Student Thesis
Specifically advertised offers for theses can be found under "Open Topics." Under "Subject Areas" you will find contact persons for fields where there are various possibilities for theses that you can discuss in detail with the respective contacts.
No Topic for you?
If there isn't a specific topic that interests you, but you would like to write a thesis with us in our thematic and research areas, feel free to contact us with your ideas, preferences, or even topic suggestions.
Additionally, it may be worthwhile to check our institute's webpage for offerings from all locations.
Subject Areas
Shape Models of Vertebrae for Robust Segmentation on MRI
Jo Lienhoop
In the diagnosis of back pain and injuries, CT and MRI scans are utilized. To process these automatically, the vertebrae in the data are segmented. This is particularly challenging in MRI scans because bones are not as clearly depicted as in CT scans. However, the procedure is free of radiation.
This work aims to improve the segmentation of vertebrae in MRI data by creating shape models of vertebrae. These shape models represent the anatomical variability of different vertebrae and can produce anatomically plausible segmentations. This is especially helpful for MRI scans with low resolution.
Estimation of Hybrid Joint States with Inertial Sensors
Aashrita Roy Potla
Motion capture systems based on inertial measurement units (IMUs) can measure the orientation of human body joints. These systems are commonly used in posture analysis, film production, and clinical movement assessment. Human joints like the knee are not ideal rotational joints but also allow translations. Translational movements are valuable for clinicians to diagnose and assess joint injuries.
In this work, we aim to develop a hybrid joint model to estimate the rotational and translational movement of simple joints based on IMU measurements. This work is a step toward the quantitative assessment of joint injuries.
Tracking with Non-Spherical Markers
Erik Immoor
Motion capture systems use spherical markers to track the position and orientation of limbs or tools. The markers reflect infrared light back to the cameras, and their position is calculated through triangulation. To estimate orientation, at least three markers in a rigid configuration (star marker) are necessary. This work will evaluate the potential of non-spherical markers in a simulation environment.
Anomaly Detection on MRI Breast Cancer Screening Data
Novruz Mammadli
Breast cancer is the most common type of cancer in women (Source). Therefore, various screening procedures are offered to women to improve early detection. High-risk patients may receive MRI scans.
In this work, we aim to use anomaly detection methods to identify abnormalities that may indicate tumors. We will analyze which approaches are suitable for the data.