Artificial Intelligence-Assisted Auscultation of Heart Murmurs: Validation by Virtual Clinical Trial.

Artificial intelligence (AI) has potential to improve the accuracy of screening for valvular and congenital heart disease by auscultation. However, despite recent advances in signal processing and classification algorithms focused on heart sounds, clinical acceptance of this technology has been limited, in part due to lack of objective performance data. We hypothesized that a heart murmur detection algorithm could be quantitatively and objectively evaluated by virtual clinical trial. All cases from the Johns Hopkins Cardiac Auscultatory Recording Database (CARD) with either a pathologic murmur, an innocent murmur or no murmur were selected. The test algorithm, developed independently of CARD, analyzed each recording using an automated batch processing protocol. 3180 heart sound recordings from 603 outpatient visits were selected from CARD. Algorithm estimation of heart rate was similar to gold standard. Sensitivity and specificity for detection of pathologic cases were 93% (CI 90-95%) and 81% (CI 75-85%), respectively, with accuracy 88% (CI 85-91%). Performance varied according to algorithm certainty measure, age of patient, heart rate, murmur intensity, location of recording on the chest and pathologic diagnosis. This is the first reported comprehensive and objective evaluation of an AI-based murmur detection algorithm to our knowledge. The test algorithm performed well in this virtual clinical trial. This strategy can be used to efficiently compare performance of other algorithms against the same dataset and improve understanding of the potential clinical usefulness of AI-assisted auscultation.

Computerized Automatic Diagnosis of Innocent and Pathologic Murmurs in Pediatrics: A Pilot Study.

OBJECTIVE: Computer-aided auscultation in the differentiation of pathologic (AHA class I) from no or innocent murmurs (AHA class III) would be of great value to the general practitioner. This would allow objective screening for structural heart disease, standardized documentation of auscultation findings, and may avoid unnecessary referrals to pediatric cardiologists. Our goal was to assess the quality of a novel computerized algorithm that automatically classifies murmurs in phonocardiograms (PCGs) acquired in a pediatric population. DESIGN: This is a pilot study testing the ability of a novel computerized algorithm to accurately diagnose PCGs compared with interpreted echocardiograms as a gold standard. SETTING: This study was performed in pediatric cardiology clinics at a tertiary care hospital. PATIENTS: All incoming patients were recruited, including patients with no murmurs, innocent murmurs, and pathologic murmurs (106 patients). INTERVENTION: Using an electronic stethoscope, PCGs were acquired by the pediatric cardiologist from each patient. The PCGs were analyzed by the algorithm and diagnoses were compared with findings by echocardiograms interpreted by pediatric cardiologists which were used as the gold standard. OUTCOME MEASURES: Sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were calculated. RESULTS: When compared with echocardiography as a gold standard in diagnosing murmurs, the computerized algorithm tested on N=34 PCGs, yielded a sensitivity of 87% and specificity of 100%, a positive predictive value of 100%, negative predictive value of 90% and an accuracy of 94%. CONCLUSION: With echocardiogram as a gold standard, this computerized algorithm can detect pathologic murmurs with high sensitivity, specificity and accuracy, comparable to if not better than published results of pediatric cardiologists and neonatologists. This study confirms the high quality and "real-world" robustness of a novel computational algorithm in the assessment of pediatric murmurs.

Biomechanical properties and microstructure of human ventricular myocardium.

In the multidisciplinary field of heart research it is of utmost importance to identify accurate myocardium material properties for the description of phenomena such as mechano-electric feedback or heart wall thickening. A rationally-based material model is required to understand the highly nonlinear mechanics of complex structures such as the passive myocardium under different loading conditions. Unfortunately, to date there are no experimental data of human heart tissues available to estimate material parameters and to develop adequate material models. This study aimed to determine biaxial extension and triaxial shear properties and the underlying microstructure of the passive human ventricular myocardium. Using new state-of-the-art equipment, planar biaxial extension tests were performed to determine the biaxial extension properties of the passive ventricular human myocardium. Shear properties of the myocardium were examined by triaxial simple shear tests performed on small cubic specimens excised from an adjacent region of the biaxial extension specimens. The three-dimensional microstructure was investigated through second-harmonic generation (SHG) microscopy on optically cleared tissues, which emphasized the 3D orientation and dispersion of the myofibers and adjacent collagen fabrics. The results suggest that the passive human LV myocardium under quasi-static and dynamic multiaxial loadings is a nonlinear, anisotropic (orthotropic), viscoelastic and history-dependent soft biological material undergoing large deformations. Material properties of the tissue components along local microstructural axes drive the nonlinear and orthotropic features of the myocardium. SHG microscopy investigation revealed detailed information about the myocardial microstructure due to its high resolution. It enabled the identification of structural parameters such as the fiber and the sheet orientations and corresponding dispersions. With this complete set of material data, a sophisticated material model and associated material parameters can be defined for a better description of the biomechanical response of the ventricular myocardium in humans. Such a model will lead to more accurate computational simulations to better understand the fundamental underlying ventricular mechanics, a step needed in the improvement of medical treatment of heart diseases. STATEMENT OF SIGNIFICANCE: Unfortunately, to date there are no experimental data of human heart tissues available for material parameter estimation and the development of adequate material models. In this manuscript novel biaxial tensile and shear test data at different specimen orientations are presented, which allowed to adequately capture the direction-dependent material response. With these complete sets of mechanical data, combined with their underlying microstructural data (also presented herein), sophisticated material models and associated material parameters can be defined for the description of the mechanical behavior of the ventricular myocardium in humans. Such models will lead to accurate computational simulations to better understand the fundamental underlying ventricular mechanics, a step needed in the improvement of medical treatment of heart diseases.

Modelling non-symmetric collagen fibre dispersion in arterial walls.

New experimental results on collagen fibre dispersion in human arterial layers have shown that the dispersion in the tangential plane is more significant than that out of plane. A rotationally symmetric dispersion model is not able to capture this distinction. For this reason, we introduce a new non-symmetric dispersion model, based on the bivariate von Mises distribution, which is used to construct a new structure tensor. The latter is incorporated in a strain-energy function that accommodates both the mechanical and structural features of the material, extending our rotationally symmetric dispersion model (Gasser et al. 2006 J. R. Soc. Interface 3, 15-35. (doi:10.1098/rsif.2005.0073)). We provide specific ranges for the dispersion parameters and show how previous models can be deduced as special cases. We also provide explicit expressions for the stress and elasticity tensors in the Lagrangian description that are needed for a finite-element implementation. Material and structural parameters were obtained by fitting predictions of the model to experimental data obtained from human abdominal aortic adventitia. In a finite-element example, we analyse the influence of the fibre dispersion on the homogeneous biaxial mechanical response of aortic strips, and in a final example the non-homogeneous stress distribution is obtained for circumferential and axial strips under fixed extension. It has recently become apparent that this more general model is needed for describing the mechanical behaviour of a variety of fibrous tissues.

Selective enzymatic removal of elastin and collagen from human abdominal aortas: uniaxial mechanical response and constitutive modeling.

The ability to selectively remove the structurally most relevant components of arterial wall tissues such as collagen and elastin enables ex vivo biomechanical testing of the remaining tissues, with the aim of assessing their individual mechanical contributions. Resulting passive material parameters can be utilized in mathematical models of the cardiovascular system. Using eighteen wall specimens from non-atherosclerotic human abdominal aortas (55 ± 11 years; 9 female, 9 male), we tested enzymatic approaches for the selective digestion of collagen and elastin, focusing on their application to human abdominal aortic wall tissues from different patients with varying sample morphologies. The study resulted in an improved protocol for elastin removal, showing how the enzymatic process is affected by inadequate addition of trypsin inhibitor. We applied the resulting protocol to circumferential and axial specimens from the media and the adventitia, and performed cyclic uniaxial extension tests in the physiological and supra-physiological loading domain. The collagenase-treated samples showed a (linear) response without distinct softening behavior, while the elastase-treated samples exhibited a nonlinear, anisotropic response with pronounced remanent deformations (continuous softening), presumably caused by some sliding of collagen fibers within the damaged regions of the collagen network. In addition, our data showed that the stiffness in the initial linear stress-stretch regime at low loads is lower in elastin-free tissue compared to control samples (i.e. collagen uncrimping requires less force than the stretching of elastin), experimentally confirming that elastin is responsible for the initial stiffness in elastic arteries. Utilizing a continuum mechanical description to mathematically capture the experimental results we concluded that the inclusion of a damage model for the non-collagenous matrix material is, in general, not necessary. To model the softening behavior, continuous damage was included in the fibers by adding a damage variable which led to remanent strains through the consideration of damage.

An automated approach for three-dimensional quantification of fibrillar structures in optically cleared soft biological tissues.

We present a novel approach allowing for a simple, fast and automated morphological analysis of three-dimensional image stacks (z-stacks) featuring fibrillar structures from optically cleared soft biological tissues. Five non-atherosclerotic tissue samples from human abdominal aortas were used to outline the multi-purpose methodology, applicable to various tissue types. It yields a three-dimensional orientational distribution of relative amplitudes, representing the original collagen fibre morphology, identifies regions of isotropy where no preferred fibre orientations are observed and determines structural parameters throughout anisotropic regions for the analysis and numerical modelling of biomechanical quantities such as stress and strain. Our method combines optical tissue clearing with second-harmonic generation imaging, Fourier-based image analysis and maximum-likelihood estimation for distribution fitting. With a new sample preparation method for arteries, we present, for the first time to our knowledge, a continuous three-dimensional distribution of collagen fibres throughout the entire thickness of the aortic wall, revealing novel structural and organizational insights into the three arterial layers.

Quantitative assessment of collagen fibre orientations from two-dimensional images of soft biological tissues.

In this work, we outline an automated method for the extraction and quantification of material parameters characterizing collagen fibre orientations from two-dimensional images. Morphological collagen data among different length scales were obtained by combining the established methods of Fourier power spectrum analysis, wedge filtering and progressive regions of interest splitting. Our proposed method yields data from which we can determine parameters for computational modelling of soft biological tissues using fibre-reinforced constitutive models and gauge the length scales most appropriate for obtaining a physically meaningful measure of fibre orientations, which is representative of the true tissue morphology of the two-dimensional image. Specifically, we focus on three parameters quantifying different aspects of the collagen morphology: first, using maximum-likelihood estimation, we extract location parameters that accurately determine the angle of the principal directions of the fibre reinforcement (i.e. the preferred fibre directions); second, using a dispersion model, we obtain dispersion parameters quantifying the collagen fibre dispersion about these principal directions; third, we calculate the weighted error entropy as a measure of changes in the entire fibre distributions at different length scales, as opposed to their average behaviour. With fully automated imaging techniques (such as multiphoton microscopy) becoming increasingly popular (which often yield large numbers of images to analyse), our method provides an ideal tool for quickly extracting mechanically relevant tissue parameters which have implications for computational modelling (e.g. on the mesh density) and can also be used for the inhomogeneous modelling of tissues.

Determination of the layer-specific distributed collagen fibre orientations in human thoracic and abdominal aortas and common iliac arteries.

The established method of polarized microscopy in combination with a universal stage is used to determine the layer-specific distributed collagen fibre orientations in 11 human non-atherosclerotic thoracic and abdominal aortas and common iliac arteries (63 ± 15.3 years, mean ± s.d.). A dispersion model is used to quantify over 37 000 recorded fibre angles from tissue samples. The study resulted in distinct fibre families, fibre directions, dispersion and thickness data for each layer and all vessels investigated. Two fibre families were present for the intima, media and adventitia in the aortas, with often a third and sometimes a fourth family in the intima in the respective axial and circumferential directions. In all aortas, the two families were almost symmetrically arranged with respect to the cylinder axis, closer to the axial direction in the adventitia, closer to the circumferential direction in the media and in between in the intima. The same trend was found for the intima and adventitia of the common iliac arteries; however, there was only one preferred fibre alignment present in the media. In all locations and layers, the observed fibre orientations were always in the tangential plane of the walls, with no radial components and very small dispersion through the wall thickness. A wider range of in-plane fibre orientations was present in the intima than in the media and adventitia. The mean total wall thickness for the aortas and the common iliac artery was 1.39 and 1.05 mm, respectively. For the aortas, a slight thickening of the intima and a thinning of the media in increasingly distal regions were observed. A clear intimal thickening was present distal to the branching of the celiac arteries. All data, except for the media of the common iliac arteries, showed two prominent collagen fibre families for all layers so that two-fibre family models seem most appropriate.