Current workflows in orthopedic surgery or orthopedic research rely on 2D radiographic analysis for monitoring bone fractures and fracture healing. However, postoperative examination of surgically treated fractures using X-ray imaging poses a diagnostic challenge when foreign materials, such as osteosynthesis plates and bone screws, obscure the anatomical structures.
With this thesis, we will investigate deep learning approaches to automatically segment plates and screws, from post-surgical radiographs of tibial and femoral fractures. These segmentations will be utilized to explore two complementary strategies: (1) Development of 3D modeling techniques to generate patient-specific representations of implanted plates and screws, and (2) investigation of 2D inpainting approaches to remove foreign material from the pre-surgical X-ray image, to reveal the underlying bone structure.
The thesis's topic direction can be tailored based on individual interests, whether focusing on 3D geometric modeling or generative image reconstruction.
Your Profile
• Enrolled in a Master's program in Computer Science, Applied Mathematics, Biomedical Engineering, or a related field
• Strong interest and sufficient background in machine learning, computer vision, and medical image analysis
• Proficient in Python and deep learning frameworks (e.g., PyTorch, TensorFlow)
• Knowledge of 3D reconstruction techniques, neural shape representations, GANs or diffusion models is a plus
• Interest in interdisciplinary work combining AI with clinical applications
What We Offer
• Close mentorship and collaboration with experienced researchers in medical AI
• Access to unique datasets and high-performance computing infrastructure at ZIB
• Collaboration with clinical partners for real-world validation and impact assessment
• Opportunity to contribute to high-impact publications in medical imaging and orthopedic research
This Master's thesis is independent of any specific university. If you are interested in this topic, find a professor who supports you and let our researchers at ZIB supervise you with passion and devotion.
How to apply
Apply to Siloé Bournez (bournez@zib.de) and/or Dr. Stefan Zachow (zachow@zib.de), including supporting documents, such as a cover letter, your CV, enrollment confirmation, and transcripts. Please mention your specific interests in medical imaging, 3D reconstruction, or orthopedic applications.
