Background: Magnetic induction tomography (MIT) is a promising imaging modality of electrical tomography. The precision of image reconstruction algorithms in MIT places the key role in its application in industrial and biomedical fields. Image reconstruction in MIT is an ill-posed inverse problem. Regularization algorithms are effective to handle the ill-posed problem. However it is difficult to select its proper regularization parameter, larger or smaller regularization parameter will both effect on the quality of the reconstructed images.
Methods: In this paper, firstly, the regularization principle of image reconstruction of MIT is analyzed. Then, we take the dimension of the Hessian matrix as the prior knowledge to obtain a novel penalty term, and add it to hybrid regularization algorithm, including Tikhonov and NOSER regularization, to propose a modified hybrid regularization algorithm. Secondly, selection for regularization parameters of image reconstruction algorithm is considered as a typical optimization problem, Particle swarm optimization with simulated annealing (PSO-SA) algorithm is used to solve the optimal solution.
Results: Numberial experiments with six typical models and cerebral hemorrhage model were carried out. The correlation coefficient (CC), relative error (RE) and condition number of Hessian matrix are used as evaluation metrics to evaluate the effectiveness of the proposed method. The results obtained from the proposed method have lower REs and higher CCs compared to the other two methods. And the condition number of Hessian matrix in the proposed method is greatly reduced.
Conclusions: The experimental results can prove the validity of the proposed method. The proposed method is capable of enhancing accuracy and noise immunity, which most closely coincides with the true conductivity distributions. Our method has better reconstructed quality compared with the other two methods, and it provides a theoretical reference for the development of application of the MIT technique in clinical decision application.