No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
A single channel phonocardiograph processing using EMD, SVD, and EFICA
Abstract
In this paper a novel approach using a trio of EMD, SVD and efficient version of FICA for extracting the fetal heart sound signals from the recorded single channel abdominal phonocardiogram (PCG) is presented. Phonocardiography is a low-cost, passive, non-invasive way of recording heart sounds and murmurs. Phonocardiography consists of recording acoustic signal on the maternal abdominal surface, and hence the recordings are heavily loaded by noise. Because of this the determination of the fetal heart rate (FHR) raises serious signal processing issues. We here propose a scheme of Blind Source Separation (BSS) technique for the fetal PCG extraction from the single channel recorded noisy abdominal phonographs. This presented method is the combination of the Empirical Mode Decomposition (EMD), Singular Value decomposition (SVD) and Independent Component Analysis (ICA).
... Many authors [3,13,15,[19][20][21][22] have looked into the design and testing of algorithms for fPCG filtration. As well as the filtering itself, some studies [3,5,15,23,24] were aimed at detecting S1 sounds, and only a few authors [17,18] looked into detecting S1 and S2 sounds. ...
... • A single-channel method combining the EMD method, singular value decomposition (SVD) and an efficient version of ICA (EFICA) was proposed in [21]. A combination of all methods was tested on real recordings and even led to effective extraction of signals burdened with high levels of interference. ...
... From the above it emerges that objective comparison of testing method performance is problematic, because authors use different signals (real or synthetic) disturbed by various levels and types of interference. Some authors [3,5,17,18,22,23,26] then evaluate the effectiveness of filtering using objective statistical parameters and some [20,21,27] only subjectively evaluate the extracted waveform. The aim of this study is to carry out an objective and uniform comparison of eight algorithms for filtering of fPCG for various types and levels of disturbance and evaluate their effectiveness using statistical parameters. ...
Fetal phonocardiography is a non-invasive, completely passive and low-cost method based on sensing acoustic signals from the maternal abdomen. However, different types of interference are sensed along with the desired fetal phonocardiography. This study focuses on the comparison of fetal phonocardiography filtering using eight algorithms: Savitzky-Golay filter, finite impulse response filter, adaptive wavelet transform, maximal overlap discrete wavelet transform, variational mode decomposition, empirical mode decomposition, ensemble empirical mode decomposition, and complete ensemble empirical mode decomposition with adaptive noise. The effectiveness of those methods was tested on four types of interference (maternal sounds, movement artifacts, Gaussian noise, and ambient noise) and eleven combinations of these disturbances. The dataset was created using two synthetic records r01 and r02, where the record r02 was loaded with higher levels of interference than the record r01. The evaluation was performed using the objective parameters such as accuracy of the detection of S1 and S2 sounds, signal-to-noise ratio improvement, and mean error of heart interval measurement. According to all parameters, the best results were achieved using the complete ensemble empirical mode decomposition with adaptive noise method with average values of accuracy = 91.53% in the detection of S1 and accuracy = 68.89% in the detection of S2. The average value of signal-to-noise ratio improvement achieved by complete ensemble empirical mode decomposition with adaptive noise method was 9.75 dB and the average value of the mean error of heart interval measurement was 3.27 ms.
... • In study [46], a method based on single-channel separation consisting of three steps was described. The proposed methodology combines the EMD method, singular value decomposition (SVD) and efficient version of fast independent component analysis (EFICA). ...
... Empirical mode decomposition (EMD) is a signal processing technique that is able to decompose any non-stationary and non-linear signal into oscillating components. These bandlimited components are also called intrinsic mode functions (IMFs) [46], [59]. Each IMF must comply with two basic conditions. ...
... First, the number of extrema and the number of zero crossings must be the same or may differ by one at most. Second, at any point, the mean value of the envelopes defined by the local maxima and the envelopes defined by the local minima is zero [46], [60]. The algorithm is iterative and can be described in the following steps [46], [59]- [61]: ...
Fetal phonocardiography (fPCG) is a non-invasive technique for detection of fetal heart sounds (fHSs), murmurs and vibrations. This acoustic recording is passive and provides an alternative low-cost method to ultrasonographic cardiotocography (CTG). Unfortunately, the fPCG signal is often disturbed by the wide range of artifacts that make it difficult to obtain significant diagnostic information from this signal. The study focuses on the filtering of an fPCG signal containing three types of noise (ambient noise, Gaussian noise, and movement artifacts of the mother and the fetus) having different amplitudes. Three advanced signal processing methods: empirical mode decomposition (EMD), ensemble empirical mode decomposition (EEMD), and adaptive wavelet transform (AWT) were tested and compared. The evaluation of the extraction was performed by determining the accuracy of S1 sounds detection and by determining the fetal heart rate (fHR). The evaluation of the effectiveness of the method was performed using signal-to-noise ratio (SNR), mean error of heart interval measurement (|ΔTi| ), and the statistical parameters of accuracy (ACC), sensitivity (SE), positive predictive value (PPV), and harmonic mean between SE and PPV (F1). Using the EMD method, ACC > 95\text {F1} = 93.12 % was achieved in 9 out of 12 types and levels of interference with average values of ACC = 97.49 %, SE = 97.89 %, PV = 99.53 % and F1 = 98.69 %. In this study, the best results were achieved using the AWT method, which provided ACC > 95 % in all 12 types and levels of interference with average values of ACC = 99.34 %, SE = 99.49 %, PPV = 99.85 % a F1 = 99.67 %.
... Vivek et al. [59,116] implemented a BSS for twin fetal phonocardiogram signals. Anil et al. implemented a PCA using Singular Value Decomposition (SVD) [117] to de-noise FPCG. Soysa et al. employed Eigen subspace filtering, a variant of PCA to improve the SNR of the FECG signal [101,118]. ...
... The EMD mainly depends on the number of frequency components and the amplitude of each component of the given signal. Anil et al. employed EMD in the study [117] as a preprocessor. The obtained Intrinsic Mode Functions (IMFs) were used as inputs to the PCA for further processing. ...
... ICA [59,83,116,117,[119][120][121][122] Extraction of FHS from noisy FPCG ...
Monitoring the well-being of a fetus through Fetal Phonocardiography (FPCG) has been occurring for more than a century. Throughout history, there have been continuous advances in sensor development, data acquisition systems, and signal processing techniques. Despite these advancements, FPCG based point of care technologies are facing serious challenges in translating from basic research to clinical trials and commercialization. This is partly due to the noisy characteristic associated with FPCG, to the lesser clinical knowledge about fetal and maternal physiological profiles, to the unavailability of gold standard databases, and to the limited application of reliable signal processing techniques. In order to understand why FPCG continues to be underutilized, it is necessary to know about the existing standards of fetal monitoring, data collection trends, and the signal processing aspects. To serve this purpose, this paper will first provide an overview of the existing standards of fetal monitoring and then provide a comprehensive survey on Fetal Phonocardiography with focus on trends in data collection, signal processing techniques and synthesis models that have been developed to date. Finally, a set of guidelines will be proposed for future research and use in signal analysis, processing and modeling based on the outlined challenges.
... However, in contrast to approaches based on wavelets, which perform the analysis by projecting the signal under consideration onto a number of predefined basis vectors, other decomposition methods, such as Empirical Mode Decomposition (EMD) and Hilbert-Huang Transform (HHT), express the signal as an expansion of basis functions which are signal-dependent, and are estimated via an iterative procedure called sifting [10]. For example, in [11], an approach based on EMD was presented, where fetal heart sounds were extracted from a recorded single channel abdominal PCG [11]. Another interesting area is the acoustical disturbances caused by heart murmurs, which can be analyzed using Mel-Frequency Cepstral Coefficients (MFCC) [12,13], but these procedures are very sensitive to artifacts or noises frequently involved in the acquisition stage [2]. ...
... However, in contrast to approaches based on wavelets, which perform the analysis by projecting the signal under consideration onto a number of predefined basis vectors, other decomposition methods, such as Empirical Mode Decomposition (EMD) and Hilbert-Huang Transform (HHT), express the signal as an expansion of basis functions which are signal-dependent, and are estimated via an iterative procedure called sifting [10]. For example, in [11], an approach based on EMD was presented, where fetal heart sounds were extracted from a recorded single channel abdominal PCG [11]. Another interesting area is the acoustical disturbances caused by heart murmurs, which can be analyzed using Mel-Frequency Cepstral Coefficients (MFCC) [12,13], but these procedures are very sensitive to artifacts or noises frequently involved in the acquisition stage [2]. ...
As cardiac murmurs do not generally appear in every area of auscultation, this paper presents an effective approach for cardiac murmur detection based on stochastic analysis of acoustic features derived from Empirical Mode Decomposition (EMD) and Hilbert-Huang Transform (HHT) of phonocardiographic (PCG) signals made up by the 4-Standard Auscultation Areas (SAA). The 4-SAA PCG database belongs to the National University of Colombia. Mel-Frequency Cepstral Coefficients (MFCC) and statistical moments of HHT were estimated over EMD components. An ergodic HMM was applied on the feature space, randomly initialized and trained by expectation maximization with a convergence at 10e-6 and a maximum iteration number of 1000. Global classification results for 4-SAA were around 98.7% with satisfactory sensitivity and specificity results, using a 30-fold cross-validation procedure (70/30 split). The representation capability of the EMD technique applied to 4-SAA PCG signals and stochastic analysis of acoustic features offered a high performance to detect cardiac murmurs.
... Décomposition modale empirique -La décomposition modale empirique s'apparente au traitement en ondelettes en réalisant une décomposition en fréquence et permettant également d'extraire l'amplitude sur des fenêtres locales. Également adaptée aux signaux non stationnaires, cette méthode permet également une auto-adaptabilité au signal d'entrée [115]. ...
La surveillance de l’asphyxie périnatale est effectuée par analyse visuelle du rythme cardiaquefoetal et des contractions utérines à l’aide d’un cardiotocographe. Cependant, cette méthode montreune forte variabilité inter-individuelle et une faible spécificité pour la détection de l’asphyxiefoetale. Pour améliorer le dépistage des foetus à risque d’asphyxie, de nombreuses recherches setournent vers l’analyse de l’activité du système nerveux autonome contrôlant l’homéostasie foetale.La variabilité de la fréquence cardiaque est un marqueur fiable et reproductible pour l’analyse del’activité du système nerveux autonome. Partant de ce constat, le Centre Hospitalier Universitairede Lille a développé un algorithme innovant d'analyse de la variabilité de la fréquence cardiaquetraduisant l’activité du système nerveux autonome. Bien que le cardiotocographe permette uneanalyse qualitative du rythme cardiaque foetal, cette technologie ne permet pas une précisionsuffisante pour l’analyse de la variabilité de la fréquence cardiaque. Actuellement, seule uneélectrode positionnée au scalp du foetus permet une analyse battement à battement du rythmecardiaque foetal mais son caractère invasif limite fortement son utilisation. Cette thèse a pourobjectif la conception d’un système non invasif permettant une acquisition précise et continue durythme cardiaque foetal. L’outil ainsi crée doit être en capacité de lever les verrous techniques desdispositifs actuellement commercialisés en permettant une acquisition battement à battement durythme cardiaque foetal pour l’analyse de la variabilité de la fréquence cardiaque. Pour ce faire,nous avons développé un dispositif multi-sources / multi-capteurs basé sur les technologiesd’électrocardiographie foetale et de phonocardiographie foetale. Après une première optimisation /validation théorique sur banc d’essais, un essai clinique effectué au CHU de Lille a permis definaliser et d’évaluer les performances du dispositif développé.
... Hadiyoso et al. (Hadiyoso et al., 2020) separately used VMD and ensemble EMD to extract multi-level features from singlelead ECG signals, thereby achieved biometric recognition and classification, and improved accuracy. Warbhe et al. (Warbhe et al., 2010) realized single-channel blind source separation of fetal heart sound signals combining EMD, SVD with ICA. Zhao et al. (Zhao et al., 2019) realized single-channel blind source separation for radio signals composed of a mixture of two cosine signals and 2ASK based on feedback VMD. ...
In industrial, biological, medical and many more scenarios, single-channel blind source separation still remains challenges. A smart universal single-channel blind source separation method, the iterative heuristic general hybrid model with evaluation parameters feedback, Loop-BAS-VMD-ICA, is proposed and verified to recover the original sources automatically in multiple scenarios. Combined with beetle antennae search algorithm based variational mode decomposition and independent component analysis, by designing a new fitness function armed with processing effects of variational mode decomposition components and the final candidate source signals in each step of beetle antennae search algorithm, this model allows us to reconstruct a new optimized vector combining the selected components of variational mode decomposition with the original observation signal through a new scheme by loop mode with evaluation parameters feedback, and then the new vector is used to separate and extract independent source signals by iterative and heuristic calculation. Experimental results show that our method is not only smart, good versatile but also outperforms the state-of-the-art traditional time–frequency-based methods in extraction accuracy and waveform integrity. Compared with the state-of-the-art deep learning based methods, our method also demonstrates its unique advantages. It provides a better and more extensive means for the analysis and application of multicomponent signals.
... PPV = 99.85% and F1 = 99.67%. • Warbhe et al. introduced a single-channel method combining EMD, singular value decomposition (SVD), and efficient version of ICA (EFICA) [56]. The combination of all methods was tested on real records and led to efficient extraction of fPCG from noisy signals. ...
Fetal phonocardiography (fPCG) is receiving attention as it is a promising method for continuous fetal monitoring due to its non-invasive and passive nature. However, it suffers from the interference from various sources, overlapping the desired signal in the time and frequency domains. This paper introduces the state-of-the-art methods used for fPCG signal extraction and processing, as well as means of detection and classification of various features defining fetal health state. It also provides an extensive summary of remaining challenges, along with the practical insights and suggestions for the future research directions.
... Further, many other fPCG signal processing techniques are explored in literature to deal with noisy fPCG and components decomposition based on: least mean square (LMS) linear prediction [20], Wavelet Transform [21], adaptive Wiener filtering [22], spectral subtraction [23], conventional filtering [24], blind source separation [25] auto or cross correlation [26], Wigner Ville Distribution -WVD [27], Short Time Fourier Transform -STFT [28], Hilbert Transform [29]. In recent times, fPCG Processing Using Empirical Mode Decomposition (EMD) and Singular Value Decomposition (SVD) is proposed in [30]. Matching Pursuit (MP) method for fPCG in telemedicine system [31] Fractal Dimension (FD) analysis in association with Wavelet transform [32] is presented. ...
Congenital cardiac anomalies of fetus are often characterized by the unprecedented changes in the auditory properties of cardiac sounds occurs during the gestation period of pregnant women. These abnormalities are often seen in inconsistent patterns of heart sounds that are driven by asynchronous variations in heart rates of mother and fetus. This hostile situation becomes severe if it is untreated and might threaten to life risk in pregnant women. In this work, we proposed a novel and automated signal processing paradigm using Variational Mode Decomposition (VMD) to detect and extract the mother and fetus heart sounds from the raw PCG signals recorded from outer surface of the maternal abdomen. The proposed framework constitutes a couple of cascaded VMD blocks: The first VMD block alleviates the raspatory noises and other artifacts from the raw PCG signal; while the Next one, decomposes the mother and fetus heart sounds obtained from its preceding block. A publicly available Shiraz University Fetal Heart Sounds Database is used to test the efficacy of the proposed model. In addition, noisy PCG corpus characterized by the additive white gaussian noise is used to test the efficacy of the proposed network. F
... The fourth step is subtraction of IMF from the original signal for creating the zero local mean. The fifth step is checking whether the output created function of the zero local mean is IMF or not based on the conditions described [5,73]. Table 2 shows a summary of different single channel methods. ...
Fetal electrocardiography is among the most promising methods of modern electronic fetal monitoring. However, before they can be fully deployed in the clinical practice as a gold standard, the challenges associated with the signal quality must be solved. During the last two decades, a great amount of articles dealing with improving the quality of the fetal electrocardiogram signal acquired from the abdominal recordings have been introduced. This article aims to present an extensive literature survey of different non-adaptive signal processing methods applied for fetal electrocardiogram extraction and enhancement. It is limiting that a different non-adaptive method works well for each type of signal, but independent component analysis, principal component analysis and wavelet transforms are the most commonly published methods of signal processing and have good accuracy and speed of algorithms.
... However, in contrast to approaches based on wavelets, other decomposition methods such as the adaptive method Empirical Mode Decomposition (EMD), introduced to analyze nonstationary signals [10], and Hilbert-Huang Transform (HHT), express the signal as an expansion of basis functions which are signal-dependent [11]. These techniques have been frequently applied for extraction of foetal heart sounds from a recorded single channel abdominal PCG [12]. However, EMD has problems, as the presence of inappropriate oscillations [10], which is attenuated adding Gauss White Noise (WGN) to ensemble the signal by a method named Ensemble Empirical Mode Decomposition (EEMD) [13]. ...
This paper presents an automatic detection system for the classification of phonocardiographic (PCG) signals using 4 standard auscultation areas (one of each cardiac valve) for heart murmur diagnosis. The database of 4-area PCG records belongs to the National University of Colombia. A set of 50 individuals were labeled as normal, while 98 were labeled as exhibiting cardiac murmurs, caused by valve disorders. With the help of medical experts, 400 representative beats were chosen, 200 normal and 200 with evidence of cardiac murmur from 4 different areas of auscultation. First, the PCG signals were preprocessed; next, four different derivations of Mel Frequency Cepstral Coefficients (MFCC) were extracted. Additionally, statistical moments of Hilbert Huang Transform (HHT) were estimated using different combinations of the signal components by means of Empirical Mode Decomposition (EMD), Ensemble EMD (EEMD) and Complete EEMD with Adaptative Noise (CEEMDAN), independently, where the computational complexity were compared. Finally, stochastic analysis of the feature space was carried out by an ergodic-HMM and the global classification result was around 98% with acceptable sensitivity and specificity scores, using a 30-fold cross-validation procedure (70/30 split).
... However, other decomposition methods, such as Empirical Mode Decomposition (EMD) and Hilbert-Huang Transform (HHT), express the signal in a better way as an expansion of signal-dependent basis functions, via an iterative procedure called sifting [8]. For example, the foetal heart sounds could be extracted from a recorded single channel abdominal PCG, using an EMD approach proposed in [9]. In another way, the acoustic analysis by Mel-Frequency Cepstral Coefficients (MFCC) [10] has been proposed to analyze the acoustical disturbances caused by heart murmurs, but these procedures are very sensitive to artifacts or noises frequently involved in the acquisition stage [1]. ...
The heart's mechanical activity can be appraised by auscultation recordings, taken from the 4-Standard Auscultation Areas (4-SAA), one for each cardiac valve, as there are invisible murmurs when a single area is examined. This paper presents an effective approach for cardiac murmur detection based on adaptive neuro-fuzzy inference systems (ANFIS) over acoustic representations derived from Empirical Mode Decomposition (EMD) and Hilbert-Huang Transform (HHT) of 4-channel phonocardiograms (4-PCG). The 4-PCG database belongs to the National University of Colombia. Mel-Frequency Cepstral Coefficients (MFCC) and statistical moments of HHT were estimated on the combination of different intrinsic mode functions (IMFs). A fuzzy-rough feature selection (FRFS) was applied in order to reduce complexity. An ANFIS network was implemented on the feature space, randomly initialized, adjusted using heuristic rules and trained using a hybrid learning algorithm made up by least squares and gradient descent. Global classification for 4-SAA was around 98.9% with satisfactory sensitivity and specificity, using a 50-fold cross-validation procedure (70/30 split). The representation capability of the EMD technique applied to 4-PCG and the neuro-fuzzy inference of acoustic features offered a high performance to detect cardiac murmurs.
The application of fiber-optic-based sensors, especially in the magnetic resonance (MR) environment and the sleep laboratory, has become an intensely discussed topic. Although these sensors offer significant benefits, their practical deployment has two very challenging issues—it is necessary to find a suitable way to construct and encapsulate sensors, and it is also required to ensure that an appropriate advanced signal processing method is chosen. This study focuses on the latter area, aiming to apply advanced methods of processing measured biosignals obtained from fiber-optic sensors that use light interference for their function. These sensors are characterized by the fact that we can classify the measured biosignals as phonocardiography (PCG). This article describes in length the determination of a patient’s heart rate (HR) as a basic parameter determining his or her state of health. The study is based on results collected from 11 test subjects (five females and six males), using the following three testing methods: empirical mode decomposition (EMD), complete ensemble EMD with adaptive noise (CEEMDAN), and wavelet transform (WT). The evaluation was conducted by determining the probability of correct detection with the use of overall accuracy (ACC), sensitivity (SE), positive predictive value (PPV), and the harmonic mean between SE and PPV (
F1
). The functionality of the system was verified against the relevant reference in the form of simultaneously measured electrocardiograms (ECGs), from which reference annotations were estimated. This work showed that WT seems to be a suitable method, when, for all 11 tested signals, it achieved an ACC of >95%, based on the evaluation parameters, and at the same time, its computational complexity was the lowest of the tested methods.
Although the noninvasive continuous fetal heart rate (FHR) monitor is often recommended, the Doppler Ultrasonographic Cardiotocography (CTG) is improper for long-term monitor due to the less safety and the requirement of professional operation skill. In this paper, we design a noninvasive, continuous and real-time FHR monitoring system based on fetal phonocardiography by stationary wavelet denoising and cyclostationary process. Good agreement with CTG is obtained by Bland Altman analysis. Besides, quantitative results show that the FHR has an average accuracy of 97% compared with CTG on clinical data sets. The proposed system provides an alternative for CTG.
RESUMEN Los sistemas de diagnóstico automatizados para la detección de soplos cardiacos, descritos en la literatura, regis-tran en intervalos cortos de tiempo la dinámica fisiológica a partir de bases de datos de señales fonocardiográficas segmentadas, obteniendo una multiplicidad de muestras del mismo paciente sin tener en cuenta la dinámica car-diaca de cada individuo, lo cual no garantiza una adecuada identificación de la anomalía cardiaca. El sistema de asistencia de diagnóstico propuesto inicia con el estudio de 1060 señales tomadas de los cuatro focos de auscul-tación cardiaca de 144 pacientes, agrupados principalmente en dos clases: normales y patológicos. Los registros se agruparon por su posición en el periodo cardiaco en 4 tipos: normales, soplo sistólico, diastólico y sistodias-tólico, obteniendo una base de datos de señales fonocardiográficas, a la cual se le realiza un preprocesamiento, sin embargo, no se segmenta para conservar la dinámica cardiaca de cada individuo en el estudio. Luego se genera un espacio de representación a partir de los PLP y los coeficientes cepstrales calculados a partir de la FFT y la STFT, desarrollando un análisis con los clasificadores estocásticos HMM y GMM y técnicas de selección de caracterís-ticas para la extracción de información relevante de la dinámica fisiológica, que permitan un adecuado y eficiente entrenamiento del sistema con una tasa adecuada de clasificación para soporte de diagnóstico clínico. Palabras Clave: Procesamiento de señales, Modelos de Mezclas Gaussianas, Modelos Ocultos de Markov, So-plos Cardiacos, Detección de Patologías Rev. Invest. Univ. Quindío.(Col.), 23(1): 8-15; 2012
This paper presents a new methodology for single-channel blind signal separation (SCBSS) of time-frequency overlapped signals in electromagnetic surveillance domain. This method combines the complete ensemble empirical mode decomposition (CEEMD) with fast independent component analysis (FastICA). Firstly, the single-channel recording is decomposed into a set of intrinsic mode function (IMF) components by the method CEEMD with adaptive noise, for the residue and the number of shifting iterations of CEEMD are smaller than that of other empirical mode decomposition approach. The IMF components become the basis representing the original data. After selecting the usefull IMF components according to their power spectrum, FastICA is used to separate the source of interest in the original signal. Simulation results obtained in evaluating the proposed methodology's performance confirmed the feasibility and effectiveness of this algorithm.
A new method for analysing nonlinear and non-stationary data has been developed. The key part of the method is the `empirical mode decomposition' method with which any complicated data set can be decomposed into a finite and often small number of 'intrinsic mode functions' that admit well-behaved Hilbert transforms. This decomposition method is adaptive, and, therefore, highly efficient. Since the decomposition is based on the local characteristic time scale of the data, it is applicable to nonlinear and non-stationary processes. With the Hilbert transform, the 'instrinic mode functions' yield instantaneous frequencies as functions of time that give sharp identifications of imbedded structures. The final presentation of the results is an energy-frequency-time distribution, designated as the Hilbert spectrum. In this method, the main conceptual innovations are the introduction of `intrinsic mode functions' based on local properties of the signal, which make the instantaneous frequency meaningful; and th
The problem of separating n linearly superimposed uncorrelated signals and determing their mixing coefficents is reduced to an Eigenvalue problem which involves the simultaneous diagonalisation of two symmetric matrices whose elements are measureable time delayed correlation functions. The diagonalisation matrix can be determined from a cost function whose number of minima is equal the number of degenerate solutions. Our approach offers the possibility to separate also nonlinear mixtures of signals. PACS numbers:02.50.+s, 05.40.+j, 06.50.-x, 87.71.-p Typeset using REVT E X The problem of source separation appears in many contexts. The most simple situation occurs for two speakers. If the mixture of their voices reaches two microphones one wants to separate both sources such that each detector registers only one voice [1]. Typical examples involving many sources and many receivers are the separation of radio or radar signals by an array of antennas [2], the separation of odors in a mi...
Doppler ultrasound, ultrasound M-mode analysis, fetal electrocardiography, and fetal magnetocardiography are methods by which the fetal heart can be monitored non-invasively. In this paper, they are evaluated and compared. Customarily, it is solely the fetal heart rate, which is monitored using the Doppler ultrasound technique since it is both simple to use and cheap. However, this method inherently produces an averaged heart rate and therefore cannot give the beat-to-beat variability. Fetal electrocardiography has similar advantages, but in addition offers the potential for monitoring beat-to-beat variability and performing electrocardiogram morphological analysis. Its disadvantage is that its reliability is only 60%, although it is the only technique that offers truly long-term ambulatory monitoring. Ultrasound M-mode analysis allows a estimation of atrial and ventricular coordination, as well as an estimation of PR intervals. Bradycardias, supraventricular tachycardias, extra systoles are readily diagnosed using this method although timing will be inaccurate. Fetal magnetocardiograms can be detected reliably and used for accurate beat-to-beat measurements and morphological analysis. Consequently, they can be used for the classification of arrhythmias and the diagnosis of a long QT syndrome and some congenital heart diseases.
FastICA is one of the most popular algorithms for independent component analysis (ICA), demixing a set of statistically independent sources that have been mixed linearly. A key question is how accurate the method is for finite data samples. We propose an improved version of the FastICA algorithm which is asymptotically efficient, i.e., its accuracy given by the residual error variance attains the Cramér-Rao lower bound (CRB). The error is thus as small as possible. This result is rigorously proven under the assumption that the probability distribution of the independent signal components belongs to the class of generalized Gaussian (GG) distributions with parameter alpha, denoted GG(alpha) for alpha > 2. We name the algorithm efficient FastICA (EFICA). Computational complexity of a Matlab implementation of the algorithm is shown to be only slightly (about three times) higher than that of the standard symmetric FastICA. Simulations corroborate these claims and show superior performance of the algorithm compared with algorithm JADE of Cardoso and Souloumiac and nonparametric ICA of Boscolo et al. on separating sources with distribution GG (alpha) with arbitrary alpha, as well as on sources with bimodal distribution, and a good performance in separating linearly mixed speech signals.
A novel technique is developed to separate the audio sources from a single mixture. The method is based on decomposing the Hilbert spectrum (HS) of the mixed signal into independent source subspaces. Hilbert transform combined with empirical mode decomposition (EMD) constitutes HS, which is a fine-resolution time-frequency representation of a nonstationary signal. The EMD represents any time-domain signal as the sum of a finite set of oscillatory components called intrinsic mode functions (IMFs). After computing the spectral projections between the mixed signal and the individual IMF components, the projection vectors are used to derive a set of spectral independent bases by applying principal component analysis (PCA) and independent component analysis (ICA). A k-means clustering algorithm based on Kulback-Leibler divergence (KLd) is introduced to group the independent basis vectors into the number of component sources inside the mixture. The HS of the mixed signal is projected onto the space spanned by each group of basis vectors yielding the independent source subspaces. The time-domain source signals are reconstructed by applying the inverse transformation. Experimental results show that the proposed algorithm performs separation of speech and interfering sound from a single mixture
In this paper we introduce a new technique for blind source separation of speech signals. We focus on the temporal structure of the signals in contrast to most other major approaches to this problem. The idea is to apply the decorrelation method proposed by Molgedey and Schuster in the time-frequency domain. We show some results of experiments with both artificially controlled data and speech data recorded in the real environment. 1 Introduction Recently, blind source separation, or BSS, within the framework of independent component analysis has attracted a great deal of attention in engineering field. It has been widely noticed that there are many possible applications such as removing additive noise from signals and images, separating crosstalk in telecommunication, preprocessing for multi-probed radar-sonar signals, and analyzing EEG (Electroencephalograph) or MEG (Magnetoencephalograph) data (see for example [12]). Blind source separation is the problem to separate independent sou...
This paper presents a novel method for the determination of the fetal heart rate by the means of phonocardiography. The two-channel acoustic recording method measures simultaneously the fetal heart sounds on the maternal abdominal surface with the environmental noises. The recorded signals are sampled and analysed by a personal computer. The proposed measurement and signal processing methods provide an accuracy which is comparable to that of ultrasound technology has. The introduced system is viable as a fully non-invasive and inexpensive portable fetal heart rate monitoring device, especially in the field of home care and telemedicine applications.
A new method for analyzing nonlinear and nonstationary data has been developed. The key pat of the method is the Empirical Mode Decomposition method with which any complicated data set can be decomposed into a finite and often small number of Intrinsic Mode Functions (IMF). An IMF is define das any function having the same numbers of zero- crossing and extrema, and also having symmetric envelopes defined by the local maxima and minima respectively. The IMF also admits well-behaved Hilbert transform. This decomposition method is adaptive, and therefore, highly efficient. Since the decomposition is based on the local characteristic time scale of het data, it is applicable to nonlinear and nonstationary processes. With the Hilbert transform, the IMF yield instantaneous frequencies as functions of time that give sharp identifications of embedded structures. The final presentation of the result is an energy-frequency-time distribution, designated as the Hilbert Spectrum. Comparisons with Wavelet and window Fourier analysis show the new method offers much better temporal and frequency resolutions.
Based on numerical experiments on white noise using the empirical mode decompo-sition (EMD) method, we find empirically that the EMD is effectively a dyadic filter, the intrinsic mode function (IMF) components are all normally distributed, and the Fourier spectra of the IMF components are all identical and cover the same area on a semi-logarithmic period scale. Expanding from these empirical findings, we further deduce that the product of the energy density of IMF and its corresponding averaged period is a constant, and that the energy-density function is chi-squared distributed. Furthermore, we derive the energy-density spread function of the IMF components. Through these results, we establish a method of assigning statistical significance of information content for IMF components from any noisy data. Southern Oscillation Index data are used to illustrate the methodology developed here.
The various noise components make the diagnostic evaluation of phonocardiographic records difficult or in some cases even impossible. This paper presents a novel wavelet-based denoising method using two-channel signal recording and an adaptive cross-channel coefficient thresholding technique. The qualitative evaluation of the denoising performance has shown that the proposed method cancels noises more effectively than the other examined techniques. The introduced method can be used as preprocessor stage in all fields of phonocardiography, including the recording of fetal heart sounds on the maternal abdominal surface.
The independent component analysis (ICA) of a random vector consists of searching for a linear transformation that minimizes the statistical dependence between its components. In order to define suitable search criteria, the expansion of mutual information is utilized as a function of cumulants of increasing orders. An efficient algorithm is proposed, which allows the computation of the ICA of a data matrix within a polynomial time. The concept of ICA may actually be seen as an extension of the principal component analysis (PCA), which can only impose independence up to the second order and, consequently, defines directions that are orthogonal. Potential applications of ICA include data analysis and compression, Bayesian detection, localization of sources, and blind identification and deconvolution.
Independent component analysis (ICA) is a statistical method for transforming an observed multidimensional random vector into components that are statistically as independent from each other as possible. In this paper, we use a combination of two different approaches for linear ICA: Comon's information-theoretic approach and the projection pursuit approach. Using maximum entropy approximations of differential entropy, we introduce a family of new contrast (objective) functions for ICA. These contrast functions enable both the estimation of the whole decomposition by minimizing mutual information, and estimation of individual independent components as projection pursuit directions. The statistical properties of the estimators based on such contrast functions are analyzed under the assumption of the linear mixture model, and it is shown how to choose contrast functions that are robust and/or of minimum variance. Finally, we introduce simple fixed-point algorithms for practical optimization of the contrast functions. These algorithms optimize the contrast functions very fast and reliably.
This paper is aimed at the selection of de-noising algorithm for de-noising of the fetal phonocardiographic (fPCG) -signals. Fourier-based analyzing tools have some limitations concerning frequency and time resolutions. Although wavelet transform (WT) overcomes these limitations, it requires selection of appropriate de-noising algorithm. The universal threshold, minimax threshold and rigorous SURE (Stein′s Unbiased Risk Estimate) threshold algorithms along with soft or hard thresholding rule have been compared for de-noising of these signals. The mean-squared error (MSE) is used to evaluate the performance of these algorithms. The results show that, the rigorous SURE threshold algorithm with soft thresholding rule has a better performance for the analysis of fPCG signals when using the fourth-order Coiflets wavelet. The proposed approach is simple and proves to be effective when applied for the selection of de-noising algorithm for the fPCG signals. These de-noised signals can be used for the accurate determination of fetal heart rate (FHR) and further diagnostic applications pertaining to the fetus.
The continued development of a computerised system for measuring the pattern of the antepartum fetal heart rate (FHR) is described. Previous work had established that measurement of FHR variation objectively detects chronic fetal hypoxaemia and the onset of metabolic acidaemia antepartum. The normal centiles were calculated for the amplitude of long-term FHR variation, in episodes of high and low variation, week by week from 24-42 weeks gestation. Reference to these (automatically by the computer) improved discrimination between normal and questionable records in 38% of records, with a small saving of time. Two types of sinusoidal rhythm were described (slow, 1 in 2-5 minutes, incidence 0.16% of subjects; and faster, 2-5 per minute, incidence 0.025%) with methods for their detection. Both may be of sufficient amplitude to induce an episode of high FHR variation. The different effects of maternal steroid (betamethasone or dexamethasone) administration of FHR variation were compared, and the clinical consequences considered. The frequency distribution of basal FHR in normal and abnormal records was measured, and the effects on basal FHR outside the normal range (120-160 bpm) on FHR variation described. Adjustment of the FHR baseline was undertaken when, exceptionally, large abrupt changes in heart rate occurred during a record. The duration and frequency of FHR record acquisition in clinical practice were reviewed, and new policies recommended. With adequate safeguards, measurement by a computer offers reliable objective information from which fetal health may be assessed, more objectively and accurately than by visual inspection.
A real-time method for fetal heart rate (FHR) monitoring based on signal processing of the fetal heart sounds is presented. The acoustic method, which utilizes an adaptive time pattern analysis to select and analyze those heartbeats that can be recorded without artefact, is guided by a number of rules involving an introduced confidence factor on the timing prediction. The algorithm was implemented in a low-power portable electronic instrument to enable long-term fetal surveillance. A large number of clinical tests have shown the very good performance of the phonocardiographic method in comparison with FHR curves simultaneously recorded with ultrasound cardiotocography. Indeed, approximately 90% of the time, the acoustic FHR curve remained inside a +/- 3 beats/min tolerance limit of the reference ultrasound method. The confidence was typically CF > 0.85. The acoustic method exceeded a +/- 5 beats/min limit relative to the ultrasound method approximately 5% of the time. Finally, no relevant FHR data was measured approximately 5% of the time.
A fundamental problem in neural network research, as well as in many other disciplines, is finding a suitable representation of multivariate data, i.e. random vectors. For reasons of computational and conceptual simplicity, the representation is often sought as a linear transformation of the original data. In other words, each component of the representation is a linear combination of the original variables. Well-known linear transformation methods include principal component analysis, factor analysis, and projection pursuit. Independent component analysis (ICA) is a recently developed method in which the goal is to find a linear representation of non-Gaussian data so that the components are statistically independent, or as independent as possible. Such a representation seems to capture the essential structure of the data in many applications, including feature extraction and signal separation. In this paper, we present the basic theory and applications of ICA, and our recent work on the subject.
Blind signal separation (BSS) and independent component analysis
(ICA) are emerging techniques of array processing and data analysis that
aim to recover unobserved signals or “sources” from observed
mixtures (typically, the output of an array of sensors), exploiting only
the assumption of mutual independence between the signals. The weakness
of the assumptions makes it a powerful approach, but it requires us to
venture beyond familiar second order statistics, The objectives of this
paper are to review some of the approaches that have been developed to
address this problem, to illustrate how they stem from basic principles,
and to show how they relate to each other
A Comparative Analysis of Algorithms for Fetal Phonocardiographjic Signals
- V Chaurasia
- A K Mitra