ITK-SNAP on 64bit Windows

[Mengda Wu]

Hi,

   Any plan to release a version for 64bit Windows? I need this because I have a very big image to segment which is larger than 2GB.
I tried to compile myself. It seemed to succeed but have the bug I mentioned yesterday.

Thanks,
Mengda

[Paul Yushkevich]

Hi Mengda

I haven't been able to recreate your bug on Windows. We don't have a
64 bit Windows machine at this time. Do you think you could go into
the debugger in visual studio to see what's causing the bug?

Also, do you know that you can select a region of interest in the
image when going into automatic segmentation mode? That will use a lot
less memory, because most of the memory is used in automatic
segmentation mode (10x the memory use in manual mode).

Paul.

--
Paul A. Yushkevich, Ph.D.
Research Assistant Professor
Penn Image Computing and Science Laboratory
Department of Radiology
University of Pennsylvania

[Mengda Wu]

Hi Paul,

   Thanks a lot for looking into this.

   Could you please tell me what versions of ITK, VTK and FLTK you compiled against for
ITK-SNAP1.6.0.1? And what special features of each these packages you enabled? So I
will try to compile with the same as you and see what happens. Is the latest CVS version
of ITK-SNAP the same as release 1.6.0.1?

   I am very interested in helping compile a 64bit version, which can address more than 4GB
memory for a very big image. Actually, I have compiled successfully both 32bit and 64bit
versions of ITK-Snap. The program seems to run fine. I think this bug I
mentioned is related to display problem. Both 32bit and 64bit versions affect.

  I attach a 32 version of ITK-SNAP I compiled with VS2005 against ITK-3.6.0, VTK-5.0.4, and FLTK-1.1.8.
So you can see the problem of my compilation. I removed the suffix .exe since Gmail does not allow
sending a executable.

Thanks,
Guanglei

2008/4/27 Paul Yushkevich <pyush...@gmail.com>:

[R Holbrey]

Hi,

I could well be wrong, but I thought one of the best things about Vista
(and one of the few good things I've read so far;) was that you could run
32bit programs and the OS would supply access to larger memory areas. Did
you try loading your 2GB file in the 32 bit program?

- regards,
Richard

[Paul Yushkevich]

Hi Mengda,

Thanks so much for sending me the binary. Indeed, I was able to
recreate the bug with your binary, but not with the version that I
built (and that's on the website). I suppose it may be some change in
ITK that's causing this. Perhaps the level set filter is initializing
the level set image differently (not setting the inside and outside to
positive/negative values as it has done before. I try building against
an older version. The version I currently build against is 2.8.1. The
VTK version is 5.0.4 and FLTK is fltk-1.1.x-r5642 (but that should
matter the least). The code currently in the HEAD branch of CVS should
match the binary on itksnap.org.

Paul.

[Mengda Wu]

Hi Richard,

   Actually you need supply a linker option /LARGEADDRESSAWARE to make the 32bit program can access up to 4GB
on a 64bit Vista. I tried my 32bit compilation with this option. The binary could access beyond 2GB but not beyond 4GB.

  I also compiled 64bit version. The binary could access beyond 20GB. My machine now has 64GB memory.

  So I think more people including me will benefit a lot if we fix that bug and release a 64bit version.

Thanks,
Guanglei

2008/4/28 R Holbrey <rich...@comp.leeds.ac.uk>:

[Mengda Wu]

Hi Paul,

   I have tried to compile against ITK-2.8.1. That bug is gone. Everything works fine.
So the bug is due to the change between ITK-2.8.1 and ITK-3.6.0 as you said.
Let me know if you have any plan to build against newest ITK. I am going to compile
the 64bit version soon. I am not sure if the ITK-2.8.1 will be compiled smoothly.
I will let you know the result.

Thanks,
Mengda

2008/4/28 Paul Yushkevich <pyush...@gmail.com>:

[Mengda Wu]

Hi Paul,

   I compiled a 64bit version. It works well.

Thanks,
Guanglei

2008/4/28 Mengda Wu <phd...@gmail.com>: