Direkt zum Inhalt springen
login.png Login    |
de | en
Technical University of Munich

Technical University of Munich

Sitemap > Bulletin Board > Diplomarbeiten, Bachelor- und Masterarbeiten > Master's thesis in Medical Image Processing: Automatic sensor-detection in dynamic videofluoroscopy image sequences
up   Back to  News Board    previous   Browse in News  next    

Master's thesis in Medical Image Processing: Automatic sensor-detection in dynamic videofluoroscopy image sequences

01.09.2023, Diplomarbeiten, Bachelor- und Masterarbeiten

Master's Thesis at Klinikum rechts der Isar in the area of Medical Data Science


Patients with benign diseases of the esophagus often have a long medical history until their final diagnosis. Depending on severity and symptoms of the disease, patients undergo a wide variety of examinations such as EGD, manometry, pH-metry, CT/MRI, videofluoroscopy, FEES, etc. While all examination results are compiled piece by piece and gradually depict the entire aspects of the disease, it is still extremely difficult to link the individual examination modalities with each other.

In order to reach a more intuitive diagnosis, a combined diagnostic approach should be developed, which provides the optimal visualization of the esophageal motility through the combination of the different modalities.


The goal of this work is the automatic projection of manometry values onto the corresponding sensor locations in the videofluoroscopy (xray) image sequences, in order to allow a synchronized and combined visualization of the two modalities.

Multiple approaches (e.g. traditional methods in the area of Computer Vision, but also Deep Learning based approaches) can be considered for this problem, which should be implemented, evaluated and compared to each other, to identify the best possible method for the present problem.


  • Literature research on similar existing solutions
  • Collection and potential annotation of required training data set
  • Development, evaluation and comparison of methods to automatically detect the manometry sensors in the cinematographic images and project the collected values onto the corresponding location within the images
  • Written documentation of the methods and results


  • Extensive Python knowledge
  • Advanced knowledge in the areas of Machine Learning & Computer Vision and experience with common libraries (e.g. Tensorflow/PyTorch, OpenCV, etc.)
  • Interest in medicine and applied research
  • Motivation and independent way of working

Please contact us if you are interested in this topic.

Kontakt: alexander.geiger@tum.de, alissa.jell@tum.de

More Information


illustration Sensor-detection in dynamic videofluoroscopy image sequences, (Type: image/jpeg, Size: 60.0 kB) Save attachment