detection supports practical brain-computer interfaces by identifying event classes and temporal boundaries directly from continuous recordings without predefined event boundaries at inference time. The primary challenge lies in extracting event-related patterns embedded in ongoing background activity. To address this, we propose \textbf{NeuroDETR}, a discrete spatiotemporal multiscale framework that combines structured tokenization with event-aware feature refinement. Multiscale temporal codebooks capture brief responses and sustained dynamics, while Spatial-Pyramid VQ hierarchically encodes spatial structure across EEG channels. These complementary representations form a compact event memory for jointly classifying events and estimating their onsets and offsets. Before tokenization, we apply FOCAL, an event-aware sparse attention module that uses predicted event presence to guide temporal feature refinement and spatial aggregation, emphasizing likely event periods while preserving context across the recording. Experiments on eight EEG datasets that span diverse paradigms demonstrate state-of-the-art performance. These findings provide a basis for the development of BCI systems that recognize meaningful neural activity during continuous interaction and monitoring.