Related Experiment Video
Updated: Sep 28, 2025
10:16
Employing Digital Droplet PCR to Detect BRAF V600E Mutations in Formalin-fixed Paraffin-embedded Reference Standard Cell Lines
Published on: October 8, 2015
13.3K
Rapid Screening Using Pathomorphologic Interpretation to Detect BRAFV600E Mutation and Microsatellite Instability in
Satoshi Fujii1,2, Daisuke Kotani3, Masahiro Hattori4
1Division of Pathology, Exploratory Oncology Research & Clinical Trial Center, National Cancer Center, Kashiwa, Japan.
Summary
Deep learning models can predict genetic abnormalities in colorectal cancer from standard pathology images, aiding precision medicine. This approach bypasses traditional gene testing for faster treatment planning.
Area of Science:
- Oncology
- Computational Pathology
- Genomics
Background:
- Precision medicine requires rapid identification of genetic abnormalities for targeted cancer therapy.
- Hematoxylin and eosin (H&E) stained images offer a potential source for predicting these abnormalities.
Purpose of the Study:
- To develop deep learning (DL) models capable of predicting genetic abnormalities in colorectal cancer from H&E images.
- To extract pathomorphologic features predictive of specific genetic mutations and microsatellite instability.
Main Methods:
- Utilized 1,657 H&E images from the SCRUM-Japan GI-SCREEN project with confirmed genetic abnormalities (BRAFV600E, KRAS, MSI-H).
- Developed DL models to predict pathomorphologic features and subsequently gene abnormalities.
- Employed training and two validation cohorts for model development and assessment.
Main Results:
- Achieved high prediction accuracies with Area Under the Curve (AUC) values exceeding 0.90 for 12 features and 0.80 for 27 features.
- Demonstrated high AUCs for predicting BRAFV600E mutations (0.851, 0.859) and MSI-High (0.923, 0.862).
Conclusions:
- Next-generation pathology methods using DL can predict genetic abnormalities from H&E images without standard gene tests.
- This approach holds promise for streamlining colorectal cancer treatment planning.
