Ultrasonic Tissue Reflectivity Function Estimation Using Correlation Constrained Multichannel FLMS Algorithm with Missing RF Data

TL;DR

This paper introduces a novel blind multichannel frequency-domain LMS algorithm with correlation constraints to improve ultrasound image resolution by estimating tissue reflectivity and system PSF, even with missing RF data.

Contribution

It proposes a new correlation constrained missing-data estimation algorithm for ultrasound image deconvolution, addressing nonstationary PSF and misconvergence issues in prior methods.

Findings

Improved ultrasound image resolution demonstrated in simulations and experiments.

Effective estimation of tissue reflectivity and PSF with missing RF data.

Enhanced convergence and robustness of the proposed algorithm.

Abstract

Poor resolution of ultrasound images due to convolution of the tissue reflectivity function (TRF) with the system point spread function (PSF) is a major issue in medical ultrasound imaging. In this paper, we propose a correlation constrained missing-data estimation based blind multichannel frequency- domain least-mean-squares (md-bMCFLMS) algorithm to undo the effect of PSF on the ultrasound radio-frequency (RF) data. In the first step, a block-based MCFLMS (bMCFLMS) algorithm is proposed to estimate the TRFs and the PSF which are used in the second step to estimate the missing data. This missing data is used in the md-bMCFLMS algorithm to construct a modified cost function for further improvement of the image resolution. To account for the nonstationarity of the PSF, unlike the blocking approach described in the literature, we introduce a time-efficient blocking method in this paper. The blocking approach described here uses a block position independent fixed size matrix and can be implemented parallely. The bMCFLMS algorithm, however, shows misconvergence due to both channel noise and propagation of TRF estimation error from the previous blocks. This phe- nomenon is more intense in the case of md-bMCFLMS algorithm because of increased estimation error. To address this problem, a novel constraint based on the correlation between the measured RF data and estimated TRF is proposed in this paper. The efficacy of our proposed blind deconvolution algorithm is measured using simulation phantom, experimental phantom and in-vivo data.