A comparison of different feature extraction methods for diagnosis of valvular heart diseases using PCG signals

This article presents a novel method for diagnosis of valvular heart disease (VHD) based on phonocardiography (PCG) signals. Application of the pattern classification and feature selection and reduction methods in analysing normal and pathological heart sound was investigated. After signal preprocessing using independent component analysis (ICA), 32 features are extracted. Those include carefully selected linear and nonlinear time domain, wavelet and entropy features. By examining different feature selection and feature reduction methods such as principal component analysis (PCA), genetic algorithms (GA), genetic programming (GP) and generalized discriminant analysis (GDA), the four most informative features are extracted. Furthermore, support vector machines (SVM) and neural network classifiers are compared for diagnosis of pathological heart sounds. Three valvular heart diseases are considered: aortic stenosis (AS), mitral stenosis (MS) and mitral regurgitation (MR). An overall accuracy of 99.47% was achieved by proposed algorithm.     

Biopsy of the heart valves. 872 cases

In this study, 872 heart valves surgically excised from 810 patients during a period of 5 years (1994 through 1998) were examined pathologically. There was a predominance of aortic (506 patients) versus mitral valves (246 pts.). While aortic valves came more often from men (364) than from women (142), in mitral valves the M:F ratio is 82/164. Isolated calcific aortic stenosis appeared as the most frequent valvular disease (418 pts.), with predominance of its sclerotic-senile type (238 pts.). Mitral stenosis (185 pts.) remains the classical post-rheumatic disease. The relative frequency of a subvalvular stenosing mitral lesion is stressed. The "pure" incompetence of both aortic (70 pts.) and mitral (56 pts.) valve was usually based on valvular myxoid degeneration. An aorto-mitral disease requiring replacement of both valves (51 pts.) presented typically as a post-rheumatic lesion, however, a combination of a post-rheumatic mitral with a degenerative-sclerotic aortic valve disease may be possible. In 30 patients, the valvular replacement was performed for infective endocarditis or a post-IE lesion, mostly of the aortic valve. With the almost non-existence of acute rheumatic fever and with the increasing average age of population in this country, we may expect a long-term decline in mitral valve disease and an increase in aortic valve disease, particularly in the sclerotic type of aortic stenosis.     

Valvulopathy in primary antiphospholipid syndrome. Prospective echocardiography study

Echocardiographic studies have demonstrated a high prevalence of valvular disease in patients with primary antiphospholipid syndrome (PAPS). However, there are no studies assessing changes over time in valvular abnormalities. We conducted a study to determine whether there are changes over time in valvular lesions as detected by serial transesophageal echocardiography (TEE). Twelve patients with a first TEE had a second evaluation after a mean period of 13.5 months. There were 10 women and two men with a mean age of 38 years. Two patients had normal TEE on both initial and follow-up studies. Ten patients (83%) had valvular abnormalities, predominantly of the mitral and aortic valves in both studies. Abnormalities consisted of thickening, nodules, regurgitation, regurgitation and stenosis, and calcification. The type and frequency of lesions changed over time. As an example, one mitral valve nodule disappeared on follow up but three new aortic nodules developed even though all patients were receiving anticoagulant therapy. Two patients with mitral and aortic nodules presented cerebral ischemia. Mitral valvuloplasty was performed in one case. These results show that valvular abnormalities in patients with PAPS resolve, appear, or persist irrespective of anticoagulant therapy. Regurgitation is often mild or moderate, but stenosis may appear.     

Cronic non-rheumatic valvular heart disease. An autopsy study.

The frequency of chronic non-rheumatic valvular heart disease in Iceland was investigated via autoposies performed from November 1965 through December 1974. During this period, about 12.400 Icelanders died at the age of 16 years and older and 28.8 per cent of these were included in the study. At autopsy, males outnumbered females by 2:1. The frequency of calcific aortic stenosis was found to be 3.63 per cent and the prevalence was calculated to be 3.17 per cent among males and 4.50 per cent among females. Calcific aortic stenosis in tricuspid valves was more frequent in females and calcific aortic stenosis in bicuspid valves was more frequent in males. Among the hearts with calcific aortic stenosis, 70.8 per cent were found to have normally tricuspid valves, 25.4 per cent bicuspid valves and 3.8 per cent tricuspid valves with an unicommissural fusion. In 0.59 per cent of the hearts the aortic valve was either bicuspid or had an uncommissural fusion without the features of calcific stenosis. However, a functional stenosis was suggested by the increased weight of most of these hearts. The frequency of bicuspid aortic valves was 1.2 per cent with a prevalence in males of 1.54 per cent and in females 0.50 per cent. A calcified mitral annulus was found in 1.98 per cent of the hearts and in most, it was either associated with calcific aortic stenosis in a tricuspid valve, or it was a single valvular disease. Rheumatic valvular disease was found in 1.08 per cent of the heart examined.     

High prevalence of amyloid in 150 surgically removed heart valves—a comparison of histological and clinical data reveals a correlation to atheroinflammatory conditions

IntroductionThe prevalence, pathophysiology, and clinical indicators of valvular amyloid deposition have not been clarified yet.MethodsOne hundred fifty surgically resected heart valve specimens [67.4±1.0 years; aortic stenosis (AS), n=100; aortic regurgitation, n=19; mitral stenosis, n=7; mitral regurgitation, n=24] were qualitatively, semiquantitatively, and immunohistochemically analyzed and correlated with clinical data.ResultsAmyloid was found in 83/150 specimens with highest prevalence in AS (74/100), intermediate prevalence in mitral stenosis (2/7) and regurgitation (7/24), and lowest prevalence in aortic regurgitation (2/19). Severe and polymorphic amyloid deposits were almost exclusively found in AS (35/100). Filamentous cloudy amyloid patterns occurred with the same frequency in AS (29/100). A combination of both was found only in AS (n=7/100). By immunohistochemistry, none of the most common amyloid proteins was identified except for a weak staining by the apolipoprotein AI antibody, but more intense adjacent to amyloid deposits. Amyloid correlated with valvular thickening (P<.05), hyperlipidemia (P=.07), coronary artery disease (P=.084), and obesity (P=.082).ConclusionsLocalized valvular amyloid is predominantly found in stenotic aortic valves. It appears to depend on atheroinflammatory conditions and high shear-stress hemodynamics. Further studies are needed to identify the underlying protein.     

Left ventricular cellular hypertrophy in pressure- and volume-overload valvular heart disease

To determine the dependence of myocyte hypertrophy in chronic valvular heart disease on the site and type of lesion, the myocardium was studied from 11 patients with either pressure-overload hypertrophy (PO; four patients with aortic stenosis and two with mixed aortic stenosis/insufficiency) or pure volume-overload hypertrophy (VO; two patients with mitral regurgitation and three with aortic insufficiency). These patients, all without coronary artery disease, died zero to 34 days after valve replacement surgery. Diameters of 25 longitudinally oriented myocytes in the circular midwall myocardium were measured with a calibrated light microscope eyepiece reticle on each of five transmural, transverse, histologic sections from the apical, anterolateral, posterolateral, anteroseptal, and posteroseptal left ventricle. Statistical analysis by modified two-way analysis of variance (ANOVA) demonstrated that mean myocyte size (based on 125 measurements) varied widely among cases but was not statistically different among sites. The myocyte diameter for PO lesions (25.9 +/- 1.1 micron, mean +/- SEM) was significantly greater (P less than 0.05) than that for pure VO lesions (20.4 +/- 0.7 micron), despite equal relative heart weights (measured/predicted from body weight: 2.5 +/- 0.2 [mean +/- SD] versus 2.5 +/- 0.5). This study suggests that 1) cellular hypertrophy in valvular heart disease occurs uniformly throughout the left ventricular myocardium; and 2) mean myocyte diameters are greater in PO than in VO hypertrophy for equivalent cardiac enlargement.     

Time—frequency analysis of heart murmurs. Part I: Parametric modelling and numerical simulations

The object of this study is to compare the performance of two new bilinear time—frequency representation techniques with the spectrogram to characterise the behaviour of heart murmurs produced by bioprosthetic heart valves implanted in the mitral or aortic position. The murmurs are those of mitral stenosis, mitral regurgitation, aortic stenosis, aortic regurgitation, a diastolic musical murmur and a systolic musical murmur. In the first part of the study, the general characteristics of the amplitude and the spectral content of these murmurs are determined by visual observation of the spectrogram of phonocardiograms obtained from several patients with known valvular pathology complemented with a literature review. A parametric model is then generated for each murmur signal. Stenotic and regurgitant murmurs are modelled as the sequential output of a bank of low-pass filters excited by a white noise input signal. The basic parameters of each filter are selected to simulate, as a function of time, the basic characteristics of random heart murmurs. Musical murmurs are modelled as a frequency-modulated deterministic sinusoid of constant amplitude. Numerical simulations of these random and musical heart murmurs are then generated and will be used in Part II to determine the best of three time-frequency representation techniques for analysing heart murmur signals.     

Valvular regurgitation impact on left ventricular 2-dimensional and Doppler echocardiographic indices in patients with essential hypertension

BackgroundBlacks have both a higher hypertension prevalence and accelerated cardiac end organ damage. Because blacks also have a higher prevalence of valvular heart disease, which occurs at a younger age than for whites, we further examined the contribution of valvular regurgitation to the severity of hypertensive heart disease in Nigerians.MethodsWe evaluated and compared echocardiograph-ic indexes in 75 essential hypertensive Nigerians with (n = 48) and without (n = 27) valvular regurgitations. Demographic and echocardiographic indices, as well as the types and severity of valvular lesions were compared between the groups using bivariate logistic regression and analysis of variance.ResultsThe 2 groups were of similar demographics, but those with regurgitations had larger cardiac size (p < .05), greater mass (147 ± 31 vs 122 ± 32 g/m², p = .01) higher volume (p < .01), and left atrial size (35.6 ± 4.6 vs 33.3 ± 4.6 mm, p < .05). Atrial size, cardiac volume, and dimension were independent correlates/predictors of regurgitation occurrence. Relative wall thickness of at least 0.6 was more common in regurgitation patients. Cardiac mass was correlated to increasing age (r = 0.23, p = .043). The valvular lesions frequencies were aortic regurgitation, 8; mitral regurgita-tion, 22; and mixed, 18. The aortic orifice dimension was significantly different among the regurgitant cases, highest in aortic regurgitation (p = .001). Aortic orifice dimension increased with hypertension duration (p = .028).ConclusionsRegurgitant lesions are common and occur early in hypertensive Africans. Apparently mild valvular regurgitation may accentuate preclinical concentric hypertrophy in hypertensive blacks.     

Role of inflammation in nonrheumatic, regurgitant heart valve disease. A comparative, descriptive study regarding apolipoproteins and inflammatory cells in nonrheumatic heart valve disease.

BACKGROUND: Nonrheumatic aortic stenosis is the predominant cause of heart valve surgery in the Western world. Aortic and mitral regurgitation account for a lesser amount of the heart valve surgery. During the 1990s, inflammatory cell infiltrates have been demonstrated in nonrheumatic stenotic aortic valves. These findings suggest an inflammatory component in the pathogenesis of nonrheumatic aortic valve stenosis. However, nonrheumatic regurgitant aortic and mitral valves have not been investigated in this respect. The aim of this study was to compare nonrheumatic regurgitant aortic and mitral valves with stenotic aortic valves regarding the presence of T lymphocytes, macrophages, apolipoprotein B, and apolipoprotein A-I. METHODS: Valve specimens were obtained from 42 patients referred to hospital for surgery because of significant heart valve disease. From these patients, 29 aortic stenotic valves, 9 aortic regurgitant, and 6 mitral regurgitant valves, all nonrheumatic, were obtained for the study. Fourteen valves collected from subjects undergoing clinical/medicolegal autopsy were used as control. In order to identify mononuclear inflammatory cells and apolipoproteins, sections were investigated with immunohistochemical analyses and then categorized semiquantitatively. RESULTS: Regurgitant and control valves showed a significantly lower degree of inflammatory cell infiltrate and a lower degree of apolipoprotein deposition as compared to stenotic aortic valves. CONCLUSIONS: The signs of inflammation seen in nonrheumatic aortic stenosis are not prominent features in the nonrheumatic, regurgitant valves. This is consistent with the multi-factorial pathogenesis of these conditions.     

Evaluation of the left ventricular reserve by dynamic exercise echocardiography after surgery for valvular heart diseases

Dynamic ergometer exercise in a supine position was applied to 64 patients more than 1 year after valvular heart surgery, and the left ventricular reserve was evaluated echocardiographically. The left ventricular reserve declined in the mitral stenosis-mitral valve replacement group, while it was better maintained in the mitral stenosis-mitral commissurotomy, aortic regurgitation and aortic stenosis groups. The patients were divided into 3 groups depending on whether the percentage increase during exercise of stroke index, an index of left ventricular pump function, increased, unchanged, or decreased. The percentage increase of mean velocity of circumferential fibre shortening (y) and that of left ventricular end-diastolic diameter (x) during exercise were plotted for each group. The increased group was isolated from the unchanged group by the line of y = -5.02x + 30.1; the unchanged group was isolated from the decreased group by that of y = -5.68x-10.0, and the increased and unchanged groups were clearly isolated from the decreased group by that of y = -6.86x-4.76. We conclude that dynamic ergometer exercise echocardiography is useful for evaluating the left ventricular reserve of postoperative patients with valvular heart disease. It was also thought that the subclinical state of cardiac failure can be effectively detected by the present method.     

Left ventricular function in aortic valve endocarditis. Echocardiographic evaluation and comparison with findings in chronic aortic regurgitation

The value of M-mode echocardiography in assessment of left ventricular (LV) function in patients with aortic regurgitation due to aortic valve endocarditis (AVE) was studied in 12 consecutive patients and compared with the findings in 30 patients with chronic aortic regurgitation (CAR). Patients with AVE had markedly increased LV end-diastolic and end-systolic diameters, whereas fractional shortening was normal. A linear correlation was found between the LV ejection fractions calculated by echocardiography and angiography, but echocardiography markedly overestimated the ejection fractions. There was a close linear correlation between the prematurity of mitral valve closure (MVC) and LV end-diastolic pressure. Patients with CAR had lower end-diastolic pressure, similarly increased LV internal diameters and none had premature MVC. Thus, M-mode echocardiography can identify patients with premature MVC and high LV filling pressure. However, echocardiographic LV function indices based on measurement of internal dimensions overestimate the LV function and these data should be interpreted with caution.     

Computer-controlledin vitro model of the human left heart

The decision for surgical intervention in the treatment of stenosis and for regurgitation of the mitral valve demands an objective and quantitative evaluation of the severity of mitral valve disease. The availability of ultrasound techniques capable of analysing flow velocities across valves and to produce representative images of valve orifices has increased the interest in the hydraulics of cardiac valves. To isolate and study the determinants of transmitral flow, an in vitro model of the human left heart was built. From the model it is possible to differentiate the influence of the different determinants of left heart performance on transmitral flow: preload, compliance of the left atrium and ventricle, peripheral resistance (afterload) and heart rate. The mechanical part of the model consists of a reservoir connected to an elastic closed circuit (Latex pulmonary veins, left atrium, left ventricle and aortic arch) with replaceable mitral and aortic valves. The electronic part of the model drives and controls the hydraulic part, allowing the independent regulation and monitoring of left atrial and left ventricular pressures p, volumes V and 'pV-loops' throughout the cardiac cycle at different cardiac rhythms. Left atrial filling pressure and aortic resistance are variable in a controlled fashion. Echo-Doppler study of the mitral valve and the transmitral valve flow is possible both from an atrial and a ventricular window in the model. This technical note describes the model.     

Prevalence of coronary artery disease in patients with valvular heart disease

To evaluate the usefulness of preoperatie coronary angiography in patients undergoing preoperative investigation because of valvular heart disease, we performed coronary angiography in a consecutive series of 329 patients. The prevalence of significant coronary artery disease was 32%. Asymptomatic coronary artery disease was present in 13%. Angina pectoris proved to be a poor predictor of coronary artery disease in aortic valve disease. In mitral valve disease, however, the specificity was high. A cost-benefit calculation was carried out in order to assess what advantage routine coronary angiography might have. According to this, coronary angiography should be performed in all patients suffering from valvular heart disease with angina pectoris, whereas it can be omitted in younger patients without angina. A cut-off point of 60 years seems appropriate for aortic valve disease and 65 years for mitral valve disease.     

Double valve disease in systemic lupus erythematosus. Case report and literature review

A young, nonhypertensive female with advanced systemic lupus erythematosus (SLE) presented in congestive cardiac failure due to aortic and mitral regurgitation. The valvular lesions resulted from organization of valvular pocket Libman-Sacks vegetations. Her clinical course mimicked infective endocarditis. She is only the third recorded patient with SLE valvular disease warranting double valve replacement. This patient, who had her valvular disease at presentation (prior to initiation of steroid therapy), illustrates that untreated SLE per se may produce severe organic valvular disease.     

Analysis of the pathological severity degree of aortic stenosis (AS) and mitral stenosis (MS) using the discrete wavelet transform (DWT)

The heart is the principal organ that circulates blood. In normal conditions it produces four sounds for each cardiac cycle. However, most often only two sounds appear essential: S1 and S2. Two other sounds: S3 and S4, with lower amplitude than S1 or S2, appear occasionally in the cardiac cycle by the effect of disease or age. The presence of abnormal sounds in one cardiac cycle provide valuable information on various diseases. The aortic stenosis (AS), as being a valvular pathology, is characterized by a systolic murmur due to a narrowing of the aortic valve. The mitral stenosis (MS) is characterized by a diastolic murmur due to a reduction in the mitral valve. Early screening of these diseases is necessary; it’s done by a simple technique known as: phonocardiography. Analysis of phonocardiograms signals using signal processing techniques can provide for clinicians useful information considered as a platform for significant decisions in their medical diagnosis. In this work two types of diseases were studied: aortic stenosis (AS) and mitral stenosis (MS). Each one presents six different cases. The application of the discrete wavelet transform (DWT) to analyse pathological severity of the (AS and MS was presented. Then, the calculation of various parameters was performed for each patient. This study examines the possibility of using the DWT in the analysis of pathological severity of AS and MS.     

Cardiac sound murmurs classification with autoregressive spectral analysis and multi-support vector machine technique

In this paper, a novel cardiac sound spectral analysis method using the normalized autoregressive power spectral density (NAR-PSD) curve with the support vector machine (SVM) technique is proposed for classifying the cardiac sound murmurs. The 489 cardiac sound signals with 196 normal and 293 abnormal sound cases acquired from six healthy volunteers and 34 patients were tested. Normal sound signals were recorded by our self-produced wireless electric stethoscope system where the subjects are selected who have no the history of other heart complications. Abnormal sound signals were grouped into six heart valvular disorders such as the atrial fibrillation, aortic insufficiency, aortic stenosis, mitral regurgitation, mitral stenosis and split sounds. These abnormal subjects were also not included other coexistent heart valvular disorder. Considering the morphological characteristics of the power spectral density of the heart sounds in frequency domain, we propose two important diagnostic features Fmax and Fwidth, which describe the maximum peak of NAR-PSD curve and the frequency width between the crossed points of NAR-PSD curve on a selected threshold value (THV), respectively. Furthermore, a two-dimensional representation on (Fmax, Fwidth) is introduced. The proposed cardiac sound spectral envelope curve method is validated by some case studies. Then, the SVM technique is employed as a classification tool to identify the cardiac sounds by the extracted diagnostic features. To detect abnormality of heart sound and to discriminate the heart murmurs, the multi-SVM classifiers composed of six SVM modules are considered and designed. A data set was used to validate the classification performances of each multi-SVM module. As a result, the accuracies of six SVM modules used for detection of abnormality and classification of six heart disorders showed 71-98.9% for THVs=10-90% and 81.2-99.6% for THVs=10-50% with respect to each of SVM modules. With the proposed cardiac sound spectral analysis method, the high classification performances were achieved by 99.9% specificity and 99.5% sensitivity in classifying normal and abnormal sounds (heart disorders). Consequently, the proposed method showed relatively very high classification efficiency if the SVM module is designed with considering THV values. And the proposed cardiac sound murmurs classification method with autoregressive spectral analysis and multi-SVM classifiers is validated for the classification of heart valvular disorders.     

Cardiovascular magnetic resonance in the evaluation of heart valve disease

Background

Over the last 25 years, cardiovascular magnetic resonance imaging (CMR) has emerged as an alternative to echocardiography for assessment of valvular heart disease (VHD). Although echo remains the first-line imaging modality for the assessment of patients with VHD, CMR can now provide a comprehensive assessment in many instances. Using a combination of techniques, CMR provides information on valve anatomy and enables quantitative analysis of the severity of the valve lesion.

Main text

In this review, the fundamentals of CMR in assessment of VHD are described, together with its strengths and weaknesses. We detail the utility of CMR for studying all aspects of VHD, including valve anatomy, flow quantification as well as ventricular volumes and function. The optimisation of CMR for evaluating the commonest valve lesions (aortic stenosis, aortic regurgitation, mitral regurgitation, mitral stenosis) as well as in right-sided VHD and prosthetic valves is summarised. The focus of this review is to enable the reader to optimise the use of CMR in his or her own evaluation of heart valve lesions in clinical practice.

Conclusions

CMR can be used for the comprehensive evaluation of VHD. This exciting, non-invasive imaging modality is likely to have increasing utility in the clinical evaluation of patients with VHD.

     

Etiology of acquired valvular heart disease in adults. A survey of 18,132 autopsies and 100 consecutive valve-replacement operations

The cardiac valve pathology in 18,132 autopsies was analyzed. A total of 1,136 patients (6.3%) had acquired valvular disease. The most commonly diseased cardiac valve was the mitral valve (49%), followed by the aortic valve (42%) and the tricuspid (9%) and pulmonary valves (0.3%). Rheumatic fever accounted for 99.7% of cases of mitral stenosis and 68.4% of mitral incompetence. The autopsy incidence of mitral stenosis remained constant over 30 years (1950 to 1979). Only 44.4% of the cases of acquired aortic stenosis were due to rheumatic fever. Review of 100 consecutive, surgically excised native valves revealed that if the pathologist is given adequate information regarding the macroscopic appearance of the intact valve prior to excision, an accurate etiopathologic diagnosis can be made in 81% of cases compared with only 35% of cases without such information.     

Computational analysis of the effect of mitral and aortic regurgitation on the function of ventricular assist devices using 3D cardiac electromechanical model

Valvular insufficiency affects cardiac responses and the pumping efficacy of left ventricular assist devices (LVADs) when patients undergo LVAD therapy. Knowledge of the effect of valvular regurgitation on the function of LVADs is important when treating heart failure patients. The goal of this study was to examine the effect of valvular regurgitation on the ventricular mechanics of a heart under LVAD treatment and the pumping efficacy of an LVAD using a computational model of the cardiovascular system. For this purpose, a 3D electromechanical model of failing ventricles in a human heart was coupled with a lumped-parameter model of valvular regurgitation and an LVAD-implanted vascular system. We used the computational model to predict cardiac responses with respect to the severity of valvular regurgitation in the presence of LVAD treatment. An LVAD could reduce left ventricle (LV) pressure (up to 34%) and end-diastolic ventricular volume (up to 80%) and maintain cardiac output at the estimated flow rate from the LVAD under the condition of mitral regurgitation (MR); however, the opposite would occur under the condition of aortic regurgitation (AR). Considering these physiological responses, we conclude that AR, and not MR, diminishes the pumping function of LVADs.     

Computational prediction of the effects of the intra-aortic balloon pump on heart failure with valvular regurgitation using a 3D cardiac electromechanical model

Intra-aortic balloon pump (IABP) is normally contraindicated in significant aortic regurgitation (AR). It causes and aggravates pre-existing AR while performing well in the event of mitral regurgitation (MR). Indirect parameters, such as the mean systolic pressure, product of heart rate and peak systolic pressure, and pressure–volume are used to quantify the effect of IABP on ventricular workload. However, to date, no studies have directly quantified the reduction in workload with IABP. The goal of this study is to examine the effect of IABP therapy on ventricular mechanics under valvular insufficiency by using a computational model of the heart. For this purpose, the 3D electromechanical model of the failing ventricles used in previous studies was coupled with a lumped parameter model of valvular regurgitation and the IABP-treated vascular system. The IABP therapy was disturbed in terms of reducing the myocardial tension generation and contractile ATP consumption by valvular regurgitation, particularly in the AR condition. The IABP worsened the problem of ventricular expansion induced as a result of the regurgitated blood volume during the diastole under the AR condition. The IABP reduced the LV stroke work in the AR, MR, and no regurgitation conditions. Therefore, the IABP helped the ventricle to pump blood and reduced the ventricular workload. In conclusion, the IABP partially performed its role in the MR condition. However, it was disturbed by the AR and worsened the cardiovascular responses that followed the AR. Therefore, this study computationally proved the reason for the clinical contraindication of IABP in AR patients.