APSIPA Transactions on Signal and Information Processing > Vol 13 > Issue 1

Simple yet Effective Video-Based Epileptic Tonic-Clonic Seizure Detection

Yoshinao Yazaki, Graduate School of BASE, Tokyo University of Agriculture and Technology, Japan, Satsuki Watanabe, Department of Psychiatry, Saitama Medical University Hospital, Japan, Yuichi Tanaka, Graduate School of Engineering, Osaka University, Japan, ytanaka@comm.eng.osaka-u.ac.jp
 
Suggested Citation
Yoshinao Yazaki, Satsuki Watanabe and Yuichi Tanaka (2024), "Simple yet Effective Video-Based Epileptic Tonic-Clonic Seizure Detection", APSIPA Transactions on Signal and Information Processing: Vol. 13: No. 1, e12. http://dx.doi.org/10.1561/116.20240012

Publication Date: 21 Aug 2024
© 2024 Y. Yazaki, S. Watanabe and Y. Tanaka
 
Subjects
 
Keywords
Epilepsytonic-clonic seizurevideo analysisimage processing, and image feature extraction
 

Share

Open Access

This is published under the terms of CC BY-NC.

Downloaded: 143 times

In this article:
Introduction 
Related Work 
Method 
Results and Discussion 
Conclusion and Future Research 
References 

Abstract

Epilepsy, a prevalent neurological disorder, often leads to tonic-clonic seizures characterized by loss of consciousness and uncontrolled motor activity. Prompt detection of these seizures is crucial for effective nursing and diagnosis. This paper introduces a novel epileptic seizure detection method leveraging low-complexity video analysis, eliminating the need for body attachments or special equipment like markers or specific clothing. Our approach is straightforward: each video frame is segmented into blocks, and the average values of these blocks are computed. We then analyze the temporal changes in these averages using spectrograms. Our findings indicate that during tonic-clonic seizures, dominant frequency components typically range from 1 to 6 Hz and decrease as the seizure progresses. By capitalizing on these clinical observations, we have formulated effective detection rules. Experimental evaluations reveal that our method not only accurately detects epileptic seizures but also operates approximately four times faster than real-time on standard desktop computers. This efficiency and accuracy underscore the potential of our method as a practical tool in epilepsy monitoring and management.

DOI:10.1561/116.20240012