Stochastic mechanical model of vocal folds for producing jitter and for identifying pathologies through real voices

Jitter, in voice production applications, is a random phenomenon characterized by the deviation of the glottal cycle length with respect to a mean value. Its study can help in identifying pathologies related to the vocal folds according to the values obtained through the different ways to measure it. This paper aims to propose a stochastic model, considering three control parameters, to generate jitter based on a deterministic one-mass model for the dynamics of the vocal folds and to identify parameters from the stochastic model taking into account real voice signals experimentally obtained. To solve the corresponding stochastic inverse problem, the cost function used is based on the distance between probability density functions of the random variables associated with the fundamental frequencies obtained by the experimental voices and the simulated ones, and also on the distance between features extracted from the voice signals, simulated and experimental, to calculate jitter. The results obtained show that the model proposed is valid and some samples of voices are synthesized considering the identified parameters for normal and pathological cases. The strategy adopted is also a novelty and mainly because a solution was obtained. In addition to the use of three parameters to construct the model of jitter, it is the discussion of a parameter related to the bandwidth of the power spectral density function of the stochastic process to measure the quality of the signal generated. A study about the influence of all the main parameters is also performed. The identification of the parameters of the model considering pathological cases is maybe of all novelties introduced by the paper the most interesting.     

Voice alterations in patients with Morquio A syndrome

Morquio A syndrome, or mucopolysaccharidosis (MPS IV A), is an inherited lysosomal storage disorder which belongs to the group of mucopolysaccharidoses (MPSs). It is caused by N-acetylgalactosamine-6-sulfatase (GALNS) activity deficiency, which results in impaired degradation of glycosaminoglycans (GAGs), including keratan sulfate (KS) and chondroitin-6-sulfate (CS). These compounds infiltrate and disrupt the architecture of the extracellular matrix, compromising the integrity of the connective tissue. Patients with Morquio A have also been noted for exhibiting abnormalities of the larynx and vocal tract. The aim of the study was to assess voice alterations using noninvasive acoustic and electroglottographic voice analysis. Electroglottographic signal and acoustic analyses revealed considerable changes in the voices of patients with Morquio A syndrome when compared to the voices of healthy controls. Affected patients tended toward tense voice, incomplete glottal closure, increased incidence of vocal fold nodules, dysphonia, and hoarse voice. Morquio A syndrome is characterized by connective tissue disease, which adversely affects voice quality. The use of objective voice analysis makes it possible to quantitatively monitor changes in the vocal apparatus over the course of disease progression, and also allows for assessment of the effects of the enzyme replacement therapy.     

Use of texture analysis to discriminate between normal livers and livers with steatosis

We describe a technique for discriminating between livers that are normal and ones that have fatty infiltrate (livers with steatosis) based upon the application of a suitably defined texture measure to the corresponding digitized ultrasonographs. In brief, this texture measure sums, from some selected optimum spatial frequency to the upper limit set by the digitizing process, the frequency components of the normalized, radial power spectral density function. The analysis was run on six cases (two normal and four with steatosis) obtained from Dunedin Public Hospital, in New Zealand. Texture measure values for these six cases were compared with the corresponding biopsy scores. The results indicate the ability of the texture measure to discriminate between the two conditions; and furthermore, to quantitatively distinguish the severity of histological change.     

Dynamic detection of rhythmic oscillations in heart-rate tracings: a state-space approach based on fourth-order cumulants

 This paper presents a new procedure specifically aimed at providing a dynamical detection of the oscillations occurring in long-term heart-rate (HR) tracings. The procedure is based on a time-variant state-space modelling of the fourth-order cumulants of the HR signal. The state-space estimator was selected because of its demonstrated capability to distinguish between deterministic and stochastic components of the signal, while the fourth-order cumulants of the signal were used as input of the model to further reduce adverse effects of coloured, white and 1/f Gaussian noise possibly present in the input data. The procedure was tested by the analysis of simulated signals and its performance was compared with the results obtained by state-space modelling applied directly on the test signals (instead of on the fourth-order cumulants of the signals) and by the more traditional auto-regressive modelling. The comparison has shown a clear superiority of the proposed procedure over the other techniques in discriminating deterministic oscillations from coloured noise. Finally, the applicability of the procedure to biological data was verified by analysing five experimental HR tracings recorded in normal subjects during laboratory and daily life conditions. Received: 17 May 1996 / Accepted in revised form: 29 November 1996     

Estimation of Frequency Resolving Power of Human Hearing by Different Methods: Roles of Sensory and Cognitive Factors

Frequency resolving power (FRP) was measured in normal listeners. FRP was estimated on the basis of the maximum resolvable ripple density in rippled-spectrum signals. Two measurement procedures were compared: detection of ripple-pattern change and comparison of rippled-spectrum signals. In the change detection method, two successive sound signals were presented to the listener: a test signal and a reference signal. The test signal contained ripple phase reversals every 400 ms; in the reference signal, the ripple phase was constant. The listener’s task was to identify the test signal. In the comparison method, three signals were presented to the listener. The ripple phase in one of the three signals was opposite to that in other two signals. The listener’s task was to identify the signal different from the other two signals. The signal frequency bands varied from 0.5 to 5 oct at a level of 0.5 of the maximum. At all frequency bands, the change-detection method yielded, on average, 1.75 oct^–1 higher FRP estimates compared to the comparison method. This difference between the two methods is supposed to be due to the greater involvement of cognitive processes (short-term memory) in the comparison method. The change-detection method is more preferable for measuring the sensory component of FRP.     

Time/frequency events of surface mechanomyographic signals resolved by nonlinearly scaled wavelets

The purpose of this investigation is to introduce a wavelet analysis designed for analyzing short events reflecting bursts of muscle activity in non-stationary mechanomyographic (MMG) signals. A filter bank of eleven nonlinearly scaled wavelets that maintain the optimal combination of time and frequency resolution across the frequency range of MMG signals (5–100 Hz) was used for the analysis. A comparison with the short-time Fourier transform, Wigner-Ville transform and continuous wavelet transform using a test signal with known time–frequency characteristics showed that the MMG wavelet analysis resolved the intensity, timing, and frequencies of events in a more distinct way without overemphasizing high or low frequencies or generating interference terms. The analysis was used to process MMG signals from the vastus lateralis, rectus femoris, and vastus medialis muscles obtained during maximal concentric and eccentric isokinetic movements. Muscular events were observed that were precisely located in time and frequency in a muscle-specific way, thereby showing periods of synergistic contractions of the quadriceps muscles. The MMG wavelet spectra showed different spectral bands for concentric and eccentric isokinetic movements. In addition, the high and low frequency bands seemed to be activated independently during the isokinetic movement. What generates these bands is not yet known, however, the MMG wavelet analysis was able to resolve them, and is therefore applicable to non-stationary MMG signals.     

Multivariate multi-scale weighted permutation entropy analysis of EEG complexity for Alzheimer’s disease

The complexity change of brain activity in Alzheimer’s disease (AD) is an interesting topic for clinical purpose. To investigate the dynamical complexity of brain activity in AD, a multivariate multi-scale weighted permutation entropy (MMSWPE) method is proposed to measure the complexity of electroencephalograph (EEG) obtained in AD patients. MMSWPE combines the weighted permutation entropy and the multivariate multi-scale method. It is able to quantify not only the characteristics of different brain regions and multiple time scales but also the amplitude information contained in the multichannel EEG signals simultaneously. The effectiveness of the proposed method is verified by both the simulated chaotic signals and EEG recordings of AD patients. The simulation results from the Lorenz system indicate that MMSWPE has the ability to distinguish the multivariate signals with different complexity. In addition, the EEG analysis results show that in contrast with the normal group, the significantly decreased complexity of AD patients is distributed in the temporal and occipitoparietal regions for the theta and the alpha bands, and also distributed from the right frontal to the left occipitoparietal region for the theta, the alpha and the beta bands at each time scale, which may be attributed to the brain dysfunction. Therefore, it suggests that the MMSWPE method may be a promising method to reveal dynamic changes in AD.     

Hilbert-Huang transform for analysis of heart rate variability in cardiac health

This paper introduces a modified technique based on Hilbert-Huang transform (HHT) to improve the spectrum estimates of heart rate variability (HRV). In order to make the beat-to-beat (RR) interval be a function of time and produce an evenly sampled time series, we first adopt a preprocessing method to interpolate and resample the original RR interval. Then, the HHT, which is based on the empirical mode decomposition (EMD) approach to decompose the HRV signal into several monocomponent signals that become analytic signals by means of Hilbert transform, is proposed to extract the features of preprocessed time series and to characterize the dynamic behaviors of parasympathetic and sympathetic nervous system of heart. At last, the frequency behaviors of the Hilbert spectrum and Hilbert marginal spectrum (HMS) are studied to estimate the spectral traits of HRV signals. In this paper, two kinds of experiment data are used to compare our method with the conventional power spectral density (PSD) estimation. The analysis results of the simulated HRV series show that interpolation and resampling are basic requirements for HRV data processing, and HMS is superior to PSD estimation. On the other hand, in order to further prove the superiority of our approach, real HRV signals are collected from seven young health subjects under the condition that autonomic nervous system (ANS) is blocked by certain acute selective blocking drugs: atropine and metoprolol. The high-frequency power/total power ratio and low-frequency power/high-frequency power ratio indicate that compared with the Fourier spectrum based on principal dynamic mode, our method is more sensitive and effective to identify the low-frequency and high-frequency bands of HRV.     

Conditional entropy approach for the evaluation of the coupling strength

A method that enables measurement of the degree of coupling between two signals is presented. The method is based on the definition of an uncoupling function calculating, by means of entropy rates, the minimum amount of independent information (i.e. the information carried by one signal which cannot be derived from the other). An estimator of the uncoupling function able to deal with short segments of data (a few hundred samples) is proposed, thus enabling the method to be used for usual experimental recordings. A synchronisation index is derived from the estimate of the uncoupling function by means of a minimisation procedure. It quantifies the maximum amount of information exchanged between the two signals. Simulations in which non-linear coordination schemes are produced and changes in the coupling strength are artificially induced are used to check the ability of the proposed index to measure the degree of synchronisation between signals. The synchronisation analysis is utilised to measure the coupling strength between the beat-to-beat variability of the sympathetic discharge and ventilation in decerebrate artificially ventilated cats and the degree of synchronisation between the beat-to-beat variability of the heart period and ventricular repolarisation interval in normal subjects and myocardial infarction patients. The sympathetic discharge and ventilation are strongly coupled and the coupling strength is not affected by manoeuvres capable of increasing or depressing sympathetic activity. The synchronisation is lost after spinalisation. The synchronisation analysis confirms that the heart period and ventricular repolarisation interval are well coordinated. In normal subjects, the synchronisation index is not modified by experimental conditions inducing changes in the sympathovagal balance. On the contrary, it strongly decreases after myocardial infarction, thus detecting and measuring the uncoupling between the heart period and ventricular repolarisation interval. Received: 29 October 1998 / Received in revised form: 4 March 1999     

Spectral jitter modeling and estimation

This paper suggests a new method for short-time jitter estimation based on a mathematical model that describes the coupling of two periodical phenomena. Specifically, jitter is modeled as the movement of one of the two periodic phenomena with respect to the other. The proposed method measures this movement indirectly by taking into account the spectral properties of the suggested model. Experiments with synthetic jitter signals showed that the suggested method produces accurate local estimates of jitter. Further evaluation was conducted on actual normal and pathological voice signals using two databases and jitter estimations from the Multi-Dimension Voice Program (MDVP) and the Praat system. Compared with the corresponding parameters from these two systems, the proposed method outperformed both in normal vs. pathological voice discrimination accuracy by at least 4%. Furthermore, it was shown that these methods actually rely on low-pass information, while the proposed method takes into account the full spectrum. The study of the short-time statistics of the jitter measurements provided by the suggested method indicates that there is a higher correlation with voice pathology compared to that obtained by the other two systems.     

一种喉部疾病口语声诊断新方法

本文提出了声带的三质量块模型,并应用这模型模拟病噪产生的嘶哑语声,这些嘶声包括有声带闭合不全,声带小结,声带麻痹,喉炎,声带淀粉样变和声门癌等十六种典型情况。采用快速傅里叶变换,线性预测,倒谱技术和离荼余弦变换等分析各类喉病引起的嘶哑语声,实验结果表明声带模型分析法是喉病诊断的一种有效方法。     

The use of backscattered electrons in analytical electron microscopy for the measurement of the mass of individual rat blood platelets

The backscattered electron signal, generated in individual cells, has been used to measure the dry mass of these cells. Absolute mass values were obtained by comparing the backscattered electron signals of cells to the signals of polystyrene-latex spheres of known mass. The technique was carried out in an automated analytical scanning transmission electron microscope and applied to rat blood platelets. The resulting mass distributions agreed well with the distribution measured with a method that uses the transmitted electron signal by means of densitometric analysis of electrographs. Also the range of masses was in agreement with values deduced from data in the literature. The fully automated technique has the advantage that it is direct, fast, and that thicker specimens can be measured than is possible using the transmitted electron signal. The method is intended for use in combination with quantitative electron probe X-ray microanalysis and is then able to produce elemental mass fractions of biological specimens at the subcellular level.     

The use of backscattered electrons in analytical electron microscopy for the measurement of the mass of individual rat blood platelets

Summary The backscattered electron signal, generated in individual cells, has been used to measure the dry mass of these cells. Absolute mass values were obtained by comparing the backscattered electron signals of cells to the signals of polystyrene-latex spheres of known mass. The technique was carried out in an automated analytical scanning transmission electron microscope and applied to rat blood platelets.The resulting mass distributions agreed well with the distribution measured with a method that uses the transmitted electron signal by means of densitometric analysis of electrographs. Also the range of masses was in agreement with values deduced from data in the literature.The fully automated technique has the advantage that it is direct, fast, and that thicker specimens can be measured than is possible using the transmitted electron signal. The method is intended for use in combination with quantitative electron probe X-ray microanalysis and is then able to produce elemental mass fractions of biological specimens at the subcellular level.In honour of Prof. van Duijn     

Detection of Abnormal Respiratory Events with Single Channel ECG and Hybrid Machine Learning Model in Patients with Obstructive Sleep Apnea

Respiratory scoring is an important step in the diagnosis of Obstructive Sleep Apnea (OSA). Airflow, abdolmel-thorax and pulse oximetry signals are obtained with the help of Polysomnography (PSG) device for the respiration scoring stage. These signals are visually scored by a specialist physician. The PSG has several disadvantages: one of them is that a technician is required to use the device. In addition, the records must be taken in the hospital environment. The aim of this study is to develop a new machine learning based hybrid sleep/awake detection method with single channel ECG alternative to respiratory scoring. For this purpose, electrocardiography (ECG) signal of 10 patients with OSA was used. The Heart Rate Variable signal was derived from the ECG signal. Then, QRS components in different frequency bands were obtained from ECG signal by digital filtering. In this way, a total of nine more signals were obtained. Each of the nine signals consists of 25 features, which amounts to a total of 225 features. Fisher feature selection algorithm and Principal Component Analysis (PCA) were used to reduce the number of features. Ultimately the features extracted from the first received signals were classified with Decision Tree, Support Vector Machines, k-Nearest Neighborhood Algorithm and Ensemble classifiers. In addition, the proposed model was checked with the Leave One Out method. At the end of the study, for the detection of apnea, 82.11% accuracy with only 3 features and 85.12% accuracy with 13 features were obtained. The sensitivity and specificity values for the 3 properties are 0.82 and 0.82, respectively. For the 13 properties, 0.85 and 0.86, respectively. These results show that the proposed model can be used for the detection of Respiratory Scoring in the OSA diagnostic process.     

A system for parallel measurement of glottis opening and larynx position

The simultaneous assessment of glottal dynamics and larynx position can be beneficial for the diagnosis of disordered voice or speech production and swallowing. Up to now, methods either concentrate on assessment of the glottis opening using optical, acoustical or electrical (electroglottography, EGG) methods, or on visualisation of the larynx position using ultrasound, computer tomography or magnetic resonance imaging techniques.The method presented here makes use of a time-multiplex measurement approach of space-resolved transfer impedances through the larynx. The fast sequence of measurements allows a quasi simultaneous assessment of both larynx position and EGG signal using up to 32 transmit–receive signal paths. The system assesses the dynamic opening status of the glottis as well as the vertical and back/forward motion of the larynx.Two electrode-arrays are used for the measurement of the electrical transfer impedance through the neck in different directions. From the acquired data the global and individual conductivity is calculated as well as a 2D point spatial representation of the minimum impedance.The position information is shown together with classical EGG signals allowing a synchronous visual assessment of glottal area and larynx position. A first application to singing voice analysis is presented that indicate a high potential of the method for use as a non-invasive tool in the diagnosis of voice, speech, and swallowing disorders.     

Voice Pathologies Classification and Detection Using EMD-DWT Analysis Based on Higher Order Statistic Features

BackgroundThe voice is a prominent tool allowing people to communicate and to change information in their daily activities. However, any slight alteration in the voice production system may affect the voice quality. Over the last years, researchers in biomedical engineering field worked to develop a robust automatic system that may help clinicians to perform a preventive diagnosis in order to detect the voice pathologies in an early stage.MethodIn this context, pathological voice detection and classification method based on EMD-DWT analysis and Higher Order Statistics (HOS) features, is proposed. Also DWT coefficients features are extracted and tested. To carry out our experiments a wide subset of voice signal from normal subjects and subjects which suffer from the five most frequent pathologies in the Saarbrücken Voice Database (SVD), is selected. In The first step, we applied the Empirical Mode Decomposition (EMD) to the voice signal. Afterwards, among the obtained candidates of Intrinsic Mode Functions (IMFs), we choose the robust one based on temporal energy criterion. In the second step, the selected IMF was decomposed via the Discrete Wavelet Transform (DWT). As a result, two features vector includes six HOSs parameters, and a features vector includes six DWT features were formed from both approximation and detail coefficients. In order to classify the obtained data a support vector machine (SVM) is employed. After having trained the proposed system using the SVD database, the system was evaluated using voice signals of volunteer's subjects from the Neurological department of RABTA Hospital of Tunis.ResultsThe proposed method gives promising results in pathological voices detection. The accuracies reached 99.26% using HOS features and 93.1% using DWT features for SVD database. In the classification, an accuracy of 100% was reached for “Funktionelle Dysphonia vs. Rekrrensparese” based on HOS features. Nevertheless, using DWT features the accuracy achieved was 90.32% for “Hyperfunktionelle Dysphonia vs. Rekurrensparse”. Furthermore, in the validation the accuracies reached were 94.82%, 91.37% for HOS and DWT features, respectively. In the classification the highest accuracies reached were for classifying “Parkinson versus Paralysis” 94.44% and 88.87% based on HOS and DWT features, respectively.ConclusionHOS features show promising results in the automatic voice pathology detection and classification compared to DWT features. Thus, it can reliably be used as noninvasive tool to assist clinical evaluation for pathological voices identification.     

Low-frequency component of photoplethysmogram reflects the autonomic control of blood pressure

The question of how much information the photoplethysmogram (PPG) signal contains on the autonomic regulation of blood pressure (BP) remains unsolved. This study aims to compare the low-frequency (LF) and high-frequency components of PPG and BP and assess their correlation with oscillations in interbeat (RR) intervals at similar frequencies. The PPG signal from the distal phalanx of the right index finger recorded using a reflective PPG sensor at green light, the BP signal from the left hand recorded using a Finometer, and RR intervals were analyzed. These signals were simultaneously recorded within 15 min in a supine resting condition in 17 healthy subjects (12 males and 5 females) aged 33 ± 9 years (mean ± SD). The study revealed the high coherence of LF components of PPG and BP with the LF component of RR intervals. The high-frequency components of these signals had low coherence. The analysis of the signal instantaneous phases revealed the presence of high-phase coherence between the LF components of PPG and BP. It is shown that the LF component of PPG is determined not only by local myogenic activity but also reflects the processes of autonomic control of BP.     

Perturbation measurements in highly irregular voice signals: Performances/validity of analysis software tools

In this paper we present results concerning validity of jitter measurement in strongly irregular voice signals (sustained vowels) moderately corrupted by noise. The performance of four tools for voice analysis is compared on synthetic signals as far as fundamental period and jitter estimation are concerned. Synthesised vowels offer the advantage of a perfect control of the amount of jitter put in.Though implementing the same formula for jitter estimation, the results obtained with these approaches become quite different for increasing jitter. The reason could be searched in the different methods used for the separation of voiced and unvoiced frames as well as for fundamental period estimation.Results show that all the tools give reliable results up to a jitter level J = 15%, that encompasses the maximum value J = 12% as obtained by expert raters by visual inspection. Hence, up to this limit, the tools presented here for jitter estimation can give a valid support to clinicians also in term of reproducibility of results and time saving.For jitter values larger than 15% all programs tend to underestimate the true jitter value, but with large differences among them. Just two methods succeed in estimating jitter values up to and larger than 20% and could thus be better suited for perturbation measure in strongly irregular voice signals.     

Dimensionality reduction for visualization of normal and pathological speech data

For an adequate analysis of pathological speech signals, a sizeable number of parameters is required, such as those related to jitter, shimmer and noise content. Often this kind of high-dimensional signal representation is difficult to understand, even for expert voice therapists and physicians. Data visualization of a high-dimensional dataset can provide a useful first step in its exploratory data analysis, facilitating an understanding about its underlying structure. In the present paper, eight dimensionality reduction techniques, both classical and recent, are compared on speech data containing normal and pathological speech. A qualitative analysis of their dimensionality reduction capabilities is presented. The transformed data are also quantitatively evaluated, using classifiers, and it is found that it may be advantageous to perform the classification process on the transformed data, rather than on the original. These qualitative and quantitative analyses allow us to conclude that a nonlinear, supervised method, called kernel local Fisher discriminant analysis is superior for dimensionality reduction in the actual context.     

Early detection of voice diseases via a web-based system

Voice is the result of the coordination of the whole pneumophonoarticulatory apparatus. The analysis of the voice allows the identification of the diseases of the vocal apparatus and currently is carried out from an expert doctor through methods based on the auditory analysis. The paper presents a web-based system for the acquisition and automatic analysis of vocal signals. Vocal signals are submitted by the users through a simple web-interface and are analyzed in real-time by using state-of-the art signal processing techniques, providing first-level information on possible voice alterations. The system offers different analysis functions to the doctors that may analyze suspected cases in detail. The system is currently being tested in the otorhinolaryngologist setting to carry out mass prevention via screening at a regional scale.