Hi Kaushik, There are at least two problems with that input volume:
1) It is rotated significantly off AC-PC line (e.g., as if nose is pointed somewhat leftward). This could definitely cause trouble with some of the Caret segmentation routines (e.g., disconnecting eye, hindbrain, contralateral hem). Whether this is the cause of the memory error isn't quite clear to me, but it is possible. To get a decent result, you will likely need to AC-PC align the volume first. It is possible to do this in Caret, but I do not know which tutorial covers it, off the top of my head.
2) While cropping, some of the frontal lobe was shorn off along the midline -- likely related to the problem above. You certainly don't want these chunks of cortex missing from your segmentation/surface.
AC-PC align the volume first; then recrop.The volume has good voxdims, resolution, and gray/white contrast. If there are non-uniformity issues, they're pretty minor. So once you AC-PC align the volume, then I think you will be in good shape for segmenting.
If you do use Caret to AC-PC align the volume, then make sure you save the resulting affine matrix for future reference. You might need it to align functional or other data that currently is aligned with the original volume with the Caret surface, and this affine matrix will be essential for this purpose.
Donna On 08/10/2009 01:43 PM, krishna chaluvadi wrote:
Hi Donna the cropped volume is Human.case10.R.Right_Hem+orig.nii.gz the prams file is Human.case10.R.params_file_22.params thanks kaushikOn Mon, Aug 10, 2009 at 11:10 AM, Donna Dierker <[email protected] <mailto:[email protected]>> wrote:Kaushik, Can you upload your cropped volume and params file again? http://pulvinar.wustl.edu/cgi-bin/upload.cgi Here are some things that come to mind: * You're trying to segment a whole brain, rather than a cropped hemisphere. * There is insufficient disk space in your working directory. (I know memory and disk space are distinct, but I have seen that error when the real issue was disk space.) * The input volume is T2 or just not something Caret can handle. Seeing your uploaded volume might give me some clues. Donna On 08/10/2009 09:41 AM, krishna chaluvadi wrote: > Thanks Donna, that worked. I am getting another error saying " out of > memory, caret terminating". This happens while i am trying to segment > the volume. Kindly let me know what could be a possible solution > > thanks > kaushik > > On Thu, Aug 6, 2009 at 4:14 PM, Donna Dierker > <[email protected] <mailto:[email protected]> <mailto:[email protected] <mailto:[email protected]>>> wrote: > > Kaushik, > > Something went horribly wrong when the volume got rescaled. See > attached capture. > > The omniscient Tim Coalson believes you started with an input > volume that was long or float that got written as char or unsigned > byte. But the data was just truncated rather than scaled. > > I didn't see the original volume in our upload area, but the > cropped volume was there. > > Try rescaling again on your end. > > Load the original anatomy volume. > > Select Volume: Edit volume attributes: Data: Rescale Voxels Values. > > Leave the current range alone. > > Next to output range, click Set and select Byte unsigned. > > Save the resulting rescaled volume by a different filename (not > yet cropped, because you'll need the rescaled version for both hems). > > Donna > > > On 08/06/2009 10:26 AM, krishna chaluvadi wrote: > > Thanks Donna > > i have uploaded the original volume file, the cropped volume > file and the params file, kindly let me know if the > problem(cropping or the thresholding?). if one of these are > the problems, kindly let me know the correct peaks and limits > that should be used > the file names are > > original volume file> 20060407_114706IM-0001-0001EPILEPSYSEIZUREDISORDAnonymizeds007a1001A.nii.gz> > cropped volume file > Human.case10.R.Right_Hem+orig.nii.gz > > params file > Human.case10.R.params_file_22.params > > thanks > kaushik > > On Thu, Aug 6, 2009 at 7:33 AM, Donna Dierker > <[email protected] <mailto:[email protected]> <mailto:[email protected] <mailto:[email protected]>> > <mailto:[email protected] <mailto:[email protected]> > <mailto:[email protected] <mailto:[email protected]>>>> wrote: > > Hi Kaushik, > > Thank you for posting to caret-users; I really appreciate it. > Asynchronous communication is a must right now, given very > tight > deadlines. But it is always preferrable. > > The easiest way for me to diagnose your trouble is for you > to upload > your cropped *anatomy* volume here: > > http://pulvinar.wustl.edu/cgi-bin/upload.cgi > > It wouldn't hurt to upload the params file, if you did take > a stab at > setting the peaks. I can give you feedback there. > > Three possible causes come to mind when I see this error: > > * You didn't crop to left or right hemisphere. > > * Your volume's origin is offset significantly from the AC. > > * There is a bridge of eye fat connecting the eyeball to > the cortex. > > Looking at your volume will tell me immediately which, if > any of these > causes, is the culprit. > > Donna > > On 08/05/2009 02:43 PM, krishna chaluvadi wrote: > > > > Hello sir > > > > This is kaushik. I am a beginner with caret and have a > few problems > > during segmentation. I am having trouble with determining the > > threshold limit for gray and white peaks. I am constantly > getting the > > error “ threshold limit reached before eye disconnected”. > Kindly > help > > me on this > > > > thanks
<<inline: shorn_frontal.jpg>>
_______________________________________________ caret-users mailing list [email protected] http://brainvis.wustl.edu/mailman/listinfo/caret-users
