This study explores using AI-based image search to improve diagnostic accuracy in histopathology. By searching the TCGA database (nearly 30,000 whole-slide images), the researchers successfully indexed and searched 20 million image patches. Employing a conservative "majority voting" approach, they achieved high accuracy in identifying cancer subtypes from both frozen section and permanent slides. The findings suggest that computational consensus is achievable for diagnosis given sufficient searchable cases per cancer subtype.
Publisher
npj Digital Medicine
Published On
Mar 10, 2020
Authors
Shivam Kalra, H. R. Tizhoosh, Sultaan Shah, Charles Choi, Savvas Damaskinos, Amir Safarpoor, Sobhan Shafiei, Morteza Babaie, Phedias Diamandis, Clinton J. V. Campbell, Liron Pantanowitz
Tags
AI-based image search
diagnostic accuracy
histopathology
cancer subtypes
TCGA database
image indexing
majority voting
Related Publications
Explore these studies to deepen your understanding of the subject.
