Adaptive clutter filtering based on sparse component analysis in ultrasound color flow imaging

Abstract

An adaptive method based on the sparse component analysis is proposed for stronger clutter filtering in ultrasound color flow imaging (CFI). In the present method, the focal underdetermined system solver (FOCUSS) algorithm is employed, and the iteration of the algorithm is based on weighted norm minimization of the dependent variable with the weights being a function of the preceding iterative solutions. By finding the localized energy solution vector representing strong clutter components, the FOCUSS algorithm first extracts the clutter from the original signal. However, the different initialization of the basis function matrix has an impact on the filtering performance of FOCUSS algorithms. Thus, 2 FOCUSS clutter- filtering methods, the original and the modified, are obtained by initializing the basis function matrix using a predetermined set of monotone sinusoids and using the discrete Karhunen-Loeve transform (DKLT) and spatial averaging, respectively. Validation of 2 FOCUSS filtering methods has been performed through experimental tests, in which they were compared with several conventional clutter filters using simplistic simulated and gathered clinical data. The results demonstrate that 2 FOCUSS filtering methods can follow signal varying adaptively and perform clutter filtering effectively. Moreover, the modified method may obtain the further improved filtering performance and retain more blood flow information in regions close to vessel walls.