Machine Learning Working Group
The Machine Learning working group deals with the processing of medical data. This mainly comes from the fields of diagnostic and interventional radiology. We place a particular focus on the translation of the developed techniques into the Clinical Department as part of clinical studies.
-
Semantic image segmentation
Many routine clinical tasks in radiology can be formulated as image segmentation tasks. In addition, this method offers the advantage of easy verifiability of the results and thus largely eliminates the problem of the "black box". The method is used, for example, for organ segmentation (lung, liver, kidney, etc.), but also for annotating other target structures such as tumors and metastases.
In cardiac imaging, fully automated segmentation enables the precise collection of therapy-critical bioparameters.
-
Translation
We work closely with industry partners to enable the newly developed methods to be used in clinical trials.
Chronic lung diseases are the fourth most common cause of death worldwide. The quantification of lung perfusion (blood flow) in chronic obstructive pulmonary disease (COPD) provides important clinical parameters for the development of new therapeutic concepts. As part of a scientific project, we have developed an evaluation pipeline that automatically processes and segments the data and calculates the relevant clinical parameters
