Projects
A selection of my work: research in medical imaging and wearable sensing at KU Leuven, and applied AI projects at Sentigrate.
Featured
Multimodal retinal imaging for early Alzheimer's detection
Can an eye scan reveal signs of amyloid in the brain before dementia sets in? I built the full pipeline, from image quality control and preprocessing to feature extraction and classification, combining hyperspectral imaging, OCT and fundus photography in 40 participants with PET-confirmed amyloid status. A new hyperspectral texture biomarker, combined with OCT, reached an AUC of 0.84 in this proof-of-concept cohort.
Applied AI at Sentigrate
Data Scientist, since November 2025
Research at KU Leuven
Research Group Ophthalmology & ESAT-STADIUS, and master's theses
Vision foundation models for glaucoma classification
Glaucoma classifiers often break down when moved to a new camera, clinic or population. I fine-tuned self-supervised vision foundation models on multiple source datasets and applied domain adaptation to make classification robust to these shifts.
Seizure detection from multimodal wearables
An automatic seizure detection algorithm for people with epilepsy, using wearable EEG, EMG and accelerometer data. Features are extracted per modality and classified with an SVM. Combining modalities improved sensitivity and reduced false alarms.
Camera-based behaviour monitoring of laying hens
Early detection of behaviour linked to poultry red mite infestation in individual hens. A YOLOv5 head detector, fine-tuned on a custom-labelled dataset, is combined with Gaussian mixture model motion detection.