[Paper Review] Expected performance of the TT-PET scanner

E. Ripiccini, D. Hayakawa+7 more|arXiv (Cornell University)|Nov 29, 2018

Medical Imaging Techniques and Applications1 references4 citations

TL;DR

This paper presents the design and Monte Carlo simulation of the TT-PET scanner, a time-of-flight (TOF) PET system for small-animal imaging using monolithic silicon pixel detectors with 30 ps RMS time resolution and 0.5×0.5×0.2 mm³ granularity. The scanner achieves sub-millimeter spatial resolution (FWHM < 0.75 mm) and high count rate capability (≤0.1% loss at 75 MBq), with significant image quality improvement via TOF reconstruction.

ABSTRACT

The TT-PET collaboration is developing a small animal TOF-PET scanner based on silicon detectors featuring 30 ps RMS time resolution and intended to be inserted in an existing MRI scanner. The TT-PET scanner makes use of a stack of layers of high-Z photon-converter and 100 $\mathrm{μm}$ thick silicon sensors, to achieve a scanner with 0.5 $\mathrm{ imes}$ 0.5 $\mathrm{ imes}$ 0.2 $\mathrm{mm^{3}}$ granularity, with precise depth-of-interaction measurement. In this paper we present the results of the Monte Carlo studies for the expected data rate, time resolution on the TOF measurements, spatial resolution and image reconstruction with and without the use of the timing information. Most of the studies have been performed according to the international standards used to assess the performance of small-animal PET system.

Motivation & Objective

• Develop a high-resolution, TOF-PET insert for integration into existing MRI scanners for small-animal imaging.
• Achieve sub-millimeter spatial resolution using high-granularity silicon pixel detectors with precise depth-of-interaction measurement.
• Optimize detection efficiency and time resolution for 511 keV gamma rays using Monte Carlo simulations.
• Enable high data rate operation with minimal count loss (<0.1%) up to 75 MBq activity.
• Demonstrate image quality improvement through time-of-flight reconstruction using the CASToR framework.

Proposed method

• Design a cylindrical PET scanner with 16 wedge-shaped towers, each containing 60 stacked layers of high-Z photon converters and 100 µm thick silicon sensors.
• Use monolithic silicon ASICs with 0.5×0.5 mm² pixel segmentation and 20 ps time binning for high time resolution and low power consumption (20 mW/cm²).
• Implement a hierarchical readout architecture with super-modules, local buffering, and serial data links (50 Mbps) for efficient signal handling.
• Perform Monte Carlo simulations to optimize layer thickness and material composition for maximum 511 keV gamma detection efficiency.
• Use the CASToR reconstruction framework with 2D-FBP and TOF filtering to evaluate spatial resolution and image quality.
• Measure time resolution using 180 GeV pion beams and validate performance with 22Na source testing.

Experimental results

Research questions

• RQ1 What is the expected time resolution for 511 keV photons in the TT-PET scanner, given a 30 ps RMS resolution for minimum ionizing particles?
• RQ2 How does the 3D granularity of the silicon detector array affect spatial resolution across the field of view?
• RQ3 What is the maximum data rate the system can handle with less than 0.1% count loss?
• RQ4 To what extent does time-of-flight reconstruction improve image signal-to-noise ratio in the Derenzo phantom?
• RQ5 How uniform is the spatial resolution performance across radial, tangential, and axial directions?

Key findings

• The expected time resolution for 511 keV photons is 24 ps RMS, derived from a 30 ps baseline for minimum ionizing particles.
• The scanner achieves a spatial resolution of 0.56–0.59 mm FWHM in the radial direction and 0.60–0.71 mm FWHM in the tangential direction across the transverse field of view.
• The axial spatial resolution remains stable at 0.45–0.51 mm FWHM, with FWTM values below 1.2 mm across all directions.
• The system maintains count loss below 0.1% even at 75 MBq activity, demonstrating high-rate capability.
• TOF reconstruction significantly improves signal-to-noise ratio in the Derenzo phantom, as shown by visual and quantitative comparison.
• The scanner exhibits uniform spatial resolution performance across the field of view, with no significant degradation at high radial positions due to the multi-layer depth-of-interaction design.