AI in Surgery

Overview

This practical course introduces students to current research topics in AI in the context of surgery. The course is offered by the Chair for Computer Aided Medical Procedures (CAMP) with co-supervision from international partner laboratories and clinical partners, including Harvard Medical School / Brigham and Women's Hospital, Imperial College London, The Chinese University of Hong Kong, INRIA, Vanderbilt University, LMU Klinikum and TUM Klinikum Rechts der Isar.

Students work in small teams on research-oriented projects that connect modern AI methods with surgical and clinical problems. The course covers topics such as perception, reasoning, and autonomy in abdominal surgery and neurosurgery; medical registration; surgical scene understanding; operating-room understanding; computer vision; vision-language models; world models; and vision-language-action models.

Project topics are jointly defined by CAMP, clinical collaborators, and international partner labs. Students may work on questions such as:

How can AI models understand surgical scenes from video, imaging, or multimodal clinical data?
How can perception and reasoning models support safer and more autonomous surgical workflows?
How can medical registration and intraoperative data be used for navigation and decision support?
What are the technical and clinical limitations of deploying AI systems in surgery?
How can research prototypes be evaluated rigorously in collaboration with clinical partners?

Preliminary meeting:

Time: Monday, July 6th, 5 pm
Link: https://teams.microsoft.com/meet/38810132761554?p=gdAOKKP4flNrLQcncr
Meeting ID: 388 101 327 615 54
Passcode: 3A392s3Z

Meeting recording: Recap: AI in Surgery - Preliminary Meeting 6. July | Meeting | Microsoft Teams

Prerequisites and Registration

Introduction to Deep Learning, Machine Learning, Computer Vision, or a comparable course
Practical programming experience in Python and deep learning frameworks such as PyTorch or TensorFlow
Interest in AI research for surgical, clinical, or medical-imaging applications
Background in one or more of the following is beneficial but not mandatory: computer vision, vision-language models, world models, vision-language-action models, medical image analysis, robotics, or clinical data analysis
Registration must be done through TUM Matching Platform (please pay attention to the Deadlines)
If you select this course in your matching, please also send an application email with CV + Transcript to maximilian.fehrentz@tum.de
Your chances to be assigned to the course increase if you give the course a higher rank in your choices.
The maximum number of participants: 15.

Objectives

After successful completion of the Praktikum, students are able to:

work in a team on an applied AI research project in a surgical or clinical context;
understand how modern AI methods can support perception, reasoning, autonomy, and decision support in surgery;
translate a clinical or surgical research question into a computational project plan;
design, implement, and evaluate AI models for medical-imaging, surgical-video, multimodal, or robotic data;
critically assess limitations related to data quality, bias, safety, interpretability, and clinical deployment;
communicate project results through a scientific report, code deliverables, and a final presentation.