Medical Image Analysis · Spatial Biology · AI/ML for Imaging
University of Illinois Urbana-Champaign · Expected 2027
I'm an MS student in Bioengineering & Imaging Computing at UIUC, focused on medical image analysis and computational approaches to spatial biology. My work connects quantitative imaging with biological context where I develop pipelines and models that extract biologically meaningful information from complex image data.
Currently:
- Active lab collaboration in computational image analysis using Cellpose
- Summer 2026 project: end-to-end CODEX/multiplex immunofluorescence analysis pipeline on the Schürch/Nolan CRC dataset — Mesmer segmentation, cell phenotyping, and Squidpy neighborhood analysis
- Deep learning for medical imaging: transfer learning with ResNet/EfficientNet, Grad-CAM visualization for model interpretability
- Statistical image analysis: quantitative methods for biomedical image data
| Domain | Tools & Libraries |
|---|---|
| Image Analysis | Cellpose · scikit-image · DICOM workflows |
| Spatial Biology | Squidpy · CODEX/multiplex IF pipelines · cell phenotyping |
| ML / Deep Learning | PyTorch · ResNet · EfficientNet · Grad-CAM · transfer learning |
| Scientific Python | NumPy · SciPy · pandas · matplotlib |
| Languages | Python, SQL |
| Workflow | Git · Jupyter · conda · PyCharm |
My BS in Biology and the MS combines coursework and research exposure in biological systems, stem cell engineering, cancer biology, and immunology. This palette shapes how I approach imaging problems. I'm not optimizing metrics in isolation; I'm thinking about what the biology demands of the analysis. That grounding is directly relevant to spatial biology applications: understanding tumor microenvironments, immune cell phenotyping, and tissue architecture requires both computational rigor and biological intuition.
Repos below represent active portfolio work. See each README for methodology, dataset details, and results.
Outside the MS curriculum, I'm deliberately building depth in deep learning for biomedical imaging including model interpretability, domain adaptation, and limited-label regimes.