I develop machine learning methods for spatial omics, with a focus on graph neural networks, self-supervised learning, and multiple instance learning for cancer biomarker discovery.
Research Interests#
- Spatial Transcriptomics — computational methods for spatially resolved gene expression data, modelling the cellular microenvironment in cancer tissue
- Graph Neural Networks — leveraging tissue architecture as graphs to capture cellular organization patterns relevant to precision oncology
- Self-supervised Learning — learning representations from spatial omics data without relying on scarce clinical labels
- Multiple Instance Learning — whole-slide reasoning over spatial structures for patient-level predictions
- Cancer Biomarker Discovery — identifying structural and molecular patterns for non-small cell lung cancer and beyond
Conference Presentations#
- Dec 2025 — Exploring Augmentation-Driven Invariances for Graph Self-supervised Learning in Spatial Omics — Workshop on Unifying Representations in Neural Models @ NeurIPS (poster)
- Sep 2025 — GRASS-MIL: Graph-based Representation and Analysis of Spatial Structures with Multiple Instance Learning — Basel Computational Biology Conference (poster)
- Sep 2024 — Spatial Omics Analysis for Hypothesis Generation Using Graph Neural Networks to Explore Cellular Organization in Non-Small Cell Lung Cancer — scverse Conference, Munich (poster)
- Jun 2024 — Spatial Omics Analysis for Hypothesis Generation Using GNNs to Explore Cellular Organization in NSCLC — Spatial Omics Conference, Ghent (poster)
Conference Service#
- Sep 2025 — Reviewer, Workshop on Unifying Representations in Neural Models @ NeurIPS