Analysis of PET neurofunctional mapping studies.

Abstract

In this work, we present a method for analyzing positron emission tomography (PET) functional mapping experiments. The method is useful for identifying statistically significant differences between two PET data sets. First, uniform-variance Z-images are created and then the statistical uncertainty in region-of-interest values are calculated using a previously published method. The Z-images are calculated from the emission sinograms only--the calculation does not use scanner normalization and attenuation corrections and hence variance from these sources is eliminated, with no decrease in validity. Next, the Z-images are analyzed for activation sites using two separate techniques: a cluster analysis method and a change distribution analysis method. Both of these techniques are shown to be effective for objectively locating significantly activated regions from the Z-images. Two advantages of these methods are that they are objective and efficient; all of the parameters necessary in the calculations can be precomputed and stored since they depend only upon the geometry of the scanner.