Blog

Anant Madabhushi: New Publication on AI-Driven Virtual Staining for Tumor-Associated Macrophages

Anant Madabhushi/X

Mar 30, 2025, 16:06

Anant Madabhushi: New Publication on AI-Driven Virtual Staining for Tumor-Associated Macrophages

Anant Madabhushi, Founding director of the Emory Empathetic AI for Health Institute, posted on LinkedIn about recent paper he and his colleagues co-authored:

“New Publication on AI-Driven Virtual Staining for Tumor-Associated Macrophages!

Excited to share our latest study, published in the European Journal of Cancer, where we introduce VISTA (Virtual Staining AI-based Translation Algorithm)—an artificial intelligence (AI)-powered virtual staining platform designed to transform Hematoxylin & Eosin (H&E) images into virtual immunohistochemistry (IHC) for identifying M2-subtype tumor-associated macrophages (M2-TAMs) in HPV+ oropharyngeal squamous cell carcinoma.

Key Highlights:

Objective: Develop and validate VISTA to virtually stain H&E images, allowing for precise detection of CD163+ M2-TAMs, which are linked to poor survival outcomes in cancer patients.
Methodology: Trained and tested VISTA on 968 patient samples across six institutional cohorts.
Compared VISTA-generated virtual CD163+ IHC images to traditional IHC staining, evaluating image quality and segmentation performance.
Findings: High M2-TAM density correlated with worse overall survival (p=0.0152, HR=1.63).
VISTA outperformed existing virtual staining methods, achieving a Structural Similarity Index of 0.72, a Peak Signal-to-Noise Ratio of 21.5, and a Fréchet Inception Distance of 41.4.
Improved segmentation of M2-TAMs (Dice score = 0.74).
Clinical Significance: Enables non-destructive, cost-effective, and scalable biomarker discovery.
Provides an alternative to traditional IHC staining, preserving tissue samples while improving diagnostic precision.

A huge congratulations to rockstar The Wallace H. Coulter Department of Biomedical Engineering at Georgia Tech and Emory University PhD student Arpit Aggarwal for leading this work as the first author! Grateful for the other co-authors include Mayukhmala J., Amritpal Singh, Tanmoy Dam, Himanshu M., Tilak Pathak, Sandra Orsulic, Kailin Yang, Deborah Chute, Justin A. Bishop, Farhoud Faraji, Wade M. Thorstad, SHLOMO A KOYFMAN, Scott Steward, Qiuying (Judy) Shi, Vlad Sandulache, Nabil Saba, James S. Lewis Jr., Germán Corredor, and many others who made this study possible.

This work demonstrates how AI-powered virtual staining can revolutionize biomarker discovery and cancer diagnostics, paving the way for more efficient and scalable pathology solutions.

Read the full paper here:

Authors: Arpit Aggarwal, Anant Madabhushi, et al.

Anant Madabhushi: New Publication on AI-Driven Virtual Staining for Tumor-Associated Macrophages