| Phonocardiograms (PCGs) ,recording of heart sounds, often reflect the malfunction of the heart.Normal heart sounds is periodically changed with the beat of the heart, but it is very different for different person with different heart diseases. In this sense, the PCG signals is non-stationary. In order to extract the characteristics of the PCG signal and understand it fully, we discussed several analysis methods.Firstly, the paper discussed the power spectrum density (PSD) of the heart sound. We use the classical Welch method. According to the sampling rate, we choose the proper window and decide its length, then calculate the PSD of the heart sound in a heart beat period. The PSD of normal heart sound demonstrates: its primary frequency is below 300 Hz, the first peak locates at between 80 and 120 Hz, the second peak locates at between 200 and 250 Hz. Secondly, the Short Time Fourier Transform (STFT) method and the wavelet method are both used to evaluate the time-frequency distribution of the heart sound. The comparison between them shows no one is absolutely more efficient than the other. At last, we calculated the Normalized Averaging Shannol Energy (NASE) of the heart sound to see its characteristics in time domain.Some typical heart sounds with heart valve defect or heart valve stenosis are analyzed using the methods presented above. The characteristics of time, frequency and energy are extracted, which show that the combined analysis of the three methods of the Welch PSD, the wavelet analysis and the NASE can classify different heart sounds properly.In this paper, we also discussed the development of the heart sound analysis system. The analysis system receives sampling data from the hardware. Due to the interference, the signal is noised. High-pass, low-pass and 50 Hz trap filters are designed to meet our demand. And the soft threshold and hard threshold method in wavelet domain, which are used to remove the white noise, are also been discussed. The analysis system is based on the windows operating system and is developed by Visual C++ 6.0. The system is made up of four modules functionally. The patient information management module is fulfilled by using ODBC which is a database technique offered by VC++. We use the Access database of the Microsoft to manage all the Patient records. This module can fulfill these performances: to previous record, to next record, add new record, find record, modify and delete record exit database and etc.. Data sampling module is fulfilled by Serial communication, which is supported by the VC++ MSComm control. We calculated the Welch PSD, wavelet analysis and the NASN in the analyzing module. Firstly we fulfill the analysis by MATLAB programming and then we integrated these methods into the system by Mixture program technique. The analysis result will show on the screen. The report module will let the patient information, the heart sounds and the diagnosis drawn by the doctor shown on the screen, and it can be printed in a predetermined format. The four modules combine smoothly to construct the whole analysis system.This study provides some analysis methods and the analysis system for further studies. And it also plays a basis role for the assistant diagnosis of the heart diseases.
|
PDF Full Text Request