Shadowed researchers at the Yuanke Lab, a neuroscience research group focused on pericytes and their role in the brain. During this experience, Ryan observed researchers performing microscopy-based data analysis on mouse samples, assisted with data collection, and gained firsthand insight into the challenges of large-scale biological image analysis.

Working closely with the research team, Ryan then developed a custom AI-powered analysis platform tailored to their workflow. The system automates multi-channel pericyte microscopy analysis through an end-to-end pipeline that integrates image preprocessing, channel organization, and deep learning–based cell detection. To address real-world laboratory challenges, the platform supports heterogeneous microscopy formats, including high-bit-depth 16-bit TIFF images, and automatically handles channel mapping inconsistencies across datasets.

At the core of the platform are channel-specific YOLOv11 computer vision models trained through transfer learning to accurately identify and quantify pericytes across thousands of microscopy images. The tool is deployed as a web-based batch processing platform, enabling researchers to analyze entire datasets in hours rather than the months of manual work previously required.

By transforming a highly labor-intensive process into a scalable and reproducible workflow, the platform accelerates pericyte research and allows scientists to focus more on biological discovery rather than manual image annotation and analysis.

​

Paper in preparation and will be submitted to the Journal of Clinical Investigation Insights

 

Li Y, Liu Y, Klouda T, Kim Y, Wang R, Moss L, Kim E, Yuan K*, Pericyte mediated vascular remodeling in response to endothelial cell injury through Mif/Ackr3 pathway, 

​

AI Platform: https://pericytes.onrender.com