FURI | Spring 2024

Integrating Machine Learning With a Cancer-On-A-Chip Model to Characterize the Morphological Response of SUM159 Cells to Infiltrative Stromal Cells

Health icon, disabled. A red heart with a cardiac rhythm running through it.

Cellular morphology is an emerging indicator of cellular state especially related to cancerous processes like metastasis and stroma invasion. Cancerous cells have been observed to undergo a pseudo epithelial mesenchymal transition in order to assume a more invasive phenotype with a stereotypically sharp morphology. Herein, it is hypothesized that machine learning can be used to correlate specific morphological features to tumor progression and invasiveness. In this case, quantitative measurements are concatenated with pixel-level features extracted from a deep learning algorithm to articulate a greater morphological profile, which can characterize how cancer cells respond to environmental modulation.

Student researcher

Ethan Hurt

Biomedical engineering

Hometown: Boise, Idaho, United States

Graduation date: Spring 2025