Related Experiment Video
Updated: Aug 11, 2026
09:32
Network Analysis of Foramen Ovale Electrode Recordings in Drug-resistant Temporal Lobe Epilepsy Patients
Published on: December 18, 2016
CosFNet: A Lightweight Epileptic EEG Detection Model Based on Cosine Convolution and FNet
Jiajun Tian1, Yazhou Zhao2, Weidong Zhou1,3
1School of Integrated Circuits, Shandong University, Jinan 250199, China.
Bioengineering (Basel, Switzerland)
|July 28, 2026
Summary
A new lightweight deep learning model, CosFNet, efficiently detects seizures using electroencephalography (EEG) with minimal parameters. This enables advanced epilepsy diagnosis and monitoring on resource-constrained devices.
Area of Science:
- Neurology
- Artificial Intelligence
- Biomedical Engineering
Background:
- Epilepsy is a common neurological disorder requiring continuous monitoring via electroencephalography (EEG).
- Current deep learning seizure detection models are computationally intensive, hindering use on limited-resource platforms.
Purpose of the Study:
- To develop a lightweight deep learning architecture for efficient and accurate automatic seizure detection.
- To enable the deployment of advanced seizure detection on resource-constrained clinical devices.
Main Methods:
- Proposed CosFNet, a hybrid lightweight model combining cosine convolution and an FNet encoder.
- Cosine convolution captures local spatiotemporal features efficiently.
- FNet encoder uses a parameter-free 2D discrete Fourier transform for global token mixing, achieving high efficiency with only 19,458 parameters.
Main Results:
- Achieved high performance on the CHB-MIT dataset: 97.60% segment sensitivity, 97.12% specificity, and 0.82/h FDR.
- Attained strong results on the SH-SDU dataset: 92.87% sensitivity and 94.74% specificity.
- Demonstrated competitive detection accuracy with significantly reduced model complexity.
Conclusions:
- CosFNet offers a viable solution for clinical epilepsy diagnosis and monitoring in resource-limited settings.
- The model's low complexity and high performance make it suitable for real-time applications on edge devices.
- This research paves the way for broader accessibility of advanced seizure detection technology.
