Version 2.0

Version 2.0 implements support for dicom format

Both, the offline as well as the online converter tool can now convert dicom files to Tissue Stack's internal raw format via dcmtk (http://support.dcmtk.org/redmine/projects/dcmtk/wiki/Overview)

Supported are:

  • gray-scale data of various bit depth
  • color data

"Partially" supported are:
  • Siemens mosaic files which will be (for now at least) displayed in the same way a time series would
  • In Multi-Frame only the first frame will be considered.
Both of the limitations above are, in part, due to the display modalities of Tissue Stack which is inherently 3D.

NOTE: jpeg 2000 compression is not supported

How to prepare dicom data for Tissue Stack conversion?

Single images can just be used/uploaded as are.

Series have to be zipped with the requirement that there only be one per zip and no sub-directories be used within the zip!


Both, the offline and online converter will unzip the contents of the zip into the /tmp directory using the zip name as a subdirectory.

Conversion may take a while depending on factors such as size of zip, use of compression, and type of data. In general, single images (obviously) and data interpreted as time series (incl. Siemens mosaic) will be much quicker than anything else which will be reconstructed into a 3D volume.





Version 1.8

This version includes support for displaying anisotropic data sets.

This feature has to still be considered somewhat beta since it has not yet been tested extensively, particularly the case where all three dimensions are not isotropic respectively.

Binary packages can be downloaded here:





Version 1.7

Version 1.7 includes bug-fixes and an upgrade of jquery and jquery mobile.

Binary Packages can be downloaded here: http://tissuestack.org/downloads/


Version 1.5

Version 1.5 is the result of a major re-factoring effort.

The former C ImageServer component has been rewritten entirely in C++ and also replaced functionality that was java code.

It is now also responsible for the configuration database and database communication as well as the http based application requests, in summation everything but the front-end bits of TissueStack.

Along the way a good deal of optimization measures were implemented and more are to come such as the memory-caching of image tiles to speed up on-the-fly tiling.

TissueStack binary packages have become much smaller since no java run-time is used any more, nor any application server is needed. JNI and inter-component communication through unix sockets is history as well.

Also: The offline pre-tiler and conversion tools got renamed. They need to be called like this:

TissueStackConverter -i IN_FILE (*.mnc,*.nii,*.nii.gz) -o OUT_FILE
The conversion is done in parallel by three processes (1 per dimension) if the platform has 3 cores or more. The pre-tiling is fine-grained enough (see command-line parameters) such that it can be started in parallel easily.


The new binaries can be found at:




Installation Notes:

The debian package has been built/tested on Ubuntu 14.04. It is very likely that it is not compatible with the latest Debian versions for differences in library versions that TissueStack depends on and/or apache/postgres configuration differences. On the plus side all dependencies can be gotten through the Ubuntu 14.04 repositories which is not the case for the redhat based distros (see below).

The install procedure is slightly tedious and involves the following steps:
sudo apt-get install apache2 postgresql  [to get dependencies that need to be configured beforehand]
sudo dpkg -i tissuestack_1.5_amd64.deb [which will fail, so don't be surprised!]
sudo apt-get install -f [which will deal with the causes of failure in the step above]

There is now an el6 version which was built and tested on CentOS 6.5.
It requires you to also download and install the builds of nifticlib (2.0.0) and minc (2.1.0) which can be found in the same location as the tissuestack rpm.

The Fedora 19 package only needs the minc (2.1.0) rpm to be installed which again can be found in the same location.

**** UPDATE *****

There is now also a Debian 7 and an OpenSUSE 13.1 package in the respective directories.


new Version 1.4 available

Latest Changes:

The internal .raw format used by Tissue Stack will store data as 8bit RGB triples while before it was limited to 8 bit grayscale.

This immediately leads to the following advantages:

  • Usage of RGB minc and NIFTI formats as input data is now possible.
  • On-the-fly image generation for a data set that frequently uses a color map may be replaced by a color data set instead.
  • Label Lookups can make use of the much larger space of lookup values (over 16 million), for instance volumes pointing to structure information encoded via RGB triples are not limited to represent 256 structures only.

  • Data Sets already converted and in use by Tissue Stack instances will still be compatible.
  • If you use RGB MINC files as input, the expected order is: spatial dimensions first, RGB channels are represented by the 4th dimension (length: 3).
  • If you use color NIFTI files, the required dimension layout is as follows: There are 3 spatial dimensions followed by a time dimension which will have length 1 in the simplest case with the RGB channels contained in the next dimension (length: 3)

Binary Installs (only 64bit):

Debian packages (compiled and tested on Ubuntu 12.10), Fedora rpms (compiled and tested on Fedora 19) and epel rpms (compiled and tested on CentOS 6.4) can be downloaded from either one of the following locations:


For the fedora packages you will need to download and install the minc library as well (same locations provided above).
The epel rpms require you to add the following repository to /etc/yum.repos.d in order to be able to satisfy the dependencies needed: http://cvlrepo.massive.org.au/repo/cvl.repo