About Correction Split Brain from Brain Mask...
About Correction Split Brain from Brain Mask...
When I run the program 'Correction Split Brain from Brain Mask' in BrainVISA , there's always no results. And there appears a warning message:'Empty volume in VipGet3DConnex'. Would anyone know that what's wrong with it? Thank you very much!!
-
Jean-Francois Mangin
- Posts: 337
- Joined: Mon Mar 01, 2004 10:24 am
- Location: Neurospin, CEA, France
- Contact:
Basically,
the process aiming at splitting the two hemispheres and the cerebellum uses a threshold to get a first raw definition of white matter. This threshold is applied to the tissue inside the brain mask. It is defined from the estimation of grey and white matter average performed by the histogram analysis process. Then, this white matter mask is eroded to get seeds of the three anatomical objects.
Your problem may stems from:
1) your brain mask is almost empty then white matter is almost empty and disappears after erosion; GOTO brain mask correction.
2) your histogram analysis failed; CORRECT the .hana file by hand using the profile window of Anatomist or play with the VipHistoAnalysis command (you need gnuplot to visualize results).
3) the splitting process iteratively increases the size of the erosion until reaching a stage where it got the three seeds it needs. The seed selection uses a template splitted brain. Sometimes the cerebellum seed is missing, either because the acquisition field of view misses cerebellum or because the bias correction did not overcome a high bias in Z direction (you may try the option high bias field in Z or play with the parameter of the process located in SHFJ low level directory). If you are in this case, if you look at brainVISA log window, you will realize that the command is making all kind of attempts that finally fail. If this is a bias problem, your best chance is to use the lowest threshold as possible, which is the lowest option in the correction list.
the process aiming at splitting the two hemispheres and the cerebellum uses a threshold to get a first raw definition of white matter. This threshold is applied to the tissue inside the brain mask. It is defined from the estimation of grey and white matter average performed by the histogram analysis process. Then, this white matter mask is eroded to get seeds of the three anatomical objects.
Your problem may stems from:
1) your brain mask is almost empty then white matter is almost empty and disappears after erosion; GOTO brain mask correction.
2) your histogram analysis failed; CORRECT the .hana file by hand using the profile window of Anatomist or play with the VipHistoAnalysis command (you need gnuplot to visualize results).
3) the splitting process iteratively increases the size of the erosion until reaching a stage where it got the three seeds it needs. The seed selection uses a template splitted brain. Sometimes the cerebellum seed is missing, either because the acquisition field of view misses cerebellum or because the bias correction did not overcome a high bias in Z direction (you may try the option high bias field in Z or play with the parameter of the process located in SHFJ low level directory). If you are in this case, if you look at brainVISA log window, you will realize that the command is making all kind of attempts that finally fail. If this is a bias problem, your best chance is to use the lowest threshold as possible, which is the lowest option in the correction list.
Excuse me for asking the other question.
The problem I asked before has been solved. But there still the other problem exists. After I run the program"Ana Split brain from brain mask",a warning message says :
!! VIP Error:
Image format or read error
------------------------------
------------------------------
!! VIP Error:
(VipReadVolumeWithBorder)Can not read this image: /home2/local/Panabase/demo/talairach/closedvoronoi
I'm sure that I input correct image files into the program.I don't know why the program can't read it. Would anyone know what's wrong with it?
Thank you so much!!
The problem I asked before has been solved. But there still the other problem exists. After I run the program"Ana Split brain from brain mask",a warning message says :
!! VIP Error:
Image format or read error
------------------------------
------------------------------
!! VIP Error:
(VipReadVolumeWithBorder)Can not read this image: /home2/local/Panabase/demo/talairach/closedvoronoi
I'm sure that I input correct image files into the program.I don't know why the program can't read it. Would anyone know what's wrong with it?
Thank you so much!!
-
Jean-Francois Mangin
- Posts: 337
- Joined: Mon Mar 01, 2004 10:24 am
- Location: Neurospin, CEA, France
- Contact:
Well,
I think what happened is that the command as been triggered without the template to guide the split. Normally, thie template filename is filled in automatically by brainVISA. It is hidden somewhere in brainVISA dependencies. If you removed it manually, it seems a local setting of the SHFJ is triggered which is no good... If the template filename is really missing, there is a pb with your install. Did you manage to process the lesson brains?
I think what happened is that the command as been triggered without the template to guide the split. Normally, thie template filename is filled in automatically by brainVISA. It is hidden somewhere in brainVISA dependencies. If you removed it manually, it seems a local setting of the SHFJ is triggered which is no good... If the template filename is really missing, there is a pb with your install. Did you manage to process the lesson brains?
-
riviere
- Site Admin
- Posts: 1361
- Joined: Tue Jan 06, 2004 12:21 pm
- Location: CEA NeuroSpin, Saint Aubin, France
- Contact:
The path to your voronoi template seems really strange: /home2/local/Panabase/demo/talairach/...
It seems to take it in your personal database, whereas it should be in <installation_dir>/SHFJ_pack-stable-2.3.x/share/shfj/hemitemplate/closedvoronoi.ima
Did you specify it by hand or something ?
The template is unique (it doesn't depend on the subject and is located in the shared database coming with BrainVisa) and should be selected automatically. You should not have to specify it by hand.
Denis
It seems to take it in your personal database, whereas it should be in <installation_dir>/SHFJ_pack-stable-2.3.x/share/shfj/hemitemplate/closedvoronoi.ima
Did you specify it by hand or something ?
The template is unique (it doesn't depend on the subject and is located in the shared database coming with BrainVisa) and should be selected automatically. You should not have to specify it by hand.
Denis
Thank you for your responses. I Checked every input in the program,then ran the program again. There was still the same warning message appearing. Now I list every input as following,please help me to find what's wrong with it.
mri_corrected = D:\BrainVISA\pin_T2wI\correction_results\IM20\bias correction\no_bias\nobias_IM20_imported.ima
variant = Barycentre 0.25
brain_voronoi = D:\BrainVISA\pin_T2wI\correction_results\IM20\bias correction\no_bias\voronoi_nobias_IM20_imported.ima
histo_analysis = D:\BrainVISA\pin_T2wI\correction_results\IM20\histogram analysis\nobias_nobias_IM20_imported.han
brain_mask = D:\BrainVISA\pin_T2wI\correction_results\IM20\brain_mask\brain_mask_IM20_imported.ima
Use_template = 0
voronoi_template = C:\BrainVISA\SHFJ_pack-stable-win32-2.3.0\share\shfj\hemitemplate\closedvoronoi.ima
Commissure_coordinates = None
readme = None
[/img]
mri_corrected = D:\BrainVISA\pin_T2wI\correction_results\IM20\bias correction\no_bias\nobias_IM20_imported.ima
variant = Barycentre 0.25
brain_voronoi = D:\BrainVISA\pin_T2wI\correction_results\IM20\bias correction\no_bias\voronoi_nobias_IM20_imported.ima
histo_analysis = D:\BrainVISA\pin_T2wI\correction_results\IM20\histogram analysis\nobias_nobias_IM20_imported.han
brain_mask = D:\BrainVISA\pin_T2wI\correction_results\IM20\brain_mask\brain_mask_IM20_imported.ima
Use_template = 0
voronoi_template = C:\BrainVISA\SHFJ_pack-stable-win32-2.3.0\share\shfj\hemitemplate\closedvoronoi.ima
Commissure_coordinates = None
readme = None
[/img]
-
riviere
- Site Admin
- Posts: 1361
- Joined: Tue Jan 06, 2004 12:21 pm
- Location: CEA NeuroSpin, Saint Aubin, France
- Contact:
Are you sure you get exactly the same error message as before ? I guess you don't, for several reasons:
- your last message has windows filenames (D:\...) whereas your previous one had unix filenames (/home2/...) so you appear to have changed system between the two messages, this doesn't help to understand where it failed...
- the previous error was on the voronoi_template filename, and it appears to be OK in your last message so I suspect it fails somewhere else now
I also have a couple of more helpful remarks:
- you didn't specify a Commissure_coordinates param filename, this suggests that you haven't done the PrepareSubject process before going to the segmentation processes (or you haven't filled this parameter), this may be a cause of failure
- the filenames you used suggest that you are not using BrainVisa database organization and specified every filename by hand (with custom subdirectories like "brain_mask" and "histogram analysis"), and, well, it's not forbidden to do so, but it is really more painful and error-prone. Moreover it's possible that the anatomical pipeline in its current state fails at the end when used outside a database. I suggest you begin by using BV database organization, and when you have obtained some results with this organization, then you could perhaps begin to do some more manual things (if you really like).
Have you created a database in BrainVisa ?
I briefly repeat the standard procedure for anatomical processing:
- create a database (if not already done) using the "preferences/database" menu
- import a T1 image into the database using the "data management/importation/anatomy/import T1 MRI" process
- use the "anatomy/pipeline/PrepareSubject" process to specify AC/PC points
- use the "anatomy/pipeline/Ana Do A Lot of things" process to do all the segmentations
Denis
- your last message has windows filenames (D:\...) whereas your previous one had unix filenames (/home2/...) so you appear to have changed system between the two messages, this doesn't help to understand where it failed...
- the previous error was on the voronoi_template filename, and it appears to be OK in your last message so I suspect it fails somewhere else now
I also have a couple of more helpful remarks:
- you didn't specify a Commissure_coordinates param filename, this suggests that you haven't done the PrepareSubject process before going to the segmentation processes (or you haven't filled this parameter), this may be a cause of failure
- the filenames you used suggest that you are not using BrainVisa database organization and specified every filename by hand (with custom subdirectories like "brain_mask" and "histogram analysis"), and, well, it's not forbidden to do so, but it is really more painful and error-prone. Moreover it's possible that the anatomical pipeline in its current state fails at the end when used outside a database. I suggest you begin by using BV database organization, and when you have obtained some results with this organization, then you could perhaps begin to do some more manual things (if you really like).
Have you created a database in BrainVisa ?
I briefly repeat the standard procedure for anatomical processing:
- create a database (if not already done) using the "preferences/database" menu
- import a T1 image into the database using the "data management/importation/anatomy/import T1 MRI" process
- use the "anatomy/pipeline/PrepareSubject" process to specify AC/PC points
- use the "anatomy/pipeline/Ana Do A Lot of things" process to do all the segmentations
Denis
-
Jean-Francois Mangin
- Posts: 337
- Joined: Mon Mar 01, 2004 10:24 am
- Location: Neurospin, CEA, France
- Contact:
-
riviere
- Site Admin
- Posts: 1361
- Joined: Tue Jan 06, 2004 12:21 pm
- Location: CEA NeuroSpin, Saint Aubin, France
- Contact:
A BrainVisa database is just a directory (empty or new directory at the beginning) where Brainvisa will import and write all data, this directory has a structure which is managed by brainvisa so it knows how to find everything within it. We still have some efforts to do in the documentation but I thought this was explained somewhere.
To create a database, just select or create a directory for the "directory" field, the other fields (hierarchy and cache) are optional, they will take default values (this is maybe the part which is unclear to you and maybe not explained clearly enough in the documentation).
Denis
To create a database, just select or create a directory for the "directory" field, the other fields (hierarchy and cache) are optional, they will take default values (this is maybe the part which is unclear to you and maybe not explained clearly enough in the documentation).
Denis