Generalized DQE analysis of dual-energy imaging using flat-panel detectors

Abstract

Dual-energy (DE) imaging is a promising x-ray modality for the screening and early detection of lung cancer but has seen limited application primarily due to the lack of an adequate image detector. Recent development of flat-panel detectors (FPDs) for advanced imaging applications provide a promising technology for DE imaging, and a theoretical framework to quantify the imaging performance of FPD-based DE imaging systems is useful for system design and optimization. Traditional methods employed to describe imaging performance in radiographic systems [i.e., detective quantum efficiency (DQE) and noise-equivalent quanta (NEQ)] are extended in this paper to DE imaging systems using FPDs. To quantify the essential advantage imparted by DE imaging, we incorporate a spatial-frequency-dependent “anatomical noise” term associated with overlying structures to yield the generalized DQE and NEQ. We estimate anatomical noise in DE images through measurements using an anthropomorphic chest phantom and parameterize the measurements using a 1/f model. Cascaded systems analysis of the generalized NEQ is shown to reveal the tradeoffs between anatomical noise and quantum noise in DE image reconstructions. The generalized dual-energy NEQ is combined with idealized task functions to compute the detectability index, providing an estimate of ideal observer performance in a variety of detection and discrimination tasks. The generalized analysis is employed to investigate optimal tissue cancellation and kVp selection as a function of dose and imaging task.