Glioblastoma (GBM) is a grade 4 and most aggressive central nervous system tumor, accounting
for ~14% of all tumors and over 50% of malignant brain tumors. Prognosis is poor, with 50% 1-
year survival and 10% surviving beyond 5 years. Incidence increases with age.
Histologically, GBM is characterized by a necrotic, hypoxic core and abnormal, highly
proliferative blood vessels with a leaky blood–brain barrier. MRI is the primary diagnostic tool:
contrast-enhanced (CE) T1 highlights tumor regions with blood–brain barrier disruption (often
surrounding necrosis), while T2/FLAIR shows non-contrast-enhancing (NCE) areas like oedema,
which may still contain infiltrating tumor cells not clearly detectable on standard imaging.
[18F]FET PET imaging uses a radiotracer that accumulates in gliomas, improving tumor
visualization, grading, prognosis assessment, and helping detect tumor spread beyond
conventional MRI-visible tumor regions. However, research has shown that advanced MRI
techniques such as ADC can capture (partly) this infiltrating tumor region present in this nonenhancing
area.

You will Investigate whether advanced MR techniques can detect active tumor beyond the CE
region seen on conventional MRI, using PET-defined active regions as ground truth, including
activity extending into NCE areas.
You will implement state-of-the-art deep learning models to segment tumor infiltration and
evaluate the performance using both quantitative metrics and visual analysis through Slicer 3D.

The following tasks are set in this project:
- Literature study on DL-based segmentation in medical imaging and for infiltration
- Identify DL-based architecture for segmentation task
- Create baseline results with conventional MRI
- Apply on advanced MR sequences and generate new results
- Analysis and comparison of new results against baseline results and current SOTA
Student will work in Python and implement training schemes using packages such as PyTorch,
MONAI, NumPy, and perform analyses with SimpleITK and visualisations on Slicer 3D.

Mandatory qualifications
- Following an MSc in a field related to one or more of the following: Computer Science,
Biomedical Engineering, Applied Computer Science - Digital Health.
- Strong programming skills (Python).
- Ability to write scientific reports and communicate research results in English.