Application

Uses artificial intelligence to analyze collagen fiber architecture in digital pathology images to predict cancer prognosis and support personalized treatment decisions.

Key Benefits

• Uses routinely acquired digital pathology images to provide prognostic information without requiring additional tissue testing.
• Delivers interpretable risk assessments based on measureable collagen architecture features.
• May provide a cost-effective alternative or complement to molecular and genomic testing for patient stratification and treatment decision support.

Market Summary

Accurate cancer prognosis remains a significant challenge in oncology. While traditional staging systems and molecular tests provide important clinical information, they may not fully capture the biological complexity of a patient’s tumor. Genomic and molecular testing can also be costly, time-consuming, and may not be readily accessible in all healthcare settings. At the same time, pathology slides are routinely collected as part of standard cancer diagnosis, creating an opportunity to extract additional clinically meaningful information from existing data. There is a growing need for scalable, affordable tools that can improve risk stratification, support treatment planning, and enhance personalized cancer care using readily available clinical resources.

Technical Summary

Researchers at Emory University have developed an artificial intelligence-based approach that analyzes collagen architecture within the tumor microenvironment using routine digital pathology images. The technology extracts quantitative features describing collagen organization, including fragmentation, bundling, rigidity, alignment, and density, and uses these features to generate prognostic assessments for cancer patients. Unlike conventional approaches that rely heavily on molecular or genomic testing, this method derives clinically relevant information directly from pathology images that are already collected during standard care. The approach may provide a more accessible, interpretable, and cost-effective tool for cancer risk stratification and treatment decision support.

Development Stage

Proof-of-concept demonstrated using retrospective pathology image datasets from multiple clinical institutions.

Publication

• Nag R, Chen C, Mejbel H, Li H, Aqeel A, Fu P, Corredor G, Khalighi S, Pathak T, Mokhtari M, Reid MD, Krasinskas AM, Pandav K, Willis JE, Madabhushi A, ”Collagen Disorder Architecture Features are associated with Genetic, Clinical, Molecular, Demographic Factors and Survival Outcomes in Colon Cancer”, NPJ Precis Oncol. 2025 Aug 28;9(1):304. doi: 10.1038/s41698-025-01098-y. (PMID: 40877439).
• Nag, R., Li, H., Corredor, G., Khalighi, S., Pathak, T., Toro, P., Koyfman, S., Fu, P., Castro, P., Frederick, M., Sandulache, V., Lewis Jr., J., & Madabhushi, A. “AI-Derived Features of Collagen Architecture Identify a Subset of High-Risk Oral Cavity Squamous Cell Carcinoma Patients with Poor Prognosis.” United States and Canadian Academy of Pathology’s 113th Annual Meeting, March 23-28, 2024, Baltimore, MD.