medical image analysis

I had a look at the results now and the first thing I have to say is that a lot of the reported defects are false positives because the check software can not handle BOOST_REQUIRE, and it also has problems with std::swap, and maybe ... [More] , std::unique_ptr::reset. It is possible that the latter also can be attributed to its use of std::swap. So fat I've identified one minor problem in the vff reader and one trivial problem in the camera class. [Less]

Something I completely forgot to mention: a paper describing the software has been accepted for publication in Source Code for Biology and Medicine of BioMed Central. It is published as Open Access and can be downloaded here

In order to make the project a bit more visible, and also to do more code checking I've connected this Sourceforge project to Ohloh and Coverty. Unfortunately, in order to display the bragging-widgets Ohloh requires scripts that can not be embedded ... [More] in this markup based forum. So, if you want to have a look at them, you will have to follow [this link](Ohloh. Coverty provides a service to do static analysis of the code. The first analysis showed that MIA has 191 defects of which 3 were dismissed, no public link exists though. This gives MIA a defect density of 0.54, lower than the average for FLOSS projects (0.69) or proprietary software (1.0). Of course the only acceptable defect density is 0.0 ;) Currently, I can not view the defects, because the projects application is still under review, but when I do, I will move on to fix them and the I will push for the 2.0.12 maintenance release. [Less]

Version 2.0.11 has now released to fix some minor bug. Especially the fix for [tickets:#123] needed to get out there in order to make it easy to run the perfusion analysis of the data now [publicly available] ... [More] (https://sourceforge.net/projects/tabseg/files/Data/). Currently, not all data sets are segmented, but I'm working on it. To describe this data, I also submitted a data paper has been submitted to ISBI 2014. Let's see if it gets accepted. [Less]

Now that version 2.0.10 has been released work will begin to overhaul the API to allow for new features, amongst these full support for orientation in 3D data sets support for IO of images and vector field stored in HDF5 and Nifti1 diffeomorphic ... [More] registration proper handling of voxel and pixel sizes in registration algorithms Meanwhile, the mia-viewit program will be prepared for Debian and the python interface package created. Most likely the minimum compiler to be used will be changed to g++-4.7 and clang++-3.3 in order to make use of more C++11 features. Limiting factor is currently support for Ubuntu 12.04 since it does only provide g++-4.6. [Less]

Right now the pre-release of 2.0.10 is compiling in the Ubuntu Launchpad servers, and I've updated the Changelog for the new release that I will probably make available by the end of the week. Hopefully by then the Debian bug #709554 will be fixed by ... [More] then so that mia-viewitgui can be uploaded without workarounds. After this all goes forward to version 2.2, which means API changes and ABI breaks. The biggest issue to be tackled is proper orientation and scaling handling. [Less]

As of version 2.0.9, MIA is now an official Debian package. The command line tools are available as mia-tools and the development files package is called libmia-2.0-dev. This means there is a good chance that MIA will it make into the next Ubuntu ... [More] release. for now, Ubuntu packages for version 12.04 are still provided via the MIA-PPA. Note, that the Ubuntu PPA may get some experimental patches then and again that will take longer to appear in the Debian release. mialmpick is also available in Debian, and currently packaging for the viewitgui package is done. The latter is already available in the Ubuntu PPA. [Less]

Version 2.0.8 has been released but unfortunately it was not really tested on i386, so if you run MIA on a 32 bit installation, some tests will fail because of lack of floating point accuracy. I guess this is because on Intel 32 bit the processor has ... [More] less floating point registers and has to store, and hence, truncate intermediate results. Interestingly, this problem doesn't appear on 33 bit PowerPC (i.e. the G4). In addition, if you want to develop with MIA, the miacore.pc file has a bug so that pkg-config will not provide the proper C-flags. Both issues have been fixed in git head, and a hotfix patch to the source code is provided. MIA is again in the Debian ftpmaster pipeline, and development will slow down a bit for a time. The 2.0.9 release should only bring incremental updates, i.e. what is listed in the 2.0.9 tracker. I've also installed the forum tool. [Less]

Additional focus is layed on [tickets:#8] - add attributes to transformations, and [tickets:#46] - make penalty terms a transformation parameter.

The Debian package was rejected in the first round because of some licencing documentation issues. For that reason I will take the chance to apply some ABI changes that would otherwise be postponed for more time. Among these are removing some unused parameters [tickets:#70], and preparing the new interface for the transformation penalty terms.