A simplicity-based fuzzy clustering approach for detection and extraction of murmurs from the phonocardiogram

Almost all cardiac abnormalities manifest themselves as murmurs in a phonocardiogram (PCG). The location of a murmur in the PCG over a cardiac cycle depends on the underlying cardiac abnormality. Locating murmurs with respect to the cardiac cycle is useful for diagnosing cardiac dysfunction. Locating murmurs is difficult due to spectral similarity and time overlap with other heart sounds. Moreover, the wide variation in murmur amplitudes and murmur spectral characteristics across different patients and abnormalities has hindered the design of a generic time segmentation algorithm to isolate murmurs within the PCG. In this paper, we present a method to locate cardiac murmurs within the PCG that is based on their visual simplicity, which does not depend upon their absolute amplitude and frequency characteristics, and hence, results in their better localization. Then, we identify fuzzy sets to cluster simplicity values, due to murmurs, to determine the time duration over which the murmur occurs. The overall accuracy of the proposed algorithm in detecting systolic murmurs is 80%. The sensitivity of the algorithm in locating systolic murmurs is 73% and its specificity is 100%. The isolated murmur can then be further processed to extract clinically relevant features to diagnose the nature of the cardiac abnormality. Experimental results with a variety of murmurs are provided.     

幼儿心脏储备两个新的无创指标的初步调查研究

目的获得幼儿心力、舒张期和收缩期时限比值两项心脏储备指标的基线统计值。方法 129名2～6岁儿童纳入该项调查研究。静息状态下记录儿童的心音图,测量第一心音幅值(S1)和第二心音幅值(S2)、心动周期、舒张期(D)和收缩期(S)时限;计算S1/S2、D/S比值和心率(HR),并分别统计2～3岁和4～6岁儿童上述3项指标的均值和标准差。结果该组儿童的S1/S2、D/S比值分别为:2～3岁1.51±0.94、1.27±0.16,4～6岁1.40±1.06、1.51±0.24;HR分别为:2～3岁102±10)次/分,4～6岁(91±12)次/分。结论上述调查结果统计值对于幼儿教师、儿童体育教练和儿科临床医生评估儿童心脏状态可能有参考价值。     

基于小波变换的心音包络提取算法及应用

背景:心音信号包含了大量心脏瓣膜活动的生理信息,心音分析对诊断心脏疾病具有重要的临床意义。目的:旨在通过心音的包络提取,分析心音信号的各种特征,进而判断心音中是否包含杂音,以改善传统听诊技术高度依赖医生经验、听诊范围受限的缺点。方法:提出了一种采用小波变换来提取心音包络的方法,通过与采用希尔伯特-黄变换、数学形态学、平均香农能量等心音包络求解方法进行对比,证明这种方法具有算法简便、曲线光滑、特征点突出等优点。结果与结论:将该方法用于临床真实心音的包络提取,利用支持向量机来训练所提取心音包络的面积和小波能量两个特征参数,判别心音信号是否明显包含杂音。选用35例心音数据对算法进行验证,结果表明该算法的准确率达到95%,具有很强的实用性。     

基于小波变换的心音包络提取算法及应用

背景：心音信号包含了大量心脏瓣膜活动的生理信息,心音分析对诊断心脏疾病具有重要的临床意义。目的：旨在通过心音的包络提取,分析心音信号的各种特征,进而判断心音中是否包含杂音,以改善传统听诊技术高度依赖医生经验、听诊范围受限的缺点。方法：提出了一种采用小波变换来提取心音包络的方法,通过与采用希尔伯特-黄变换、数学形态学、平均香农能量等心音包络求解方法进行对比,证明这种方法具有算法简便、曲线光滑、特征点突出等优点。结果与结论：将该方法用于临床真实心音的包络提取,利用支持向量机来训练所提取心音包络的面积和小波能量两个特征参数,判别心音信号是否明显包含杂音。选用35例心音数据对算法进行验证,结果表明该算法的准确率达到95%,具有很强的实用性。     

Noise detection during heart sound recording using periodicity signatures

Heart sound is a valuable biosignal for diagnosis of a large set of cardiac diseases. Ambient and physiological noise interference is one of the most usual and highly probable incidents during heart sound acquisition. It tends to change the morphological characteristics of heart sound that may carry important information for heart disease diagnosis. In this paper, we propose a new method applicable in real time to detect ambient and internal body noises manifested in heart sound during acquisition. The algorithm is developed on the basis of the periodic nature of heart sounds and physiologically inspired criteria. A small segment of uncontaminated heart sound exhibiting periodicity in time as well as in the time-frequency domain is first detected and applied as a reference signal in discriminating noise from the sound. The proposed technique has been tested with a database of heart sounds collected from 71 subjects with several types of heart disease inducing several noises during recording. The achieved average sensitivity and specificity are 95.88% and 97.56%, respectively.     

Assessment of systolic and diastolic function in heart failure using ambulatory monitoring with acoustic cardiography

INTRODUCTION. The circadian variation of heart function and heart sounds in patients with and without heart failure (HF) is poorly understood. We hypothesized HF patients would exhibit less circadian variation with worsened cardiac function and sleep apnea. METHODS. We studied 67 HF patients (age 67.4 ± 8.2 years; 42% acute HF) and 63 asymptomatic control subjects with no history of HF (age 61.6 ± 7.7 years). Subjects wore a heart sound/ECG/respiratory monitor. The data were analyzed for sleep apnea, diastolic heart sounds, and systolic time intervals. RESULTS. The HF group had significantly greater prevalence of the third heart sound and prolongation of electro-mechanical activation time, while the control group had an age-related increase in the prevalence of the fourth heart sound. The control group showed more circadian variation in cardiac function. The HF subjects had more sleep apnea and higher occurrence of heart rate non-dipping. CONCLUSIONS. The control subjects demonstrated an increasing incidence of diastolic dysfunction with age, while systolic function was mostly unchanged with aging. Parameters related to systolic function were significantly worse in the HF group with little diurnal variation, indicating a constant stimulation of sympathetic tone in HF and reduction of diurnal regulation.     

Assessment of systolic and diastolic function in heart failure using ambulatory monitoring with acoustic cardiography

Abstract

Introduction. The circadian variation of heart function and heart sounds in patients with and without heart failure (HF) is poorly understood. We hypothesized HF patients would exhibit less circadian variation with worsened cardiac function and sleep apnea. Methods. We studied 67 HF patients (age 67.4 ± 8.2 years; 42% acute HF) and 63 asymptomatic control subjects with no history of HF (age 61.6 ± 7.7 years). Subjects wore a heart sound/ECG/respiratory monitor. The data were analyzed for sleep apnea, diastolic heart sounds, and systolic time intervals. Results. The HF group had significantly greater prevalence of the third heart sound and prolongation of electro-mechanical activation time, while the control group had an age-related increase in the prevalence of the fourth heart sound. The control group showed more circadian variation in cardiac function. The HF subjects had more sleep apnea and higher occurrence of heart rate non-dipping. Conclusions. The control subjects demonstrated an increasing incidence of diastolic dysfunction with age, while systolic function was mostly unchanged with aging. Parameters related to systolic function were significantly worse in the HF group with little diurnal variation, indicating a constant stimulation of sympathetic tone in HF and reduction of diurnal regulation.     

Characterization of the Korotkoff sounds using joint time-frequency analysis

The sounds associated with the five classical Korotkoff phases are clinically important for measuring systolic and diastolic blood pressures. The frequency ranges of the sounds have already been described simply using the overall peak frequencies within each phase by Fourier methods. However, such analysis may be missing potentially useful clinical information. The aim of this study was to compare features associated with the different phases of the Korotkoff sounds obtained during blood pressure measurement using a joint time-frequency analysis (JTFA) technique. A single operator recorded Korotkoff sounds from 25 healthy subjects using a measurement system comprising cardiology stethoscope, microphone, amplifier and recording system for computer sound digitization, and a MiniDisc system for playback to the cardiologist for Korotkoff phase classification. We have shown that using this system the phase classification by the cardiologist is repeatable, with no significant differences found in the number of sounds allocated to phases on two separate recording assessments. The digitized sounds were processed using a MATLAB-based short-time Fourier transform JTFA technique and differences in time, frequency and amplitude characteristics between the phases compared. It was found that on average, phase III had the largest overall amplitude and high frequency energy. Phase II had the greatest high frequency component and longest murmur, and was visibly the most complex phase in terms of time and frequency content. In contrast, phases IV and V had the lowest amplitude and frequency components. Overall, the statistically significant transitions between phases were: phase I to II with increases in high frequency (224 to 275 Hz) (p < 0.01) and sound duration (49 to 98 ms) (p < 0.0001), II to III with a significant decrease in sound duration (to 37 ms) (p < 0.0001), III to IV with decreases in maximum amplitude (0.95 to 0.25), highest frequency (262 to 95 Hz), and relative high frequency energy of the sounds (0.61 to 0.10) (all p < 0.0001), and IV to V with decreases in the maximum amplitude (0.25 to 0.13) (p < 0.0002) and high frequency energy (0.10 to 0.03) (p < 0.005). This study has demonstrated that joint time-frequency analysis of Korotkoff sounds was able to identify characteristic differences associated with the different phases classified by the expert cardiologist. Ultimately, exploiting the joint time and frequency characteristics of the sounds may improve blood pressure measurement and help to assess the stiffness of the peripheral arteries.     

Use of a heart sound simulator in teaching cardiac auscultation

The use of a heart sound simulator in teaching recognition of heart sounds was tested in 37 graduate students. Subjects were randomly divided into two groups; one group received both auditory and tactile cues and the other group received only auditory cues. Students were tested at three times: (1) before use of the simulator, (2) after a short teaching session using the simulator, and (3) 2 weeks later. Recognition of heart sounds increased significantly after the teaching session with the simulator, and this knowledge was retained. There was no difference in sound recognition between the two groups.     

Remote laser-speckle sensing of heart sounds for health assessment and biometric identification

Assessment of heart sounds is the cornerstone of cardiac examination, but it requires a stethoscope, skills and experience, and a direct contact with the patient. We developed a contactless, machine-learning assisted method for heart-sound identification and quantification based on the remote measurement of the reflected laser speckle from the neck skin surface in healthy individuals. We compare the performance of this method to standard digital stethoscope recordings on an example task of heart-beat sound biometric identification. We show that our method outperforms the stethoscope even allowing identification on the test data taken on different days. This method might allow development of devices for remote monitoring of cardiovascular health in different settings.     

The effect of environmental sound and communication on CCU patients' heart rate and blood pressure

The effects of high ambient stressors (equipment sounds) and social stressors (conversation) on heart rate (HR) and blood pressure (BP) were examined in coronary care patients. Simultaneously occurring sound level in decibels, actual sound, electrocardiogram, and BP were recorded three times/day over 2 days for 20 subjects. Repeated measures ANOVA and pairwise comparisons revealed maximum HR to be significantly higher during conversation than during low ambient sounds (quiet). High ambient stressors did not affect HR for these subjects. BP did not significantly change during any of the sound conditions. Further research is needed to delineate cardiovascular effects of specific social stressors.     

A dynamic method to estimate the time split between the A2 and P2 components of the S2 heart sound

The time interval between the aortic (A2) and the pulmonary (P2) components of the second heart sound (S2) is an indicator of pulmonary arterial pressure. However, knowledge of the A2 and P2 components of the S2 sound is difficult to obtain due to their temporal overlap and significant spectral similarity. In this work, we aim to extract the A2 and P2 components from the phonocardiogram to estimate the time interval between them. We attain our objective by first isolating the S2 sound from the phonocardiogram by utilizing the mode complexity of the heart. Then, we assume the statistical independence of the A2 and P2 components and extract them from the S2 sound by the application of blind source separation techniques. Once separated, the time interval between the A2 and P2 components is estimated with a time-centroid-based method. Experimental results using simulated data show excellent performance of the proposed algorithm to extract the A2 and the P2 components from the S2 sound and to estimate the time interval between them. Results obtained from real data are also encouraging and show promise for utilizing the proposed method in a clinical setting to non-invasively tract pulmonary hypertension.     

A low-cost,high-fidelity fm wireless precordial radiostethoscope for continuous monitoring of heart and breath sounds

We constructed and tested an inexpensive (less than S50) FM wireless, acoustically shielded, precordial radiostethoscope that enables the anesthetist to follow the heart tones and breath sounds of the patient regardless of the anesthetist’s location in the operating room. We compared our acoustically shielded device with a similar, but acoustically unshielded, commercially available device. We found the sound quality of our radiostethoscope to be superior to that of the commercial device; the signal-to-noise ratio of our device was 7.6 for heart tones and 8.4 for breath sounds, whereas the commercial device had a signal-to-noise ratio of 2.7 and 3.9 for heart tones and breath sounds, respectively. Our device offers all of the advantages of a radiostethoscope and has the added advantages of low cost and high fidelity.     

利用颈部听诊法对健康青年人自主控制吞咽时吞咽音变化的分析

目的利用颈部听诊法分析健康青年人在自主控制吞咽时吞咽音的参数变化。方法37 名健康青年志愿者分别进行3ml 和10 ml 水的吞咽,吞咽方式设定为用力吞咽、正常吞咽以及轻吞咽,利用颈部听诊技术采取志愿者的吞咽声,并将采取的声音进行电脑声学参数分析,主要选择吞咽音间期(SSD)、平均吞咽音振幅(ASA)和平均声谱频率(MSF)3 个参数数据,对不同吞咽方式下以及不同水容量下的吞咽音进行比较。结果用力吞咽时MSF 高于其他两种吞咽方式(P<0.05);10 ml 水吞咽时SSD、ASA和MSF 均明显高于3 ml 水吞咽(P<0.01)。结论健康人对吞咽动作进行主动控制时,吞咽声音的频率体现健康青年人对吞咽方式主动控制的变化,下咽液体容量变化时吞咽声音的声音间期、平均频率以及平均振幅均发生变化。     

Automatic timing of aortic valve closure in apical tissue Doppler images

Ultrasound color tissue Doppler imaging (TDI) can be used to estimate velocities of moving left ventricular cardiac tissue. Aortic valve closure (AVC) can be observed as a notch in apical TDI velocity/time curves occurring after ejection, but before early relaxation. This work sought to evaluate automatic and automated algorithms using TDI for timing AVC. Mitral valve position and the time point of early relaxation were extracted and used to accomplish the task. To test the algorithms, phonocardiogram of the second heart sound was recorded simultaneously with TDI and used as a reference method. The algorithms were tested on apical views of 16 healthy subjects. In 98% of the cardiac cycles, the automatic algorithm estimated the time point of AVC within 25 ms of the reference. Automatic detection of AVC might save manual effort and provide a marker separating ejection and diastole for further automated analysis.     

Cardiac aspect of Cooley's anemia.

The short- and long-term effect of multiple versus regularly spaced blood transfusions on cardiac size, electrocardiographic alterations, heart sounds, murmurs, and carotid pulse were studied in 18 children with Cooley's anemia. These parameters were evaluated before transfusions, one week after multiple blood transfusions (when hemoglobin levels had reached normal), and during the following year, when conventionally spaced transfusions were given. Results showed that enlarged hearts may regress to normal size after multiple transfusions. Chamber hypertrophies and second heart sound were inconstant, and the average of all mean QRS and T vectors was not altered significantly after transfusions. There was good correlation between hemoglobin levels and tachycardia, intensity of first heart sound, heart murmurs, and ejection systolic clicks. Upstroke time of carotid pulse increased with restoration of normal hemoglobin levels. It is concluded that patients maintained at a higher hemoglobin level are generally in better cardiovascular status.     

Enhancing BOLD response in the auditory system by neurophysiologically tuned fMRI sequence

Auditory neuroscience has not tapped fMRI's full potential because of acoustic scanner noise emitted by the gradient switches of conventional echoplanar fMRI sequences. The scanner noise is pulsed, and auditory cortex is particularly sensitive to pulsed sounds. Current fMRI approaches to avoid stimulus-noise interactions are temporally inefficient. Since the sustained BOLD response to pulsed sounds decreases with repetition rate and becomes minimal with unpulsed sounds, we developed an fMRI sequence emitting continuous rather than pulsed gradient sound by implementing a novel quasi-continuous gradient switch pattern. Compared to conventional fMRI, continuous-sound fMRI reduced auditory cortex BOLD baseline and increased BOLD amplitude with graded sound stimuli, short sound events, and sounds as complex as orchestra music with preserved temporal resolution. Response in subcortical auditory nuclei was enhanced, but not the response to light in visual cortex. Finally, tonotopic mapping using continuous-sound fMRI demonstrates that enhanced functional signal-to-noise in BOLD response translates into improved spatial separability of specific sound representations.     

Optimal Programming of the Atrioventricular Delay Using the Phonocardiogram

Purpose: To predict the optimal atrioventricular (AV) delay using the phonocardiogram (PCG).
Methods: We studied 12 recipients of cardiac resynchronization therapy (CRT) system and eight recipients of dual-chamber pacemakers implanted for AV block with normal left ventricular (LV) function. The amplitude of the first heart sound (S1) was recorded by PCG and the LV outflow tract (OT) time-velocity integral (TVI) was measured by pulsed Doppler echocardiography. The AV delay was prolonged in 20-ms increments, from 60 ms to 240 ms. Ishikawa's method was used for the echocardiographic optimization of the AV delay. The relation between S1 amplitude and the AV delay was analyzed.
Results: The correlation between the amplitude of S1 and the length of AV delay showed an S-shaped curve. The AV delay at the inflection point of each patient's S-shaped curve (161.2 ± 19.5 ms) was positively correlated with the optimal AV delay determined by echocardiography (148.3 ± 16.9 ms, r = 0.83, P < 0.001). In addition, there was a positive correlation between the AV delay at the maximal TVI of LVOT (150.8 ± 22.7 ms) and the AV delay at the inflection point of the S-shaped curve (159.5 ± 24.9 ms, r = 0.87, P < 0.001). In two CRT system recipients, an optimal AV delay could not be found by echocardiography; however, an optimal AV delay could be determined by PCG.
Conclusions: A high correlation was observed between the optimal AV delay determined by phonocardiography versus echocardiography.     

General discrete modelling of lung sound production in normal subjects

Lung sound analysis is of a major importance in diagnostic malfunctions of the respiratory system. In normal subjects, it is known that these sounds are caused by the interaction of the respiratory flows with the bronchial tree structure. However, the detailed knowledge of the reasons for the spectral characteristics of such sounds remains to be elucidated. In this paper we propose a model for normal lung sound production based on a discretization of air flow in particle-like elements. Their transport with the involved interactions is implemented using a pseudo-molecular dynamics Monte Carlo procedure. General physical principles were considered for the interaction of these elements with the bronchial tree as well as a two-body interaction potential. The particle-tree interactions and the particle-particle interactions represent the flow-tree and the internal flow interactions, respectively. According to the model, sound is produced in each bronchus with the pitch frequency inversely proportional to its dimensions and with amplitude proportional to the intensity of the interaction, also a function of the bronchus dimensions. The lung sound is then the composition of the sounds produced in each bronchus. The model was successful in approximating the spectral characteristics reported by Gavriely et al (1981, 1995) as a direct consequence of the fractal properties of the bronchial tree and the considered internal fluid interactions. Thus, the reported high-frequency spectrum with its affine property as well as the low-frequency irregularity could be reproduced.