No abstract

No abstract

Abstract-This paper is concerned with the problem of cancellation of heart sounds from the acquired respiratory sounds using a new joint time-delay and signal-estimation (JTDSE) procedure. Multiresolution discrete wavelet transform (DWT) is first applied to decompose the signals into several subbands. To accurately separate the heart sounds from the acquired respiratory sounds, time-delay estimation (TDE) is performed iteratively in each subband using two adaptation mechanisms that minimize the sum of squared errors between these signals. The time delay is updated using a nonlinear adaptation, namely the Levenberg-Marquardt (LM) algorithm, while the function of the other adaptive system-which uses the block fast transversal filter (BFTF)-is to minimize the mean squared error between the outputs of the delay estimator and the adaptive filter. The proposed methodology possesses a number of key benefits such as the incorporation of multiple complementary information at different subbands, robustness in presence of noise, and accuracy in TDE. The scheme is applied to several cases of simulated and actual respiratory sounds under different conditions and the results are compared with those of the standard adaptive filtering. The results showed the promise of the scheme for the TDE and subsequent interference cancellation.

Abstract-In this correspondence, a new time delay estimation procedure is proposed using the multiresolution analysis framework through a discrete wavelet transform (DWT). Once the signals are decomposed, the time delays are estimated iteratively in each sub-band using two different adaptation mechanisms that minimize the mean squared error (MSE) between the reference and primary signals in the corresponding sub-band and level. The localization of the minima of the MSE curves at different levels and subbands is used in order to arrive at the time delay estimates. The proposed scheme is then applied to a real-life problem of underwater target detection from the acoustic backscttered data.

In this work the capability of a fetal Phonocardiogram (PCGfl processing method to generate reliable time series has been evaluated Twenty-nine simultaneous PCG and ECG signals were recordedfrom 18 adults and 11 fetus. Each PCG signal was processed to extract the starting (S), maximum (M), and ending points (E) fiom the envelope of SI and S2. The time series (SS, M, EE) were compared with the pattern signal by ECG using time and frequency heart rate variability indexes derived fiom each time serie. Data were analysed using ANOVA and post hoc test for repeated measures. Results on adults have shown that with the proposed method SSI, MI, SS2 and MS2 are reliable time references to generate time series @0.05). On fetal signals, because of the recording setup, the method peformance has been limited to detect Us2 as the only reliable time reference. Thus, the PCGf is a reliable source to produce at least one time serie equivalent to the pattern signal. IntroductionDuring the last two decades, the fetal Phonocardiogram (PCGf) has been proposed as an alternative way to produce the fetal Cardiotachogram (SCTGf) in order to get information about fetal wellbeing [1,2,3]. To achieve this goal, it is necessary to extract reliable time references ftom the cardiac sounds (SI and/or S2). This is not an easy task because of the poor signal to noise ratio (SNR) of the PCGf, and the cardiac sounds morphology. To contend with these problems, different filtering and envelope generation techniques have been used to extract the maximum point from the cardiac sounds envelope [1,3]. Specifically, we have proposed and evaluated a filtering scheme based on multiresolution analysis and de-noising techniques to filter the PCGf prior to extract time references tiom the Hilbert Transform envelope [4]. Our results, obtained on short simulated PCGfS (1 50 cardiac cycles), have shown that the best performance of the method is carried out when three levels of decomposition, wavelet Daubechies of sixth order, and adaptive thresholds, are used [5]. This methodology improves the SNR, reduces the deterioration of the cardiac sounds morphology, and therefore increases the precision to extract time references fi-om the starting (S), maximum (M), and ending (E) points on the envelope associated to S1 and S2. Nevertheless, the knowledge about the precision of the time references extracted from short simulated signals is not enough to assure the method capability to produce reliable SCTGfs according to clinical standards [6] that use the ECG to generate the Cardiotachogram fiom the R wave (RCTG). The goal of this paper is to extend our observations on longer real PCGs, through the evaluation of the performance of our time references extraction method (EM) [4,5], by comparison of the heart rate variability indexes [6] derived fiom RR and the fetal cardiac sounds periods. MaterialThe group of signals was composed of twenty-nine recordings from 18 adults (age: 18-42 years old) and 1 1 fetus (age: 28-41 weeks old) using 16-bit accuracy resolution and 1000 Hz ...