Multiscalar Machine Learning for the Integration of Quantitative in vivo MRI with ex vivo Analysis to assess Pathological Changes of the Extracellular Matrix
You are here:
The ECM is highly abundant in the extracellular space and has important and complex biological functions and roles.
Therefore, it represents an optimally suited and highly relevant target for molecular and biophysical MRI, enabling the
visualization and quantification of disease progression and response to therapy.
Currently, quantitative in vivo MRI and ex vivo data, including histology, immunohistopathology, Laser Ablation Inductively
Coupled Plasma Mass Spectrometry (La-ICP-MS) are analyzed separately using individual analysis techniques. For
correlation analysis, single parameters are extracted and related to each other. Deriving and integrating all information
from in vivo imaging and ex vivo analysis, for the prediction of disease progression and response to therapy, has the
potential to improve the identification of novel predictive multiparametric biomarkers.
The hypotheses for this project are, that the integration of advanced multiscalar imaging and machine learning (ML)
techniques enables (1) the identification of the most promising quantitative imaging biomarkers and ex vivo parameters
for the prediction of disease-progression and response-to-therapy in the disease models investigated within CRC 1340; (2) the assessment of the influence and interactions of the different ECM-components on quantitative imaging
parameters as well as ex vivo data; (3) the discovery of common principles and specific differences regarding the ECM-composition
between the different disease entities; (4) the development of intelligent tools for the advanced image
analysis, including structure identification, segmentation, registration and correlation of data-sets from X-ray based
and other methods in cooperation with other projects.
To test these hypotheses, we will first focus on the data from molecular/biophysical imaging and ex vivo analysis of
ECM components in the different disease models investigated within CRC 1340. All available in vivo and ex vivo datasets
from the first funding period, from the A (basic research) and B (biomedical applications) projects will be aggregated
and used to train and build advanced machine learning models. The developed machine learning models will be
tested retro- and prospectively regarding the prediction of disease progression and, based on data from the second
funding period, response-to-therapy.
The aspired gain in knowledge is, the discovery of novel insights into common principles and specific differences regarding
the role of the ECM-composition between the different disease entities and the interactions of in vivo and ex
vivo ECM biomarkers during disease development and progression including their combined predictive value.
Prof. Dr. rer. nat. Ingolf Sack
Prof. Dr. rer. nat. Anja Hennemuth
Dr. rer. nat. Frank Wiekhorst
Prof. Dr. med. Dipl.-Phys. Matthias Taupitz
Prof. Dr. rer. nat. Robert Bittl
Dr. rer. nat. Christian Teutloff
Prof. Dr. rer. nat. Birgit Kanngießer
Dr. rer. nat. Christian Seim
- Project A05 Prof. Dr. rer. nat. Kevin Pagel
Prof. PhD Peter Seeberger
Prof. Dr. nat. Dipl.-Pharm. Sarah Hedtrich
Prof. Dr. rer. nat. Jürgen Braun
Dr. med. Tobias Penzkofer
Prof. Dr. rer. nat. Fabian Praßer