Performance Analysis of Brain Imaging Using Enriched CGLS and MRNSD in Microwave Tomography

摘要

A trade-off between computational time complexity and more number of sensing antennas is a hurdle in high-resolution microwave tomography image reconstruction process. This paper deliberates the efficacy of Krylov subspace- based gradient regularization methods such as Enriched Conjugate Gradient Least Square (Enriched CGLS) and Modified Residual Norm Steepest Descent (MRNSD) method imposed in the reconstruction algorithm which effectively handles the above impediment. The performance of the proposed methods has been tested with varying the number of antennas, operating frequency and the levels of Gaussian noise in brain phantom and mean square error (MSE) and number of iterations are the parameters used for the analysis. MRNSD method has proved its betterment in all the criteria. It achieves 77% accuracy within five iterations.