Evaluation of Iterative Methods MLEM and OSEM for Bone SPECT Image Reconstruction Employing Monte Carlo Simulations

Abstract

Bone scintigraphy is one of the main exams recommended for evaluating the skeletal system. In order to identify alterations more accurately, the SPECT technique (Single Photon Emission Computed Tomography) may be indicated. However, the quality of the SPECT image is dependent on the reconstruction methods and parameters employed. Among the reconstruction methods, the iterative methods, the Maximum Likelihood Expectation Maximization (MLEM) and Ordered Subset Expectation Maximization (OSEM) stand out, capable of achieving better spatial resolution and contrast-to-noise ratio. This study aims to evaluate the iterative methods which are MLEM and OSEM for the reconstruction of bone SPECT images, in order to identify optimized reconstruction parameters for each technique. The acquisition of projection images was performed using the Monte Carlo method using SIMIND software for computational simulation of signal acquisition in a gamma camera. Image reconstruction was performed using the MLEM and OSEM methods with different numbers of iterations, with the latter using 2, 4, 8, and 16 subsets. Analysis of the reconstructed images was based on the signal-to-noise ratio (SNR), contrast (C), and contrast-to-noise ratio (CNR) between regions of high and low radionuclide uptake. The results show that the OSEM technique provides images with higher contrast and CNR values than MLEM, for a lower number of iterations. In particular, OSEM with 8 and 16 subsets provided the highest CNR values at 6 and 2 iterations, respectively. Increasing the number of iterations beyond these values tends to reduce image quality due to increased noise levels. Overall, the results obtained highlight the importance of optimizing the reconstruction parameters of iterative methods.