ESTIMATION OF THE PROBABILITIES OF 3D CLUSTERS IN FUNCTIONAL BRAIN IMAGES

The interpretation of functional brain images is often hampered by the presence of noise. This problem is most commonly solved by using a st atistical method and only considering signals that are unlikely to occ ur by chance. The method used should be specific and sensitive, specif ic because only true signals are of interest and sensitive because thi s will enable more information to be extracted from each experiment. H ere we present a modification of the cluster analysis proposed by Rola nd ct al. (Human Brain Mapping 1: 3-19, 1993). A covariance model is u sed to test hypotheses for each voxel. The generated statistical image s are searched for the largest clusters. From the same data set noise images are generated. For each of these noise images the autocorrelati on function is estimated. These estimates are subsequently used to gen erate simulated noise images, from which a distribution of cluster siz es is derived. The derived distribution is used to estimate probabilit ies for the clusters detected in the statistical images generated by t esting the hypothesis. This presented method is shown to be specific a nd is further compared with SPM96 and the nonparametric method of Holm es ct al. (C) 1998 Academic Press.