Coarse crackles symptoms are shown as inflammation, such as pneumonia, pulmonary congestion, early pulmonary edema, bronchitis, alveolar inflammation and other diseases. The common coarse crackles in patients with pulmonary interstitial fibrosis, bronchiectasis and chronic bronchitis last for about 5s. There are also some other lung sounds, chest rubs for instance.įor a long time there have been a lot of researches on lung sounds and its pathologic causes. Coarse crackle is discontinuous sound called moist rale while the others are continuous sound called dry rale. In 1985, adventitious sounds, a kind of additional noise in periodic signal, was principally identified by The 10th International Respiratory Voice Association as: coarse crackle, fine crackle, wheeze, and rhonchus. The abnormal lung sounds consists of weaken and vanished sound, prolongement expiratoire and tubular breathing sound. The normal lung sounds primarily include three categories: vesicular breathing sound, bronchial breathing sound and tracheal sound. Weaken, vanished sound, prolongement expiratoire, tubular breathing sound Vesicular sound, tracheal sound, bronchial sound Table 1 shows the lung sound categories in detail. The abnormal and the adventitious lung sounds are produced when there exist problems or something unfit in human body. Normal lung sound is mainly generated from the airflow through principle bronchi, bronchiole and lung in the respiratory system. Taking health condition into consideration, lung sounds are mainly classified into three types: normal, abnormal and adventitious. With more researches on its classification there came various supplement and modification on lung sound categories. The classification of lung sounds could be dated back in 1819 when Leannec published Indirect Auscultation sorting it into five kinds. During diagnosis process, it is helpful for accurately judging characteristic vectors and determining disease reasons to de-noise lung sounds and abstract vibration frequency, sound waves amplitude and amplitude gradient. Therefore, it will be helpful to develop an efficient method to recognize lung sounds and to diagnose related diseases based on the analysis of the more informative lung sounds data collected by electronic stethoscope and develop an efficient way to recognize lung sounds and its related diseases. However, conventional auscultation of lung sounds has a very limited usage because of the limitation of information it can obtain, its dependence on physician's experience, and its high rate of misdiagnosis. With the increasing incidence of respiratory system diseases, more attention is being paid on the harmless diagnosis method based on analysis of lung sounds which contains abundant information on lung condition. It is meaningful to research on lung diseases to prevent death and promote treatment and clinical level. Figure 1 is the crude death rates caused by some main diseases in domestic city and rural, which indicates that the respiratory disease is non-negligible and noteworthy. In 2015, the death rate due to respiratory system diseases was 7.336 and 7.996 per one thousand people among urban and rural domestic residents respectively. Respiratory disease is now listed as one of the leading causes of death of domestic residents according to statistics from the National Health Bureau. In recent years, the incidence of respiratory system diseases such as chronic obstructive pulmonary disease, bronchitis and asthma tend to rise gradually due to the atmosphere contamination, for instance, frequent haze. Keywords: lung sound, category recognition, wavelet de-noising, linear discriminant analysis, BP neural network Introduction Finally, we use BP neural network to carry out lung sounds recognition where comparatively high-dimensional characteristic vectors and low- dimensional vectors are set as input and lung sounds categories as output with a recognition accuracy of 82.5% and 92.5%. In addition, we use linear discriminant analysis (LDA) to reduce the dimension of characteristic vectors for comparison in order to obtain a more efficient way for recognition. Considering the problem that lung sounds characteristic vectors are of high dimensions after wavelet decomposition and reconstruction, a new method is proposed to transform the characteristic vectors from reconstructed signals into reconstructed signal energy. Wavelet de-noised method is adopted to reduce noise of collected lung sounds and extract wavelet characteristic coefficients of the de-noised lung sounds by wavelet decomposition. In this paper, a method of characteristic extraction and recognition on lung sounds is given. Select the file that you have just downloaded and select import option Reference Manager (RIS). A Lung Sound Category Recognition Method Based on Wavelet Decomposition and BP Neural Network.
0 Comments
Leave a Reply. |