Dynamic Imaging for Time-Varying Distribution Using Electrical/Ultrasonic Dual-Modality Tomography

Abstract

A dynamic imaging method using electrical/ ultrasonic dual-modality tomography is proposed to reconstruct the time-varying distribution based on the Bayesian inversion framework. The dynamic inversion problems of electrical resistance tomography (ERT) and ultrasonic transmission tomography (UTT) are constructed on a state-space model, formulating with the state evolution and observation update equations. The dynamic inversion problems are solved by the Kalman filter, and accelerated by the spatial dimensionality reduction method. The Kalman smoother is used to post-process the Kalman filter results to improve the imaging quality. The logistic regression method is proposed to fuse the ERT and UTT dynamic reconstructed images by nonlinear compression mapping. Experimental tests are conducted to evaluate the performance of the proposed methods. The results show that the proposed electrical/ultrasonic dual-modality dynamic imaging method has a better imaging accuracy than the traditional single-modality ERT and UTT dynamic imaging methods.