Dear FreeSurfers,
We are running FreeSurfer on CentOS 4.2 and are currently in the process of updating from CentOS 4.2 to 4.3. To check stability, we ran the same data on multiple machines with the different versions of CentOS using recon-all. We also ran the same data twice on the same machine (on the following day, with no changes made). We actually found slight differences in the ROI stats for the data run twice on the same machine, (in addition to slight differences between machines.) While these differences do not seem to be significant, we are concerned about them because we do not know of any reason to expect random effects in our analysis.
here is an example of the differences found: [root@bwy-1-113mm subjects]$ grep -s caudalmid 2606/stats/lh.aparc.stats caudalmiddlefrontal 3873 2648 6807 2.155 0.735 0.119 0.023 41.641 3.388 [root@bwy-1-113mm subjects]$ grep -s caudalmid apr11run2606/stats/lh.aparc.stats caudalmiddlefrontal 3877 2644 6994 2.212 0.736 0.118 0.023 41.228 3.411
Does anyone know what may cause these differences? Is there any standard method you would use to test for stability other than looking at data such as these? Thanks! -Sasha
Disclaimer: The materials in this e-mail are private and may contain Protected Health Information. Please note that e-mail is not necessarily confidential or secure. Your use of e-mail constitutes your acknowledgment of these confidentiality and security limitations. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying, distribution, or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this e-mail in error, please immediately notify the sender via telephone or return e-mail.
Sasha,
Some of the algorithms rely on a random number generator, which is why you will see small variations in results when re-processing a subject. What we do for our own testing scheme is to pass a seed value to the binaries that rely on random number generation. You can do this also by creating a file called expert-options, containing the following:
mris_fix_topology -seed 1234 mris_sphere -seed 1234 mris_ca_label -seed 1234
and then include this parameter with recon-all:
-expert expert-options
which will cause the three binaries to seed their random number generator with the same number, thus ensuring an identical run from run to run.
It is not strictly necessary to do this, though, as the differences resulting from the random effects are small.
Nick
On Fri, 2006-04-14 at 14:39 -0400, Sasha Wolosin wrote:
Dear FreeSurfers,
We are running FreeSurfer on CentOS 4.2 and are currently in the process of updating from CentOS 4.2 to 4.3. To check stability, we ran the same data on multiple machines with the different versions of CentOS using recon-all. We also ran the same data twice on the same machine (on the following day, with no changes made). We actually found slight differences in the ROI stats for the data run twice on the same machine, (in addition to slight differences between machines.) While these differences do not seem to be significant, we are concerned about them because we do not know of any reason to expect random effects in our analysis.
here is an example of the differences found: [root@bwy-1-113mm subjects]$ grep -s caudalmid 2606/stats/lh.aparc.stats caudalmiddlefrontal 3873 2648 6807 2.155 0.735 0.119 0.023 41.641 3.388 [root@bwy-1-113mm subjects]$ grep -s caudalmid apr11run2606/stats/lh.aparc.stats caudalmiddlefrontal 3877 2644 6994 2.212 0.736 0.118 0.023 41.228 3.411
Does anyone know what may cause these differences? Is there any standard method you would use to test for stability other than looking at data such as these? Thanks! -Sasha
Disclaimer: The materials in this e-mail are private and may contain Protected Health Information. Please note that e-mail is not necessarily confidential or secure. Your use of e-mail constitutes your acknowledgment of these confidentiality and security limitations. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying, distribution, or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this e-mail in error, please immediately notify the sender via telephone or return e-mail. email message attachment On Fri, 2006-04-14 at 14:39 -0400, Sasha Wolosin wrote:
logo Oxford University Press | USA The World's Largest University Press: Excellence, Tradition,Innovation Online Catalog | Catalogs by Mail | Subscribe Today | About Us | Search Author Title ISBN # Keyword Corporate Info
Chodová Planá Coordinates: 49 50 N 12 44 E Annual beer production: 2,378,000 gallons (90,000 hectoliters) Geographic Theme: Location
Clever advertisers may have come up with the slogan: “Heineken refreshes the parts other beers cannot reach” for a Dutch company, but it took Czech ingenuity to turn a tagline into reality. Perhaps in an effort to cater to a local population that leads the world in per capita beer consumption, the Chodovar Family Brewery in Western Bohemia recently began offering a unique type of therapy in the form of Real Beer Baths. Bizarre as the notion of soaking in a hot tub of yeast and herbs may sound to some, the owners certainly chose their location well—the geothermal activity beneath this forested region of the country has drawn people to its salubrious spas and mineral springs for centuries.
Explore a little closer to home with the new Atlas of the United States.
View a sample page of our featured Regions in the News.
Marvel at the magnitude of Texas here.
For an example of Oxford's city mapping, click here.
Plus, browse all the other great titles that Oxford publishes by visiting our website.
To send a message to Ben Keene, email him at: atlas.us@oup.com.
Fun Stuff: Our very own Ben Keene created this quiz so you can test your geography knowlege. Take it now but don't look at the answers until you're finished! Oxford University Press | 198 Madison Avenue | New York, NY 10016 | Update Your Preferences | Unsubscribe | Customer Service
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Great, thanks!
Nick Schmansky nicks@nmr.mgh.harvard.edu 4/14/2006 3:29 pm >>>
Sasha,
Some of the algorithms rely on a random number generator, which is why you will see small variations in results when re-processing a subject. What we do for our own testing scheme is to pass a seed value to the binaries that rely on random number generation. You can do this also by creating a file called expert-options, containing the following:
mris_fix_topology -seed 1234 mris_sphere -seed 1234 mris_ca_label -seed 1234
and then include this parameter with recon-all:
-expert expert-options
which will cause the three binaries to seed their random number generator with the same number, thus ensuring an identical run from run to run.
It is not strictly necessary to do this, though, as the differences resulting from the random effects are small.
Nick
On Fri, 2006-04-14 at 14:39 -0400, Sasha Wolosin wrote:
Dear FreeSurfers,
We are running FreeSurfer on CentOS 4.2 and are currently in the process of updating from CentOS 4.2 to 4.3. To check stability, we ran the same data on multiple machines with the different versions of CentOS using recon-all. We also ran the same data twice on the same machine (on the following day, with no changes made). We actually found slight differences in the ROI stats for the data run twice on the same machine, (in addition to slight differences between machines.) While these differences do not seem to be significant, we are concerned about them because we do not know of any reason to expect random effects in our analysis.
here is an example of the differences found: [root@bwy-1-113mm subjects]$ grep -s caudalmid 2606/stats/lh.aparc.stats caudalmiddlefrontal 3873 2648 6807 2.155 0.735 0.119 0.023 41.641 3.388 [root@bwy-1-113mm subjects]$ grep -s caudalmid apr11run2606/stats/lh.aparc.stats caudalmiddlefrontal 3877 2644 6994 2.212 0.736 0.118 0.023 41.228 3.411
Does anyone know what may cause these differences? Is there any standard method you would use to test for stability other than looking at data such as these? Thanks! -Sasha
Disclaimer: The materials in this e-mail are private and may contain Protected Health Information. Please note that e-mail is not necessarily confidential or secure. Your use of e-mail constitutes your acknowledgment of these confidentiality and security limitations. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying, distribution, or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this e-mail in error, please immediately notify the sender via telephone or return e-mail. email message attachment On Fri, 2006-04-14 at 14:39 -0400, Sasha Wolosin wrote:
logo Oxford University Press | USA The World's Largest University Press: Excellence, Tradition,Innovation Online Catalog | Catalogs by Mail | Subscribe Today | About Us | Search Author Title ISBN # Keyword Corporate Info
Chodová Planá Coordinates: 49 50 N 12 44 E Annual beer production: 2,378,000 gallons (90,000 hectoliters) Geographic Theme: Location
Clever advertisers may have come up with the slogan: "Heineken refreshes the parts other beers cannot reach" for a Dutch company, but it took Czech ingenuity to turn a tagline into reality. Perhaps in an effort to cater to a local population that leads the world in per capita beer consumption, the Chodovar Family Brewery in Western Bohemia recently began offering a unique type of therapy in the form of Real Beer Baths. Bizarre as the notion of soaking in a hot tub of yeast and herbs may sound to some, the owners certainly chose their location well*the geothermal activity beneath this forested region of the country has drawn people to its salubrious spas and mineral springs for centuries.
Explore a little closer to home with the new Atlas of the United States.
View a sample page of our featured Regions in the News.
Marvel at the magnitude of Texas here.
For an example of Oxford's city mapping, click here.
Plus, browse all the other great titles that Oxford publishes by visiting our website.
To send a message to Ben Keene, email him at: atlas.us@oup.com.
Fun Stuff: Our very own Ben Keene created this quiz so you can test your geography knowlege. Take it now but don't look at the answers until you're finished! Oxford University Press | 198 Madison Avenue | New York, NY 10016 | Update Your Preferences | Unsubscribe | Customer Service
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Nick Schmansky -
Sasha Wolosin