in

#Mapping histomorphological cancer phenotypes with self-supervised learning on pathology slides

Mapping the landscape of histomorphological cancer phenotypes using self-supervised learning on unannotated pathology slides

The research conducted complies with ethical regulations, and specimens from NYU were collected under specific protocols approved by the Investigational Review Board. The study used datasets from The Cancer Genome Atlas (TCGA) and NYU for lung tumor type prediction and overall survival analysis. The methodology involved deep learning models and self-supervised learning to analyze whole slide images and patient data. The clustering of representations was done using Leiden community detection, and the evaluation of the models was performed through logistic regression and Cox proportional hazards regression. The study also examined the correlation between Histomorphological Phenotype Clusters (HPCs) and immune landscape features, as well as the enrichment of cell types and lung adenocarcinoma subtypes in different HPCs. The histological assessment of the HPCs was conducted by expert pathologists, and the results were summarized based on their observations. The study provides detailed insights into the methodology, analysis, and results obtained from the research. Further information on the research design is available in the Nature Portfolio Reporting Summary linked to the article.

Source link

Source link: https://www.nature.com/articles/s41467-024-48666-7

What do you think?

Leave a Reply

GIPHY App Key not set. Please check settings

Top 10 AI-Powered Robotic Cleaners for Every Budget in 2024 | by AI2U | Jun, 2024

Top 10 AI-Powered Robotic Cleaners for Every Budget in 2024 #AIcleaning2024

Apple's AI overhaul: How the tech giant's new products will ALL have creepy inbuilt AI that can remove people from photos, track your family on flights, and reply to emails for you

Apple’s new products will have creepy inbuilt AI features. #AIoverhaul