Advertisement

Time–Frequency Methods and Brain Rhythm Signal Processing

  • Jesse Gillis
Chapter
Part of the Springer Series in Computational Neuroscience book series (NEUROSCI, volume 2)

Abstract

Physiological signal analysis could encompass any of the methods of signal analysis being brought to bear upon a problem in physiology. In practice, physiological signal analysis is a narrower field than this because physiological signals share common features which limit the applicability of different methods. An important property of physiological signals is that they typically possess features determined by their historical contribution to organism fitness. This gives rise to the problem of identifying and classifying features present in a physiological signal. The primary statistical feature of physiological signals is that they are nonstationary, meaning their statistical characteristics change with time, and a variety of different techniques exist to analyze such data. The suitability of such techniques is data dependent, and particularly upon the degree and manner in which the signal changes in time and frequency. This is a well-known problem and specialized techniques have been developed for particular data sets (e.g., Achermann and Borbély, 1998). Developing data-specific methodologies presents its own difficulties in terms of hypothesis and verifiability (e.g., separation into training and testing sets for the methodology). In this article, I focus on introducing standard and well-defined techniques relevant to signal processing of brain rhythms as well as the organization of time–frequency activities in a hippocampal rhythm. In general, analyses consist of three stages: (1) Transformation into the time–frequency domain; (2) Quantification of properties in the time–frequency domain; and (3) Grouping together of these properties. For example, wavelet transform (time–frequency transform) coefficients (quantification) might be extracted and clustered (grouping) (Quiroga et al., 2002). Filtering of a neurophysiological signal might be performed (an implicit time–frequency step), a threshold chosen (an implicit quantification), and the remaining data averaged (grouping). Below, I suggest a generally useful way to incorporate these three steps into analyses with a minimum of additional assumptions, i.e., a time–frequency transformation, followed by characterization of the time–frequency domain’s local activity, and then mixture distribution analysis.

Keywords

Frequency Resolution Mother Wavelet Wigner Distribution Brain Rhythm Spike Sorting 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

References

  1. Abeyeskera R, Bolton R, Westphal L, and Boashash B (1986) Patterns in Hilbert transforms and Wigner-Ville distributions of electrocardiograms data. Proc IEEE ICASSP 1793–1796.Google Scholar
  2. Achermann P, and Borbély AA (1998) Coherence analysis of the human sleep electroencephalogram. Neuroscience 85: 1195–208.PubMedCrossRefGoogle Scholar
  3. Akay M (Ed.), (1998) Time Frequency and Wavelets in Biomedical Signal Processing. New York: IEEE, Inc.Google Scholar
  4. Baase S, Van Gelder S (2000) Computer Algorithms: Introduction to Design and Analysis. Reading, Massachusetts: Addison-Wesley, 3rd edn.Google Scholar
  5. Baraniuk RG, Jones DJ (1993) Signal-dependent time-frequency representation: Optimal kernel design. IEEE Trans Signal Proc 41: 1589–1602.CrossRefGoogle Scholar
  6. Brooks RR, Grewe L, Lyengar SS (2001) Recognition in the wavelet domain: a survey. J Electron Imaging 10: 757–784.CrossRefGoogle Scholar
  7. Brown ML, Williams WJ, Hero III AO (1994) Nonorthogonal Gabor representations of biological signals. Proc Int Conf Acoustics, Speech, Signal Process 4: 305–308.Google Scholar
  8. Buzsáki G, Buhl DL, Harris KD, Csicsvari J, Czh B, Morozov A (2003) Hippocampal network patterns of activity in the mouse. Neuroscience 116: 201–211.CrossRefGoogle Scholar
  9. Choi HI, Williams WJ (1989) Improved time-frequency representation of multicomponent signals using Exponential kernels. IEEE Trans Acoust, Speech, Signal Proc ASSP-37 6: 862–871.CrossRefGoogle Scholar
  10. Claasen TC, Mecklenbrauker WG (1980) The Wigner distribution – A tool for time-frequency signal analysis. Philips J Res 35: 276–300, 217–250.Google Scholar
  11. Cohen, L (1989) Time-frequency distributions – A review. Proc IEEE 77: 941–981.CrossRefGoogle Scholar
  12. Cohen L, and Posch TE (1985) Generalized ambiguity function. IEEE Int Conf Acoust, Signal Process 1033–1036.Google Scholar
  13. Daubechies I (1990) The wavelet transform, time-frequency localization and signal analysis. IEEE Trans Inf Theory 36: 961–1005.CrossRefGoogle Scholar
  14. Dempster AP, Laird NM, Rubin DB (1977) Maximum likelihood from incomplete data via the EM Algorithm. J R Stat Soc B 39:1–38.Google Scholar
  15. Devedeux D, Duchene J (1994) Comparison of various time/frequency distributions applied to synthetic uterine EMG signals. IEEE Int Symp Time-Frequency Time-Scale Anal 572–575.Google Scholar
  16. Everitt BS, and Hand DJ (1981) Finite Mixture Distributions. London, New York: Chapman and Hall.Google Scholar
  17. Flury B (1997) A First Course in Multivariate Statistics. New York: Springer-Verlag.Google Scholar
  18. Gray CM, Konig P, Engel AK, Singer W (1989) Oscillatory responses in cat visual cortex exhibit inter-columnar synchronization which reflects global stimulus properties. Nature 338: 334–337.PubMedCrossRefGoogle Scholar
  19. Jones DL, Parks TW (1992) A resolution comparison of several time-frequency representations. IEEE Signal Proc Mag 9: 21–67.CrossRefGoogle Scholar
  20. Jeong J, Williams WJ (1990) Variable windowed spectrograms: connecting Cohen’s class and the wavelet transform. IEEE ASSP Workshop on Spectrum Estimation and Modeling, 270–273.CrossRefGoogle Scholar
  21. Kadambe S, Boudreaux-Bartels GF (1992) A comparison of the existence of “cross terms” in the Wigner distribution and the squared magnitude of the wavelet transform and the short time Fourier transform. IEEE Trans Signal Proc 40: 2498–2517.CrossRefGoogle Scholar
  22. Letelier JC, Weber P (2000) Spike sorting based on discrete wavelet transform coefficients. J Neurosci Meth 101: 93–106.CrossRefGoogle Scholar
  23. Lingras P, Huang X (2005) Statistical, evolutionary and neurocomputing clustering techniques: Cluster-based vs. object-based approaches. Artif Intell Rev 23: 3–29.CrossRefGoogle Scholar
  24. Loughlin P, Pitton J, Atlas L (1994) Construction of positive time-frequency distributions. IEEE Trans Signal Proc 42: 2697–2705.CrossRefGoogle Scholar
  25. Marr D, Hildreth E (1980) Theory of edge detection. Proc Roy Soc London B 207: 187–217.CrossRefGoogle Scholar
  26. Navarro R, Tabernero A, Cristobal G. (1995) Image representation with Gabor wavelets and its applications. In P. W. Hawkes (Ed.), Advances in Imaging and Electron Physics. Orlando, FL: Academic.PubMedCrossRefGoogle Scholar
  27. Nicolelis MA, Baccala LA, Lin RCS, Chapin JK (1995) Sensorimotor encoding by synchronous neural ensemble activity at multiple levels of the somatosensory system. Science 268: 1353–1358.PubMedCrossRefGoogle Scholar
  28. Papandreou A, Boudreaux-Bartles G (1993) Distributions for time-frequency analysis: A generalization of Choi-Williams and the Butterworth distribution. IEEE Trans Signal Proc 41: 463–472.CrossRefGoogle Scholar
  29. Quian Quiroga R, Sakowica O, Basar L, Schürmann M (2002) Wavelet Transform in the analysis of the frequency composition of evoked potentials. Brain Res Protoc 8: 16–24.CrossRefGoogle Scholar
  30. Redner RA, Walker HF (1984) Mixture densities, maximum likelihood and the EM algorithm. SIAM Rev 26: 195–239CrossRefGoogle Scholar
  31. Rieke F, Bialek W, Warland D (1997) Spikes: Exploring the Neural Code. Boston: MIT Press.Google Scholar
  32. Rioul O, Flandrin P (1992) Time-scale energy distributions: A general class of extending wavelet transforms. IEEE Trans Signal Proc 40: 1746–1757.CrossRefGoogle Scholar
  33. Sharma SC, Werner Al (1981) Improved method of grouping provincewide permanent traffic counters. Transp Res Rec 815: 12–18.Google Scholar
  34. Snider RK, and Bonds AB (1988) Classification of nonstationary neural signals. J Neurosci Meth 84: 155–166.CrossRefGoogle Scholar
  35. Teolis, A (1998) Computational Signal Processing with Wavelets. Boston, MA: Birkhauser.Google Scholar
  36. Ville, J (1948) Theorie et applications de la notion de signal analytique. Cables et Transmissions 20AA: 61–74.Google Scholar
  37. Whittaker SG, Siegfried JB (1983) Origins of the wavelet in the visual evoked potential. Electroencephalogr Clin Neurophysiol 55: 91–101.PubMedCrossRefGoogle Scholar
  38. Wigner E (1932) On the quantum correction for thermodynamic equilibrium. Phys Rev 40: 749–759.CrossRefGoogle Scholar
  39. Williams W (1998) Recent advance in time-frequency representations: Some theoretical foundations. In Akay M, (Ed.), Biomedical Signal Processing. Amsterdam: IEEE.Google Scholar
  40. Williams WJ, Brown ML, Zaveri HP, Shevrin H (1994) Feature extraction from time-frequency distributions. In C. H. Dagli (Ed.), Intelligent Engineering Systems through Artificial Neural Networks. 4, pp. 823–829. New York: ASME Press.Google Scholar
  41. Williams WJ, Jeong J (1992) Reduced interference time-frequency distributions. In B. Boashash (Ed.), Time-Frequency Signal Analysis: Methods and Application. Melbourne: Longman and Cheshire.Google Scholar
  42. Zhao Y, Atlas LE, Marks R (1990) The use of cone shaped kernels for generalized time frequency representations of nonstationarity signals. IEEE Trans Acoust, Speech, Signal Proc 38: 1084–1091.CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media, LLC 2009

Authors and Affiliations

  1. 1.The Centre for High-Throughput Biology (CHiBi), 177, Michael Smith LaboratoriesUniversity of British ColumbiaVancouverCanada

Personalised recommendations