Software Article: A generalized cascaded linear system model implementation for x-ray detectors

Area of Science:

• Medical Imaging Physics
• Computational Imaging
• Detector Modeling

Background:

• Cascaded linear models are crucial for x-ray imaging system development.
• Existing Python implementations for these models are lacking.
• Accurate modeling of detector performance is essential for image quality.

Purpose of the Study:

• Introduce CASYMIR, a novel, open-source Python package for simulating x-ray imaging detectors.
• Provide a flexible tool for modeling both direct and indirect conversion detectors.
• Enable computation of Modulation Transfer Function (MTF) and Noise Power Spectrum (NPS) across various acquisition conditions.

Main Methods:

• Modular software design utilizing generalized frequency-domain expressions.
• Modeling of individual detection processes as serial or parallel blocks.
• Derivation of gain factors and parameters from user-defined detector characteristics, system geometry, and x-ray spectra.

Main Results:

• Experimental validation with a-Se and CsI detectors showed high accuracy.
• Modeled MTF errors below 6% and normalized NPS errors below 3% for a-Se detectors.
• Modeled MTF errors of 5.4% and normalized NPS errors of 5.8% for CsI detectors.

Conclusions:

• CASYMIR offers a flexible and accurate solution for x-ray detector modeling.
• The open-source package facilitates integration into simulation pipelines and virtual clinical trials.
• Ideal for optimizing x-ray system design and acquisition techniques through batch simulations.