Hi Po-he, Often, Update Handle Count still shows handles when none are visible on the surface. If I can find them by scrutinizing the inflated surface from multiple views/angles, zooming and rotating around, then I proceed to Prepare for Flattening and do Surface:Measurements: Crossover check on the ellipsoid surface in caret. Almost always, it reports no crossovers, so I proceed to flatten.
But in your case, you're not ready to do that for two reasons: * Much of the contralateral (right) hemisphere remains connected to the left hemisphere via the corpus callossum -- i.e., the huge bridge you mention in question #1. This is shown in the attached capture cc_bridge.jpg * It isn't worth fixing this problem until you resample your original *LR*mnc volume to cubic 1mm voxels and start over in SureFit. It's clear from the attached sagittal screen capture that the input volume was not cubic 1mm. You included your original Analyze volume, so it was easy to find out what the voxels actually were. Running Analyze2Minc chengcc1100001.hdr actually tells you what you need to do: Warning: pixel dimensions are not equal to 1, as SureFit requires. To resample to 1x1x1, run the following command: Resample -pixdim_in_x 0.898000 -pixdim_in_y 0.898000 -pixdim_in_z 1.500000 chengcc1100001.mnc Many if not most input volumes are in radiological convention (i.e., right-on-left), so they must be flipped to neurological (left-on-left). If needed, right-left flip the volume after resampling using the Volume Orientation tools. Or, use Raw2Minc with the -flipx option instead of Analyze2Minc: Raw2Minc -flipx chengcc1100001.img 256 256 128 Resample -pixdim_in_x 0.898000 -pixdim_in_y 0.898000 -pixdim_in_z 1.500000 -outfname chengcc1100001_111.mnc chengcc1100001.mnc Then, use chengcc1100001_111.mnc -- not chengcc1100001.mnc -- as input to SureFit. Because the nose is pointing toward the right, the y axis is rotated significantly off AC-PC line, so you'll still have trouble with a CC bridge, but it looks like you could crop maybe five slices further to the left without slicing off medial portions of the frontal cortex. (If your input volume was in radiological convention, then you won't have this trouble on the left hemisphere, but it will be a problem in the right hemisphere.) Questions 1 and 2: Using the Interactive Error Correction tab, setting masks with, say, an xmask of 2, ymask ranging between 10 and 60, and zmask of 30, you can place the crosshairs along the midline in horizontal view; set mask center; and zap long cuts along the midline. But you'll need a series of these cuts, varying the crosshairs location along the corpus callossum, before you'll be able to Flood Fill Volume 2 and disconnect the contralateral hemisphere. I won't bother with specific locations until the volume has been resampled. Question 3: Posterior portions of the ventricle filled, while the anterior portions did not. I don't know how to unfill the filled regions or avoid them filling in the first place. Caret needs the ventricles filled to inflate the hemisphere to a sphere. Could you explain why you want it unfilled? (SureFit and Caret are not optimized for analyzing subcortical structures.) Question 4: There is a patch that improves the transparency of the segmentation volume. I'll email the list with details in a separate message. Question 5: SureFit aims for layer 4 of the cortex -- i.e., midway between the WM/GM boundary and the GM/CSF boundary (pial surface). Your segmentation looks a tad biased toward the white matter, but it is clearly not tracking the WM/GM boundary (nor does SureFit try to). This is a difference between SureFit and Freesurfer. Freesurfer does aim for the WM/GM boundary, and it generates a cortical thickness metric and pial surface. We plan to enhance SureFit/Caret to generate cortical thickness metrics and pial surface in the future. But SureFit was designed to find layer 4. If my answers to questions 3 and 5 don't scare you off, then I can try providing more detailed instructions for addressing the CC bridge problem. First, though, please specify the left-right orientation of chengcc1100001.hdr/chengcc1100001.img. Thanks, Donna On 11/16/2004 01:27 AM, pohetsn wrote: > Dear Donna > > Thanks very much for your answer in the previous mail. > > now i encounter another problem - Fixed handles appeared again after > "Run Surefit" with "correct errors". > > I uploaded my mnc files, chengcc.tar.gz, on > http://pulvinar.wustl.edu/cgi-bin/upload.cgi > MRI file is /chengcc/chengcc1100001.L.full.sMRI.mnc > "/chengcc/SEGMENTATION/chengcc1100001.L.full.segment_corr_corr_corr.patch_corr_corr_corr.mnc" > > was produced after correcting > "/chengcc/SEGMENTATION/chengcc1100001.L.full.segment_corr_corr_corr.patch_corr_corr.mnc". > > Both's number of handles are 4, although I correct 2 small handles of > the 1st one. > > Other questions are: > 1. Is there any tips to correct handles, especially those looks like > huge "bridge". > 2. Is there any way to further crop the image? > Because SureFit doesn't have rotating function while cropping, some > part of the other hemisphere cannot be cropped when image is not > straight. > 3. I didn't select "fill ventricle", but the lateral ventricle was > segmented as white matter. > any recommendation for fixing it? > 4. while drawing volume 2 on volume1, is it possible to adjust the > transparency of volume? > It will be beneficial for us to visually exam the segmentation result. > 5. Obtaining surface with precise boundary between white matter and > grey matter is critical for our further study, > taking the file I uploaded for example, any suggestion for it? > > Thank you very much > > Sincerely, > Po-he > > > > > 在 Thu, 11 Nov 2004 12:00:02 -0600, > <[EMAIL PROTECTED]>寫道: > >> Send caret-users mailing list submissions to >> caret-users@brainvis.wustl.edu >> >> To subscribe or unsubscribe via the World Wide Web, visit >> http://pulvinar.wustl.edu/mailman/listinfo/caret-users >> or, via email, send a message with subject or body 'help' to >> [EMAIL PROTECTED] >> >> You can reach the person managing the list at >> [EMAIL PROTECTED] >> >> When replying, please edit your Subject line so it is more specific >> than "Re: Contents of caret-users digest..." >> >> >> Today's Topics: >> >> 1. about painting error on surface (pohetsn) >> 2. Re: about painting error on surface (Donna Hanlon) >> >> >> ---------------------------------------------------------------------- >> >> Message: 1 >> Date: Thu, 11 Nov 2004 16:27:56 +0800 >> From: pohetsn <[EMAIL PROTECTED]> >> Subject: [caret-users] about painting error on surface >> To: caret-users@brainvis.wustl.edu >> Message-ID: <[EMAIL PROTECTED]> >> Content-Type: text/plain; format=flowed; charset=big5 >> >> Dear all >> >> What I want to do is >> 1. depressing "Run Surefit" >> 2. draw segment errors on surface >> 3. correcting some handle errors >> 4. saving the new edit >> 5. depressing "Run SureFit" again (default setting) >> 6. checking the surface of lately fixed handle errors >> 7. draw the other segment errors which I havn't fixed >> >> However, at step7, I can't find some file looks like >> *.segment.errors.xxxxx.RGB_paint >> except the one which generated at step 2. >> >> e.g. >> I can find files as follows: >> Test.L.occipital.segment_corr2.patch.38499.surface_shape >> Test.L.occipital.segment_corr2.patch.38499.topo >> Test.L.occipital.segment_corr2.patch.ellipsoid.38499.coord >> Test.L.occipital.segment_corr2.patch.ellipsoid.38499.vtk >> Test.L.occipital.segment_corr2.patch.fiducial.38499.coord >> Test.L.occipital.segment_corr2.patch.fiducial.38499.vtk >> Test.L.occipital.segment_corr2.patch.geography.38499.paint >> Test.L.occipital.segment_corr2.patch.geography.38499.RGB_paint >> Test.L.occipital.segment_corr2.patch.inflated.38499.coord >> Test.L.occipital.segment_corr2.patch.inflated-50.38499.vtk >> Test.L.occipital.segment_corr2.patch.raw.38499.coord >> Test.L.occipital.segment_corr2.patch.raw.38499.vtk >> Test.L.occipital.segment_corr2.patch.sulci.38499.RGB_paint >> Test.L.occipital.segment_corr2.patch.Surface.38499.spec >> >> but I cannot find anything like >> Test.L.occipital.segment_corr2.patch.errors.38499.RGB_paint >> in the SURFACE folder. >> >> Anyway, how to do step.7? >> It's important while having lots handle errors. >> >> thank you >> >> Sincerely, >> phtseng >> >> >> >> ------------------------------ >> >> Message: 2 >> Date: Thu, 11 Nov 2004 08:08:53 -0600 >> From: Donna Hanlon <[EMAIL PROTECTED]> >> Subject: Re: [caret-users] about painting error on surface >> To: "Caret, SureFit, and SuMS software users" >> <caret-users@brainvis.wustl.edu> >> Message-ID: <[EMAIL PROTECTED]> >> Content-Type: text/plain; charset=Big5 >> >> Hi Phtseng, >> >> You have drawn attention to a well-known deficit in SureFit: There is no >> quick, easy way to update the ERROR_CORRECTION_INTERMEDIATES/*errors*mnc >> file, from which the SURFACES/*errors*RGB_paint is generated. This is >> the same mnc file that is the default file selection when you press the >> Locate Objects button on the Interactive Error Correction tab. >> >> This *errors*mnc file is generated only when you run automatic error >> correction, which is the most time-consuming portion of the Run SureFit >> routine. When vol2 is loaded (e.g., after you run segmentation and/or >> have been patching), then the defaults on the Run SureFit menu change to >> skip error correction and prepare to flatten. This is because typically >> one doesn't use the Run SureFit menu until you have patched all the >> handles and are ready to Prepare to Flatten. >> >> You can change the settings to Correct Errors and Review Surface First, >> but this will take a long time to run -- certainly not something you >> want to do after fixing each handle. Instead, learn to rely less on the >> Locate Objects output and the SURFACES/*errors*RGB_paint coloring on the >> surface, and just LOOK for the handles on the inflated surface. Usually, >> you can see them if you rotate the surface, zooming and panning as >> needed to get a close-up view of all areas. If you can't see the >> handles, then they're often not worth fixing. >> >> Is this the SureFit demo volume you are trying to patch? It doesn't have >> many handles. What is your handle count? If more than one or two, then >> perhaps your patching has introduced some handles. Learning to patch is >> certainly the hardest part of learning to use SureFit, and it is >> probably harder than flattening, registering, or mapping in Caret. If >> you like, I can have a look at your segmentation volume and help you >> figure out exactly what to do to fix the handles. If this is not the >> demo volume, or if you used cropping parameters (xmin, xmax, ymin, ymax, >> zmin, zmax values on the Define VOI tab during Volume Preparation) other >> than those specified in the tutorial, then I'll also need your *sMRI.mnc >> and *sMRI.params file. Upload these files here: >> >> http://pulvinar.wustl.edu/cgi-bin/upload.cgi >> >> Right now, both John Harwell and I are working on integrating Caret and >> SureFit. After SureFit's functions are merged into Caret, then we >> envision numerous improvements -- including making patching easier and >> more intuitive. >> >> Donna >> >> On 11/11/2004 02:27 AM, pohetsn wrote: >> >>> Dear all >>> >>> What I want to do is >>> 1. depressing "Run Surefit" >>> 2. draw segment errors on surface >>> 3. correcting some handle errors >>> 4. saving the new edit >>> 5. depressing "Run SureFit" again (default setting) >>> 6. checking the surface of lately fixed handle errors >>> 7. draw the other segment errors which I havn't fixed >>> >>> However, at step7, I can't find some file looks like >>> *.segment.errors.xxxxx.RGB_paint >>> except the one which generated at step 2. >>> >>> e.g. >>> I can find files as follows: >>> Test.L.occipital.segment_corr2.patch.38499.surface_shape >>> Test.L.occipital.segment_corr2.patch.38499.topo >>> Test.L.occipital.segment_corr2.patch.ellipsoid.38499.coord >>> Test.L.occipital.segment_corr2.patch.ellipsoid.38499.vtk >>> Test.L.occipital.segment_corr2.patch.fiducial.38499.coord >>> Test.L.occipital.segment_corr2.patch.fiducial.38499.vtk >>> Test.L.occipital.segment_corr2.patch.geography.38499.paint >>> Test.L.occipital.segment_corr2.patch.geography.38499.RGB_paint >>> Test.L.occipital.segment_corr2.patch.inflated.38499.coord >>> Test.L.occipital.segment_corr2.patch.inflated-50.38499.vtk >>> Test.L.occipital.segment_corr2.patch.raw.38499.coord >>> Test.L.occipital.segment_corr2.patch.raw.38499.vtk >>> Test.L.occipital.segment_corr2.patch.sulci.38499.RGB_paint >>> Test.L.occipital.segment_corr2.patch.Surface.38499.spec >>> >>> but I cannot find anything like >>> Test.L.occipital.segment_corr2.patch.errors.38499.RGB_paint >>> in the SURFACE folder. >>> >>> Anyway, how to do step.7? >>> It's important while having lots handle errors. >>> >>> thank you >>> >>> Sincerely, >>> phtseng >>> >>> >>> _______________________________________________ >>> caret-users mailing list >>> caret-users@brainvis.wustl.edu >>> http://pulvinar.wustl.edu/mailman/listinfo/caret-users >>> >> >> >> >> ------------------------------ >> >> _______________________________________________ >> caret-users mailing list >> caret-users@brainvis.wustl.edu >> http://pulvinar.wustl.edu/mailman/listinfo/caret-users >> >> >> End of caret-users Digest, Vol 16, Issue 3 >> ****************************************** > > > > > _______________________________________________ > caret-users mailing list > caret-users@brainvis.wustl.edu > http://pulvinar.wustl.edu/mailman/listinfo/caret-users >
<<inline: cc_bridge.jpg>>
<<inline: needs_resampling.jpg>>
