Resolution Recovery with Pre-Reconstruction Fourier Transforms Filtering for a High Resolution Animal SPECT System

Abstract

Purpose: Spatial resolution and accurate quantification in animal SPECT plays a critical role in preclinical imaging. Appropriate Projections filtering can help to address the issue of resolution compensation and recovery coefficients. The main aim of current study was to explore a new version of MLEM algorithm with pre-reconstruction Fourier transform filtering to improve the Recovery Coefficient RC by filtering the projections.
Materials and Methods: A complete set of measurements was performed using HiReSPECT system. Different phantom studies were performed using point sources, NEMA NU 4 and dedicated hot rod phantom. All data was reconstructed using the developed Maximum-Likelihood Expectation-Maximization (MLEM) algorithm with and without resolution recovery information and also with our presented pre-reconstruction Fourier transform filter. Image quality was assessed by different parameters such as Coefficient Of Variation (COV), Recovery Coefficient (RC) and FWHM.
Results: FWHM in tangential direction was improved 0.6 and 0.2 for our presented algorithm against MLEM without and with CDRF, respectively. Fourier filtering image led to improvement of image quality where hot rods with 1.6 and 1.8 mm in diameter could be recognized clearly in our presented algorithm in dedicated hot rod phantom. Our presented algorithm resulted in rapid convergence and growth of RC in lower iteration.
Conclusion: The purpose of the current study was to introduce new pre-reconstruction filter. The results of this investigation showed that the inclusion of our proposed filter can lead to better image quality and quantification in lower iteration. These data suggest that current filter can provide higher RC in lower iteration.