Dual matrix ordered subsets reconstruction for accelerated 3D scatter compensation in single-photon emission...

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Image Reconstruction

Background:

• Three-dimensional (3D) iterative Maximum Likelihood Expectation Maximization (ML-EM) algorithms in single-photon emission tomography (SPET) correct for image degradations like non-uniform attenuation, distance-dependent response, and scatter.
• These corrections improve quantitation, resolution, and signal-to-noise ratio (SNR) but require substantial computational resources.

Purpose of the Study:

• To introduce and evaluate a novel acceleration method for ML-EM algorithms in 3D SPET.
• To assess the computational efficiency and image quality preservation of the new method compared to conventional ML-EM.

Main Methods:

• Development of the Dual Matrix Ordered Subsets (DM-OS) algorithm, combining distinct projection/back-projection matrices with ordered subsets.
• Comparison of DM-OS with ML-EM using simulated and physical thorax phantom data for both 180- and 360-degree orbits.
• Quantitative analysis using contrast, normalized standard deviation, and mean squared error for digital phantom experiments.

Main Results:

• DM-OS achieved speed-up factors of up to 200 (120 projections) and 50 (60 projections) compared to ML-EM, with similar image quality.
• DM-OS with scatter compensation demonstrated significant SNR improvement over ML-EM without scatter compensation.
• The DM-OS algorithm simplifies the integration of complex image formation models into the computational code.

Conclusions:

• DM-OS is a computationally efficient and relatively simple algorithm for 3D iterative scatter compensation in SPET.
• It offers comparable image quality to conventional ML-EM while drastically reducing reconstruction time for both 180- and 360-degree data acquisition.
• DM-OS represents a significant advancement for accelerating iterative reconstruction in SPET imaging.