Evaluation of weighted FDK algorithms applied to four-dimensional CT (4D-CT)

Progress of CT technology has realized four-dimensional (4D) imaging with dynamic scanning of multi-detector CT (MD-CT). However the axial coverage of commercial CT scanner is 32 mm at maximum, which impose a limitation on application of 4D imaging. In order to take 4D images for a wider field of view, our group developed a prototype CT scanner. To improve temporal resolution of the prototype scanner, we compared weighted Feldkamp-Devis-Kress (FDK) algorithms proposed by Taguchi. In this comparison, we employed as weighting functions, half-scanning (HS) and new half-scanning (NHS) as well as full-scanning (FS) that was the original FDK. We evaluated (1) image noise and uniformity, (2) artifacts and (3) temporal resolution with stationary and moving phantoms. Results were as follows. The magnitudes of image noise were larger at the X-ray tube side than at the detector side in the direction corresponding to scan time center for HS and NHS. Uniformities were similar for FS and HS. Slanting streak artifacts were observed in longitudinal images of a multiple disk phantom for HS and NHS, while HS showed more severe one. In the images of a moving phantom or a normal chest motion artifacts were observed for FS, while HS and NHS showed fewer artifacts with the fewest artifacts for HS. The results suggest HS and NHS may be useful for imaging of moving organs.