動作電位訊號分類錯誤對於放電時間序列的頻率特性之影響

神經動作電位波形分類(spike sorting)為神經電生理領域中相當重要的一個議題。目前已知演算法均無法正確的分類動作電位波形，不同程度的分類錯誤(spike sorting error)均有可能發生，而這些錯誤對於神經訊號的各種分析也會造成各種不同程度的影響，如因果性、消息量(Entropy)等等。放電頻率的分析是了解神經訊號特性的重要基礎，動作電位分類的錯誤究竟會在分析放電頻率上產生多大影響，是本論文關心的議題，我們將利用時頻分析的方法，針對動作電位分類錯誤是否會破壞時頻圖的特徵來進行研究。因為動作電位訊號的非等間隔取樣特性，我們無法利用一般應用於均勻間隔取樣的時頻分析方式來進行分析。近幾年來在天文領域中，發展出了針對非等間隔取樣的行星週期訊號之時頻分析方法:加權小波Z轉換(Weighted Wavelet Z-transform, WWZ )。在本研究當中，我們將此方法引進神經科學領域當中，驗證其在神經動作電位訊號時頻分析的可靠性。我們將試著比較不同種類的分類誤差模型與錯誤率的動作電位訊號之時頻圖，歸納出給予神經科學家在神經動作電位分類上的建議。

關鍵字

神經動作電位；神經動作電位分類錯誤；非等間隔取樣訊號；時頻分析；加權小波Z轉換

並列摘要

Spike sorting as a topic in the field of neuron science is a crucial portion on analyzing neuronal activities. For various algorithms and conditions, the spike sorting is impossible free of errors during the classification. The spike sorting errors produce a great impact on the analysis of neural signals, such as causality and entropy. Here, we concerned on the effects of the spike sorting errors on influencing the frequency characteristic of a spike train whenever a time-frequency analysis used to be applied on the analysis of the pattern on spectrogram of spike train. In the time-frequency analysis, it is not straight forward for an uneven spike train which means that the sampling intervals are not identical, for instance, Short Time Fourier Transform (STFT) cannot be directly applied. Fortunately, a new method of time–frequency analysis for un-evenly sampled signals called “Weighted Wavelet Z-transform (WWZ)” has been developed for analyzing the period of a pulsar in astronomy in recent. In this study, we will introduce WWZ to neuron science and demonstrate its performance and reliability for time-frequency analysis on neural spike train through simulations. We construct some neural spike train model and introduce several types of errors on the proposed models. Then using WWZ to analyze them, we further compare the spectrograms with difference spike sorting errors. Through this study, our observation results could be a useful guideline for neuroscientists on spike sorting approach.

並列關鍵字

Action Potential ； Spike Sorting Error ； Time-Frequency Analysis ； Weighted Wavelet Z-transfrom

參考文獻

[1] Hammond, Constance, Hagai Bergman, and Peter Brown. "Pathological synchronization in Parkinson's disease: networks, models and treatments." Trends in neurosciences, vol.30, pp.357-364, 2007.

[3] Beeler, Go W.; Reuter, H. “Reconstruction of the action potential of ventricular myocardial fibres,” The Journal of physiology, vol.268, pp.177-210, 1977.

[5] Brown, Emery N.; Kass, Robert E.; Mitra, Partha P. “Multiple neural spike train data analysis: state-of-the-art and future challenges.” Nature neuroscience, vol.7, pp.456-461, 2004.

[6] Shao, Pei-Chiang, et al. “Effects of spike sorting error on the Granger causality index.” Neural Networks, vol.46, pp249-259, 2013.

[7] Won, Deborah S.; Chong, David Y.; Wolf, Patrick D. “Effects of spike sorting error on information content in multi-neuron recordings.” In: Neural Engineering, 2003. Conference Proceedings. First International IEEE EMBS Conference on. IEEE., pp.618-62. ,2003.

國際替代計量

動作電位訊號分類錯誤對於放電時間序列的頻率特性之影響

全文下載

主題瀏覽