Minimisation of data transfer losses in the display of digitised scintigraphic images.

The transfer of scintigraphic data to any type of display, and ultimately to the eye, is a data compression procedure which invariably leads to information loss. Unless this information loss can be minimised, valuable image data may well be discarded. Attempts to achieve data loss minimisation have led to procedures such as the use of statistically equal levels for display, and the more recent histogram modification techniques such as equalisation and hyperbolisation. The method discussed here uses information theory to obtain the mean uncertainty (or entropy) per pixel in the intensity of the displayed image. Statistical noise and the characteristics of the image are taken into account. Calculation of the best choice of grey or colour levels for transfer loss minimisation is then possible: that is, the best conditions for data transfer to the display can be set up. A computer algorithm which carries out this function has been written. Several different types of simulated phantom, and clinical images, have been investigated. Improved perceptibility of many of these images has been obtained, correlating with reduction in mean pixel uncertainty. Although the technique still requires some refinement, it appears that optimisation of display characteristics for any transmitted image is potentially feasible.