| Authors |
Berdibaeva Gul'mira Kuanyshevna, doctorate degree student, Kazakh National Research Technical University named after K. I. Satpayev (22a Satpayev street, Almaty, Republic of Kazakhstan), horli@mail.ru
Bodin Oleg Nikolaevich, doctor of technical sciences, professor, sub-department of information-measuring equipment and metrology, Penza State University (40 Krasnaya street, Penza, Russia), bodin_o@inbox.ru
Firsov Dmitriy Sergeevich, master degree student, Penza State University (40 Krasnaya street, Penza, Russia), firsov.d.7.58@gmail.com
|
| Abstract |
Background. The subject of the study are the respiratory sounds of asthmatic patients and healthy individuals. The subject of the study is the analysis of sound segments of respiratory sounds using discrete wavelet transform (DWT) and wavelet-packet transformation (WPT). The aim of the work is to classify the sound signals of normal breathing and, correspondingly, to the level of asthmatic diseases (mild asthma, moderate asthma and severe asthma) using an artificial neural network (ANN).
Materials and methods. The algorithm of sequential processing of a signal through a filter bank is described, taking into account the psychoacoustic nature of the hearing and the results of classification of the derived attribute vectors using the apparatus of artificial neural networks.
Results. Normal and asthmatic sound signals of respiration are divided into segments, which include one cycle of breathing as inhalation
and exhalation. Analysis of these sound segments is carried out using both a discrete wavelet transform (DWT) and wavelet-packet transformation (WPT). Each audio segment is divided into frequency subbands using DWT and WPT. Functional vectors are created by extracting statistical characteristics from subbands. The results of the classification of DWT and WPT are compared with each other in terms of classification accuracy.
Conclusions. Respiratory sound analysis using signal processing techniques is important for the diagnosis of lung diseases such as asthma. There are many studies on the analysis of respiratory sounds. In these studies, it is shown that neural networks give a high success rate. Thus, we use DWT analysis methods, WPT and ANN classifier to analyze our breathing sounds. We compare these methods of analysis in terms of the accuracy of classification. As a result, as can be seen from the results, the DWT is slightly better than the WPT in our study. The results obtained are very promising for the detection of asthma.
|
