A measure for radioisotope scan image quality.

A perennial difficulty in the field of radioisotope scanning* is the fact that the requirements for resolu tion and number of counts are specified independ ently although everyone knows that both contribute to the ultimate picture quality. This difficulty results from the fact that the image degradation due to blurring and that due to statistics are not directly comparable. Blurring introduces positional uncer tainty whereas statistical fluctuations introduce un certainty in brightness, that is to say, dot population density. We assume that ultimately positional uncertainty of picture features is what we mean by quality (or rather the lack of quality). However, brightness Un certainties, far from being irrelevant, contribute to the overall positional uncertainties since it is only through some systematic brightness contrast that features are revealed to the observer. “Brightness― can only be measured by evaluating the dot population over a sampling area of some significant extent. The statistical uncertainty of the brightness given by N counts can be decreased if N is made larger, but this can be done only by increas ing the sampling area, and thus the positional uncertainty related to that reading. What we are after, then, is a formula giving the amount of posi tional uncertainty perceived by a human observer as a result of the brightness uncertainties and bright ness contrast present in the picture. Combining this positional uncertainty due to statistics with the posi tional uncertainty due to blurring, one can then create a measure of overall positional uncertainty which presumably will reflect overall subjective qual ity of the scan. The existence of such a measure will permit us to optimally make the resolution sensitivity trade-off embodied in any instrument.