Hi Soichi

that is too big a defect to expect it to fix - more than half the number of vertices in a typical hemisphere. It means something is badly wrong that you need to correct prior to the topo fixer. You can look at our wiki for this kind of thing, but I would start by looking at the rh.orig.nofix and the rh.inflated.nofix. Something big is wrong, like large parts of the skull attached to the surface or something like that

cheers Bruce

On Sun, 24 Mar 2019, Soichi Hayashi wrote:

External Email - Use Caution

Hello. I am trying to troubleshoot an issue where recon_all is not finishing after >24 hours. It's getting stuck when it tries to run the following command.

$ mris_fix_topology -rusage ./output/touch/rusage.mris_fix_topology.rh.dat -mgz -sphere qsphere.nofix -ga -seed 1234 output rh  INFO: assuming .mgz format reading spherical homeomorphism from 'qsphere.nofix' using genetic algorithm with optimized parameters setting seed for random number genererator to 1234

---

Topology Correction Parameters retessellation mode:           genetic search number of patches/generation : 10 number of generations :        10 surface mri loglikelihood coefficient :         1.0 volume mri loglikelihood coefficient :          10.0 normal dot loglikelihood coefficient :          1.0 quadratic curvature loglikelihood coefficient : 1.0 volume resolution :                             2 eliminate vertices during search :              1 initial patch selection :                       1 select all defect vertices :                    0 ordering dependant retessellation:              0 use precomputed edge table :                    0 smooth retessellated patch :                    2 match retessellated patch :                     1 verbose mode :                                  0

---

$Id: mris_fix_topology.c,v 1.50.2.1 2016/10/27 22:25:58 zkaufman Exp $   $Id: mrisurf.c,v 1.781.2.6 2016/12/27 16:47:14 zkaufman Exp $ reading input surface /home/hayashis/Downloads/output/surf/rh.qsphere.nofix... before topology correction, eno=-696 (nv=191488, nf=384368, ne=576552, g=349) reading brain volume from brain... reading wm segmentation from wm... using quasi-homeomorphic spherical map to tessellate cortical surface...

Correction of the Topology Finding true center and radius of Spherical Surface...done Surface centered at (0,0,0) with radius 100.0 in 9 iterations marking ambiguous vertices... 133741 ambiguous faces found in tessellation segmenting defects... 63 defects found, arbitrating ambiguous regions... analyzing neighboring defects...       -merging segment 16 into 0       -merging segment 23 into 0       -merging segment 48 into 0       -merging segment 32 into 22       -merging segment 49 into 47       -merging segment 54 into 51 57 defects to be corrected  0 vertices coincident Computing Initial Surface Statistics       -face       loglikelihood: -9.3123  (-4.6561)       -vertex     loglikelihood: -6.5427  (-3.2713)       -normal dot loglikelihood: -3.5794  (-3.5794)       -quad curv  loglikelihood: -6.0564  (-3.0282)       Total Loglikelihood : -25.4907

CORRECTING DEFECT 0 (vertices=65141, convex hull=8287, v0=517)

---

After displaying the last line, the process seems to be stuck running with 100% cpu (single core) for >24 hours until our batch scheduler kills it. I think the issue is with my input data, but it looks clean and well ACPC aligned, and I don't know what's wrong with it. The input image dimensions are 182 x 218 x 182.

Does anyone have any suggestion on how to troubleshoot this problem?

Thank you,

Soichi Hayashi Indiana University