Dose reduction through variable dose CT scanning: optimality of the filtered backprojection algorithm.

Abstract

A novel approach to patient radiation dose reduction in X-ray computed tomography (CT) (variable dose scanning) is proposed and evaluated. This technique spares radiosensitive tissues from unnecessary radiation if they are not in the area of immediate clinical interest. Specifically evaluated is the potential for reduction of dose to the eye and active bone marrow in head transmission CT. Using computer simulations, doses are calculated for an easily implementable two level exposure variation, and results are discussed for the fully optimized case in which the intensities of all pencil beams are independently variable. Dose reductions of more than 80% in peak dose are possible in head scanning. To determine whether the filtered backprojection algorithm is optimal in terms of signal-to-noise ratio of the final images for this novel scanning technique, a Cramer-Rao lower bound to noise in the image (which is independent of reconstruction algorithm) is derived and calculated for several types of scan. The bound uses the photon distributions determined from mathematical optimization and is compared with the actual noise of the reconstructions. The comparison indicates that little if any improvement can be expected in signal-to-noise ratio using reconstruction algorithms other than the filtered backprojection.