Analysis of the Multi-Channel Epileptiform EEG Using the Markov Chains Formalism

Abstract:

A multi-level scheme of syntactic reduction of the epileptiform EEG data is briefly discussed and the possibilities it opens up in describing the dynamic behaviour of a multi-channel system are indicated. A new algorithm for the inference of a Markov network from finite sets of sample symbol strings is introduced. Formulae for the time-dependent state occupation probabilities, as well as joint probability functions for pairs of channels, are given. An exemplary case of analysis in these terms, taken from an investigation of anticonvulsant drug effects on EEG seizure patterns, is presented.

Key-Words:

Markov Chains - Finite Automaton Inference - Syntactic Analysis - Epilepsy - Multi-Channel EEG

• REFERENCES

• 1 Gersch W, Goddard GV. Epileptic focus location: spectral analysis method. Science 1970; 769: 701-2.
• 2 Brazier MAB. Electrical seizure discharges within the human brain: the problem of spread. In Brazier M A B. ed Epilepsy: its Phenomenon in Man . New York: Academic Press; 1973
• 3 Gotman J. Measurement of small time differences between EEG channels: method and application to epileptic seizure propagation. Electroencephal Clin Neurophysiol 1983; 56: 501-14.
• 4 Mars NJI, Lopes da Silva FH. Propagation of seizure activity in kindled dogs. Electroencephal Clin Neurophysiol 1983; 56: 194-209.
• 5 Mars NJI, van Arragon GW. Time delay estimation in nonlinear systems. IEEE Trans Acoust Speech Sign Proc. 1981. ASSP-29 619-21.
• 6 Pijn JPM, Vijn P, Lopes da Silva FH. Evolution of the interaction between brain structures during epileptic seizures. Neurol Neurosurg Suppl I 1987; 89: 104
• 7 Grochulski W, Penczek P. Syntactic analysis of the epileptic electroencephalogram. Int J Bio-Med Comp 1986; 19: 219-34.
• 8 Grochulski W, Penczek P, Posielski J. Segmentation of the epileptic EEG by means of a finite state automaton. Int J Bio-Med Comp 1986; 18: 35-44.
• 9 Penczek P, Grochulski W, Grzyb J, Kowalczyk M. The use of a multi-channel Kaiman filter algorithm in structural analysis of the epileptic EEG. Int J Bio-Med Comp 1987; 20: 135-51.
• 10 Kemp B, Kamphuisen HAC. Simulation of human hipnograms using a Markov chain model. Sleep 1986; 9: 405-14.
• 11 Lewis HR, Papadimitriou CH. Elements of the theory of computation. Englewood Cliffs: Prentice Hall; 1981
• 12 Biermann AW, Feldmann JA. On the synthesis of finite-state machines from samples of their behaviour. IEEE Trans Comp 1972; 21: 592-7.
• 13 Richetin M, Vernadat F. Efficient regular grammatical inference for pattern recognition. Pattern Recogn 1984; 17: 245-50.
• 14 Thomason MG, Granum E. Dynamic programming inference of Markov Networks from finite sets of sample strings. IEEE Trans Pattern Anal Mach Intell. 1986. PAMI-8 491-501.
• 15 Hopcroft J. An n log n algorithm for minimizing states in a finite automaton. In Kohavi Z, Paz A. ed Theory of Machines and Computations . New York: Academic Press; 1971
• 16 Rao CR. Linear statistical inference and its applications. New York: John Wiley & Sons; 1973
• 17 Seneta E. Nonnegative Matrices and Markov Chains. . New York: Springer-Verlag; 1981
• 18 Isaksson A. SPARK – a sparsely updated Kaiman filter with application to EEG signals. Techn. Report 120, Royal Inst of Technology, Dept of Telecomm Theory, Stockholm. 1976
• 19 Franaszczuk PJ, Blinowska ICI. Linear model of brain electrical activity – EEG as a superposition of damped oscillatory modes. Biol Cybern 1985; 53: 19-25.
• 20 Späth H. Cluster Analysis Algorithms for Data Reduction and Classification of Objects. England: Ellis Horwood Ltd; 1980
• 21 Beale EML. Cluster analysis. London: Scientific Control Systems; 1969
• 22 Rump S, Kowalczyk M, Penczek P. Syntactic analysis of the experimental epileptic EEG. Pol J Pharmacol Pharm (to be published)