Related Experiment Video
Updated: Jul 2, 2025
Quantitation of Protein Expression and Co-localization Using Multiplexed Immuno-histochemical Staining and Multispectral Imaging
Published on: April 8, 2016
A deep learning based holistic diagnosis system for immunohistochemistry interpretation and molecular subtyping
Lin Fan1, Jiahe Liu2, Baoyang Ju2
1School of Integrated Circuit Science and Engineering (Industry-Education Integration School), Nanjing University of Posts and Telecommunications, Nanjing 210023, PR China; State Key Laboratory of Bioelectronics, Jiangsu Key Laboratory for Biomaterials and Devices, School of Biological Science and Medical Engineering & Collaborative Innovation Center of Suzhou Nano Science and Technology, Southeast University, Nanjing 210096, PR China; Medical School of Nanjing University, Nanjing 210093, PR China.
A new intelligent breast cancer diagnosis system accurately identifies molecular subtypes using artificial intelligence. This AI system improves upon manual interpretation for HER2, ER, PR, and Ki67 markers, enhancing efficiency and precision in breast tumor analysis.
Area of Science:
- Oncology
- Biomedical Engineering
- Computational Pathology
Background:
- Breast cancer molecular subtypes (HER2, ER, PR, Ki67) influence treatment and prognosis.
- Immunohistochemistry is crucial for subtype determination but limited by manual interpretation subjectivity and efficiency.
- Existing methods struggle with stability and operational efficiency in tumor marker analysis.
Purpose of the Study:
- To develop a holistic intelligent breast tumor diagnosis system for automated tumor-markeromic analysis.
- To combine automatic interpretation with clinical suggestions for improved breast cancer subtyping.
- To overcome the limitations of conventional manual interpretation in breast cancer diagnosis.
Main Methods:
- Developed a system with two modules: interpretation and subtyping.
- Utilized convolutional neural networks (CNNs) for multi-feature extraction from immunostaining images.
- Encoded interpreting results into low-dimensional representations for efficient molecular subtype output.
Main Results:
- The system accurately determined overexpression rates of HER2, ER, PR, and Ki67.
- Achieved high average sensitivity (97.6%) and specificity (96.1%) for molecular subtype determination.
- Demonstrated exceptional performance for HER2 interpretation with 99.8% sensitivity and 96.9% specificity.
Conclusions:
- The intelligent diagnosis system surpasses pathologist-level performance in interpreting immunohistochemical images.
- The system offers improved efficiency, precision, and repeatability compared to manual methods.
- This AI-driven approach holds potential to revolutionize breast cancer diagnosis and management.
