Single Sensor Compressive Ultrasound Imaging: A Study of Affecting Factors

Abstract

Compressed sensing has enabled 2D and 3D ultrasound imaging using a single transducer by encoding lateral and elevation spatial information as temporal variations in the transmitted and received ultrasound signal through a coded aperture in the form of a pseudo-random delay mask. This technology has become increasingly important with the development of ultrasound techniques as it allows for a reduction in machinery size and power consumption. In this article, we develop a model for compressive ultrasound imaging using a single coded sensor to investigate the factors that affect image quality and enable computationally-efficient simulation of the system. We provide a step-by-step guide to creating synthetic data and demonstrate compressive ultrasound experiments of scenes with varying levels of sparseness generated according to a linear image formation model. We then calculate qualitative and quantitative measurements and solve the inverse problem using several sparse recovery solutions to achieve faithful reconstruction of the scene under different signal-to-noise ratios (SNR) and coded sensor geometries. Our model analysis reveals that failure to consider preferable conditions results in degraded peak signal-to-noise ratio (PSNR), mean squared error (MSE), and structural similarity (SSIM) indexes related to the quality of the reconstruction.