Disturbances in the Medical Application of Artificial Intelligence

The processing of medical data is a key area of application for artificial intelligence (AI). However, data anomalies pose significant challenges for the development of high-performance AI models and their safe clinical application. We are conducting research and development into high-performance AI methods and technologies for the automatic detection and reliable processing of data anomalies.

We are carrying out this work as part of the Research Training Group “Disturbances in the Medical Application of Artificial Intelligence”, in collaboration with Mannheim University of Applied Sciences at the Institute for Medical Technology (IMT). In this project, we are investigating data anomalies in the collection of medical data, their representation in clinical patient cases, and their processing in diagnostics.

In radiology, artefacts in image data can significantly reduce the quality of the radiomics and AI results derived from them. Novel methods for data augmentation are designed to perform reliably in the presence of data anomalies, thereby ensuring high-quality results. In data integration centres, anomalies in case data represent potential risk factors for the clinical treatment processes based on them. Innovative technologies are designed to automatically quantify these anomalies and ensure that the origin and processing of the case data are traceable. In radiotherapy, atypical patterns and trends in case data may indicate rare but particularly significant clinical conditions. The aim is to reliably identify and plausibly analyse these using artificial intelligence methods (XGB, ANN, XAI, etc.).

Using these innovative methodological and technological solutions, the Research Training Group aims to bring about a comprehensive improvement in the clinical data flows and workflows of radiology and radiotherapy.

Project partners

• Department “Complex Data Processing in Medical Informatics”, MIISM, Medical Faculty in Mannheim – University of Heidelberg
• Department for Radiology and Nuclear Medicine, University Hospital Heidelberg – University Clinic Mannheim
• Department “Biomedical Informatics”, MIISM, Medical Faculty in Mannheim – University of Heidelberg
• Research Group „Applied Artificial Intelligence in Radiation Oncology”, MIISM, Medical Faculty in Mannheim – University of Heidelberg
• Department for Radiotherapy and Radiation Oncology, University Hospital Heidelberg – University Clinic Mannheim
• Research Group “Biomedical Chemistry”, Medical Faculty in Mannheim – University of Heidelberg
• Centre for Mass Spectrometry and Optical Spectrometry (CeMOS), Mannheim University of Applied Sciences
• Institute for Applied Artificial Intelligence and Robotics (A²IR), Mannheim University of Applied Sciences
• Faculty of Computer Sciences, Mannheim University of Applied Sciences