M.Sc. Lorena Krames

B.Sc. Antonia Neidhardt

Motivation
Laparoscopic liver surgery is a minimally invasive technique offering several advantages over traditional open surgery, such as smaller incisions, reduced pain, and less scarring. However, it is technically more challenging and demands highly skilled surgeons. To assist during surgery, we are developing a navigation tool that overlays preoperative data onto laparoscopic images. Currently, the required methods are implemented in Python. In the next phase, this software will be integrated into real laparoscopic hardware.

Project Description / Tasks
The main task is to convert different components (e.g., the liver segmentation method) of the existing registration pipeline into the ONNX (Open Neural Network Exchange) format. This enables integration into a real hardware system.
Parts of the project:
• Implement a test environment to compare model performance across formats (ONNX vs. Python).
• Define suitable performance metrics.
• Document the performance comparison in a report.