Abstract.
The focus of the paper is the problem of developing
effective methods for automated segmentation of epileptic seizures in EEG
recordings intended for the diagnosis of epilepsy. As a solution to the problem
the new method is proposed. This method is based on the analysis of
non-stationary, containing quasi-periodic fragments signal, comprising local
period estimation procedure followed by identification of period dynamics
characteristics in order to segment fragments with repeating oscillations. The
method proposed combines two basic ideas of existing approaches – multiscale (wavelet)
analysis and synthesis of the quadratic time–frequency representations. The
union of these approaches could be carried out in the framework of the
previously proposed methodology of multiscale correlation analysis (MCA) and
the analytical spectra technique.
A hallmark of the main tools of the method
– special MCA representations – is that they, in contrast to classical
approaches, are focused on the time domain, rather than on frequency domain.
The paper discusses in detail the successive steps of synthesis of such
representations in the time domain, from MCA estimation of the autocorrelation
function to linear procedures such as window weighting and matched filtering
with a special sliding window. The fact that the effective realization of the synthesized
representations is based on shifted analytical spectra, should be considered
nothing more than a convenient technique.
All MCA synthesis steps and segmentation
procedure characteristics are illustrated by real rat EEG recording marked by a
neurophysiologist–expert. Taking into account the results obtained by a real
recording, we conclude that the synthesized method will be one of the most
promising tools for the automatic segmentation of epileptic seizures.
Key words:
time−frequency EEG analysis, automated epileptic
seizure detection methods, multiscale correlational analysis, analytic spectra.
References
1.
Tzallas A.T., et al. Automated
Epileptic Seizure Detection Methods: A Review Study.
In Epilepsy - Histological, Electroencephalographic and
Psychological Aspects, Ed. by Dr. Dejan Stevanovic, InTech, 2012, pp.
75-98, DOI: 10.5772/31597.
2.
Guerrero-Mosquera C.,
et al.
Time-Frequency EEG analysis in epilepsy: What is more suitable?
Proceedings
of the 5-th IEEE International Symposium on Signal Processing and
Information Technology (ISSPIT), Athens, Greece, 2005, pp.
202-207, DOI: 10.1109/ISSPIT.2005.1577096.
3.
Kannathal N.,
et al.
Entropies for detection of epilepsy in EEG.
Computer Methods and Programs in
Biomedicine, 2005, Vol. 80, No. 3, pp. 187–194, DOI:
10.1016/j.cmpb.2005.06.012.
4.
Williams W.J.
et al.
Time-frequency analysis of electrophysiology signals in epilepsy.
IEEE Eng. Med. Biol.,
1995, Vol. 14, No. 2, Mar./Apr. pp. 133–143, DOI: 10.1109/51.376750.
5.
Adeli H., Ghosh-Dastidar S., Dadmehr,
N. A wavelet-chaos methodology for analysis of EEGs and EEG subbands
to detect seizure and epilepsy.
IEEE Trans Biomed Eng, 2007, Vol. 54, No 2, pp. 205-211, DOI:
10.1109/TBME.2006.886855.
6.
Orosco L., Garces A., Laciar E.
Review: A Survey of Performance and Techniques for Automatic Epilepsy
Detection. Journal of Medical
and Biological Engineering, 2013, Vol. 33, No 6, pp. 526-537, DOI:
10.5405/jmbe.1463.
7.
Antsiperov V.E., Obukhov Y.V.
Epileptic Seizure Detection and Segmentation Based on Multiscale
Correlation Analysis of EEG recordings.
Sbornik materialov III-go
Evraziyskogo kongressa po meditsinskoy fizike i inzhenerii
"Meditsinskaya fizika - 2010" [Materials of 3-rd Eurasian Congress
on Medical Physics and Engineering.
"Medical
Physics -
2010"], Moscow,
Moscow State University, 2010, Vol. 3, pp.188-190.
8.
Antsiperov V.E. Multiscale correlation
analysis of nonstationary signals containing quasi-periodic fragments.
Journal of Communications
Technology and Electronics, 2008, Vol. 53, No 1, pp 65–77,
DOI:10.1134/S1064226908010099.
9.
Antsiperov V.E. Human Brain Rhythms Detection based on Correlation of
EEG Analytical Spectra in the Main Frequency Bands.
Zhurnal Radioelektroniki -
Journal of Radio Electronics,
2014, No. 5. Available at
http://jre.cplire.ru/jre/may14/13/text.pdf.
(In Russian).
10.Antsiperov
V.E., Obukhov Y.V., et al. Analysis
of Brain Structures Hypersynchronization during Epileptic Seizure on
the basis of Conical Representations of EEG Signal.
Zhurnal Radioelektroniki -
Journal of Radio Electronics, 2014, No. 11. Available at
http://jre.cplire.ru/jre/nov14/18/text.pdf,
(In Russian).
11.Jeong
J., Williams W. J. Kernel design for reduced interference
distributions. IEEE Transactions
on Signal Processing, 1992, Vol. 40, No 2, pp. 402 – 412, DOI:
10.1109/78.124950.
12.Sachs
J. Handbook of Ultra-Wideband Short-Range Sensing: Theory, Sensors,
Applications. Wiley-VCH, Berlin, 2012.
13.Antsiperov
V.E. A new method for correlation functions estimation in problems of
radar and communication signals detection and processing.
Doklady III Vserossiyskoy konferentsii "Radiolokatsiya and radiosvyaz"
[Proc. III Nat. Scien. Techn. Conf. Radiolocation and Radio
Communications], Moscow, Russia, 2009, pp. 887-881 (in Russian).
14.Antsiperov
V.E. Scale-frequency representation of signals and their computation
by means of discrete analytical spectra.
Doklady 18 Mezhdunarodnoy
konferentsii "Tsifrovaya obrabotka signalov i ee primenenie -
DSPA-2016" [Proc. 18-th International Conference “Digital Signal
Processing and its Applications” (DSPA-2016)], Moscow, Russia, 2016,
pp. 135-140 (in Russian).