Google and the Department of Defense are building an AI-powered microscope to help doctors spot cancer
ai-healthcarepathologycomputer-visioncancer-detection
Abstraction: AI Augmented Reality Microscope overlays cancer detection onto traditional pathology workflow
Key points:
- The Augmented Reality Microscope (ARM) overlays AI-generated green outlines and heat maps directly into the pathologist's eyepiece in real time without disrupting workflow
- Only 13 ARMs exist as of Sep 2023; costs health systems $90,000–$100,000 each; hardware by Jenoptik, AI software by Google
- Google has deployed four cancer-detection algorithms: breast, cervical, prostate cancer and mitosis detection; models trained on DoD data, encrypted and inaccessible to Google
- Breast cancer metastasis detection in lymph nodes showed reasonable performance in a 2022 peer-reviewed study, but accuracy varied by cell type
- Designed for resource-limited settings where pathologists lack easy access to a second opinion; obvious use case is remote/small labs and training residents
- Initiative originated from a 2016 email from Navy pathologist Niels Olson to a Google contact; DoD Defense Innovation Unit selected Google after evaluating 39 companies
Connections: Google · Department Of Defense · AI In Healthcare · Computer Vision