From vladimir.fonov at gmail.com Tue Aug 10 09:03:13 2010 From: vladimir.fonov at gmail.com (Vladimir Fonov) Date: Tue, 10 Aug 2010 09:03:13 -0400 Subject: [MINC-users] ecattominc - segfault in 64bit mode, works fine in 32bit Message-ID: Hello, looks like ecattominc have issues when compiled for 64bit: on the bic system: 32bit: file /home/bic/vfonov/data/proj02/quarantine/bin/ecattominc /home/bic/vfonov/data/proj02/quarantine/bin/ecattominc: ELF 32-bit LSB executable, Intel 80386, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.4.1, not stripped /home/bic/vfonov/data/proj02/quarantine/bin/ecattominc -version program: 2.0.18 libminc: 2.0.18 netcdf : 3.6.1 of Sep 4 2009 18:14:50 $ HDF5 : 1.6.9 /home/bic/vfonov/data/proj02/quarantine/bin/ecattominc /data/scratch/scratch1/ecat/test.v /tmp/test.mnc Reading headers. Copying frames:.Done 64bit: file /home/bic/vfonov/data/proj02/quarantine_x64/bin/ecattominc /home/bic/vfonov/data/proj02/quarantine_x64/bin/ecattominc: ELF 64-bit LSB executable, x86-64, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.8, not stripped /home/bic/vfonov/data/proj02/quarantine_x64/bin/ecattominc -version program: 2.0.18 libminc: 2.0.18 netcdf : 3.6.1 of Feb 19 2010 15:20:14 $ HDF5 : 1.6.10 /home/bic/vfonov/data/proj02/quarantine_x64/bin/ecattominc /data/scratch/scratch1/ecat/test.v /tmp/test.mnc Segmentation fault I don't use MINC2 environment variable. Running in gdb shows that it fails in get_long_value (from=0x7fffe7d40008, to=0x7fffe7d3f348) at conversion/ecattominc/machine_indep.c:25 -- Best regards, ?Vladimir S. Fonov ~ v.s.fonov <@> ilmarin.info From claude at bic.mni.mcgill.ca Tue Aug 10 09:30:10 2010 From: claude at bic.mni.mcgill.ca (Claude LEPAGE) Date: Tue, 10 Aug 2010 09:30:10 -0400 Subject: [MINC-users] ecattominc - segfault in 64bit mode, works fine in 32bit In-Reply-To: Message-ID: <201008101330.o7ADUAQJ032590@grumio.bic.mni.mcgill.ca> Vlad, Can you try with minc-2.1.00 that Andrew should have released by now? You know how to get it from cvs head. I remember fixing some similar problem before for a converter (check ChangeLog maybe) since long=8bytes in 64 bits and the code is expecting 4 bytes. Claude From anthonin at biospective.com Tue Aug 10 10:28:49 2010 From: anthonin at biospective.com (Anthonin Reilhac) Date: Tue, 10 Aug 2010 10:28:49 -0400 Subject: [MINC-users] ecattominc - segfault in 64bit mode, works fine in 32bit In-Reply-To: References: Message-ID: HI, I had fixed this issue some times ago. I thought the fixed code was part of the distribution. I can send it to you Claude? Anthonin On 10 August 2010 09:03, Vladimir Fonov wrote: > Hello, > > looks like ecattominc have issues when compiled for 64bit: > on the bic system: > > > 32bit: > > file /home/bic/vfonov/data/proj02/quarantine/bin/ecattominc > /home/bic/vfonov/data/proj02/quarantine/bin/ecattominc: ELF 32-bit LSB > executable, Intel 80386, version 1 (SYSV), dynamically linked (uses > shared libs), for GNU/Linux 2.4.1, not stripped > > /home/bic/vfonov/data/proj02/quarantine/bin/ecattominc -version > program: 2.0.18 > libminc: 2.0.18 > netcdf : 3.6.1 of Sep 4 2009 18:14:50 $ > HDF5 : 1.6.9 > > /home/bic/vfonov/data/proj02/quarantine/bin/ecattominc > /data/scratch/scratch1/ecat/test.v /tmp/test.mnc > Reading headers. > Copying frames:.Done > > > 64bit: > file /home/bic/vfonov/data/proj02/quarantine_x64/bin/ecattominc > /home/bic/vfonov/data/proj02/quarantine_x64/bin/ecattominc: ELF 64-bit > LSB executable, x86-64, version 1 (SYSV), dynamically linked (uses > shared libs), for GNU/Linux 2.6.8, not stripped > > /home/bic/vfonov/data/proj02/quarantine_x64/bin/ecattominc -version > program: 2.0.18 > libminc: 2.0.18 > netcdf : 3.6.1 of Feb 19 2010 15:20:14 $ > HDF5 : 1.6.10 > > /home/bic/vfonov/data/proj02/quarantine_x64/bin/ecattominc > /data/scratch/scratch1/ecat/test.v /tmp/test.mnc > Segmentation fault > > > I don't use MINC2 environment variable. > > Running in gdb shows that it fails in get_long_value > (from=0x7fffe7d40008, to=0x7fffe7d3f348) at > conversion/ecattominc/machine_indep.c:25 > > -- > Best regards, > Vladimir S. Fonov ~ v.s.fonov <@> ilmarin.info > _______________________________________________ > MINC-users at bic.mni.mcgill.ca > http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users > From a.janke at gmail.com Wed Aug 11 00:22:35 2010 From: a.janke at gmail.com (Andrew Janke) Date: Wed, 11 Aug 2010 14:22:35 +1000 Subject: [MINC-users] MINC 2.1.00 available Message-ID: Hi all, MINC 2.1.00 is available here: http://packages.bic.mni.mcgill.ca/tgz/ Note that there are some potentially big changes in this release with the use of HDF 1.8.x over 1.6.x, shared libraries are now also built alongside static versions. Note that no changes to MINC analyses should happen just compiling and linking differences. A full list of changes (from NEWS) are below, binary packages will follow in a week or three once this release has been tested more in the "wild". This release includes a few more tools also, minccmp and xfm2def. New in Release 2.1.00 --------------------- * Improved convergence and accuracy for application of non-linear transformation (especially for 2-d slices) - Claude * Set default volume_io caching to none * Added pod2man for man page generation of help for perl scripts * Added minccmp and xfm2def man pages * Added checks for outfiles to minccalc * Added taking first time point for 4D files, -test_size and -sagittal_offset_perc to mincpik * ported some HDF calls to 1.8.x (must now use 1.8.X; 1.6.X no longer supported from now on) * Added libtoolize/glibtoolize logic in autogen.sh (thanks Sean) * Fixed bug in multidim_array_is_alloced for correct check of memory allocation of image data. Return volume=NULL when memory allocation fails. * Fixed a bug in upet2mnc on 64 bit platforms (thanks John Cupitt) -- Andrew Janke (a.janke at gmail.com || http://a.janke.googlepages.com/) Canberra->Australia +61 (402) 700 883 From laurence at bic.mni.mcgill.ca Tue Aug 10 14:53:11 2010 From: laurence at bic.mni.mcgill.ca (Laurence Mercier) Date: Tue, 10 Aug 2010 14:53:11 -0400 Subject: [MINC-users] Would you have a MINC tool to extract sulci from a gad-enhanced T1? Message-ID: <4C61A017.7010303@bic.mni.mcgill.ca> Hello minc-users, I would like to know if one of you has coded a tool to extract the sulci from a gad-enhanced T1? I am programming a script to convert a gad-enhanced T1 MRI into an ultrasound looking image. In ultrasound, the sulci look very bright and I need to model that. I programmed some sort of a hack, but it is not very elegant and I was wondering if anyone would have something better. My version works as follows: I get the CSF map with classify_clean. I then attempt to remove the ventricles from there to only keep the CSF in the sulci. As the ventricles of the patients I work on are far from normal (patients with large tumors) I have to dilate the ventricle mask that I get because for my application it is very important not to have any "false sulci". In attachment you can see the result (sulci in red) which is not bad, but is still not sensitive enough. ------------------------- For those that prefer the code, it goes as follows: #Dilate ventricles to make sure they all connect do_cmd("mincchamfer $ventricles_mask ventricles_mask_chamfer.mnc"); do_cmd("minccalc -expression 'if(A[0]>6){1}else{0}' ventricles_mask_chamfer.mnc larger_ventricles_inv.mnc"); do_cmd("mincmath -not larger_ventricles_inv.mnc larger_ventricles.mnc"); #Add that to the csf map do_cmd("mincmath -seg -const2 1 1 $classif_tal csf_map.mnc -clob"); do_cmd("mincmath -or larger_ventricles.mnc csf_map.mnc csf_map_larger_vent.mnc"); #Only keep the largest connected group #ED is needed in order to diminish the number of groups, otherwise there are more than 256 do_cmd("mincmorph -successive 'EDB[0.5:1.5:1:0]GB[0.5:1.5:1:0]' csf_map_larger_vent.mnc final_vent_mask.mnc"); #Substract ventricles from csf to keep the sulci only do_cmd("minccalc -expression 'if(A[0]>0.5 && A[1]>0.5){0}else{A[0]}' csf_map.mnc final_vent_mask.mnc sulci_mask.mnc -clob"); ---------- Thanks for your help! Laurence -- Laurence Mercier McConnell Brain Imaging Center Montreal Neurological Institute, room: WB-320 3801 University St Montreal (QC), H3A 2B4 CANADA Office Phone: +1 514.398.1573 Office FAX: +1 514.398.2975 Web: http://www.bic.mni.mcgill.ca/~laurence From a.janke at gmail.com Sun Aug 22 23:35:12 2010 From: a.janke at gmail.com (Andrew Janke) Date: Mon, 23 Aug 2010 13:35:12 +1000 Subject: [MINC-users] register and Display issues on Nvidia hardware + later versions of Ubuntu/Fedora. Message-ID: Hi all, I have had a number of reports of "cannot get requested colour_map_mode(x,x)" with later distributions and nvidia hardware. In an attempt to fix this I have made a number of changes/simplifications to the GLUT code in register in order to try and get around this. Can those with this problem give this package a try: http://packages.bic.mni.mcgill.ca/tgz/Register-1.4.0.tar.gz I'll be most interested in bug reports from it. Thanks -- Andrew Janke (a.janke at gmail.com || http://a.janke.googlepages.com/) Canberra->Australia +61 (402) 700 883 From mishkind at gmail.com Wed Aug 25 15:30:18 2010 From: mishkind at gmail.com (Mishkin Derakhshan) Date: Wed, 25 Aug 2010 15:30:18 -0400 Subject: [MINC-users] configure options Message-ID: Hi, I'm having some problems understanding how to control ./configure. How do you tell configure to look in a custom place for libraries that it can't find? When trying to compile I get an error like this: ./configure --prefix=/home/mishkin/development/source_code/bicpl-1.4.6/ . . . checking for library m... yes checking for library netcdf... yes checking for library minc... no configure: error: cannot find required library minc I'm guessing this is because my system doesn't have the minc1 libraries installed in a default location. (For the record, are these the .so, la or .a files that it is looking for?) I thought I could have configure look in a custom location by using the LDFLAGS (I tried CFLAGS as well): export LDFLAGS=-L/home/mishkin/development/libs/ ./configure --prefix=/home/mishkin/development/source_code/bicpl-1.4.6/ But I still get the same error. /home/mishkin/development/libs/ contains: opus[~/development/libs]$ ll libminc.* -rwxr-xr-x 1 mishkin cta 70K 2008-09-22 17:58 libminc.so.0.7 -rwxr-xr-x 1 mishkin cta 63K 2008-09-22 17:58 libminc.so.0.6 -rwxr-xr-x 1 mishkin cta 70K 2008-09-22 17:58 libminc.so.0 -rwxr-xr-x 1 mishkin cta 70K 2008-09-22 17:58 libminc.so -rwxr-xr-x 1 mishkin cta 770 2008-09-22 17:58 libminc.la -rw-r--r-- 1 mishkin cta 388K 2008-09-22 17:58 libminc.a I also tried ./configure --prefix=/home/mishkin/development/source_code/bicpl-1.4.6/ --with-build-path=/home/mishkin/development/libs/ but same error. thanks, mishkin From a.janke at gmail.com Wed Aug 25 18:55:32 2010 From: a.janke at gmail.com (Andrew Janke) Date: Thu, 26 Aug 2010 08:55:32 +1000 Subject: [MINC-users] configure options In-Reply-To: References: Message-ID: Hi Mishkin On Thu, Aug 26, 2010 at 05:30, Mishkin Derakhshan wrote: > I also tried > ./configure --prefix=/home/mishkin/development/source_code/bicpl-1.4.6/ > --with-build-path=/home/mishkin/development/libs/ What you want is: ./configure --prefix=/home/mishkin/development/source_code/bicpl-1.4.6 --with-build-path=/home/mishkin/development Presuming you want to build from the libraries in /home/mishkin/development/{include,lib} and want to install the resulting binaries in /home/mishkin/development/source_code/bicpl-1.4.6/{bin,lib,include} -- Andrew Janke (a.janke at gmail.com || http://a.janke.googlepages.com/) Canberra->Australia +61 (402) 700 883