Model for perception of pulsed fluoroscopy image sequences.

Pulsed fluoroscopy at reduced frame rates can be used to lower x-ray dose with equivalent detection (hereafter called equivalent perception) of low-contrast, stationary objects. Experimentally average dose savings of 22%, 38%, and 49%, for pulsed fluoroscopy at 15, 10, and 7.5 acquisitions per second, respectively, are documented. Dose savings depend on object size, with fewer savings for smaller objects. To explain these data, we extend the framework of an ideal observer with three models for the spatiotemporal response of the human visual system (HVS). They are model 1, separable; model 2, nonseparable; and model 3, nonseparable with internal observer noise. With no free parameters, model 1 predicts the average dose savings within a 3% difference but does not describe the effect of object size. Models 2 and 3 explain the influence of size, and model 3, with a single free parameter, fits the measurements best. Perception of pulsed fluoroscopy is thus well described in terms of spatiotemporal processing by the HVS.