M.D. Alekhin, L.N. Anishchenko, A.V. Zhuravlev

One of the priority areas of sleep medicine nowadays is an examination of breathing malfunctions, which are character to many disorders of different types. Universally recognized standard of diagnostics in this field is supposed to be polysomnography. Results of such procedure give a very informative aspect of sleep dysfunctions but require a great amount of sensors and electrodes to be used. So application of new non-contact methods in somnology is an up-to-date scientific task. One of these approaches is bio-radiolocation technology, which gives an opportunity to carry out remote non-contact monitoring of patients breath disorders during sleep.

Early recognition and proper classification of apnea episodes are both very important stages in treatment strategy planning and opportune preventive measures taking. Application of artificial neural networks is considered to be an effective instrument in the tasks of classification of breathing patterns, recorded from abdominal and thoracic belt perimetric sensors. Multiresolution wavelet analysis is a perspective mathematical apparatus which is extremely needed in processing of non-stationery multicomponent signals with noise, which have high-frequency components of short duration and prolonged low-frequency components. Based on two mentioned approaches, a new method of bio-radiolocation signal breathing patterns classification was developed.

The proposed scheme of bio-radiolocation signal breathing patterns classification consists of three main parts. At first data preprocessing step for each signal quadrature fragment moving average filtering was performed and all the constant components were deleted. Then for selected breathing patterns attribute space was formed, based on utilization of series of absolute values of the third level wavelet decomposition detail coefficients of signal quadratures. A multilayer perceptron with a backpropogation training algorithm, one hidden layer and a bipolar sigmoid activation function of neurons was applied as a classifier.

For estimating of the proposed method efficiency a sample of 240 modeled bio-radiolocation signal realizations corresponding to three types of breathing patterns (obstructive sleep apnea, central sleep apnea and normal calm sleeping) was used. The total classification accuracy came to the value of 92,5%% on a test sample. Thus, the proposed approach for bio-radiolocation signal breathing patterns classification turned out to be an efficient one and satisfies standard recommendations for sleep apnea syndrome diagnostics.

1. Вейн А.М. Синдром апноэ во сне и другие расстройства дыхания, связанные со сном: клиника, диагностика, лечение. М.: Эйдос Медиа. 2002. 310 с.
2. American Academy of Sleep Medicine Task Force. Sleep-related breathing disorders in adults: recommendations for syndrome definition and measurement techniques in clinical research // Sleep. 1999. V. 22. P. 667-689.
3. American Sleep Disorders Association Standards of Practice Committee Practice parameters for the indications for polysomnography and related procedures. Polysomno­graphy Task Force // Sleep. 1997. V. 20. P. 406-422.
4. Черникова А.Г., Слепченкова И.Н., Ковров Г.В. Оценка качества сна по данным бесконтактной регистрации физиологических сигналов и сравнение с результатами полисомнографии // Сб. трудов VII Всеросс. конф. «Актуальные проблемы сомнологии». Москва. 2010. С. 69-70.
5. Биорадиолокация. Коллективная монография / под. ред. А.С. Бугаева, С.И. Ивашова, И.Я. Иммореева. М.: Изд-во МГТУ им. Н.Э. Баумана. 2010. 396 с.
6. Алёхин М.Д., Анищенко Л.Н., Корчагина Д.А. Метод биорадиолокации в анализе перемещений грудной клетки при спокойном дыхании // Биомедицинская радиоэлектроника. 2009. № 10. С. 56-61.
7. Alekhin M.D., Anishchenko L.N. Non-contact Remote Bio-Radiolocation Method of Sleep Monitoring // Proceedings of the 5th Russian-Bavarian Conference on Biomedical Engineering. Munich, Germany. 2009. P. 145-146.
8. Sezgin N., Tagluk M. Energy Based Feature Extraction for Classification of Sleep Apnea // Computers in Biology and Medicine. 2009. V. 39. P. 1043-1050.
9. Fontenla-Romero O., Guijarro-Berdinas B., Alonso-Betanzos A., Moret-Bonillo V. A New Method for Sleep Apnea Classification Using Wavelets and Feedforward Neural Networks // Artificial Intelligence in Medicine. 2005. V. 34. P. 65-76.
10. Терехов С.А. Лекции по нейроинформатике. Снежинск: Всероссийский НИИ технической физики. 2008. 40 с.
11. Дьяконов В.П. Вейвлеты: от теории к практике. М.: Солон-П. 2002. 448 с.
12. Яковлев А.Н. Основы вейвлет-преобразования сигналов. М.: Радиотехника. 2003. 80 с.
13. Митрофанов Д.Г., Сафонов А.В. Применение вейвлет-анализа для сохранения структуры дальностных портретов воздушных целей при повышении уровня шумов // Электромагнитные волны и электронные системы. 2005. № 9. С. 19-24.
14. Alekhin M.D., Anishchenko L.N. Wavelet Analysis in Bio-radiolocation Signal Movement Artefacts Discerning // Proceedings of the 6th Russian-Bavarian Conference on Biomedical Engineering. Munich, Germany. 2010. P. 11-13.
15. Parallel Distributed Processing: Explorations in the Microstructures of Cognition / ed. by D.E. Rumelhart and J.L. McClelland. M.: MIT Press. 1986 .1190 p.
16. Хайкин С. Нейронные сети: полный курс. Изд-е 2-е / пер. с английского. М.: Вильямс. 2006. 1104 с.
17. Потапов А.С. Распознавание образов и машинное восприятие. М.: Изд-во Политехника. 2007. 552 с.
18. Охотников Д.А. Особенности радиолокационного обнаружения целей, совершающих возвратно-поступательное движение // Всеросс. радиофизические чтения-конферен­ции памяти Н.А. Арманда. Муром. 2010. С. 176-181.
19. Sheen D., McMakin D., Hall E. Three-Dimensional Millimeter-Wave Imaging for Concealed Weapon Detection // IEEE Transactions on Microwave Theory and Techniques. 2001. V. 49. № 9. P. 1581--1592.
20. Muehlsteff J, Pinter R., Morren G. Comparison of Respiration Rate Monitoring with a Low-cost Doppler-Radar Sensor and Inductive Thorax-Plethysmography // IFMBE Proceedings 25/VII. 2009. P. 768-771.
21. MIT-BIH Polysomnographic Database (http://www.physionet.org/physiobank/database/slpdb)