Modeling of Detector Performance

Recent advances in medical imaging include the development of large-area flat-panel x-ray detectors for radiography, fluoroscopy, mammography, and cone-beam CT as well as multi-detector and photon counting systems for diagnostic CT. Such advances have enabled new capabilities (for example, improved spatial resolution and reduced radiation dose) and propelled a revolution in 3D imaging systems over the last two decades. Quantitative understanding of the performance of various detector technologies - and optimizing their performance for a particular clinical application - benefits tremendously from mathematical models of imaging performance, including analytical models of spatial resolution, noise, and detective quantum efficiency. Such models give insight on the performance of each element of the imaging chain in terms of its spatial-frequency-dependent transfer characteristics. In turn, these characteristics can be related to the performance of a particular imaging task by considering spatial-frquency-dependent signal and noise with respect to the spatial-frequencies associated with performance of a particular task - i.e., "task-based" models. Such analysis has provided a foundation for imaging chain optimization (including scintillator thickness, pixel pitch, and electronic readout noise) and helped to accelerate the development of new imaging systems for 2D (projection) and 3D (volumetric) imaging systems for a variety of applications in medical diagnosis and interventional guidance. In this presenationa, we review the essentials of such task-based models of imaging performance, study examples of model-based design of new imaging systems, and consider future challenges in modeling of nonlinear imaging systems.