The delineation of P and T waves is important for the medical interpretation of ECG signals. We propose a Bayesian algorithm for simultaneous detection, delineation, and estimation of P and T waves. A block Gibbs sampler exploits the strong local dependencies in ECG signals by imposing block constraints on the P and T wave locations. The proposed algorithm is evaluated on the annotated QT database and compared with two classical algorithms.

Coastal altimetric waveforms may be corrupted by peaks. A simple parametric model was recently introduced to model peaky altimetric waveforms. This model assumes that the received altimetric waveform is the sum of a Brown echo and a Gaussian peak. This model has provided interesting results for symmetric peaks affecting altimetric signals. However, it is not appropriate for altimetric signals corrupted by asymmetric peaks. This paper introduces a Brown with asymmetric Gaussian peak model for altimetric waveforms. The parameters of this model are estimated by a maximum likelihood estimator. The performance of the proposed model and the resulting estimation strategy are evaluated via simulations conducted on synthetic and real data.

The problem of detecting T-wave alternans (TWA) in ECG signals has received considerable attention in the biomedical community. This paper introduces a Bayesian model for the T waves contained in ECG signals. A block Gibbs sampler was recently studied to estimate the parameters of this Bayesian model (including wave locations, amplitudes and shapes). This paper shows that the samples generated by this Gibbs sampler can be used efficiently for TWA detection via different statistical tests constructed from odd and even T-wave amplitude samples. The proposed algorithm is evaluated on real ECG signals subjected to synthetic TWA and compared with two classical algorithms.

Coastal altimetric waveforms may be corrupted by peaks. A simple parametric model was recently introduced to model peaky altimetric waveforms. This model assumes that the received altimetric waveform is the sum of a Brown echo and a Gaussian peak. This model has provided interesting results for symmetric peaks affecting altimetric signals. However, it is not appropriate for altimetric signals corrupted by asymmetric peaks. This paper studies a Brown with asymmetric Gaussian peak model for parameter estimation of altimetric waveforms. The parameter estimation problem is solved by a maximum likelihood estimator relying on an optimization algorithm. The performance of the proposed model and the resulting estimation strategy is evaluated via simulations conducted on synthetic and real data.

The delineation of P and T waves is important for the interpretation of ECG signals. We propose a Bayesian detection-estimation algorithm for simultaneous detection, delineation, and estimation of P and T waves. A block Gibbs sampler exploits the strong local dependencies in ECG signals by imposing block constraints on the P and T wave locations. The proposed algorithm is evaluated on the annotated QT database and compared with two classical algorithms.

A simple parametric model was recently introduced to model peaky altimetric waveforms [1] [2]. This model assumes that the received altimetric waveform is the sum of a Brown echo and Gaussian peaks. A maximum likelihood estimator for the parameters of this Brown + peak model was studied in [2] in the simple case where altimetric signals are corrupted by a single peak. However, an analysis conducted on real altimetric waveforms from the PISTACH project [3] shows it is also interesting to consider multi-peak models. This paper studies a generalization of the algorithm presented in [2] to estimate the parameters of multi-peak altimetric signals. The main contribution of this paper is a method allowing one to estimate the number of peaks which are present the Brown + peak model. The effects of model order mismatch will also be studied. Simulation results conducted on synthetic and real altimetric waveforms allow one to appreciate the performance of the proposed multi-peak model and its interest related to the single-peak version. Note that comparisons between the different proposed algorithms for altimetric waveform parameter estimation is done based on a 3 parameter Brown model estimating the amplitude, the epoch and the significant wave height of the echo. When dealing with peaky waveforms, the classical algorithm (MLE3) can fail to fit the altimetric signal, as shown in Fig. 1 (black curve). The single-peak model provides interesting results (left figure - red curve) but cannot model accurately the presence of multiple peaks in the observed signal. The multi-peak algorithm proposed in this paper clearly shows significant improved performance

The Maseng-Bakken model has shown interesting properties to model rain attenuation for Ka and Q/V broadband satellite systems. This paper derives the optimal rain attenuation predictor based on the Maseng-Bakken model. The optimal predictor is obtained by minimizing the mean square error between the rain attenuation and its estimate. We show that this predictor reduces to a bank of filters whose parameters depend on the rain attenuation power spectral density. Simulation results allow us to appreciate the performance of the resulting rain attenuation prediction that is compared with more traditional strategies.

This paper addresses the problem of classifying altimetric signals according to their shapes. The proposed classifier is divided into three steps. A one-class support vector machine method is first used to isolate the large amount of Brown-like echoes from others signals which are considered as outliers. The second step extracts pertinent features from the the remaining echoes (which cannot be well described by the Brown model). These features are projected onto discriminant axes using linear discriminant analysis. The final step classifies the projected feature vectors using a standard Bayesian classifier. The proposed three step classification strategy is evaluated on supervised real altimetric echoes.

In condition monitoring a part of the information necessary for decision-making comes from scrutinizing a time measure or a transform of this measure. Frequency domain is commonly exploited; lag domain is not, albeit advantages of the autocorrelation function have long been known. In this paper, we dwell on the autocorrelation function in order to extract some interesting properties of the measure. We propose two indicators in order to characterize the periodicity of a signal. First is based on the non-biased autocorrelation function and indicates a fundamental periodicity rate. Second is based on the biased autocorrelation and gives a dominant-power periodicity rate. The study of the 2Dplane defined by these two indicators allows the definition of regions attached to one type of periodicity from periodic to aperiodic through almost-periodic and quasi-periodic. Combined with an estimation of the correlation support, a final decision about the periodicity of the signal is given. In case of a periodic signal, a way of estimating the global signal ratio is proposed. These new outputs are valuable for initializing more complex processing. All the algorithms proposed are fully automatic, one click use! Relevance of these indicators is shown on real-world signals, current and vibration measures mainly.

Considering the emergence of telemedicine applications, different links such as fixed access network (PSTN), mobile access network (GSM/GPRS and future UMTS) or satellite interfacing (DVB-RCS technology) are involved in e-health applications. These are liable to induce errors and/or missing packets on the received data. Therefore the recovering of missing samples for biomedical signals is of great interest. This paper proposes a reconstruction method for ECG signals which is a combination of a left-sided and right-sided autoregressive (AR) model, and the well-known Gerchberg-Papoulis (GP) method. The proposed interpolation algorithm takes into account the samples before and after the missing ones to estimate a forward and a backward AR model. These estimates are used as an initialization of the original GP method. Results show that this interpolation method represents a really suitable technique to ECG reconstruction in a possible corrupted transmission.