An Improved Reconstruction Method of MIT Based on One-Step NOSER

Abstract

Magnetic induction tomography (MIT) is a new biologic tomography technology with the feature of harmless, non-invasive and convenience. The resolution and speed of image reconstruction algorithm is critical for improving the performance of MIT system and its application. Newton-One-Step Error Reconstruct or (NOSER) is a common reconstruction algorithm in MIT, but the slight variations of original data will impact the reconstructed images because of Hessian matrix which is ill-posed in the process of NOSER. In this paper, the NOSER was improved by eigen value threshold(ET) which was used to modify Hessian matrix. Compared with NOSER and Tikhonov regularization, the algorithm might improve the image resolution and anti-noise characteristic. Because the algorithm has no iterative procedure, it also was able to enhance imaging speed. The reconstruction results of varied imaging models demonstrate that the algorithm could potentially improve the performance of the MIT system and promote the application of the technology.