Real-time Epileptic Seizure Detection Using Hybrid SNN and ConvSNN Architectures

#Biomedical #engineering #EEG #Epileptic #seizure #embedded #systems #Spiking #neural #networks #Neuromorphic #computing. #monitoring #device
Share

Epileptic seizure detection is a long-standing challenge in consumer and clinical neurotechnology. Conventional deep-learning models offer high accuracy but remain computationally heavy and unsuitable for real-time, low-power deployment in wearable and bedside monitoring devices.
This talk presents a new neuromorphic processing framework that combines event-based spike encoding with lightweight spiking neural architectures for fast, reliable, real-time EEG analysis.

Key Topics Covered:

  • Event-based EEG preprocessing using hybrid Signal-to-Spike encoding
  • Design of Hybrid SNN and ConvSNN architectures for streaming 0.5-second windows
  • Real-time seizure detection without lookahead or future context
  • Comparative study vs. 1D-CNN baseline on the CHB-MIT dataset
  • Deployment pathways toward neuromorphic hardware and consumer health devices

Why It Matters:
This neuromorphic approach achieves millisecond-scale decision latency and is designed for continuous monitoring applications such as patient-side seizure alarms, wearable neurotech, and embedded smart EEG systems.


  Date and Time

  Location

  Hosts

  Registration


  • Add_To_Calendar_icon Add Event to Calendar

Loading virtual attendance info...

  • Contact Event Host
  • Starts 25 November 2025 01:00 PM UTC
  • Ends 01 December 2025 01:00 PM UTC
  • No Admission Charge

  Speakers

Topic:

Real-time Epileptic Seizure Detection Using Hybrid SNN and ConvSNN Architectures

Overview:
Epileptic seizure detection is a long-standing challenge in consumer and clinical neurotechnology. Conventional deep-learning models offer high accuracy but remain computationally heavy and unsuitable for real-time, low-power deployment in wearable and bedside monitoring devices.
This talk presents a new neuromorphic processing framework that combines event-based spike encoding with lightweight spiking neural architectures for fast, reliable, real-time EEG analysis.

Key Topics Covered:

  • Event-based EEG preprocessing using hybrid Signal-to-Spike encoding
  • Design of Hybrid SNN and ConvSNN architectures for streaming 0.5-second windows
  • Real-time seizure detection without lookahead or future context
  • Comparative study vs. 1D-CNN baseline on the CHB-MIT dataset
  • Deployment pathways toward neuromorphic hardware and consumer health devices

Why It Matters:
This neuromorphic approach achieves millisecond-scale decision latency and is designed for continuous monitoring applications such as patient-side seizure alarms, wearable neurotech, and embedded smart EEG systems.

 

Address:Australia