Related Experiment Video
Updated: May 24, 2025
07:15
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
6.6K
Artificial Intelligence-Assisted Detection of Breast Cancer Lymph Node Metastases in the Post-Neoadjuvant Treatment
Tony Xu1, Dina Bassiouny2, Chetan Srinidhi3
1Department of Medical Biophysics, University of Toronto, Toronto, Ontario.
Laboratory Investigation; a Journal of Technical Methods and Pathology
|February 28, 2025
Summary
This study developed a deep learning tool to accurately detect breast cancer metastasis in lymph nodes after neoadjuvant systemic therapy (NAT). The AI pipeline improves pathologist accuracy and efficiency in identifying metastatic lymph nodes.
Area of Science:
- Oncology
- Pathology
- Artificial Intelligence
Background:
- Lymph node metastasis assessment is crucial but challenging for pathologists.
- Existing deep learning models lack evaluation for patients receiving neoadjuvant systemic therapy (NAT).
Purpose of the Study:
- To develop and evaluate a generalizable deep learning pipeline for detecting breast cancer metastasis in lymph nodes post-NAT.
- To assess the impact of including post-NAT treatment effects in training data.
Main Methods:
- Created a 1027-slide dataset of breast cancer patients treated with NAT.
- Developed an interpretable deep learning pipeline for slide classification and metastasis probability heatmap generation.
- Evaluated pipeline performance with and without post-NAT treatment effect data.
Main Results:
- Training with post-NAT treatment effects improved pipeline specificity.
- The final pipeline achieved an AUC of 0.986 for slide-level classification.
- The patch-level heatmap aided pathologists in correcting erroneous predictions.
Conclusions:
- Deep learning models incorporating post-NAT treatment effects enhance lymph node metastasis detection accuracy.
- Interpretable AI tools can assist pathologists, improving diagnostic efficiency and reliability.
