Image Reconstruction of Electrical Impedance Tomography Based on Optical Image-Guided Group Sparsity

The low spatial resolution of Electrical Impedance Tomography (EIT) makes it challenging to conduct quantitative analysis of the electrical properties of imaging targets in biomedical applications. We in this paper propose to integrate optical imaging into EIT to improve EIT image quality and report a dual-modal image reconstruction algorithm based on optical image-guided group sparsity (IGGS). IGGS receives an RGB microscopic image and EIT measurements as inputs, extracts the structural features of conductivity distribution from optical images and fuses the information from the two imaging modalities to generate a high-quality conductivity image. The superior performance of IGGS is verified by numerical simulation and real-world experiments. Compared with selected single-modal EIT image reconstruction algorithms, i.e., the classical Tikhonov regularization and the state-of-the-art Structure-Aware Sparse Bayesian Learning and Enhanced Adaptive Group Sparsity with Total Variation, the proposed method presents superiorities in terms of shape preservation, background noise suppression, and differentiation of conductivity contrasts.

I. Introduction

Electrical Impedance Tomography (EIT) is a tomographic imaging modality, which estimates the conductivity distribution within a 2D or 3D bounded domain from a sequence of boundary voltage measurements [1]–[3]. EIT offers non-radiative, non-intrusive and high-temporal-resolution imaging capabilities and is investigated in many research domains, such as flow velocity field measurement [4], chemical engineering [5], multiphase flow monitoring [6], [7], and non-destructive cell culture imaging in tissue engineering [8]–[10]. However, its low image quality as a longstanding challenge has become a critical issue that prevents its wider adoption in different fields.