Fluoroscopic x-ray beams can be produced in either a continuous or a pulsed fashion (Fig. 1). The usually quoted dose rate for fluoroscopy is simply the time-averaged output of the tube (e.g., mGy/s or R/min). By definition, the instantaneous and average dose rates are the same for continuous fluoroscopy. Producing an average dose rate for pulsed fluoroscopy higher than that of continuous fluoroscopy is accomplished by using a much higher instantaneous dose rate during the pulses (Fig. 1). Increasing or decreasing the instantaneous dose rate or the number of pulses per second (frame rate) or both will increase or decrease the average pulsed fluoroscopic dose rate (Figs. 2 and 3).

Fluoroscopists are presented a series of images while the beam is on. The observer's eye continuously integrates all of the incoming images with a typical overlapping integration time of about one fifth of a second. The average radiation dose rate used to produce the images affects the radiologist's perception of noise. The effect of eye integration is easily shown during a clinical procedure by comparing the appearance of a fluoroscopic last-image-hold image with that of live fluoroscopy or the replay of a stored fluoroscopic loop of the same patient and anatomy [3]. Because of eye integration, for equal perceived noise in pulsed fluoroscopy, the required dose rate approximately scales as the square root of frame rate [4]. For equal noise perception, the average dose rate will only drop from 1.0 to about 0.7 when the frame rate is lowered from 15 frames per second (fps) to 7.5 fps. The dose for each remaining pulse is increased to maintain perception. Doubling the frame-rate from 15 to 30 fps yields an average dose rate increase of 1.4 via a decrease of dose for each of the doubled number of pulses. Figure 4 schematically shows these concepts. Most fluoroscopes can be separately programed for dose per pulse and pulses (frames) per second. The relation between frame and dose rate is not obvious from the control settings. A qualified medical physicist should measure actual dose rates in each laboratory for applicable clinical technique sets and conditions.

Noise integration by the eye is often enhanced when the fluoroscope provides temporal averaging of successive image frames. This originally occurred because of the intrinsic properties of analog video tubes. A video camera tube was often selected for fluoroscopic use because it had enough inherent integration time (lag) to provide desirable noise integration. Charge-coupled device (CCD) cameras and flat-panel fluoroscopic detectors do not have any appreciable lag. The CCD camera was regarded as excessively noisy when it was first introduced. Digital image processing adds “lag” to a fluoroscopic series using temporal averaging. The recursive filtering algorithm combines newly acquired data with the previous image. The influence of each acquisition diminishes over time as it becomes a smaller and smaller fraction of the latest image (Fig. 5). Retaining a larger fraction of the previous image increases the number of images over which any particular acquisition provides a useful contribution.

A series of pulsed images will appear to be a continuous image at high pulse rates (frequencies) and a flickering series of discrete images at low pulse rates. The transition between continuous and flickering images occurs at the critical flicker frequency (CFF). Many factors affect the CFF [5, 6]; however, the rate is around 15 fps at light levels consistent with television viewing. Avoiding flicker at low frame rates was a problem for motion pictures and analog video. Early motion pictures were called the “flicks” because of visual flicker induced by low frame rates. The 24 fps film standard is below the CFF for some observers. Theatrical motion-picture projectors eliminate flicker by showing each frame two or three times before advancing the film. This raises the perceived frame rate to 48 or 72 fps. Digital video accomplishes the same effect by displaying frames at a high rate. Each fluoroscopic image acquired at a lower rate is shown multiple times at the higher display rate (Fig. 6). Thus, a smoothly moving object will appear to move in an increasingly discontinuous manner as the frame rate decreases. The effects of displaying exactly the same noise pattern multiple times (via gap fill) on noise perception are not well known.

For pulsed fluoroscopy, the dose rate is determined by the dose per frame and the acquisition frame rate. Decreasing the dose rate by decreasing the frame rate is a good strategy for viewing noise-limited static objects because of both eye integration and image processing, such as recursive filtering. However, optimizing the imaging of moving objects is more complex. Recursive filtering temporally averages images; it therefore increases the unsharpness of moving objects. Gap filling by replicating images minimizes critical flicker-fusion frequency effects. However, as the acquisition frame rate decreases, the appearance of a smoothly moving object (e.g., an angiographic guidewire) becomes increasingly intermittent. This may interfere with the radiologist's eye-hand coordination. The optimum dose rate for a given procedure is obtained by selecting a frame rate that is compatible with adequately controlling moving objects, a recursion rate that does not produce unacceptable blur of moving objects, and a dose per frame that is just high enough to minimize the effects of image noise on the procedure. Such optima are procedure specific and may also be operator specific. Caution is indicated when performing a procedure using a fluoroscopic mode tuned to one procedure for a substantially different type of procedure.