Intuitive Large Scale Image Processing for Biologists

OMERO and RAPID were pre-existing lines of development, see below. The challenges was to explore whether the principles and architecture unnderpinning RAPID would yield substantial benefits and improve usability for the global community of biologists who use OMERO. A particular goal was to handle very large 3D images, and long time sequences of images, which were emerging as essential for leading biology research and medical procedures, particularly pathology. For example, the measurement of the evolution of the density of various proteins marked with a florescent dye that responded to one of the available laser frequencies in a 3D space of >100,000x100,000x1000 voxels, or measuring the proportion of proteins currently bound in a reaction by using FLIM imaging, where the florence decay has to be analysed for each voxel. other examples, include counting the cells that were in various states, e.g., division or apoptosis. The form of RAPID needed to evolve, e.g. be disaggregated to embed in the existing web interfaces of OMERO and extended to use clouds and clusters. It was also necessary to automate data movement - the image sequences could occupy many gigabytes. With these adaptions, it was possible to demonstrate that the sought for benefits could be achieved for compelling examples and to provide well-tested code to be incorporated in OMERO releases.

OMERO, developed by Jason Swedlow at the University of Dundee, is a sophisticated and widely used biological image processing system that performs a wide range of advanced microscopy image processing methods, in order to extract measurements. RAPID, developed by the data-intensive research group, led by Atkinson, is a web-based job submission system, that hides details of the underlying technology and automates the job management, for a wide variety of target job servers by exploiting standards. THe purpose of RapidOMERO was to bring these two systems together to make it easy for biology researchers to carry out more computationally expensive processes than their own facilities could conveniently handle, e.g. using other clusters and cloud services. This in turn would releave load on the OMERO servers and open up the possibility of more advanced image analysis, which would previously have been too computationally demanding.

An integration of automated multi-target job management with automated data movement when necessary that could be integrated with OMERO and would be applicable in other such systems.
A demonstration of computationally demanding image processing being automatically offloaded from the image and microscopy servers.
A simple interface for biologists using advanced microscopy.
Facilities to run the most demanding image computations remotely.