Theses and Research Projects

You must fulfill all requirements of your FPO before applying to a thesis!

Please directly contact a team member of SPARC if you have a particular interest in a project or topic for a thesis. Do not forget to include a transcript of records with all applications!

Master theses (30 ECTS):

Description

The objective of this thesis is to develop and evaluate methods for three-dimensional (3D) reconstruction based on stereo images acquired with a stereo endoscope. Such reconstructions are essential for enhanced navigation, visualization, and surgical assistance in minimally invasive procedures. The work includes calibration of the stereo endoscope, development of stereo-matching and depth-estimation algorithms, and generation of accurate 3D surface models. Depending on the thesis type (Bachelor, Master, or Research Project), the focus can range from fundamental implementation to advanced optimization and validation in surgical scenarios.

Tasks

  • Calibration of the stereo endoscope: Determination of intrinsic and extrinsic camera parameters. Correction of distortions and alignment of stereo image pairs.

  • Stereo image preprocessing: Rectification and illumination normalization. Handling of reflections and challenging endoscopic imaging conditions.

  • Depth estimation and stereo reconstruction: Implementation and comparison of stereo matching algorithms (e.g., block matching, semi-global matching, deep-learning-based methods). Generation of dense disparity maps and conversion into 3D point clouds or surface meshes.

  • Evaluation and validation: Testing on synthetic datasets and real endoscopic images. Quantitative and qualitative assessment of reconstruction accuracy.

Supervisor:

Dr.-Ing. Christian Kunz

Group Leader / Postdoctoral Fellow

Department Artificial Intelligence in Biomedical Engineering (AIBE)
Surgical Planning and Robotic Cognition Lab

Research Projects and Master Thesis Projects (10 ECTS):

The goal of this project is to develop a reinforcement learning agent for autonomous guidewire navigation in simulated vascular anatomies. The work will build on the open-source stEVE_training framework.

Unlike existing approaches that mainly use positional information, the agent will receive force-based observations generated by interactions between the guidewire and the vessel wall. The aim is to investigate whether force signals can be used to navigate safely toward predefined targets.

The student will adapt the simulation environment, define suitable force-based reward functions, train the agents, and evaluate its performance. Results will be compared with a position-based baseline using metrics such as navigation success, procedure time, applied forces, vessel-wall contacts, and generalization to unseen anatomies.

Tasks

  • Extend stEVE_training with force-based observations
  • Adapt the reinforcement learning environment and reward function
  • Train and optimize the navigation agent
  • Evaluate safety, performance, and generalization
  • Compare the results with a position-based approach

Requirements

  • Master’s student in Medical Engineering
  • Good Python programming skills
  • Excellent knowledge of machine learning and reinforcement learning
  • High interest in medical robotics and simulation
JF

Contact