Evaluation of Monte-Carlo-based System Response Matrix Completeness and its Impact on Image Quality in Positron Emission Tomography

TL;DR

This paper evaluates how the statistical variance in Monte Carlo-generated system response matrices affects image quality in PET imaging, highlighting the importance of matrix completeness for accurate image reconstruction.

Contribution

It introduces an analysis of Monte Carlo simulation variance in system response matrices and its impact on PET image quality, which was not thoroughly examined before.

Findings

Higher statistical variance degrades image quality.

Complete and accurate SRMs improve reconstruction accuracy.

Monte Carlo simulation parameters significantly influence PET image outcomes.

Abstract

A major strength of iterative algorithms used in positron emission tomography (PET) lies in their abilities to introduce precise models of the physics at play, which includes the statistical nature of the detection processes, and a detailed description of the radiation-matter interactions. The process of data acquisition by the imaging system is described in a system response model, or system response matrix (SRM). In PET, elements of this matrix correspond to a probability of a coincident pair of gamma emitted from a certain element of the imaged volume (voxel) to be detected by the apparatus along a given pair of detection elements (e.g. a pair of scintillating crystals), or line of response (LoR). The number of voxels involved for each line of response and the statistical error on each matrix element are directly dependent on the number of coincident events simulated to generate the SRM. The main goal of this paper is to first evaluate the behavior of these two parameters and secondly to estimate their impact on the overall image quality in preclinical PET imaging. Our results show the direct impact of the statistical variance of the Monte-Carlo generated System Response Matrices used in iterative reconstruction algorithms on image quality.