 |
 |
Hi - I suggest that for a good intro to GRF-based cluster inference you see Matthew Brett's excellent page: http://www.mrc-cbu.cam.ac.uk/Imaging/randomfields.html The simple explanation is: If you increase the Z threshold, the actual number of voxels in a given cluster may not decrease very much, depending on the sharpness of the edges of the cluster in the Z image. Then, becuase the actual significance calculated for the cluster depends on both the number of voxels in the cluster AND the threshold that was used (the higher the threshold, the more significant the cluster), it is possible that this cluster will pass the final significance test when it didn't for lower Z threshold. Thanks, Steve. On Mon, 24 Mar 2003, Andrew Bleicher wrote: > I was hoping that somebody could explain the parenthetical statement made at the end of the previous message in this string, "the significance of a cluster depends on its size _and_ the threshold that was used" > > How and why is that? I have found that for my higher level analyses, the > higher my z threshold, the greater the area of activation in almost > every case. For this reason, I have had difficulty understanding my > results. > > I appreciate any help, > > Andrew Bleicher, MD > Albert Einstein College of Medicine > Montefiore Medical Center > Dept of Radiology > Bronx, New York > Stephen M. Smith MA DPhil CEng MIEE Associate Director, FMRIB and Analysis Research Coordinator Oxford University Centre for Functional MRI of the Brain John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK +44 (0) 1865 222726 (fax 222717) [log in to unmask] http://www.fmrib.ox.ac.uk/~steve