Abstract
241508
Introduction: Advances in positron emission tomography instrumentation have pursued smaller detector sizes to achieve better spatial resolution and overall image quality. The inter-crystal scatter (ICS) hampers the positioning accuracy of gamma ray detection. In this work, we aim to investigate the effect of ICS recovery in sensitivity and image quality for the NeuroEXPLORER, a high-performance dedicated brain PET/CT scanner, and further demonstrate that using both energy and DOI information can substantially improving the accuracy of recovered ICS events.
Methods: .
The NeuroEXPLORER aims to optimize scanner parameters for brain imaging by utilizing both time-of-flight (TOF) and depth-of-interaction (DOI) information. The core of the detector block comprises four U-Shaped crystals read out by 2×2 silicon photomultipliers (SiPM). While the ICS inside this block degrade positioning, the event will be registered with energy close to photopeak interaction due to the light sharing. The scattered events between adjacent micro-blocks should be handled carefully to recover the first interaction position. In this, work we used Monte Carlo method and simulated the 6-ring scanner, where each ring has 20 detector modules, and each module comprises 12×12 micro-blocks. An energy resolution of 10.5%, DOI resolution of 3 mm and coincidence timing resolution of 240 ps was introduced. A micro-Derenzo phantom was placed at the center. The positioning method for ICS events between two micro-blocks were calculated by three different methods: 1) micro-block with lower energy, 2) micro-block with higher energy, 3) micro-block with lower DOI and 3) based on combination of DOI and energy: selecting as the first interaction position when ΔE <= -T, or abs(ΔE) < T && ΔDOI <= 0.
Here, ΔE = E1 - E2, ΔDOI = DOI1 – DOI2, T is a threshold. T can be selected based on energy resolution, e.g., T = n σ, σ = FWHM / 2.355, FWHM is energy resolution. For a 10.5% energy resolution, FWHM = 511 keV * 10.5% = 53.65 keV.
Results: Results show that without ICS recovery, the scanner loses 43% of sensitivity. Correct ICS identification percentages are 61%, 38%, 70% and 72% for low energy, high energy, low DOI and combined energy and DOI based methods, respectively. The reconstructed images of the micro-Derenzo phantoms show similar spatial resolution for no ICS recovery and combined energy and DOI ICS recovery method, the latter has higher sensitivity. The combined DOI and energy method provides an optimal decision boundary for identifying the first interaction event.
Conclusions: The simulation model for the NeuroEXPLORER brain scanner with small crystal sizes and DOI information, was created to assess the impact of ICS events on the image quality and possible solutions to recover ICS events. Simulation results show that sensitivity of the scanner have increased by >75% after the ICS is properly processed. The combined DOI and energy criteria provide a perfect balance between positioning accuracy and timing performance to localize the recovered events, which increases the system sensitivity without degrading spatial resolution.
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.