I am having some issues running SurfSmooth.
I ran it initially on a dataset about a month or so ago, and it worked alright, using the command:
SurfSmooth -spec /jukebox/kastner/surfaces/sm/SUMA/sm_lh.spec -surf_A lh.pial.asc -surf_B lh.smoothwm.asc -met HEAT_07 -target_fwhm 4 -blurmaster sm_lh_r1_vrud_surf_norm.niml.dset -detrend_master -b_mask sm_lh_r1_vrud_surf_mask.1D.dset -input sm_lh_r1_vrud_surf_norm.niml.dset -output sm_lh_r1_vrudssm4_surf_norm.niml.dset
and the 64 bit afni version: ani_2008_07_18_1710
I tried running it again on a new dataset using the 32 bit afni version afni_2008_07_18_1710.
I received an error message when running the same command on a different data set for both hemispheres i tried. here it the command:
SurfSmooth \
-spec /jukebox/kastner/surfaces/sb/SUMA/sb_{$hemi}.spec \
-surf_A {$hemi}.pial.asc \
-surf_B {$hemi}.smoothwm.asc \
-met HEAT_07 \
-target_fwhm 4 \
-blurmaster sb_{$hemi}_{$run}_vrud_surf_norm.niml.dset \
-detrend_master \
-b_mask sb_{$hemi}_{$run}_vrud_surf_mask.1D.dset \
-input sb_{$hemi}_{$run}_vrud_surf_norm.niml.dset \
-output sb_{$hemi}_{$run}_vrudssm4_surf_norm.niml.dset
and the error message:
++ Notice SurfSmooth (SUMA_SurfSmooth.c:2060):
Smoothing kernel bandwidth (sigma) = 0.500535
oo Warning SUMA_Chung_Smooth_Weights_07 (SUMA_GeomComp.c:2748):
Weights sum < 1.0f+1e-8 at node 50454
Mesh may be too coarse for kernel
bandwidth of 0.500535 in float precision.
Consider decreasing your number of iterations.
Future similar warnings are muted, but
a count is issued at the end.
oo Warning SUMA_Chung_Smooth_Weights_07 (SUMA_GeomComp.c:2773):
3 precision warnings out of 143977 nodes forming surface (0.00208 %).
Nodes with possible precision problems:
50454, 108889, 115869
Iteration 0, fwhm = 1.884529; target 4.000000
-- Error SUMA_Chung_Smooth_07_toFWHM_dset (SUMA_GeomComp.c:5536):
Failed to get mean fwhm
-- Error SurfSmooth (SUMA_SurfSmooth.c:2074):
Failed to blur master data dset
------------------------------------------------------------------------------------
I went in and checked the node numbers it outputed, these nodes are actually not in the mask dataset, so i am not sure why they are being smoothed.
the other hemisphere gave an error also, but did not list any nodes.
++ Notice SUMA_SigForFWHM (SUMA_GeomComp.c:5783):
Low Sigma/AvgLe, increased it to 0.500
Expected niter now: 59
Kernel Bandwidth / Average Edge Distance = 0.500000/1.000241 = 0.500121
Corresponding Kernel Numerator = 0.135335
++ Notice SurfSmooth (SUMA_SurfSmooth.c:2060):
Smoothing kernel bandwidth (sigma) = 0.500121
Iteration 0, fwhm = 1.874566; target 4.000000
-- Error SUMA_Chung_Smooth_07_toFWHM_dset (SUMA_GeomComp.c:5536):
Failed to get mean fwhm
-- Error SurfSmooth (SUMA_SurfSmooth.c:2074):
Failed to blur master data dset
--------------------------------------------------------------------------------------------
After having issues, I tried running my old dataset with this version of afni, it no longer ran, and gave the following message.
++ Notice SUMA_SigForFWHM (SUMA_GeomComp.c:5783):
Low Sigma/AvgLe, increased it to 0.500
Expected niter now: 63
Kernel Bandwidth / Average Edge Distance = 0.500000/1.026747 = 0.513374
Corresponding Kernel Numerator = 0.135335
++ Notice SurfSmooth (SUMA_SurfSmooth.c:2060):
Smoothing kernel bandwidth (sigma) = 0.513374
Iteration 0, fwhm = 1.961647; target 4.000000
++ Notice SUMA_estimate_FWHM_1dif (SUMA_GeomComp.c:6202):
Distribution of data is possibly random noise (p=0.223467)
Expect fwhm to be no different from 0
FWHM values up to 0.57(segments) or 0.57(mm)
are likely meaningless (at p=0.01) on this mesh.
-- Error SUMA_Chung_Smooth_07_toFWHM_dset (SUMA_GeomComp.c:5536):
Failed to get mean fwhm
-- Error SurfSmooth (SUMA_SurfSmooth.c:2074):
Failed to blur master data dset
----------------------------------------------------------------------------------------
After switching to the current 64 bit version we have, AFNI_2008_08_21_64bit, both the old and the new data set ran alright. However, the hemisphere that outputed the node warnings still did so, but finished anyway:
## SUMA_afnidset2sumadset:
Clearing dset ...
++ Notice SUMA_SigForFWHM (SUMA_GeomComp.c:5774):
Low Sigma/AvgLe, increased it to 0.500
Expected niter now: 60
Kernel Bandwidth / Average Edge Distance = 0.500000/1.001070 = 0.500535
Corresponding Kernel Numerator = 0.135335
++ Notice SurfSmooth (SUMA_SurfSmooth.c:2060):
Smoothing kernel bandwidth (sigma) = 0.500535
oo Warning SUMA_Chung_Smooth_Weights_07 (SUMA_GeomComp.c:2739):
Weights sum < 1.0f+1e-8 at node 50454
Mesh may be too coarse for kernel
bandwidth of 0.500535 in float precision.
Consider decreasing your number of iterations.
Future similar warnings are muted, but
a count is issued at the end.
oo Warning SUMA_Chung_Smooth_Weights_07 (SUMA_GeomComp.c:2764):
3 precision warnings out of 143977 nodes forming surface (0.00208 %).
Nodes with possible precision problems:
50454, 108889, 115869
Iteration 0, fwhm = 1.884437; target 4.000000
Iteration 10, fwhm = 3.461009; target 4.000000
++ Notice SUMA_WriteSmoothingRecord (SUMA_GeomComp.c:5403):
Writing FWHM progression history to sb_rh_r1_vrudssm4_surf_norm.niml.dset.1D.smrec ...
#Final smoothing parameters via master:
#Niter Sigma OutputFWHM
20 0.5005 -1.000
----------------------------------------------
so i guess i have two issues. 1. not sure why the 32 bit failed to run
2. why are nodes outside of the mask being considered?
Thanks,
Stephanie