(306g) Quantifying the Effect of Aortic Valve Degradation Using Signal Processing Techniques

Heart valves malfunction through stenosis or regurgitation (Otto C.M, 1999). Stenosis is the narrowing of the valve opening causing increased resistance to blood flow and subsequent thickening of the heart muscle, leading to heart failure. Regurgitation refers to the backflow of blood through the valve due to leakage and may occur due to a congenital defective valve, wear and tear of the valve, endocarditis, atherosclerosis, valve sclerosis, valve damage by radiation, and high blood pressure (Otto, 1999).

Detection of valve disorders is an important area of research to facilitate early treatment and prevent sudden heart failures. The invasive (cardiac catheterization, angiography) and non-invasive methods (ultrasound imaging, Doppler technique) used currently for diagnosis are based on experience and judgment of the consulting physician. Also, the progression of aortic stenosis is nonlinear and the patient might not show symptoms for years. The deterioration in the performance is very rapid and surgery becomes urgent at the stage where symptoms are clear. Therefore, early detection of the aortic valve disorder is important. Many a times, it is unclear whether the primary cause of disease is ventricular or valvular (Otto, 1999). Various spectral analysis methods like Fast Fourier techniques, auto regressive moving average, moving average, and wavelet methods have been suggested (Güler et. al, 2001, 1996). The results can be used to extract features from Doppler signals and detect variations. Wavelet technique gives better time and frequency resolution than other ones because it uses short time window for high frequency and vice versa (Aydin N., 1999). Also, wavelet domain allows us to view information simultaneously in both time and frequency domain. However, most of the studies have been preliminary and quantification of the valve condition has not been addressed.

In this work we will report on experimental and simulation studies aimed at developing early warning systems for valve malfunction detection. Aortic valves of varying stiffness (to simulate valve deterioration) were cast using a series of silastic rubber. The valves were then mounted in a pulse duplicator to simulate blood flow from left ventricle to the aorta using a 40% solution of glycerol in distilled water. Pressures and flow were recorded at various instants of time. The mean arterial pressure (which is a measure of the aortic compliance) was mimicked by applying pressure on the flow loop downstream of the valve and the peripheral resistance of the arteries was simulated by clamping the flow tube downstream. Data was recorded and averaged over a large number of beats to offset statistical variations from beat to beat. Doppler Ultrasound was also used to record the instantaneous velocity through the valve.

The pressure and flow waveforms were used to calculate the instantaneous aortic valve area and hemodynamic resistance to flow under various conditions. Analysis reveals that the valve resistance increases with stiffness of the valve, blood pressure, and peripheral resistance downstream. Pressure drop and flow waveforms indicate that valves with high stiffness show higher pressure drops for the same or lower flow rates. These observations are summarized in the figure below:

The patterns in pressure drop and flow waveforms are analyzed using wavelets. The major differences in the valve dynamics are observed during the valve closing phase, in the wavelet domain.

These waveforms are further being analyzed to extract features of the flow and identify the ones that are unique to a valve of certain stiffness. The idea is to isolate the effect of valve degradation or failure from the effect of other conditions, using signal processing techniques like wavelets and statistical pattern recognition models. Simulations of blood flow through valves of varying stiffness are also being carried out to better understand the experimental results.

References

Otto C.M. Valvular Heart disease. W.B. Saunders Company, Pennsylvania, 1999, pp 1-32, pp 43, pp 60, pp 179, pp191.

Güler İ; Kara S; Güler N F; Kiymik M K. Application of Autoregressive and Fast Fourier Transform spectral analysis to tricuspid and mitral valve stenosis. Computer Methods and Programs in Biomedicine, 49, (1996) 29-36.

Güler İ; Hardalac F; Muldur S. Determination of aorta failure with the application of FFT, AR, and wavelet methods to Doppler technique. Computers in Biology and Medicine 31 (2001) 229-238.

Aydin N; Padayachee S; Markus H S. The use of the wavelet transform to describe embolic signals. Ultrasound in Medicine and Biology, 25 (6), 1999, pp 953-958.