Maria Babak, Head of The Babak Lab and Assistant Professor at City University of Hong Kong, shared a post on LinkedIn by The Babak Lab, adding:
“A great example of AI being used to extract new biological information from data we already routinely collect.”
Quoting The Babak Lab’s post:
“Scientific Wednesdays.
Can a routine pathology slide reveal the spatial organization of a tumor?
The tumor microenvironment is not just a mixture of cancer, immune, and stromal cells. Their spatial organization can influence tumor progression and treatment response, but mapping it usually requires complex and expensive technologies.
Key Insights
- Using 41-plex spatial proteomics, researchers mapped more than 18 million cells from 457 patients with NSCLC.
- They identified 10 reproducible cellular neighborhoods formed by tumor, immune, and stromal cells.
- CANVAS, an AI platform, was trained to recognize these neighborhoods directly from routine H&E slides.
- The approach was evaluated in more than 5,000 patients across 9 cancer types.
- CANVAS-derived spatial patterns were associated with prognosis, tumor ecotypes, and immunotherapy outcomes.
- Neutrophil-rich neighborhoods were associated with poorer outcomes and resistance to immune checkpoint blockade.
Conclusion
CANVAS brings spatial tumor profiling to routine H&E slides, making it possible to study tumor ecosystems at a much larger scale. The immunotherapy findings still require clinical validation, but the approach could make spatial biomarkers easier to develop and test.”
Title: Cellular architecture and neighborhood-informed virtual spatial tumor profiling from histopathology
Authors: Yuchen Li, Zhe Li, Ryan Quinton, Yuanfeng Ji, Xiaoming Zhang, Jinxi Xiang, Xiyue Wang, Sen Yang, Feyisope Eweje, Yijiang Chen, Xiangde Luo, Yuanyuan Li, Jonathan Mulholland, Siwei Chen, Colin Bergstrom, Ted Kim, Francesca Maria Olguin, Sierra Willens, Steven H. Lin, Jeffrey J. Nirschl, Robert West, Joel Neal, Maximilian Diehn, Ruijiang Li
Read the full article here.
Other articles featuring Maria Babak on OncoDaily.