350 rub
Journal Neurocomputers №3 for 2016 г.
Article in number:
The study of artifical neural networks in the problem personal identification by electrocardiosignals, registered CardioQVARK devices
Authors:
R.V. Isakov - Ph.D. (Eng.), Associate Professor, Vladimir State University O.V. Suntsova - Project Coordinator, Hedical Profect CordioQvqrk (Moscow)
Abstract:
Nowadays, electrocardiography (ECG) is one of the leading methods of instrumental studies of the cardiovascular system. For remote ECG monitoring is a problem to the error log ECS from other people, recorded by the same device. This will cause inconsistency and unreliability of the analysis. Therefore it is required to develop special methods and means for verification of ECG recordings. In this case, the most convenient biometric indicator is electrocardiosignal (ECS). In this work it was studied approach to the identification (verification) person by his ECS, registered by CardioQVARK device in I lead. Technology of perceptron artificial neural networks (ANN) was used to obtain the model individual to individual. For the study was used a data set consisting of 60 ECS records. In this work developed two approaches to the con-struction of the space of the input features for the neural network system: by the form of a generalized a cardiocycle and ECS correlation rhythmogram. As the basic model of an artificial neural network (ANN) was chosen as a multilayer perceptron with four varieties of structures. To find the optimal number of neurons was performed computer experiment by training and testing ANN series with different sizes. The results showed that the identifying by ECG form gives significantly better results than of the heart rhythm. This is a strong dependence on the current heart rate of the human function condition that brings a lot of noise to the identification information. Recommended for practical application is a hybrid neural network structure of a two-layer perceptron of equal size hidden layers with the addition linear output. It is also recommended to update of trained network to the new, verified data from the identified person.
Pages: 31-38
References

 

  1. Suvorov A.V. Klinicheskaja ehlektrokardiografija. N.Novgorod: Izd-vo NMI. 1993. 124 s.
  2. Medicinskijj proekt CardioQVARK www.qvark-med.ru Data obrashhenija: 19.06.2015
  3. Semchukov JU.B., Sosedko JU.I., Karandashev A.A. Sudebno-medicinskaja identifikacija ehlektrokardiogramm lic bez grubojj patologii serdechno-sosudistojj sistemy // Mater. II-go Vseros. sezda sudebnykh medikov: tezisy dokladov. Irkutsk-M. 1987. S. 285-287.
  4. Patent 2286713 (RF). Sposob identifikacii cheloveka po biologicheskojj aktivnosti serdca / V.B. Korolev, A.I. Reshetnikova, D.B. Osipov, 2004.
  5. Korchinskaja Z.A., Fajjnzilberg L.S. Sistema podderzhki prinjatija reshenijj pri identifikacii lichnosti po fazovomu portretu EHKG // Materialy VIInauch.-prakt. konf. «Sistemy podderzhki prinjatija reshenijj. Teorija i praktika»(SPPR\'2011). Kiev. 2011. S. 119-122.
  6. Proekt personalnojj autentifikacii www.nymi.com Data obrashhenija: 19.06:20159.
  7. Biel L., Petersson O., Philipson L., Wide P. ECG analysis: a new approach in human identification // IEEE Trans. on Instrumentation and Measurement. 2001.V. 50.№3.P. 808-812.
  8. Shen T. W., Tompkins W. J., Hu Y. H. One-lead ECG for identity verification // IEEE EMBS. 2002. V. 1.P. 62-63.
  9. Israel S. A., Irvine J. M., Cheng A., Wiederhold M. D., Wiederhold B. K. ECG to identify individuals // Pattern Recognition. 2005. V. 38.№ 1.P.133-142.
  10. Kruglov V.V., Borisov V.V. Iskusstvennye nejjronnye seti. Teorija i praktika. Izd. 2-e, stereotip. M.: Gorjachaja linija-Telekom. 2002. 382 s.
  11. Fast Artificial Neural Network Library http://leenissen.dk/fann/wp/ Dataobrashhenija: 19.06.2015