Proceedings of the KOSOMBE Conference (대한의용생체공학회:학술대회논문집)
- Volume 1997 Issue 11
- /
- Pages.99-102
- /
- 1997
Classification of Multi-Unit Neural Action Potential by Template Learning
학습 가능한 실시간 다단위 신경 신호의 분류에 관한 연구
Abstract
A neural spike sorting technique has been developed that also has the capability of template learning. A system of software has been written that first obtains the templates by learning, and then performs the sorting of the spikes into single units. The spike sorting can be done in real time. The template learning consists of spike detection based on the discrete Haar transform (DHT), feature extraction by clustering of spike amplitude and duration, classification based on rms error, and fabrication of templates. The developed algorithms can be implemented into real time systems using digital signal processors.
Keywords