Hemodynamics-based Heart Sound Generation Model And Simulations

Heart sounds are mechanical vibrations produced by interplay between abrupt hemodynamic changes and cardiovascular organs including heart valves, vessels and heart wall. In the sense of heart sound generation, heart sound has the potential to reflect physiological and pathological state of the cardiovascular system. Abnormal heart sounds can be observed when lesions occur in the cardiovascular system but not yet enough to cause significant pathological characteristics. Hemodynamics-based heart sound generation model can be used as a simulation platform to simulate the mechanisms of heart sound generation under normal physiological or some pathological conditions. This model can provide a way for study of relationships between heart sounds and hemodynamics, which is an important reference for diagnosis of cardiovascular diseases. The main work of this paper is focused on three aspects as follows.(1) A hemodynamics-based first and second heart sound generation model is proposed. First, the analog circuit model of cardiovascular system is established, including four heart chambers, the systemic/pulmonary arteries, capillaries, veins, and physiological control of heart rate and cardiac contractibility. The model are represented by differential equations and solved by finite-difference method. According to the simulation results, the aortic valve, pulmonary valve, mitral valve and tricuspid valve closure timings and pressure gradient between both sides of the valves can be obtained. Four delays, i.e., the timing delay between mitral and aortic valve closure (TDMA), the timing delay between aortic and mitral valve closure (TDAM), the timing delay between aortic and pulmonary valve closure (TDAP) and the timing delay between mitral and tricuspid valve closure (TDMT), are further defined in a heart cycle. Relationships between the four delays and hemodynamics are studied. The valves are simulated as forced damped spring oscillator. The driven force calculated according to pressure gradient is further fitted as Ramp function, by which the analytical expressions of the valves vibration can be easily solved, and the heart sounds are then synthesized. Frequency-domain and time-frequency-domain features of synthetic heart sounds are consistent with those of experimental heart sounds.(2) Decomposition of S1and S2methods are proposed. It is assumed that both A and P components of S2are subject to envelope modulated chirp model and are characterized by18parameters. Under mean squared-error criterion, the optimal parameters of A component and P component are estimated by using SA (Simulated Annealing) algorithm. The A and P components are then reconstructed. A and P components are extracted from S2signals of10 subjects. which shows the effectiveness of the proposed algorithm. It is assumed that S1signal is represented by several Gaussian functions, each Gaussian function has3parameters. Similarly, the optimal parameters are estimated by using SA algorithm. The algorithm is further verified by S1signals of10subjects. The correlation coefficients of original and reconstructed signals are greater than0.99.(3) The cardiovascular hemodynamic changes and corresponding heart sound changes under exercise and respiration conditions are simulated by using the proposed model. Results of exercise conditions indicate that during the recovery phase after exercise, aortic systolic blood pressure (SBP) and heart rate are decreased. TDMA, TDAM and TDAP have strong negative correlations with SBP; meanwhile. TDMT has a slightly negative relationship with SBP. During the recovery phase, the amplitudes of S1and S2are decreased. Results of respiration conditions show that during inspiration, systemic blood pressure is reduced, heart rate is increased, TDMA is significantly decreased and TDAP is significantly increased, TDMT does not change significantly, The amplitudes of S1and S2are not significantly changed. To further validate the results, physiological signals including heart sound, ECG. respiratory wave and blood pressures of6healthy male subjects are experimentally acquired under corresponding conditions. The heart sound signals are analyzed by using the proposed heart sound decomposition methods. The experimental results are in good agreement with the simulated results.