Open Positions
Postdoctoral position in Computerized Image Processing with focus on Applications in data-driven precision medicine
Application deadline: Dec 5, 2025
Master’s thesis projects offered for VT 2026
At our lab we host a limited number of master’s thesis projects, listed below.
Extending AI-Based Cell Graph Analysis of Cancer Tumor Microenvironments to 3D
3D reconstruction of tissue by registration of 2D images of tissue slices acquired by multichannel (multiplex) immuno-fluorescence microscopy. AI-driven analysis of cell graphs constructed on these images, in 2D and 3D, on a selected classification task.
AI-supported Survival Prediction from Multichannel Microscopy Images of Cancer Tissue
Performance evaluation of selected state-of-the-art models (e.g., foundation models) applicable to multichannel microscopy images (e.g. multiplex immunofluorescence, mIF) on the task of cancer survival prediction.
Impact of Mixed Witness Rates on Malignant Cell Detection in Cytology
Explore techniques on how to best train deep networks on heterogeneous medical data without biasing the learning towards the most frequently appearing objects/cases, to maximize performance also for more rare cases.
Beyond Grids: Dynamic Placement of Tokens in Vision Transformers for Efficient Learning on Sparse Biomedical Data
Novel tokenization strategies for vision transformers (ViT), that position tokens continuously within images instead of the typical regular grid, promise substantial performance gains for exploiting sparse image information. This project aims to utilize such techniques for improved learning on sparse whole slide cytology image data.
Pipeline for Fast and Robust Multimodal Image Registration of Whole Slide Images
Pipeline for Fast and Robust Multimodal Image Registration of Whole Slide Images Develop an image registration pipeline for aligning whole slide images, acquired by brightfield and fluorescence microscopy. Identify the best methods available, build a practical useful tool, and integrate it into our existing computational pathology workflow.
Deep Learning Based Focus Interpolation for Microscopy
To explore to what extent it is possible to confidently predict an image at a focus level in between two or more images acquired, i.e., to perform focus interpolation, possibly relying on modern generative models such as variational autoencoders or stable diffusion.
Deep Learning in the Browser – AI-supported Cancer Detection for the Masses
To explore if and how efficient Deep Learning inference and training for AI-supported early cancer detection, can be performed on the browser of the medical doctor instead of, as today, on a central server.