Development

We have a number of software development projects under way to further the aims of the INCF Swiss node.

Web-based Annotation of Image Data

Stephan Gerhard, Stephan Saalfeld and Mark Longair are working on a new version of the CATMAID web framework. This is a web-based framework for annotating massive image data sets, such as those generated with ssTEM or SBFSEM, from any web browser.  The current development efforts are focussed on:

• Providing a new "treeline" annotation types to support tracing skeletons of neurons.
• Developing a more flexible in-browser windowing system
• Providing support for semantic querying of relationships based on annotations
• Adding 3D visualizations of annotated structures

The primary aim of these development efforts is to create a system to allow many users to annotate the same data set in parallel, seeing changes reflected in real time in their browser, without the requirement for expensive hardware.

A screenshot of the current development version can be seen below, showing project statistics in the left hand pane, the logical structure of annotations in the middle pane and the image with annotations on the right hand side.  Three types of annotation are shown: text labels, nodes making up treelines (in blue) and markers of synapses (red dots).

The annotations are stored with a data model that represents these as Subject-Relationship-Object triples, so that ultimately we can allow semantic querying of this data.  A draft description of this data model is available, although there are many changes pending to this document so that the naming matches the standard concepts recommended by the INCF's Program on Ontologies of Neural Structures.

We have now published the source code for this new version of CATMAID on GitHub.  The "svgoverlay" branch is where development is currently active.

You can see examples of image data sets viewable using the previous version of CATMAID via the Swiss Node's data page

We anticipate making a public release of this new version of CATMAID early in 2011.  If you would like more information please contact Mark Longair.

Fiji / ImageJ

Fiji is a widely used distribution of ImageJ that packages many useful plugins for biomedical image analysis.  We are active developers of Fiji, in particular of the TrakEM2 and Simple Neurite Tracer components, and are happy to provide support, bug fixes and advice on using this software for image analysis.  (To report a bug in Fiji, just go to "Help > Report a Bug".)

In particular we are working on making it easier to produce and distribute plugins for Fiji that were originally developed using non-Java libraries and programming languages, such as C++, Matlab and ITK, to enable a faster transfer of technologies from computer vision and image processing groups to biological researchers.