We describe MANTIS (Monte carlo x-rAy electroN opTical Imaging
Simulation), a tool for simulating imaging systems that tracks x
rays, electrons, and optical photons in the same geometric model.
The x-ray and electron transport and involved physics models are
from the PENELOPE package and include elastic and inelastic
scattering, and bremsstrahlung from 100 eV to 1 GeV. The optical
transport and corresponding physics models are from DETECT-II and include Fresnel refraction and reflection at material
boundaries, bulk absorption and scattering. X rays are generated
using the flexible source description from PENELOPE. When x
rays or electrons interact and deposit energy in the scintillator,
the code generates a number of optical quanta at that location,
according to a model for the conversion process. The optical photons
are then tracked until they reach an absorption event that in some
cases contributes to the electronic signal. We demonstrate the
capabilities of the new tool with respect to x-ray source, object to
be imaged, and detector models. Of particular importance is the
improved geometric description of structured phosphors that can
handle tilted columns in needle-like phosphor screens. Examples of
the simulation output with respect to signal blur and pulse-height
distributions of the scintillation light are discussed and compared
with previously published experimental results.
Previously we used a simple 2-D model to evaluate the imaging performance of a digital radiographic system while varying input parameters such as transducer blur and signal size. We extend this work using a realistic phosphor simulation to explore the effect of the incident x-ray spectrum and the depth dependence of the point spread function and optical collection efficiency. Initially we investigate one Swank screen type representative of modern powder phosphor design. Images resulting from these simulations are used to get an estimate of the impact of these factors on lesion detectability. Results show that the simple 2-D model gives optimistic estimates of detectability.
The usefulness of Fourier-based measures of imaging performance has come into question for the evaluation of digital imaging systems. Figures of merit such as detective quantum efficiency are relevant for linear, shift-invariant systems with stationary noise. However, no digital imaging system is shift invariant, and realistic images do not satisfy the stationarity condition. Our methods for task- based evaluation of imaging systems, based on lesion detectability, do not require such assumptions. We have computed the performance of Hotelling and nonprewhitening matched-filter observers for the task of lesion detection in digital radiography.