MSc Thesis: Prediction of Points-of-Interest on CT Vertebrae

The position of attachment points of ligaments and muscles on vertebral bodies are crucial for biomechanical simulations.

Your task would be to compare existing methods and develop a model to accurately predict these points onto given vertebrae. The task is similar to Landmark prediction (, which is used in deep-fake, for example.

You will be working with 3D CT images and segmentations. You are given a hand-made Ground-Truth for the points as well as a prediction of points using registration with some selected “known-goods”.

The tasks are:

  • Using the registrated data, make a preselection of the dataset.
  • Make a first baseline approach using the registrated data/points.
  • After a initial research phase, develop and train a DL model to outperform the registration approach.

Later steps could involve using the predicted POIs to measure the form of the vertebrae, i.e. for fracture detection.

Your Qualifications:

  • Most important: solid coding skills and familiarity with PyTorch and Numpy
  • Strong background in machine learning
  • Motivated master student in Informatics, Mathematics, or a closely related field
  • Ability to thoroughly answer a research question
  • Strong research mindset

How to Apply

Send an email to, with a short CV and your grade report. We promise to get back to you within days.

Hendrik Möller
Hendrik Möller
Affiliated Researcher

Transitional Vertebrae detection and labeling in medical images.