 |
 |
Re: Can one make localizing inferences when a cluster reaches significance but no voxels reach voxel-level significance? We have seen a number of PET publications (the imaging modality we use) that attribute regional significance to data defined at a cluster-level. However, having read what we think are the relevant publications, we still have some uncertainty and doubt about the ability to associate a specific location on the basis of a cluster when the significance of the cluster is less than 0.05 (corrected) and when there are no voxels within the cluster that reach statistically significant level. The SPM99 manual seems fairly clear, stating that the cluster level significance means that the size of the cluster exceeds that expected by chance. Nothing is said about the cluster location. In Human Brain Function, page 91, it says that within a cluster, nothing can be said at the voxel level [yet] the interpretation of results will clearly be different depending on threshold. This seems to be a clear reference to the ROC curves discussed in that chapter and in NeuroImage 1996;4:223. A high threshold increases the chance that the voxels within a cluster are a part of an underlying signal, without making any claim as to where that signal might occur, from an anatomical perspective. However, in NeuroImage 1996;4:223, nearly the last sentence says this example highlights the potential benefit of set-level inferences; in that the entire activation profile can be described anatomically and characterized as significant, therein providing a complete and comprehensive picture of activations ... (emphasis added). This implies that one can make anatomical inferences based on set-level statistics. If one can do this at the set level, then why not at the cluster level as this is a stronger test? In the SPM output, the cluster data are shown as the cluster size at the specified analytical threshold. No anatomical data are provided, although one can find the cluster on the glass-brain projections. Any statement about a specific anatomical location would seem to take the analysis further toward the voxel level. Thus, if a cluster achieves significant size and appears on the glass-brain images in a location that indicates a specific region of the brain, say the occipital lobe, can one say that the contrast utilized yields a significant difference in the occipital lobe? I would think not. If one wanted to make an anatomical inference, then one of two strategies or conditions should be employed: 1) there should be significance at a voxel level, or, 2) there should be an a priori regional hypothesis, in which case the investigator would create a mask to define the region, thus enabling the investigator to use small volume correction option to seek significance in the region. It would not seem to be appropriate to create the mask using post hoc information based on the appearance of the glass-brain projections. We would appreciate a clarification on this issue. As always, we are very appreciative of your help. Alan Lockwood [log in to unmask] voice: (716) 862-8788 voice mail (716) 862-8788 fax: (716) 862-8766