recon-all fails complaining about missing T1 data

[wmvanvliet]

When trying to run the freesurfer pipeline in 01-run_anatomy.ipy (uncommenting the appropriate lines), it complains with:

16:57 [master] gyrus:~/data/mne-biomag-group-demo/scripts/processing> ipython 01-run_anatomy.ipy
processing sub001
recon-all -all -s sub001 -i /data/marijn/gdw_faces/ds117/sub001/anatomy/highres001.nii.gz > /data/marijn/gdw_faces/ds117/sub001/my-recon-all.txt
reading mef05_0.mgz...TE = 1.85, TR = 20.00, flip angle = 5.00
reading mef05_1.mgz...TE = 4.15, TR = 20.00, flip angle = 5.00
reading mef05_2.mgz...TE = 6.45, TR = 20.00, flip angle = 5.00
reading mef05_3.mgz...TE = 8.75, TR = 20.00, flip angle = 5.00
reading mef05_4.mgz...TE = 11.05, TR = 20.00, flip angle = 5.00
reading mef05_5.mgz...TE = 13.35, TR = 20.00, flip angle = 5.00
reading mef05_6.mgz...TE = 15.65, TR = 20.00, flip angle = 5.00
reading mef30_0.mgz...TE = 1.85, TR = 20.00, flip angle = 30.00
reading mef30_1.mgz...TE = 4.15, TR = 20.00, flip angle = 30.00
reading mef30_2.mgz...TE = 6.45, TR = 20.00, flip angle = 30.00
reading mef30_3.mgz...TE = 8.75, TR = 20.00, flip angle = 30.00
reading mef30_4.mgz...TE = 11.05, TR = 20.00, flip angle = 30.00
reading mef30_5.mgz...TE = 13.35, TR = 20.00, flip angle = 30.00
reading mef30_6.mgz...TE = 15.65, TR = 20.00, flip angle = 30.00
using 2 FLASH volumes to estimate tissue parameters.
parameter rms = 0.202
non-equal flip_angle found for the volume 1.
Flip_angle is set to zero.
TR = 20.00, TE = 1.85, Flip_angle = 0.00 are used
writing PD estimates to parameter_maps/PD.mgz...
writing T1 estimates to parameter_maps/T1.mgz...
writing residual sse to parameter_maps/sse.mgz...
writing synthetic images to parameter_maps/flash5.mgz...
writing registration matrix to parameter_maps/flash5.lta...
writing synthetic images to parameter_maps/flash30.mgz...
writing registration matrix to parameter_maps/flash30.lta...
estimating T2* with 2 different acquisitions, each with 7 echoes...
writing T2star estimates to parameter_maps/T2star.mgz...
parameter estimation took 1 minutes and 27 seconds.



$Id: fsl_rigid_register,v 1.31 2011/03/02 20:16:39 nicks Exp $
Tue Jul  4 16:59:30 EEST 2017
/data/marijn/gdw_faces/subjects/sub001/mri/flash/parameter_maps/flash5_reg.mgz
--------------------------------------
/data/marijn/mne-biomag-group-demo/scripts/processing
mri_convert /data/marijn/gdw_faces/subjects/sub001/mri/T1 /data/marijn/gdw_faces/subjects/sub001/mri/flash/parameter_maps/fsl_rigid_register.28198/refvol.fsl_rigid_register.nii
mri_convert /data/marijn/gdw_faces/subjects/sub001/mri/T1 /data/marijn/gdw_faces/subjects/sub001/mri/flash/parameter_maps/fsl_rigid_register.28198/refvol.fsl_rigid_register.nii 
ERROR: file /data/marijn/gdw_faces/subjects/sub001/mri/T1 does not exist

Where would the T1 data come from and why can't Freesurfer find it?

[jasmainak]

Thanks for the bug report. Let me try to reproduce first ...

[jasmainak]

@agramfort any clues? We are able to reproduce the problem. I guess you ran recon-all on the data from the ftp site which happens to be different from the data on OpenfMRI?

[agramfort]

I took it from openfmri with the fetch script online

[jasmainak]

@wmvanvliet me and @jaeilepp couldn't figure out what's wrong. We trust you'll figure it out and make a PR :) Otherwise, we'll get back to it some time later.

[jaeilepp]

This is probably related to ContinuumIO/anaconda-issues#152.
For me, the recon-all segfaulted because readline threw errors. conda remove --force readline fixed it for me.

[wmvanvliet]

for me, it's not a segfault (I have done conda remove --force readline, because I use the rpy2 package regularly). Freesurfer just can't find the T1 images, and neither can I :)

[jaeilepp]

Yeah, I had the same problem. This is probably because the recon-all did not finish without errors, so it never gets created. I believe fsl_rigid_register is called by make_flash_bem?

[wmvanvliet]

aaah, I will double check for errors other than T1 not being found...

[jasmainak]

@wmvanvliet you should check the recon-all log file carefully what it says.

[wmvanvliet]

Found it. The environment was not set up correctly. Strangely, no error messages were written to the log file.