For the health monitoring of epilepsy patients in multiple scenarios,a wireless wearable EEG sensor is designed that can iden-tify seizures locally and in real time. Firstly,the analog-to-digital converter of ADS1299 is used to sample EEG signals. Simultaneously, EEG signals are filtered in real time using a 50-Hz notch filter and a band-pass filter by a local microprocessor of STM32F4. Afterward, Hjorth features are extracted periodically in different frequency bands of the filtered EEG signals to obtain a feature vector. Eventually, the epileptic seizure status is recognized by using Support Vector Machine( SVM) and is sent to external devices via low-power Bluetooth ( BLE) . External devices will take necessary actions to inform guardians when epileptic seizures are appearing. The data transmission and algorithm execution are optimized by taking the advantage of the architecture and interrupt mechanism of STM32 to realize real-time seizure detection. The validation is tested on CHB-MIT epileptic dataset. The result shows that the sensor can recognize the epileptic seizure status in real time and effectively with an average recognition accuracy of 97％.
维普
万方数据