Nonparametric cluster analysis of autoradiographic images

Describes a novel non-parametric statistical method based on cluster analysis for localizing significant differences in autoradiographic image data sets under two conditions. By thresholding cluster-size rather than pixel-values to reject false positives, this approach enhances statistical power. This test makes no assumption as to probability distribution or other properties of the statistical parametric map (SPM). The computational burden entailed by the Monte-Carlo method is also greatly reduced by a randomization method. An experiment compared differences in autoradiographic local blood flow in rats.