Strokes are the 2nd most common cause of death worldwide. For accurately treating acute stroke, a fast diagnosis is critical as the prognosis for the patient worsens with a delay of treatment. The Institute for Enterprise Systems is engaged in the publicly funded project Rettungsnetz 5G where a mobile stroke unit is implemented in Mannheim to facilitate rapid stroke care. Machine learning tools for stroke diagnosis and treatment decision support are already commercially exploited and used in modern hospitals. However, these tools lack insights into the functionality of the models, especially for specific underrepresented subpopulations.
In this Team Project, a decision support tool for acute stroke will be developed. Participants will get familiar with the state of the art in neural networks for CT image data and how to develop a deep learning model to be used in a real-world, critical care environment. Moreover, participants will analyze the application context and decision environment in the mobile stroke unit. Major aspects will be getting familiar with medical imaging data types, deep learning models, model behavior visualization as well as methods to how get insights into neural network predictions. Four students, with a strong interest in machine learning will work together with two software engineering students from Cluj-Napoca.