Digital radiology using active matrix readout of amorphous selenium: Theoretical analysis of detective quantum efficiency

TLDR

Digital radiology with flat‑panel detectors that use amorphous selenium for direct conversion and an active‑matrix readout is being explored, but their high intrinsic resolution and pixelated readout cause inherent undersampling and aliasing. The study proposes strategies for maximizing detector DQE for each x‑ray imaging application. The authors theoretically analyze the frequency‑dependent detective quantum efficiency of self‑scanned a‑Se detectors by modeling signal and noise propagation, using presampling MTF and aliased NPS, and apply the analysis to mammography, chest radiography, and fluoroscopy while examining pixel fill factor, geometry, and electronic noise effects. The analysis shows that the aliased noise power spectrum is white, causing the DQE(f) shape to follow the square of the presampling MTF, and that DQE(0) scales with the pixel fill factor.

Abstract

A flat-panel x-ray imaging detector using a layer of amorphous selenium (a-Se) for direct conversion of x rays (to charge) and an active matrix for self-scanned readout is being investigated for digital radiology. A theoretical analysis of the spatial frequency dependent detective quantum efficiency (DQE(f)) of the self-scanned a-Se detector is performed based on a model of signal and noise propagation in a cascaded imaging system. Because of the high intrinsic resolution of a-Se and the pixelated active matrix readout method, such detectors are inherently undersampled and aliasing is present. The presampling modulation transfer function (MTF) and aliased noise power spectrum (NPS) of the detector were used in the analysis of DQE(f). It is proven that the aliased NPS for the self-scanned a-Se detectors is white. Since the shape of DQE(f) is determined by the ratio of MTF squared and the NPS, the shape of DQE(f) follows the square of the presampling MTF of the detector as a result of the white NPS. The analysis also shows that DQE(0) is proportional to the pixel fill factor, i.e., the fraction of each pixel area used for image charge collection. The DQE analysis is applied to detector parameters for three x-ray imaging applications: mammography, chest radiography, and fluoroscopy. The effects of pixel fill factor, imaging geometry (i.e., incident angle of x rays), and various sources of electronic noise on the detector DQE(f) are discussed. Strategies for maximizing detector DQE for each x-ray imaging application are proposed.