Forschung

The Computational Pathology group seeks to understand the relationships between cell morphology, function, and underlying molecular features. Deep learning and quantitative image analysis are used to link various features to morphology.
We aim to closely link our research to the clinic and improve diagnostic decision making. To this end, we aim to develop tools that support medical professionals and quickly provide information on additional features of a tumor based on routinely generated H&E sections.

In gastric cancer, four molecular subclasses are defined (CIN, EBV, GS and MSI), which will be predicted based on morphological features extracted from H&E stained tissue sections using convolutional neural networks (cNNs). Here we also focus on tumor heterogeneity and its influence on the training process and we will use multiplex imaging for further validation.
Publication: 
Flinner N*, Gretser S*  et.al. 2022; PMID: 35119111

We are studying different types of lymphomas (cHL and ALCL) that share some molecular characteristics (e.g. positivity for CD30) but have different clinical courses. We quantify cell morphological structures used for cell migration, quantify cell movement, and use cNNs to characterize migration type to gather information that could help understand early spread of ALCL. 
Publications: 
Bein J*, Flinner N* et al. 2022; PMID: 35586951
Goncharova*, Flinner* et al. 2019; PMID: 31581676

The filamentous cyanobacterium Anabaena sp. PCC7120 is abele to differentiate single cells into heterocysts in case of nitrogen deficiency. Morphological features should be extracted from EM images (acquired at different timepoints of the differentiation process) and are correlated to the respective gene expression profile in order to identify the genes which are involved in the morphological changes using different techniques like e.g. Person, Spearman or MIC/MINE. At the end the whole differentiation process is modelled using the POTTS model. This project is part of the CMMS.

„Zellbasierte Modellierung – Von den Daten zum Modell“ in cooperation with Prof. Dr. Franziska Matthäus OLAT-link 

Projects for Bachelor or Master thesis are constantly offered to interested students.