We perform research activities on different aspects of AI technology. Here you find the overview of ongoing projects.
Intelligent Robots for Autonomous Ultrasound Imaging
Medical ultrasound imaging provides radiation-free, real-time, and cost-effective imaging of internal organs. However, a high amount of expertise and experience is required to acquire and interpret images in a meaningful way. In addition, image acquisition is hardly reproducible even among experts. Automating ultrasound acquisition using a collaborative robot can take over the often repetitive processes of image acquisition while increasing the reproducibility of examinations. The goals of this project are to reduce the workload of medical staff in the clinic, to enable extensive screening by autonomous systems, and to use the system in environments that are hostile to humans, such as radiation therapy.
Our artificial intelligence research in this project can be divided into two central areas. In the area of image-based robot control (visual servoing), we are developing deep neural networks (deep learning) for autonomous analysis of ultrasound images, which are then translated into robot movements. This can be used, for example, to track specific target structures, for automated diagnostics, or for imaging large organs that require coverage of larger body regions.
The second major project area focuses on methods that enable the robot to continuously learn sensorimotor skills, e.g. probabilistic reinforcement learning. This makes it possible to independently position the ultrasound probe on the patient and, in particular, to react to unforeseen external influences such as movements of the patient.
Prof. Dr. Floris Ernst (firstname.lastname@example.org)
- Daniel Wulff, Jannis Hagenah, Svenja Ipsen und Floris Ernst, Learning Local Feature Descriptions in 3D Ultrasound, in: The 20th IEEE International Conference on BioInformatics and BioEngineering, IEEE Computer Society, Seiten 323-330, 2020
- Felix von Haxthausen, Jannis Hagenah, Mark Kaschwich, Markus Kleemann, Verónica García-Vázquez und Floris Ernst, Robotized ultrasound imaging of the peripheral arteries – a phantom study, in: Current Directions in Biomedical Engineering, 20200033, De Gruyter, 2020
- Svenja Ipsen, Sven Böttger, Holger Schwegmann und Floris Ernst, Target tracking accuracy and latency with different 4D ultrasound systems – a robotic phantom study, in: Current Directions in Biomedical Engineering, 6:1, 20200038, De Gruyter, 2020
- Sven Böttger, Tolga-Can Çallar, Elmar Rueckert und Achim Schweikard, Medical robotics simulation framework for application-specific optimal kinematics, in: Current Directions in Biomedical Engineering, 5:1(145-148), De Gruyter, 2019