Super-resolution variable-dose imaging in digital radiography: quality and dose reduction with a fluoroscopic flat-panel detector

Abstract

Super-resolution (SR) image processing produces a high-resolution image from a series of low-resolution images. The aim of this study was to evaluate SR-images based on fluoroscopic flat-detector (FD) acquisition at different frame rates. Fluoroscopic FD-sequences with 20 frames were obtained with varying pulse frequencies of (1) a line pair resolution phantom; (2) a low-contrast resolution phantom, and (3) a human knee specimen. Super-resolution digital radiographs (SR-Radiographs) were generated from each sequence. Variable-dose images were simulated by constructing SR-Radiographs using 6 and 12 frames from the corresponding fluoroscopic sequence. "Single Shot" and Computed Radiography (CR) images were obtained for comparison based on dynamic range and sharpness of bone detail structures. Patient-derived SR-Radiographic images were constructed to demonstrate clinical examples. The spatial resolution of SR-radiographs obtained at 12.5 frames per second (fps) and 6 fps were comparable with CR and "Single Shot" images, providing ~3.5 line pairs per mm (l p/mm). Similarly, low-contrast resolution of SR-radiographs obtained at 12.5, 6, and 30 fps were equivalent to CR and "Single Shot" images. The human knee specimen SR-radiograph obtained using 12 FD images at 12.5 fps was superior to a CR image in overall image quality, with a dose reduction of 75%. Variable-dose SR-radiographic simulations suggest a dose saving potential of 90-95% when using 6 FD images at 12.5 fps or 6 fps, respectively. The phantom testing images and simulation results demonstrate that diagnostic quality SR-radiographic images of skeletal extremities can be synthesized using a flat-panel detector system designed primarily for angiography. SR-images obtained with substantially reduced radiation dose are feasible, and this technology may improve digital radiography for pediatric, neonatal radiography, or mammography applications. Further testing is needed to validate super-resolution techniques in other body regions and for different flat-detector systems.