Master’s thesis: Deep learning-based segmentation of cerebrospinal fluid at brainstem level in human MRI








Uploaded on 16 Nov 2023

Background: Recent fMRI studies in humans have shown that the reduction of correlations between cerebrospinal fluid (CSF) flow in the brainstem and global activity in the brain’s grey matter is linked to impaired waste product clearance from the brain in neuropsychiatric diseases. Our group has developed a semi-automated pipeline to extract brainstem CSF and grey matter signals from fMRI images. We are interested in optimizing these processes, particularly in a fully automated, multi-modal segmentation of CSF voxels at the brainstem level based on both functional and structural MRI data.



Zimmermann et al., bioXiv 2023

Han et al., 2021








Project description: The project aims to develop a deep learning-based method for slice-wise automatic segmentation of CSF voxels in fMRI signals.

What we offer to you:

  • We offer a master’s thesis on “Deep learning-based segmentation of cerebrospinal fluid at brainstem level in human MRI”.
  • You are an active member of our international and interdisciplinary research team working on a cutting-edge research question.
  • You work in close collaboration with medical experts from neuroradiology at Klinikum rechts der Isar as well as computational scientists from TUM.
  • You will learn to know and apply the latest segmentation architectures and neuroimaging tools.

What sets you apart:

  • You are studying computer science, biomedical image computing (BMC), or a comparable course in a master’s degree.
  • You have gained experience in Python programming with libraries such as PyTorch, Pandas, and/or OpenCV.
  • You have a good understanding of mathematics and image processing.
  • You have some knowledge about neuroimaging and its data formats.
  • You are willing to work in an interdisciplinary team at the intersection of computation and medicine at Klinikum rechts der Isar.
  • You are fluent in English.
  • You are characterized by an independent way of working, a high degree of motivation and commitment, and the willingness to take on responsibility quickly.

This project is a collaboration between Departments of Radiology, Neuroradiology and Anesthesiology at Klinikum rechts der Isar, TUM.

Contact Information:

Fabian Bongratz (, Artificial Intelligence in Medical Imaging Lab, Department of Radiology, TUM

Juliana Zimmermann (, Department of Anesthesiology, TUM

Christian Sorg (, Department of Neuroradiology and Psychiatry, TUM