Project KI-Morph

AI-supported automated segmentation of 3D image data for the analysis of morphological structures

The digitization of morphological characteristics and the correlation of large amounts of morphological data with other data are two of the central challenges of imaging in the life sciences.

The rapid progress currently being made in 3D imaging on large devices is enabling ever more efficient and faster data acquisition, including increasingly automated data analysis. However, the resulting petabytes of data lead to a data backlog affecting scientific usability since parts of the evaluation process are not yet able to be automated.

In this context, KI-Morph aims to jointly develop strategies through the use of artificial intelligence (AI) and with the involvement of representatives of user communities, measuring station operators, computer sciences and data facilities that are intended to enable the morphological analysis of large amounts of data through the automation of image analysis at a manageable overall cost. The aim is to be able to correlate existing metadata on a large scale with 3D morphological data series.

The University Computing Centre supports these research projects by providing IT services, developing the necessary processing pipelines and providing support services. The remote visualization service bwVisu, the high-performance computer bwForCluster Helix and the online storage service SDS@hd are used for this purpose. Furthermore work is being done on the technical and organizational processes for publishing and archiving the large volume of research data.