64 bits and 32 bits
Locked
 
|
Hello,
We use ImageJ to quantify the muscle surface area of our patients at the third lumbar vertebra. I use a 64bits version of ImageJ but some of my colleagues use a 32 bits version and our results, for a same image, differ of at least 25 cm2, even if I do the reading on both computers. Does anyone know where that problem comes from and what version I should trust? Kind regards, Guillaume Ulmann Praticien Attaché Service de Biochimie Hôpital Cochin - HUPC - APHP -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html |
|
|
On Friday, 20 October 2017 09:23:32 BST you wrote:
> We use ImageJ to quantify the muscle surface area of our patients at the > third lumbar vertebra. I use a 64bits version of ImageJ but some of my > colleagues use a 32 bits version and our results, for a same image, differ > of at least 25 cm2, even if I do the reading on both computers. Does anyone > know where that problem comes from and what version I should trust? Are both versions of the jar file the same? Are the Java versions the same? Update the jar for *both* to the the Daily Build via the Help menu and see if that changes anything. Cheers Gabriel -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html |
|
|
In reply to this post by ULMANN Guillaume
Bonjour Guillaume,
I'm pretty sure that the described effect is not due to 32/64bit-versions of ImageJ! I guess that the scale setting is different on the computers or that some other options are set differently. I would first check "Show Info..." for the same image on the different computers, then check the settings made in "Edit >> Options". Good luck Herbie :::::::::::::::::::::::::::::::::::::::::::::: Am 20.10.17 um 10:23 schrieb ULMANN Guillaume: > Hello, > > We use ImageJ to quantify the muscle surface area of our patients at > the third lumbar vertebra. I use a 64bits version of ImageJ but some > of my colleagues use a 32 bits version and our results, for a same > image, differ of at least 25 cm2, even if I do the reading on both > computers. Does anyone know where that problem comes from and what > version I should trust? > > Kind regards, > > Guillaume Ulmann Praticien Attaché Service de Biochimie Hôpital > Cochin - HUPC - APHP > > > -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html > -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html |
|
|
In reply to this post by ULMANN Guillaume
Hi Guillaume,
there should be absolutely no difference between the results obtained with the 32-bit and 64-bit versions of Java. All calculations are done with the same accuracy, regardless of the Java version. If you see a difference, it could be: - different protocols or other differences how different people work - different versions of plugins or macros used - different settings of ImageJ preferences (Edit>Options>Reset restores them to the defaults) - different settings of preferences used by some plugin (not only the ImageJ core, also plugins can save settings in IJ_Prefs.txt. In case of doubt, copy the IJ_Prefs.txt to the other computers (usually in the home directory in a .imagej hidden folder, on Macs in the user's Library/Preferences) - maybe some settings file of a plugin, if there is one - different versions of ImageJ (Help>Update ImageJ) - very unlikely, a bug that is effective only for your 64-bit system or only for the 32-bit systems. (for my feeling, the most likely points are at the top, the least likely at the bottom) Michael ________________________________________________________________ On 20/10/2017 10:23, ULMANN Guillaume wrote: > Hello, > > We use ImageJ to quantify the muscle surface area of our patients at the third lumbar vertebra. > I use a 64bits version of ImageJ but some of my colleagues use a 32 bits version and our results, for a same image, differ of at least 25 cm2, even if I do the reading on both computers. > Does anyone know where that problem comes from and what version I should trust? > > Kind regards, > > Guillaume Ulmann > Praticien Attaché > Service de Biochimie > Hôpital Cochin - HUPC - APHP > -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html |
|
|
Dear Michael and Guillaume,
On top reason, I would as well add a difference within the reading of the scaling calibration settings. Maybe one version is using by default an importer that is correctly reading the meta data as the other version is using a reader where the meta data are not well interpreted. Have a nice week-end, Philippe -----Message d'origine----- De : ImageJ Interest Group [mailto:[hidden email]] De la part de Michael Schmid Envoyé : vendredi 20 octobre 2017 11:05 À : [hidden email] Objet : Re: 64 bits and 32 bits Hi Guillaume, there should be absolutely no difference between the results obtained with the 32-bit and 64-bit versions of Java. All calculations are done with the same accuracy, regardless of the Java version. If you see a difference, it could be: - different protocols or other differences how different people work - different versions of plugins or macros used - different settings of ImageJ preferences (Edit>Options>Reset restores them to the defaults) - different settings of preferences used by some plugin (not only the ImageJ core, also plugins can save settings in IJ_Prefs.txt. In case of doubt, copy the IJ_Prefs.txt to the other computers (usually in the home directory in a .imagej hidden folder, on Macs in the user's Library/Preferences) - maybe some settings file of a plugin, if there is one - different versions of ImageJ (Help>Update ImageJ) - very unlikely, a bug that is effective only for your 64-bit system or only for the 32-bit systems. (for my feeling, the most likely points are at the top, the least likely at the bottom) Michael ________________________________________________________________ On 20/10/2017 10:23, ULMANN Guillaume wrote: > Hello, > > We use ImageJ to quantify the muscle surface area of our patients at the third lumbar vertebra. > I use a 64bits version of ImageJ but some of my colleagues use a 32 bits version and our results, for a same image, differ of at least 25 cm2, even if I do the reading on both computers. > Does anyone know where that problem comes from and what version I should trust? > > Kind regards, > > Guillaume Ulmann > Praticien Attaché > Service de Biochimie > Hôpital Cochin - HUPC - APHP > -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html |
|
|
Thanks everyone for your answer.
It was indeed different preferences on the two computers. I did a reset on both and now the results are matching. Have a nice weekend, Guillaume -----Message d'origine----- De : ImageJ Interest Group [mailto:[hidden email]] De la part de Philippe CARL Envoyé : vendredi 20 octobre 2017 11:17 À : [hidden email] Objet : Re: 64 bits and 32 bits Dear Michael and Guillaume, On top reason, I would as well add a difference within the reading of the scaling calibration settings. Maybe one version is using by default an importer that is correctly reading the meta data as the other version is using a reader where the meta data are not well interpreted. Have a nice week-end, Philippe -----Message d'origine----- De : ImageJ Interest Group [mailto:[hidden email]] De la part de Michael Schmid Envoyé : vendredi 20 octobre 2017 11:05 À : [hidden email] Objet : Re: 64 bits and 32 bits Hi Guillaume, there should be absolutely no difference between the results obtained with the 32-bit and 64-bit versions of Java. All calculations are done with the same accuracy, regardless of the Java version. If you see a difference, it could be: - different protocols or other differences how different people work - different versions of plugins or macros used - different settings of ImageJ preferences (Edit>Options>Reset restores them to the defaults) - different settings of preferences used by some plugin (not only the ImageJ core, also plugins can save settings in IJ_Prefs.txt. In case of doubt, copy the IJ_Prefs.txt to the other computers (usually in the home directory in a .imagej hidden folder, on Macs in the user's Library/Preferences) - maybe some settings file of a plugin, if there is one - different versions of ImageJ (Help>Update ImageJ) - very unlikely, a bug that is effective only for your 64-bit system or only for the 32-bit systems. (for my feeling, the most likely points are at the top, the least likely at the bottom) Michael ________________________________________________________________ On 20/10/2017 10:23, ULMANN Guillaume wrote: > Hello, > > We use ImageJ to quantify the muscle surface area of our patients at the third lumbar vertebra. > I use a 64bits version of ImageJ but some of my colleagues use a 32 bits version and our results, for a same image, differ of at least 25 cm2, even if I do the reading on both computers. > Does anyone know where that problem comes from and what version I should trust? > > Kind regards, > > Guillaume Ulmann > Praticien Attaché > Service de Biochimie > Hôpital Cochin - HUPC - APHP > -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html |
|
|
Hi Guillaume,
to avoid similar problems in the future, in case you can identify which of the settings causes the difference between different computers, you could set that value in the StartupMacros file and distribute it on all computers that are used for this task. (use Plugins>Macros>Record to get the macro command corresponding to an action performed manually) Here are some options that are stored in the Prefs and may affect the quantitative result of image operations, also for operations performed by macros (the list is probably not complete, but these come into my mind): - Edit>Options>Conversions ("scale when converting" and "weighted RGB") - Process>Binary>Options ("black background" and "pad edges") - For line profiles, in Edit>Options>Plots the 'sub-pixel resolution' - In Edit>Options>Misc, the 'divide by zero value' Michael ________________________________________________________________ On 20/10/2017 11:40, ULMANN Guillaume wrote: > Thanks everyone for your answer. > It was indeed different preferences on the two computers. I did a reset on both and now the results are matching. > > Have a nice weekend, > > Guillaume > > > -----Message d'origine----- > De : ImageJ Interest Group [mailto:[hidden email]] De la part de Philippe CARL > Envoyé : vendredi 20 octobre 2017 11:17 > À : [hidden email] > Objet : Re: 64 bits and 32 bits > > Dear Michael and Guillaume, > On top reason, I would as well add a difference within the reading of the scaling calibration settings. > Maybe one version is using by default an importer that is correctly reading the meta data as the other version is using a reader where the meta data are not well interpreted. > Have a nice week-end, > Philippe > > -----Message d'origine----- > De : ImageJ Interest Group [mailto:[hidden email]] De la part de Michael Schmid Envoyé : vendredi 20 octobre 2017 11:05 À : [hidden email] Objet : Re: 64 bits and 32 bits > > Hi Guillaume, > > there should be absolutely no difference between the results obtained with the 32-bit and 64-bit versions of Java. All calculations are done with the same accuracy, regardless of the Java version. > If you see a difference, it could be: > > - different protocols or other differences how different people work > - different versions of plugins or macros used > - different settings of ImageJ preferences (Edit>Options>Reset restores them to the defaults) > - different settings of preferences used by some plugin (not only the ImageJ core, also plugins can save settings in IJ_Prefs.txt. In case of doubt, copy the IJ_Prefs.txt to the other computers (usually in the home directory in a .imagej hidden folder, on Macs in the user's > Library/Preferences) > - maybe some settings file of a plugin, if there is one > - different versions of ImageJ (Help>Update ImageJ) > - very unlikely, a bug that is effective only for your 64-bit system or only for the 32-bit systems. > > (for my feeling, the most likely points are at the top, the least likely at the bottom) > > Michael > ________________________________________________________________ > On 20/10/2017 10:23, ULMANN Guillaume wrote: >> Hello, >> >> We use ImageJ to quantify the muscle surface area of our patients at the third lumbar vertebra. >> I use a 64bits version of ImageJ but some of my colleagues use a 32 bits version and our results, for a same image, differ of at least 25 cm2, even if I do the reading on both computers. >> Does anyone know where that problem comes from and what version I should trust? >> >> Kind regards, >> >> Guillaume Ulmann >> Praticien Attaché >> Service de Biochimie >> Hôpital Cochin - HUPC - APHP >> > > -- > ImageJ mailing list: http://imagej.nih.gov/ij/list.html > > -- > ImageJ mailing list: http://imagej.nih.gov/ij/list.html > > -- > ImageJ mailing list: http://imagej.nih.gov/ij/list.html > -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html |
|
Hi everyone,
There is an entry on the Troubleshooting page discussing the problem of results differing across machines or environments: https://imagej.net/Troubleshooting#The_same_plugin_gives_different_results_on_different_machines.21 If you notice anything there which can be improved, it would be very helpful if you could improve it! Thanks, Curtis -- Curtis Rueden LOCI software architect - https://loci.wisc.edu/software ImageJ2 lead, Fiji maintainer - https://imagej.net/User:Rueden Did you know ImageJ has a forum? http://forum.imagej.net/ On Fri, Oct 20, 2017 at 6:15 AM, Michael Schmid <[hidden email]> wrote: > Hi Guillaume, > > to avoid similar problems in the future, in case you can identify which of > the settings causes the difference between different computers, you could > set that value in the StartupMacros file and distribute it on all computers > that are used for this task. > (use Plugins>Macros>Record to get the macro command corresponding to an > action performed manually) > > Here are some options that are stored in the Prefs and may affect the > quantitative result of image operations, also for operations performed by > macros (the list is probably not complete, but these come into my mind): > > - Edit>Options>Conversions ("scale when converting" and "weighted RGB") > - Process>Binary>Options ("black background" and "pad edges") > - For line profiles, in Edit>Options>Plots the 'sub-pixel resolution' > - In Edit>Options>Misc, the 'divide by zero value' > > > Michael > ________________________________________________________________ > > On 20/10/2017 11:40, ULMANN Guillaume wrote: > >> Thanks everyone for your answer. >> It was indeed different preferences on the two computers. I did a reset >> on both and now the results are matching. >> >> Have a nice weekend, >> >> Guillaume >> >> >> -----Message d'origine----- >> De : ImageJ Interest Group [mailto:[hidden email]] De la part de >> Philippe CARL >> Envoyé : vendredi 20 octobre 2017 11:17 >> À : [hidden email] >> Objet : Re: 64 bits and 32 bits >> >> Dear Michael and Guillaume, >> On top reason, I would as well add a difference within the reading of the >> scaling calibration settings. >> Maybe one version is using by default an importer that is correctly >> reading the meta data as the other version is using a reader where the meta >> data are not well interpreted. >> Have a nice week-end, >> Philippe >> >> -----Message d'origine----- >> De : ImageJ Interest Group [mailto:[hidden email]] De la part de >> Michael Schmid Envoyé : vendredi 20 octobre 2017 11:05 À : >> [hidden email] Objet : Re: 64 bits and 32 bits >> >> Hi Guillaume, >> >> there should be absolutely no difference between the results obtained >> with the 32-bit and 64-bit versions of Java. All calculations are done with >> the same accuracy, regardless of the Java version. >> If you see a difference, it could be: >> >> - different protocols or other differences how different people work >> - different versions of plugins or macros used >> - different settings of ImageJ preferences (Edit>Options>Reset restores >> them to the defaults) >> - different settings of preferences used by some plugin (not only the >> ImageJ core, also plugins can save settings in IJ_Prefs.txt. In case of >> doubt, copy the IJ_Prefs.txt to the other computers (usually in the home >> directory in a .imagej hidden folder, on Macs in the user's >> Library/Preferences) >> - maybe some settings file of a plugin, if there is one >> - different versions of ImageJ (Help>Update ImageJ) >> - very unlikely, a bug that is effective only for your 64-bit system or >> only for the 32-bit systems. >> >> (for my feeling, the most likely points are at the top, the least likely >> at the bottom) >> >> Michael >> ________________________________________________________________ >> On 20/10/2017 10:23, ULMANN Guillaume wrote: >> >>> Hello, >>> >>> We use ImageJ to quantify the muscle surface area of our patients at the >>> third lumbar vertebra. >>> I use a 64bits version of ImageJ but some of my colleagues use a 32 bits >>> version and our results, for a same image, differ of at least 25 cm2, even >>> if I do the reading on both computers. >>> Does anyone know where that problem comes from and what version I should >>> trust? >>> >>> Kind regards, >>> >>> Guillaume Ulmann >>> Praticien Attaché >>> Service de Biochimie >>> Hôpital Cochin - HUPC - APHP >>> >>> >> -- >> ImageJ mailing list: http://imagej.nih.gov/ij/list.html >> >> -- >> ImageJ mailing list: http://imagej.nih.gov/ij/list.html >> >> -- >> ImageJ mailing list: http://imagej.nih.gov/ij/list.html >> >> > -- > ImageJ mailing list: http://imagej.nih.gov/ij/list.html > -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html |
«
Return to ImageJ
|
1 view|%1 views
| Free forum by Nabble | Edit this page |