Related Experiment Video
Updated: Sep 14, 2025
05:33
Introduction of an Integrated Pathology Image Management, Artificial Intelligence, and Reporting System
Published on: July 11, 2025
255
Interpretable AI Framework for Secure and Reliable Medical Image Analysis in IoMT Systems.
IEEE Journal of Biomedical and Health Informatics
|July 23, 2025
Summary
This study introduces an Explainable AI (XAI) framework for medical cyber-physical systems (MCPS) to enhance AI in the Internet of Medical Things (IoMT). The novel approach improves diagnostic accuracy and security against cyber threats.
Area of Science:
- Medical Image Analysis
- Artificial Intelligence in Healthcare
- Cybersecurity in IoMT
Background:
- Artificial intelligence (AI) integration in medical imaging offers enhanced diagnostic precision but faces challenges in transparency, trustworthiness, and security within the Internet of Medical Things (IoMT).
- Existing AI models often lack clinical interpretability, hindering trust and adoption in critical healthcare applications.
- The dynamic and interconnected nature of Medical Cyber-Physical Systems (MCPS) necessitates robust security measures against evolving threats.
Purpose of the Study:
- To introduce a novel Explainable AI (XAI) framework designed for Medical Cyber-Physical Systems (MCPS).
- To address critical challenges of transparency, trustworthiness, and security in AI-driven medical image analysis within IoMT.
- To enhance clinical interpretability and robustness of AI models in healthcare settings.
Main Methods:
- The framework combines deep neural networks with symbolic knowledge reasoning for clinically interpretable insights.
- An Enhanced Dynamic Confidence-Weighted Attention (Enhanced DCWA) mechanism refines attention maps using adaptive normalization and multi-level confidence weighting.
- A Resilient Observability and Detection Engine (RODE) employs sparse observability principles to detect and mitigate adversarial threats.
Main Results:
- The framework demonstrated a 15% increase in lesion classification accuracy on benchmark datasets.
- A 30% reduction in robustness loss was achieved, indicating improved resilience against adversarial attacks.
- A 20% improvement in the Explainability Index was observed compared to existing state-of-the-art methods.
Conclusions:
- The proposed XAI framework significantly enhances AI performance and trustworthiness in medical image analysis within IoMT environments.
- The integration of Enhanced DCWA and RODE provides robust and interpretable AI solutions for MCPS.
- This work paves the way for more secure, transparent, and reliable AI applications in digital healthcare.
