← ResourcesPublications
Training and evaluation of an AI-assisted HER2 scoring in breast carcinoma-stained whole-slide images
01 SEPT 2024
Abstract
Background & objectives
To improve the accuracy of HER2 scoring, we have developed an artificial intelligence (AI) algorithm designed for precise and consistent quantification of HER2 protein overexpression, supporting pathologists in the assessment of HER2 immunohistochemistry scores.
Methods
The algorithm detects invasive tumor regions and quantifies and characterizes tumor cells, trained on over 5000 labeled images. It segments invasive areas and cells, analyzing cell membranes. Its performance was validated against ground-truth scores established by three senior pathologists through majority voting. Post a two-month washout, slides were rescored with AI assistance to compare inter-observer agreement.
Results
The test dataset was built using 68 routine cases of HER2 immunostained slides of Breast Cancer, without any further selection criteria, in order to reflect the routine practice. The model exhibited an overall balanced accuracy of 90.2% on this dataset, thus demonstrating its capability to assist pathologists in their routine practice while providing an interpretable HER2 Score. Notably, 100% of the HER2 3+ scores and 93.3% of the HER2-Low cases were identified by the model, yielding promising results to identify patients eligible to targeted therapy. Moreover, the inter-observer agreement rose from 57% to 75% with the assistance of AI.
Conclusion
This study demonstrates the efficacy of the proposed fully-automated AI system in accurately assessing HER2-immunostained WSI, in accordance with the 2018 ASCO/CAP Guidelines. By providing pathologists with easily interpretable scores, such an AI tool could be used as a regular aid in clinical decision-making, improving reproducibility and aiding in the assessment of challenging cases due to the detailed information it provides for each cases.