Related Experiment Videos
Explainable Artificial Intelligence (AI) for Medical Imaging: A Framework for Bridging the AI Trust Gap
Cody H Savage1, Jeremias Sulam2,3, Chien-Ming Huang2,4
1Department of Diagnostic Radiology & Nuclear Medicine, University of Maryland School of Medicine, Baltimore, MD.
AJR. American Journal of Roentgenology
|May 13, 2026
Summary
Explainable AI (XAI) in healthcare needs to be robust, user-adapted, and clinically aligned to build trust. Current methods are insufficient, requiring a new framework for trustworthy AI adoption in medical imaging.
Area of Science:
- Medical Imaging
- Artificial Intelligence in Healthcare
- Explainable AI (XAI)
Background:
- Artificial intelligence (AI) adoption in healthcare is rising, yet faces significant trust barriers among clinicians and patients.
- Existing Explainable AI (XAI) methods often fail to provide transparent, interpretable, and clinically actionable insights, hindering effective integration.
- Trust is a critical factor for the successful implementation of AI tools in clinical settings, particularly in medical imaging.
Purpose of the Study:
- To propose guiding principles for developing clinically useful and trustworthy Explainable AI (XAI) in medical imaging.
- To introduce a conceptual framework for XAI design and deployment that addresses current limitations.
- To foster shared responsibilities among AI developers, vendors, and healthcare institutions for successful AI adoption.
Main Methods:
- The study presents a perspective on current XAI limitations and proposes a conceptual framework.
- Key principles for effective XAI include technical robustness, end-user adaptation, and clinical task alignment.
- The framework emphasizes a shift from one-size-fits-all solutions to task- and user-centered design.
Main Results:
- Current XAI methods are inadequate for clinical utility and trust-building in healthcare.
- Three core principles—robustness, user adaptation, and clinical alignment—are essential for effective XAI.
- A conceptual framework is introduced to guide the development and deployment of trustworthy XAI.
Conclusions:
- Explainable AI (XAI) must be technically robust, adapted to end-users, and aligned with clinical tasks to be effective in medical imaging.
- A new conceptual framework is proposed to guide the design and deployment of XAI, emphasizing user-centered approaches.
- Implementing these principles and framework can enhance trust and facilitate the adoption of AI in healthcare.
