Machine Learning and Computational Modelling at TUM (Prof. Alim's group)
SegmentationDeveloped and trained a Vision Transformer for automated segmentation of vascular structures in microscopy images as part of machine learning and computational modeling research. The work involved creating training data for semantic segmentation of vascular anatomy and evaluating model performance on microscopy-derived targets. This pipeline supported reproducible preprocessing, simulation, and statistical evaluation steps for the segmentation task. • Labeling focus: vascular structure masks/regions in microscopy images for segmentation ground truth. • Model training: Vision Transformer architecture for segmentation. • Tooling: MATLAB and Python used for simulations and computational workflows. • Output: trained segmentation model for automated detection of vascular structures.