Singular value decomposition filtering in high-frame-rate cardiac vector flow imaging

Michiya Mozumi,Ryo Nagaoka,Hideyuki Hasegawa

doi:10.11591/eei.v9i1.1858

Abstract

Dysfunction of the left ventricle (LV) weakens the cardiac function and affects the physical activity. Echocardiagraphy has been used to visualize the blood flow dynamics and to evaluate the cardiac function. However, the signal processing to suppress the clutter signals should be employed. In this study, we employed the singular value decomposition (SVD) clutter filtering to obtain the cardiac blood speckle images. We also employed the adaptive thresholding metric to determine the proper cutoff values at each phase during the cardiac cycle. Moreover, we employed a depth-dependent SVD clutter filter for more accurate estimation of the cardiac blood echo signals. The 2D blood flow velocity vectors were estimated by applying the block matching method to obtained blood speckle images. The obtained results show that the proposed filter suppressed the clutter signals from left ventricular wall significantly, and the contrast-to-noise ratio (CNR) was improved from -0.5 dB to 13.8 dB by the proposed SVD clutter filtering.

Highlights

Echocardiography has been used to evaluate the cardiac function
The color-Doppler imaging is a credible technique for visualization of blood flow [1] and used for visualizing of the blood flowing in and out from the cardiac cavity
The high pass filter, which chooses the cut -off frequency adaptively based on the velocity of the hea rt wa ll [8], could visua lize the blood B -mode ima ges a t a high-fra me ra te [9, 10]

Summary

Introduction

Echocardiography has been used to evaluate the cardiac function. The color-Doppler imaging is a credible technique for visualization of blood flow [1] and used for visualizing of the blood flowing in and out from the cardiac cavity. The color-Doppler imaging is employed to visualize an abnormal blood flow such as mitra l regurgitation [2]. In the color-Doppler imaging, the flow velocity component only in the a xia l direction is estima ted, a nd the velocity component pa ra llel to the a xia l direction cannot be obtained. In this sense, the color-Doppler image doesn’t stand for the true velocities. The block matching method is able to estimate 2D velocities [3 -6] In this method, the motion of speckle patterns between successive frames is tracked and, blood speckle images are required to obtain flow velocities. The high pass filter, which chooses the cut -off frequency adaptively based on the velocity of the hea rt wa ll [8], could visua lize the blood B -mode ima ges a t a high-fra me ra te [9, 10]

Methods

Results

Conclusion