Spatial transcriptomics allows researchers to study gene expression in the context of tissue structure, but the microscopy images collected during these experiments are often not fully used after cell segmentation. My project will use scPortrait, a single-cell imaging framework, to extract image-based features from cells across a pan-cancer Xenium dataset, including squamous cell carcinoma, colorectal cancer, and other tumor types. Using a self-supervised vision model, I will test whether differences in cell shape and organization can help identify biological states such as injury, senescence, and immune activation, especially when RNA measurements are weak. By comparing these patterns across tumor types, this project asks whether changes in cell appearance can serve as a measurable readout of cell state beyond gene expression alone.

I am participating in SuperUROP because I want to build upon the computational biology skills I have developed through my coursework and apply them to meaningful biological questions. Through classes such as 20.260 (Computational Analysis of Biological Problems), I have gained experience analyzing biological datasets and thinking computationally about complex biological systems. I am excited to strengthen these skills through a longer-term research project, contribute to my lab’s work, and grow as a computational biologist.