Hi All,
We have recently uncovered problems with version 5.2. Based on these problems, we are recommending that people stop using 5.2, mainly in regards to usage of the surface data. A new version, 5.3, will be released within two weeks, and a beta version for each platform will be available within a couple days.
There were two main problems that were uncovered. First, a change was made that was only meant to affect the white surface placement in high-resolution data sets, but unfortunately is being applied to all data sets. This causes the white surface to contract a small amount. This causes the surface area to be smaller and cortex to be measured as thicker relative to version 5.1. Visually, the surfaces do not necessarily look less accurate, but we feel that the systematic differences with previous versions are ill-advised without a clear reason for the change.
Second, a change made to the placement of the pial surface in order to improve the positioning based on T2/FLAIR that caused it to extend too far in some data sets during the T1-based pial surface deformation. This unfortunately became the default behavior, which was not intended. None of the dozens of test data sets we ran displayed this problem. The behavior was reported by other users which led us to the bug.
In general, we rigorously test each version. However, our systematic comparison of a new version with previous versions which would have caught these problems was conducted prior to these changes. We are now tightening and regimenting our testing procedures to ensure this doesnt happen again. Note that there are no problems with the volumetric segmentation (aseg) in version 5.2.
We realize that this will cause a substantial inconvenience for users. We have isolated and fixed the problems and have run two rounds of alpha testing and have one more to go, after which we will release a beta version in a couple of days to anyone that wants it.
sorry for the problems.
Bruce
Hi Bruce,
Thanks a lot for the info. Would it mean we have to completely reprocess the data processed with 5.2? Or at which point of the preprocessing pipeline we should step in to fix the problems of 5.2...? I just have finished to process about 200 data sets with 5.2...
Best wishes,
Luke
Am 03.05.2013 um 02:58 schrieb "Bruce Fischl" fischl@nmr.mgh.harvard.edu:
Hi All,
We have recently uncovered problems with version 5.2. Based on these problems, we are recommending that people stop using 5.2, mainly in regards to usage of the surface data. A new version, 5.3, will be released within two weeks, and a beta version for each platform will be available within a couple days.
There were two main problems that were uncovered. First, a change was
made
that was only meant to affect the white surface placement in high-resolution data sets, but unfortunately is being applied to all data sets. This causes the white surface to contract a small amount. This
causes
the surface area to be smaller and cortex to be measured as thicker relative to version 5.1. Visually, the surfaces do not necessarily look less accurate, but we feel that the systematic differences with previous versions are ill-advised without a clear reason for the change.
Second, a change made to the placement of the pial surface in order to improve the positioning based on T2/FLAIR that caused it to extend too
far
in some data sets during the T1-based pial surface deformation. This unfortunately became the default behavior, which was not intended. None
of
the dozens of test data sets we ran displayed this problem. The behavior was reported by other users which led us to the bug.
In general, we rigorously test each version. However, our systematic comparison of a new version with previous versions which would have
caught
these problems was conducted prior to these changes. We are now
tightening
and regimenting our testing procedures to ensure this doesnt happen
again.
Note that there are no problems with the volumetric segmentation (aseg) in version 5.2.
We realize that this will cause a substantial inconvenience for users. We have isolated and fixed the problems and have run two rounds of alpha testing and have one more to go, after which we will release a beta
version
in a couple of days to anyone that wants it.
sorry for the problems.
Bruce _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
The information in this e-mail is intended only for the person to whom it
is
addressed. If you believe this e-mail was sent to you in error and the
contains patient information, please contact the Partners Compliance
HelpLine at
http://www.partners.org/complianceline . If the e-mail was sent to you in
error
but does not contain patient information, please contact the sender and
properly
dispose of the e-mail.
Luke,
It is just mris_make_surfaces that has changed (and a bug was found and fixed in mri_mask, but that bug only affected certain edits, and you would get a blank volume if you encountered that bug) so you can do the following to patch 5.2 to match 5.3:
get mris_make_surfaces (and mri_mask) for your platform from here:
ftp://surfer.nmr.mgh.harvard.edu/pub/dist/freesurfer/misc/
then run:
recon-all -s <subjid> \ -white \ -autorecon3 \ -nosphere \ -nosurfreg
which should take about an hour. This regenerates the white and pial surfaces, and creates new stats files.
Nick
Hi Bruce,
Thanks a lot for the info. Would it mean we have to completely reprocess the data processed with 5.2? Or at which point of the preprocessing pipeline we should step in to fix the problems of 5.2...? I just have finished to process about 200 data sets with 5.2...
Best wishes,
Luke
Am 03.05.2013 um 02:58 schrieb "Bruce Fischl" fischl@nmr.mgh.harvard.edu:
Hi All,
We have recently uncovered problems with version 5.2. Based on these problems, we are recommending that people stop using 5.2, mainly in regards to usage of the surface data. A new version, 5.3, will be released within two weeks, and a beta version for each platform will be available within a couple days.
There were two main problems that were uncovered. First, a change was
made
that was only meant to affect the white surface placement in high-resolution data sets, but unfortunately is being applied to all data sets. This causes the white surface to contract a small amount. This
causes
the surface area to be smaller and cortex to be measured as thicker relative to version 5.1. Visually, the surfaces do not necessarily look less accurate, but we feel that the systematic differences with previous versions are ill-advised without a clear reason for the change.
Second, a change made to the placement of the pial surface in order to improve the positioning based on T2/FLAIR that caused it to extend too
far
in some data sets during the T1-based pial surface deformation. This unfortunately became the default behavior, which was not intended. None
of
the dozens of test data sets we ran displayed this problem. The behavior was reported by other users which led us to the bug.
In general, we rigorously test each version. However, our systematic comparison of a new version with previous versions which would have
caught
these problems was conducted prior to these changes. We are now
tightening
and regimenting our testing procedures to ensure this doesnt happen
again.
Note that there are no problems with the volumetric segmentation (aseg) in version 5.2.
We realize that this will cause a substantial inconvenience for users. We have isolated and fixed the problems and have run two rounds of alpha testing and have one more to go, after which we will release a beta
version
in a couple of days to anyone that wants it.
sorry for the problems.
Bruce _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
The information in this e-mail is intended only for the person to whom it
is
addressed. If you believe this e-mail was sent to you in error and the
contains patient information, please contact the Partners Compliance
HelpLine at
http://www.partners.org/complianceline . If the e-mail was sent to you in
error
but does not contain patient information, please contact the sender and
properly
dispose of the e-mail. _______________________________________________
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Hi Luke
Nick will post a command line you can run that should take about 2 hours/dataset. Rerunning shouldn't be necessary. Sorry for the trouble, I really, really wish we had more engineering resources.
Bruce
On Sat, 4 May 2013, Lukas.Scheef@ukb.uni-bonn.de wrote:
Hi Bruce,
Thanks a lot for the info. Would it mean we have to completely reprocess the data processed with 5.2? Or at which point of the preprocessing pipeline we should step in to fix the problems of 5.2...? I just have finished to process about 200 data sets with 5.2...
Best wishes,
Luke
Am 03.05.2013 um 02:58 schrieb "Bruce Fischl" fischl@nmr.mgh.harvard.edu:
Hi All,
We have recently uncovered problems with version 5.2. Based on these problems, we are recommending that people stop using 5.2, mainly in regards to usage of the surface data. A new version, 5.3, will be released within two weeks, and a beta version for each platform will be available within a couple days.
There were two main problems that were uncovered. First, a change was made that was only meant to affect the white surface placement in high-resolution data sets, but unfortunately is being applied to all data sets. This causes the white surface to contract a small amount. This
causes
the surface area to be smaller and cortex to be measured as thicker relative to version 5.1. Visually, the surfaces do not necessarily look less accurate, but we feel that the systematic differences with previous versions are ill-advised without a clear reason for the change.
Second, a change made to the placement of the pial surface in order to improve the positioning based on T2/FLAIR that caused it to extend too far in some data sets during the T1-based pial surface deformation. This unfortunately became the default behavior, which was not intended. None of the dozens of test data sets we ran displayed this problem. The behavior was reported by other users which led us to the bug.
In general, we rigorously test each version. However, our systematic comparison of a new version with previous versions which would have caught these problems was conducted prior to these changes. We are now tightening and regimenting our testing procedures to ensure this doesnt happen again. Note that there are no problems with the volumetric segmentation (aseg) in version 5.2.
We realize that this will cause a substantial inconvenience for users. We have isolated and fixed the problems and have run two rounds of alpha testing and have one more to go, after which we will release a beta
version
in a couple of days to anyone that wants it.
sorry for the problems.
Bruce _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
The information in this e-mail is intended only for the person to whom it
is
addressed. If you believe this e-mail was sent to you in error and the
contains patient information, please contact the Partners Compliance
HelpLine at
http://www.partners.org/complianceline . If the e-mail was sent to you in
error
but does not contain patient information, please contact the sender and
properly
dispose of the e-mail.
freesurfer@nmr.mgh.harvard.edu
-
Bruce Fischl -
Lukas.Scheef＠ukb.uni-bonn.de -
Nick Schmansky