Master Thesis: Neural Surrogate Models for Biomechanical Forces in the Spine

08.09.2025, Abschlussarbeiten, Bachelor- und Masterarbeiten

Master Thesis: Neural Surrogate Models for Biomechanical Forces in the Spine Accurately determining biomechanical forces in the spine is essential for many medical applications. Current approaches rely on complex multi-body simulations [1] that are computationally demanding. In this thesis, you will investigate neural surrogate models trained on large synthetic datasets to predict biomechanical forces from CT images. The aim is to achieve accurate and fast predictions that can make these methods applicable in clinical workflows. AI for simulation [2] is a rapidly emerging field and will play a major role in the future of medical technology. With state-of-the-art methods such as transformers, there are powerful tools available that can be applied to this problem. The exact direction of the thesis is flexible, and extensions such as body movement simulation or related applications can also be explored depending on your interests. Your qualifications: - Strong motivation to apply AI for impactful medical applications - Background in computer science, physics, biomedical engineering, or a related field - Solid Python programming skills, ideally with PyTorch - Prior experience in medical imaging, biomechanics, or machine learning is a plus We offer: - Supervision in collaboration with experts in AI and biomechanics - Access to advanced computational resources - The opportunity to contribute to cutting-edge interdisciplinary research How to apply: Please send your CV and transcripts to j.weidner@tum.de and tanja.lerchl@tum.de Group: AI-IDT, Prof. Benedikt Wiestler, https://ai-idt.github.io/ [1] Lerchl, Tanja, et al. "Musculoskeletal spine modeling in large patient cohorts: how morphological individualization affects lumbar load estimation." Frontiers in Bioengineering and Biotechnology 12 (2024): 1363081. [2] Alkin, Benedikt, et al. "Universal physics transformers: A framework for efficiently scaling neural operators." Advances in Neural Information Processing Systems 37 (2024): 25152-25194.

AI for simulation [2] is a rapidly emerging field and will play a major role in the future of medical technology. With state-of-the-art methods such as transformers, there are powerful tools available that can be applied to this problem. The exact direction of the thesis is flexible, and extensions such as body movement simulation or related applications can also be explored depending on your interests.

Your qualifications:

- Strong motivation to apply AI for impactful medical applications

- Background in computer science, physics, biomedical engineering, or a related field

- Solid Python programming skills, ideally with PyTorch

- Prior experience in medical imaging, biomechanics, or machine learning is a plus

We offer:

- Supervision in collaboration with experts in AI and biomechanics

- Access to advanced computational resources

- The opportunity to contribute to cutting-edge interdisciplinary research

How to apply:
Please send your CV and transcripts to j.weidner@tum.de and tanja.lerchl@tum.de

Group:
AI-IDT, Prof. Benedikt Wiestler, https://ai-idt.github.io/



[1] Lerchl, Tanja, et al. "Musculoskeletal spine modeling in large patient cohorts: how morphological individualization affects lumbar load estimation." Frontiers in Bioengineering and Biotechnology 12 (2024): 1363081.

[2] Alkin, Benedikt, et al. "Universal physics transformers: A framework for efficiently scaling neural operators." Advances in Neural Information Processing Systems 37 (2024): 25152-25194.

Kontakt: j.weidner@tum.de and tanja.lerchl@tum.de