Procedures for a highly automated assessment and classification of cells in tissue sections by means of spatial marker profiles

In TisQuant we aim for a highly integrated computational tissue analysis comprising the identification of specific antibodies and adjusted staining methods, high-performance image analysis and pattern recognition techniques as well as new user interaction work flows.  For instance, for selected tumors this shall allow us an accurate computational assessment of tumor dignity and an assignment to their respective class of aggressiveness.

Inital situation and objectives 

In many biomedical research areas and in various biomedical laboratories the analysis of microscopy images of cellular structures is of high importance. New diagnostic procedures, in the course of a personalized medicine, often imply automated analysis systems fed with large data sets which shall support the diagnosis in combination with computerized microscopes. However, systems currently on the market provide accurate results with an acceptable effort for the user only for specific types of tissue and have shortcomings with respect to their functional usability. Moreover, they currently do not provide adequate results in the area of histological differential diagnosis.  

In project TisQuant we are developing an integrated analysis system for providing clinical research and routine laboratories with a highly automated application for the qualitative and quantitative analysis of microscopic cell and tissue compounds. By combining experimental procedures for visualizing cellular structures of interest with computational procedures for the analysis of the resulting images we aim to achieve a new level in the automated and computational analysis of tissue samples. Within the frame of project TisQuant this new technology shall be applied in clinically and commercially highly relevant use cases.

Subprojects and use cases 

On the one hand, the subprojects of project TisQuant are comprising experimental problems such as the search for specific antibodies, on the other hand, they address the conception of interactive image analysis approaches and work flows. The new techniques are applied within the project to use cases from the domains of tumor diagnosis, virology, and investigation of the effectiveness of anti-cancer drugs.


TisQuant’s results shall be commercially exploited by the company TissueGnostics GmbH, Vienna, as well as by the project partner microDimensions GmbH, Munich. For that matter it is planned to integrate the software modules developed in TisQuant into the respective Hard- and Software-Systems of these companies and to test them under laboratory conditions for the project’s use cases.


The first published results of project TisQuant concern

  • a work flow for the user guided adaption of parameters which are relevant for the segmentation of cell nuclei of DAPI stained cytopreparations as well as
  • a work flow for the quantification of cell type occurrences based on the segmentation of images of fluorescently labeled cytopreparations or tumor sections, respectively.

F. Kromp, S.Taschner-Mandl, M. Schwarz, J. Blaha,T. Weiss, P. F. Ambros, and M. Reiter. Semiautomated segmentation of neuroblastoma nuclei using the gradient energy tensor: A user driven approach. In SPIE Proceedings of ICMV 2014, Milano, 2014. 

F. Kromp, M. Reiter, S. Taschner-Mandl, P. F. Ambros, and A. Hanbury. Classification of cellular populations using Image Scatter-Plots, 20th Computer Vision Winter Workshop, February 9th to 11, Schloss Seggau, Graz, Austria