Perception of detail and greyscale range in X-ray fluoroscopy images captured with a personal computer and frame-grabber

Abstract

Objective: To assess the dynamic range of radiological images captured with a personal computer frame-grabbing system coupled to an X-ray fluoroscopy machine. Methods: A 386DX-40 MHz, IBM compatible, computer with an SVGA monochrome graphics subsystem and a 387 co-processor, installed with a Screen Machine frame-grabber and controlled by a program specially written was used. Various systems were examined and the observer's perceptions of the results assessed. Results: The dynamic range available to an ordinary X-ray fluoroscopy system was found to be restricted to about 750 mV. Similar measurements showed that the dynamic range was always restricted to 3/4–1/2 of the full available signal because of a high value of the dark voltage of the TV camera's target on all seven systems measured. The dynamic range of the computer — frame grabber system was found to be significantly wider than the Image Intensifier — TV camera chain but, surprisingly, it was affected by the type of file format used for image storing on disk. Clinical images from a barium meal examination as well as CT images captured after optimisation of the frame-grabber were found to contain large quantities of noise in the first two least significant bit planes making them redundant and limiting the grey levels needed for image display to less than 64. This number was also less than the 80 grey levels that could be discriminated by the human eye on the computer monitor. Conclusions: It was concluded that 6 bit digitisation would have been sufficient for image capture. The advantages of the wider dynamic range of the frame-grabber and the processing capabilities of the computer were tested for the possibility of improving the perception of detail. However, the results were negative. The limiting spatial resolution measured with a variable density bar pattern at all magnifications was about 0.4 lp/mm lower from that measured directly on the fluoroscopic screen. A detail perception test had the same result. The perception success was significantly lower with the digital images at all but the highest of the exposure rates and despite the use of image processing filters.