Optimization of bone SPECT image reconstruction based on a hybrid framework

Ordered subset expectation maximization (OSEM) is one of the most widely used reconstruction algorithm for Singlephoton emission computed tomography (SPECT) images reconstruction because of their efficiency in providing a better image quality. However, by increasing the number of subsets of this method, the convergence of this algorithm is speeded which can lead to undesirable noise levels amplification and inaccurate detection of lesion in low activity image regions. This paper presents a new algorithm for bone SPECT image reconstruction based on ordered subset expectation maximization (OSEM) algorithms and can remove the noise from images with the best degree of accuracy. In our proposed method, a de-noising pre-processing Butterworth filter is applied on the projections followed by OSEM algorithm to reconstruct 128 axial slices from a 128 enhanced sinograms, and finally we extract the coronal and sagittal slices from the enhanced axial slices volume. Our method was compared to Maximum Likelihood Expectation Maximization (MLEM) and OSEM techniques used only. Each method was tested on a three dimensional Shepp-Logan phantom and a bone SPECT database and evaluated qualitatively and quantitatively. The results show that the proposed method kept quantitative accuracy with preservation of the singularity and exhibited lower noise in low-activity regions while achieving high-resolution recovery in structures with high activity uptake in comparison to other methods.