Comparison analysis between SART and ART algorithm for microwave imaging

Abstract

Human cells normally grow, divide and replace into new cells every single minute. Nowadays, improper daily lifestyle leads to cell growth and divide faster than cell replacement, leading to tumor development in the body. The early detection system is currently extensively studied to diagnose the condition of human body faster. We are then focusing on the development of a non-invasive and non-intrusive tomographic imaging system. Microwave imaging is a good candidate for tomography technology to early detect the tumor, due to its advantages compared to other tomography technologies such as low health risk (non-ionization), low cost in implementation and operation, and ease of use. In this paper, we simulate an imaging system by using a simple numerical phantom, in which the phantom dielectric constant is divided into a normal and an abnormal tissue. Dipole antennas are used for transmitting and receiving microwaves signals at 3 GHz. In this case, a translational and rotational method is applied for data acquisition system. The scattering S 21 parameter at the receiving antenna is then used as the acquired data for reconstructing an image. In the previous work, we have developed an algebraic reconstruction technique (ART) algorithm that was used for reconstructing an image. As a simplest iterative technique, the image is not stable. Hence, we are now developing a simultaneous algebraic reconstruction technique (SART) algorithm to improve the image quality. SART is supposed to be agile to noise, allowing to having a smoother reconstruction image compared with the ART. This paper will focus on qualitatively and quantitatively analyzing SART algorithm in comparison with the ART. A two-layer-cylindrical phantom model is used for validating the imaging system by CST Microwave Studio™ EM simulator. The outer phantom is set by 53.53 of relative permittivity with the diameter of 14 cm, representing a normal tissue and the inner layer phantom's permittivity is 78 with the diameter of 6 cm, representing benign or malignant tissue. The reconstructed image shows that the SART image is smoother than the ART qualitatively. We also discuss quantitatively about the peak signal-to-noise ratio (PSNR), mean squared error (MSE), normalized cross-correlation (NCC), structural content (SC), maximum difference (MD), and normalized absolute error (NAE).