Rapid Monte Carlo simulation of detector DQE(f)

Area of Science:

• Medical Physics
• Detector Science
• Computational Imaging

Background:

• Indirect x-ray detector performance optimization requires accurate modeling of photon transport.
• Monte Carlo simulations are crucial for detector physics but are hindered by long simulation times.
• Detective Quantum Efficiency (DQE) is a key performance metric, traditionally requiring extensive computation for Noise Power Spectrum (NPS) estimation.

Purpose of the Study:

• To demonstrate that reduced gamma and optical photon counts can accurately predict NPS in simulations.
• To develop a novel Monte Carlo algorithm for efficient DQE, Modulation Transfer Function (MTF), and NPS calculation.
• To significantly decrease simulation time for indirect x-ray detector characterization.

Main Methods:

• Derived Mean Squared Error (MSE) for simulated NPS, showing independence from input fluence above a minimum threshold.
• Proposed reducing input fluence to generate point-spread functions per event, enabling simultaneous MTF, NPS, and DQE(f) generation.
• Investigated reducing optical photon counts in scintillator simulations and compared simulation results with experimental data from a Varian AS1000 imager.

Main Results:

• Only 10-100 gamma photons per flood image were needed to avoid NPS bias, enabling a 10^7 reduction in fluence.
• Increasing flood image quantity (up to 500) improved signal-to-noise ratio (SNR) more than increasing gammas per image.
• Simulation time for thick pixelated arrays reduced from 2.5 x 10^6 CPU min to 3.1 CPU min, with DQE(f) matching experimental results within 11%.

Conclusions:

• Accurate modeling of x-ray detector DQE(f) is feasible in minutes using a single CPU.
• Optimized photon fluence significantly reduces computational burden without sacrificing accuracy.
• GEANT4 facilitates convenient simulation by offering both gamma and optical photon transport capabilities.