Related Experiment Video
Updated: Jun 28, 2026
05:33
Introduction of an Integrated Pathology Image Management, Artificial Intelligence, and Reporting System
Published on: July 11, 2025
Interpretable CRAM‑Enhanced Lightweight Dual‑Branch CNN for Real‑Time Breast Cancer Histopathology in
Roseline Oluwaseun Ogundokun1,2,3, Rotimi-Williams Bello2, Pius Adewale Owolawi2
1Department of Multimedia Engineering, Kaunas University of Technology, Kaunas, Lithuania.
Small (Weinheim an Der Bergstrasse, Germany)
|March 19, 2026
Summary
A new hybrid deep learning model offers accurate and efficient breast cancer diagnosis using histopathology images. This interpretable AI is suitable for real-time medical applications, enhancing diagnostics in resource-limited settings.
Area of Science:
- Digital pathology
- Artificial intelligence in healthcare
- Medical imaging analysis
Background:
- Breast cancer diagnosis relies on manual histopathology, which is time-consuming and subjective.
- Current deep learning models for medical imaging are often too complex for real-time Internet of Medical Things (IoMT) applications.
Purpose of the Study:
- To develop an interpretable and lightweight hybrid deep learning model for efficient breast cancer histopathology image analysis.
- To enable real-time, trustworthy AI diagnostics in resource-limited and point-of-care settings.
Main Methods:
- A hybrid deep learning model combining MobileNetV2 and EfficientNet-B0 was developed.
- A novel contextual recurrent attention module (CRAM) was integrated to refine features.
- Interpretability was achieved using Grad-CAM and SHAP analyses.
Main Results:
- The model achieved 99.9% classification accuracy and an Area Under the Curve (AUC) of 1.00.
- The model is efficient with approximately 12 million parameters, suitable for IoMT deployment.
- Interpretability methods highlighted diagnostically relevant malignant tissue features, aligning with pathologist judgment.
Conclusions:
- The proposed model offers a balance of high accuracy, efficiency, and transparency for breast cancer diagnosis.
- This interpretable AI system is well-suited for real-time applications in digital pathology, especially in resource-limited environments.
- The system represents a significant advancement in making AI in digital pathology more accessible and explainable.