Key points are not available for this paper at this time.
A proper analysis and comparison of medical images requires an objective criterion for image discrimination. We developed a statistical approach for color image comparison and showed how it can be applied to discriminate cancer histology images from four treatment groups. We applied a well-established multivariate analysis of variances (MANOVA) on the logit scale so that image comparison was reduced to the classic chi-square test.
Eugene Demidenko (Sat,) studied this question.