Hi, 

I am trying to use the longitudinal pipeline of the “hippocampal subfields and nuclei of amygdala”. Keep getting error messages when running it longitudinal but with no error message running it cross-sectional. I want to know if there is any way to fix the error and to run a longitudinal pipeline successfully. 

I am using Freesurfer/dev and matlab_runtime/2014b.

I got two timepoints for each subject and finished recon-all longitudinal with no errors. 

1) First Try

segmentHA_T1_long.sh SUBJ_base

>> run for about 15 mins and exited with error message:

“Error using kvlGEMSMatlab

/autofs/space/panamint_005/users/iglesias/software/freesurfer/GEMS2/kvlAtlasMeshDeformationOptimizer.cxx:327:

itk::ERROR: AtlasMeshDeformationLBFGSOptimizer(0x2ac1fead7740): search direction is not a descent direction!

Error in kvlStepOptimizer (line 11)

Error in SegmentSubfieldsT1Longitudinal”

>> no mri/lh.amygNucVolumes-T1.v21 and mri/lh.hippoSfVolumes-T1.v21 generated. 

2) Second Try

segmentHA_T1.sh SUBJ1_timepoint1.long.SUBJ1_base

segmentHA_T1.sh SUBJ1_timepoint2.long.SUBJ1_base

>> run for about 16 mins, most successful but a few with errors:

“Error in segmentSubjectT1_autoEstimateAlveusML (line 1539)”

Or

>>

“Open file failed because file '/tmp/MCR_2841219388/.mcrCache8.4/segmen0/autofs/space/panamint_005/users/iglesias/software/freesurfer.GEMS2.terrier/bin/kvlSetMeshCollectionPositions.m' has invalid content. Details: 'file is closed.failbit is set.'

The file

 "/tmp/MCR_2841219388/.mcrCache8.4/segmen0/autofs/space/panamint_005/users/iglesias/software/freesurfer.GEMS2.terrier/bin/kvlSetMeshCollectionPositions.m"

cannot be executed.

Error in segmentSubjectT1_autoEstimateAlveusML (line 538)”

 

>> successful done subject or hemisphere generated mri/lh.amygNucVolumes-T1.v21 and mri/lh.hippoSfVolumes-T1.v21.

 

3) Third Try cross-sectional

segmentHA_T1.sh SUBJ1_timepoint1

segmentHA_T1.sh SUBJ1_timepoint2

>> run for about 15 mins, all successful with no errors:

>> all subjects have mri/lh.amygNucVolumes-T1.v21 and mri/lh.hippoSfVolumes-T1.v21 generated. 

Manual inspection with freeview, it seems fine with all successful longitudinal and cross sectional subfield segmentation outputs mgz.

I wonder if I can fix first try and second try errors so that longitudinal output can be generated.

 

Thank you for your help!

Best,

Qi

--