Dear Cyril,
it is disabled per default. By modifying spm_default.m / defaults.stats.rft.nonstat it is turned on and then it works for any of the single-subject or group models when hitting the "Results" button.
Concerning version history, please check http://www.fil.ion.ucl.ac.uk/spm/download/spm8_updates/README.txt It states
***
r3822 | karl | 2010-04-16 19:43:08 +0100 (Fri, 16 Apr 2010) | 2 lines
Changed paths:
M /trunk/spm_P_RF.m
M /trunk/spm_bwlabel.m
M /trunk/spm_clusters.m
M /trunk/spm_est_smoothness.m
M /trunk/spm_getSPM.m
M /trunk/spm_get_data.m
M /trunk/spm_list.m
M /trunk/spm_max.m
M /trunk/spm_resels.m
M /trunk/spm_spm.m
RPV images is now implemented to provide cluster specific volume estimators in resels. This allows for non stationariness in smoothness when inferring on spatial extent.
spm_P_RF now releases extent-based p-values for all statistics.
r3953 | guillaume | 2010-06-28 17:58:48 +0100 (Mon, 28 Jun 2010) | 3 lines
Changed paths:
M /trunk/spm_P_RF.m
M /trunk/spm_defaults.m
M /trunk/spm_figure.m
M /trunk/spm_list.m
Non stationarity in smoothness is disabled by default, as well as extent-based p-values for F-statistics.
This is still work in progress, that will be released later.
***
spm_est_smoothness.m references the Hayasaka paper, for the necessary m files a search for "nonstat" should do it.
For SPM5 there was a toolbox http://fmri.wfubmc.edu/cms/software#NS , and the VBM8 toolbox also had (still has) an option for the correction.
Best,
Helmut
|