Multi Classification of Sleep Sounds using Support Vector Machines

Yıl 2020, , 2474 - 2485, 15.12.2020

Erkin Kılıç Aykut Erdamar

https://doi.org/10.21597/jist.723161

Cited By: 1

Öz

Sleep continuity and sleep hygiene directly affect people's daily lives. The sounds that cause awakening such as snoring, coughing, and obstruction during sleep are generally related to sleep diseases. Noisy sound patterns, such as snoring, can negatively affect the sleep quality of other people who sleep in the same environment as the patient. Physiological signals and sleep sounds of patients are recorded by polysomnography. Then all results are examined by the sleep physician and appropriate diagnosis is made according to the results. Visual or auditory scoring is a very difficult, time-consuming and tiring process that requires professional experience. Hence, studies on the automatic classification of sleep sounds become important. In the presented study, it is aimed to develop a computer-aided diagnostic algorithm that can analyze sleep sounds quickly and reliably and classify them automatically. Six different sleep sound patterns (breathing / exhaling, coughing, simple snoring, duplex low frequency snoring, duplex high frequency snoring and triplex snoring) are automatically classified with an algorithm based on machine learning using the time-domain features. The proposed algorithm consists of three stages: In the first stage, raw sound signals are checked and pre-processed. In the second stage, features are obtained with waveform analysis. At the last stage, classification is done by using support vector machines. As a result of the study, six different sleep patterns were classified with an average accuracy rate of 90.20 %.

Anahtar Kelimeler

Sleeping Sounds, Simple Snoring, Duplex Snoring, Triplex Snoring, Classification

Destek Vektör Makineleri Kullanarak Uyku Seslerinin Çoklu Sınıflandırılması

Öz

Uyku sürekliliği ve uyku hijyeni, insanların günlük yaşantısını doğrudan etkilemektedir. Uyku sırasında ortaya çıkan horlama, öksürme, tıksırma gibi uyanmaya neden olan sesler genellikle uyku hastalıklarıyla ilintilidir. Horlama gibi gürültülü ses paternleri hasta ile aynı ortamda uyuyan diğer insanların da uyku kalitesini olumsuz yönde etkileyebilmektedir. Hastaların fizyolojik sinyalleri ve uyku sesleri polisomnografi ile kayıt edilir. Ardından tüm sonuçlar uzman doktor tarafından incelenir ve sonuçlarına göre uygun teşhis konulur. Görsel veya işitsel skorlama mesleki deneyim gerektiren, oldukça zor, zaman alan ve yorucu bir süreçtir. Bu nedenle, uykudaki seslerin otomatik sınıflandırılması üzerine yapılan çalışmalar önem kazanmaktadır. Sunulan çalışmada, uyku seslerini hızlı ve güvenilir bir şekilde analiz edebilen, otomatik olarak sınıflandırabilen bilgisayar destekli tanı algoritmasının geliştirilmesi amaçlanmıştır. Altı farklı uyku ses paterni (nefes alma/verme, öksürme, basit horlama, dubleks düşük frekans horlama, dubleks yüksek frekans horlama ve tripleks horlama) zaman bölgesinden elde edilen öznitelikler kullanılarak makine öğrenmesine dayanan bir algoritmayla otomatik olarak sınıflandırılmaktadır. Önerilen algoritma üç aşamadan oluşur: Birinci aşamada ham ses sinyallerine kontrol ve ön işleme yapılır. İkinci aşamada dalga formu analizleri yapılarak öznitelikler edilir. Son aşamada ise destek vektör makineleri kullanılarak sınıflandırma işlemi yapılır. Çalışma sonucunda, altı farklı uyku sesi paterni ortalama % 90.20 doğruluk oranıyla sınıflandırılmıştır.

Anahtar Kelimeler

Uyku Sesleri, Basit Horlama, Dubleks Horlama, Tripleks Horlama, Sınıflandırma

Kaynakça

• Alshaer H, Pandya A, Bradley TD, Rudzicz F, 2014. Subject Independent Identification of Breath Sounds Components Using Multiple Classifiers. 2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Florence, Italy, May 04-09, 2014, pp: 3577–3581.
• Ayhan S, Erdoğmuş Ş, 2014. Destek Vektör Makineleriyle Sınıflandırma Problemlerinin Çözümü İçin Çekirdek Fonksiyonu Seçimi. Eskişehir Osmangazi Üniversitesi İktisadi ve İdari Bilimler Dergisi, 9 (1): 175–201.
• Azarbarzin A, Moussavi ZMK, 2011. Automatic and Unsupervised Snore Sound Extraction From Respiratory Sound Signals. IEEE Transactions on Biomedical Engineering, 58(5): 1156–1162.
• Bazzani A, Bevilacqua A, Bollini D, Brancaccio R, Campanini R, Lanconelli N, Romani D, 2001. An SVM Classifier to Separate False Signals From Microcalcifications in Digital Mammograms. Physics in Medicine and Biology, 46 (6): 1651–1663.
• Berry RB, Budhiraja R, Gottlieb DJ, Gozal D, Iber C, Kapur VK, Tangredi M, 2012. Rules For Scoring Respiratory Events in Sleep: Update of the 2007 AASM Manual for the Scoring of Sleep and Associated Events. Journal of Clinical Sleep Medicine, 8 (5): 597–619.
• Burges CJ, 1998. A Tutorial on Support Vector Machines for Pattern Recognition. Data Mining and Knowledge Discovery, 2(2): 121-167.
• Cavusoglu M, Kamasak M, Erogul O, Ciloglu T, Serinagaoglu Y, Akcam T, 2007. An Efficient Method for Snore/nonsnore Classification of Sleep Sounds. Physiological Measurement, 28 (8): 841–853.
• Cortes C, Vapnik V, 1995. Support-Vector Networks. Machine Learning, 20(3): 273-297.
• Counter P, Wilson JA, 2004. The Management of Simple Snoring. Sleep Medicine Reviews, 8 (6): 433–441.
• Dafna E, Tarasiuk A, Zigel Y, 2013. Automatic Detection of Whole Night Snoring Events Using Non-Contact Microphone. PLoS ONE, 8 (12): e84139.
• Deller JR, Hansen JHL, Proakis JG, 1999. Discrete-Time Processing of Speech Signals. Wiley-IEEE Press, pp. 223-285, New York-United States.
• Doukas C, Petsatodis T, Boukis C, Maglogiannis I, 2012. Automated Sleep Breath Disorders Detection Utilizing Patient Sound Analysis. Biomedical Signal Processing and Control, 7 (3): 256–264.
• Duckitt WD, Tuomi SK, Niesler TR, 2006. Automatic Setection, Segmentation and Assessment of Snoring from Ambient Acoustic Data. Physiological Measurement, 27 (10): 1047–1056.
• Fawcett T, 2006. An Introduction to ROC Analysis. Pattern Recognition Letters, 27 (8): 861–874.
• Fiz JA, Abad J, Jané R, Riera M, Mañanas MA, Caminal P, Morera J, 1996. Acoustic Analysis of Snoring Sound in Patients with Simple Snoring and Obstructive Sleep Apnoea. European Respiratory Journal, 9 (11): 2365–2370.
• Jane R, Fiza JA, Sola-Soler J, Blanch S, Artis P, Morera J, 2003. Automatic Snoring Signal Analysis in Sleep Studies. Proceedings of the 25th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, Cancun, Mexico, September 17-21, 2003, pp:366–369.
• Jolliffe IT, Cadima J, 2016. Principal Component Analysis: a review and recent developments. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 374 (2065): 1-16.
• Jones TM, 2005. Acoustic Analysis of Snoring Before and After Palatal Surgery. European Respiratory Journal, 25 (6): 1044–1049.
• Karunajeewa AS, Abeyratne UR, Hukins C, 2008. Silence–breathing–snore Classification from Snore-related Sounds. Physiological Measurement, 29 (2): 227–243.
• Kecman V, 2001. Learning and Soft Computing: Support Vector Machines, Neural Networks, and Fuzzy Logic Models. MIT Press, pp. 121-189, Cambridge, MA-USA.
• Keenan SA, Hirshkowitz M, Casseres H, 2013. Monitoring and Staging Human Sleep. In Encyclopedia of Sleep, 5 (1): 71–79
• Kılıç E, 2017. Uyku Esnasında Çıkan Seslerin Sınıflandırılması. Başkent Üniversitesi Fen Bilimleri Enstitüsü Biyomedikal Mühendisliği Anabilim Dalı, Yüksek Lisans Tezi (Basılmış).
• Lee LA, Yu JF, Lo YL, Chen YS, Wang DL, Cho CM, Li HY, 2012. Energy Types of Snoring Sounds in Patients with Obstructive Sleep Apnea Syndrome: A Preliminary Observation. PLoS ONE, 7 (12): e53481.
• Proakis JG, Monolakis DG, 1996. Digital Signal Processing: Principles, Algorithms, and Applications. Pentice Hall, USA.
• Qian K, Xu Z, Xu H, Wu Y, Zhao Z, 2015. Automatic Detection, Segmentation and Classification of Snore Related Signals from Overnight Audio Recording. IET Signal Processing, 9 (1): 21–29.
• Rangayyan RM, 2015. Biomedical Signal Analysis. John Wiley & Sons, Inc, pp.720, Hoboken, NJ-USA
• Ugur TK, Erdamar A, 2019. An Efficient Automatic Arousals Detection Algorithm in Single Channel EEG. Computer Methods and Programs in Biomedicine, 173: 131-138.
• Vapnik VN, 2000. Methods of Pattern Recognition: In The Nature of Statistical Learning Theory, Springer, pp. 123–180, New York-USA ).
• Wilson K, Stoohs RA, Mulrooney TF, Johnson LJ, Guilleminault C, Huang Z, 1999. The Snoring Spectrum: Acoustic Assessment of Snoring Sound Intensity in 1,139 Individuals Undergoing Polysomnography. Chest, 115 (3): 762–770.
• Wu X, Kumar V, Ross QJ, Ghosh J, Yang Q, Motoda H, Steinberg D, 2008. Top 10 Algorithms in Data Mining. In Knowledge and Information Systems, 14 (1): 1-37.
• Yu JF, Chen YS, Li HY, 2012. The Characteristics of Snoring at Pharyngeal Anatomy in Natural Sleep: Snoring Duration. Journal of Mechanics, 28 (1): 91–95.