TY - JOUR
T1 - eSeiz 2.0
T2 - An Optimized Pulse Exclusion Mechanism for Accurate and Energy-Efficient Seizure Detection in the IoMT
AU - Sayeed, Md Abu
AU - Nasrin, Fatahia
AU - Mohanty, Saraju P.
AU - Kougianos, Elias
N1 - Publisher Copyright:
© 2024, This is a U.S. Government work and not under copyright protection in the US; foreign copyright protection may apply.
PY - 2024/1
Y1 - 2024/1
N2 - Approximately, 50 million people worldwide are impacted by epilepsy, necessitating the development of a seizure detection system that is low power, low latency, and capable of providing accurate and real-time monitoring to address the issue. In this paper, a low-power wearable device for epilepsy has been presented that uses a novel pulse exclusion (PEM) algorithm to characterize seizure and normal activities. PEM resolves the issue of heavy computational burden by reducing the number of channels or features. The feature vectors of reduced size become input to an optimized deep neural network (DNN) classifier for seizure identification. The proposed PEM-based approach has been extensively validated with 10 epileptic subjects obtained from the CHB-MIT Scalp datasets. PEM in combination with DNN classifier shows huge potential in eliminating false detections, and the average specificity of the specified subjects is recorded as 100%, which may be useful for seizure detection and subsequent epilepsy treatment.
AB - Approximately, 50 million people worldwide are impacted by epilepsy, necessitating the development of a seizure detection system that is low power, low latency, and capable of providing accurate and real-time monitoring to address the issue. In this paper, a low-power wearable device for epilepsy has been presented that uses a novel pulse exclusion (PEM) algorithm to characterize seizure and normal activities. PEM resolves the issue of heavy computational burden by reducing the number of channels or features. The feature vectors of reduced size become input to an optimized deep neural network (DNN) classifier for seizure identification. The proposed PEM-based approach has been extensively validated with 10 epileptic subjects obtained from the CHB-MIT Scalp datasets. PEM in combination with DNN classifier shows huge potential in eliminating false detections, and the average specificity of the specified subjects is recorded as 100%, which may be useful for seizure detection and subsequent epilepsy treatment.
KW - Deep neural network
KW - Electroencephalography (EEG)
KW - Energy efficiency
KW - Feature vector
KW - Internet-of-medical-things (IoMT)
KW - Pulse exclusion mechanism (PEM)
UR - http://www.scopus.com/inward/record.url?scp=85181662579&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85181662579&partnerID=8YFLogxK
U2 - 10.1007/s42979-023-02454-0
DO - 10.1007/s42979-023-02454-0
M3 - Article
AN - SCOPUS:85181662579
SN - 2662-995X
VL - 5
JO - SN Computer Science
JF - SN Computer Science
IS - 1
M1 - 165
ER -