Regularized Dual Averaging Image Reconstruction for Full-Wave Ultrasound Computed Tomography

TL;DR

This paper introduces a structured optimization approach combining regularized dual averaging with source encoding for ultrasound computed tomography, achieving high-resolution images with reduced computational time and improved noise suppression.

Contribution

It proposes integrating regularized dual averaging with source encoding to enhance image quality and efficiency in USCT reconstruction, overcoming limitations of stochastic gradient descent.

Findings

Images with less noise and comparable resolution to SGD methods.

Reduced reconstruction times due to source encoding.

Effective incorporation of non-smooth regularization penalties.

Abstract

Ultrasound computed tomography (USCT) holds great promise for breast cancer screening. Waveform inversion-based image reconstruction methods account for higher order diffraction effects and can produce high-resolution USCT images, but are computationally demanding. Recently, a source encoding technique was combined with stochastic gradient descent to greatly reduce image reconstruction times. However, this method bundles the stochastic data fidelity term with the deterministic regularization term. This limitation can be overcome by replacing stochastic gradient descent (SGD) with a structured optimization method, such as the regularized dual averaging (RDA) method, that exploits knowledge of the composition of the cost function. In this work, the dual averaging method is combined with source encoding techniques to improve the effectiveness of regularization while maintaining the reduced reconstruction times afforded by source encoding. It is demonstrated that each iteration can be decomposed into a gradient descent step based on the data fidelity term and a proximal update step corresponding to the regularization term. Furthermore, the regularization term is never explicitly differentiated, allowing non-smooth regularization penalties to be naturally incorporated. The wave equation is solved by use of a time-domain method. The effectiveness of this approach is demonstrated through computer-simulation and experimental studies. The results suggest that the dual averaging method can produce images with less noise and comparable resolution to those obtained by use of stochastic gradient descent.