Filtered Back Projection and Simultaneous Algebraic Reconstruction Technique for Image Formation on Square-Shaped Physical Phantom Aimed at Microwave Imaging Applications

Abstract

Microwave Imaging is one of the emerging modalities, which has potential to become supplement to the existing modalities such as Magnetic Resonance Imaging (MRI) and Computed Tomography (CT). Microwave Imaging is suitable for early tumor detection due to its advantages of being low cost, non-invasive, and portable. In this paper, we simulate the microwave imaging system using prototype scanner that utilize two dipole antennas for transmitting and receiving the microwave signal at 3 GHz and square-shaped homogeneous phantom as object scanned. This simulation uses translational and rotational method for data acquisition scheme which is based on first generation of CT. Data acquisition consist of 21 translational step and 21 rotational step. The $S_{21}$ parameter on receiving antenna is used as acquired data and adopted as a sinogram. This $21 \times 21$ sinogram matrix becomes the input of reconstruction algorithm. Filtered Back Projection (FBP) is widely used as an image reconstruction algorithm for analytically perception in medical imaging. In this paper, we apply a Simultaneous Algebraic Reconstruction Technique (SART) algorithm as well in order to have comparison performance between SART and FBP method, especially for microwave imaging system. The performance comparison is conducted both qualitatively and quantitatively for reconstructing a square-shape physical homogeneous phantom. The Results show that both FBP and SART can successfully reconstruct the image of the square-shaped phantom. To evaluate the performance of FBP, Hann window is chosen as filter because it gives better result than other filters. As for the SART algorithm, five (5) times iteration is chosen for optimal result. In terms of qualitative perception, the reconstructed image of the SART has smoother quality and more easily recognizable (clearer edge) than the FBP. Quantitatively, the SART shows better PSNR (lower absolute error) rather than the FBP. However, the FBP is better SSIM, processing time and histogram, allowing FBP to have better perceptual quality and faster reconstruction time. This study shows that SART shows better qualitative parameters while FBP provides better quantitative parameters. From the results, further investigation is required to be conducted for other types of physical phantoms.