Low-power processors for implantable epileptic seizure detection system
Epilepsy is a neurological disorder affecting around 50 million people in the world. It is characterized by seizure which results in the loss of patient consciousness. Despite several treatments are available to suppress the seizure such as anti-epileptic drugs and surgery, around 25% of the epilept...
Saved in:
Main Author: | |
---|---|
Other Authors: | |
Format: | Final Year Project |
Language: | English |
Published: |
2013
|
Subjects: | |
Online Access: | http://hdl.handle.net/10356/54581 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Institution: | Nanyang Technological University |
Language: | English |
Summary: | Epilepsy is a neurological disorder affecting around 50 million people in the world. It is characterized by seizure which results in the loss of patient consciousness. Despite several treatments are available to suppress the seizure such as anti-epileptic drugs and surgery, around 25% of the epileptic patients still experiences the seizure. It does mean that one quarter of the total patients still suffer incurable seizure. A reliable real time epileptic seizure detection system is required to help and alert the patients about the incoming seizure. This final year project explores the design of the epileptic seizure detection system and utilizing the energy of EEG signal as the feature for detection system.
Extreme Learning Machine (ELM) has gained some attentions recently, due to the fact that learning speed of ELM is really fast compared to the traditional learning algorithm. In this project ELM is used as the classifier for epileptic seizure detection system. Implementing second order filter decreases the percentage of false alarm detection as compared to first order filter. The result shows that 100% sensitivity, 3 seconds latency and 9.4% false detection percentage are achieved using 21 hours of EEG data having 6 seizures for second order filter.
The design of epileptic seizure detection system is implemented using operational transconductance amplifier (OTA-C) filter. Second order low pass filter is designed using OTA-C filter. The operational transconductance amplifier is operated under sub-threshold condition. Cascode technique is employed to increase the output resistance of the operational transconductance amplifier, and thus it does increase the open-loop gain of the operational transconductance amplifier. Harmonic distortion is also become the issue when operational transconductance amplifier is designed. In this project, a technique called bump linearization circuit is employed to reduce the total harmonic distortion of transconductance amplifier during the transient analysis. The result shows that the total harmonic distortion is greatly reduced after bump linearization circuit is implemented to the transconductance amplifier. |
---|