Demystifying Artificial Intelligence: A Systematic Review of Explainable Artificial Intelligence in Medical Imaging
Sensors, 2026
This systematic review examines how explainable artificial intelligence (XAI) could make AI in medical imaging more understandable and more useful in clinical practice. The authors aim to map the main explainability methods used in the field, examine how they are applied across tasks such as classification, segmentation, and diagnosis, and highlight the gaps that still limit their use in real healthcare settings. The paper is built around a key issue in medical AI: even when models perform well, it is often difficult to understand how they reach their conclusions, which can make trust and adoption harder.
To explore this, the authors reviewed the recent literature on XAI in medical imaging and compared a wide range of methods, including both built-in and post-hoc approaches. They also looked at how these methods are used in different imaging applications and considered how explanation quality is assessed. One of the paper’s main contributions is its effort to bring more structure to this area by proposing a clearer way to evaluate explainability methods.
Overall, the review shows that XAI can play an important role in making medical imaging AI more transparent, helping clinicians better interpret model outputs and supporting greater accountability. At the same time, the authors show that the field is still developing. Many approaches are difficult to compare, their clinical value is not always clear, and evaluation remains inconsistent. The paper concludes that future progress will depend not only on creating more explainability tools, but on making them more reliable, meaningful, and relevant to clinical practice.