Abstract

Artificial intelligence (AI) has significantly transformed radiology by enabling automated medical image classification, particularly for detecting abnormalities in chest X-rays and other imaging modalities. While deep learning models achieve remarkable accuracy, their black-box nature limits interpretability, raising concerns among clinicians and regulatory bodies [1]. Explainable AI (XAI) techniques aim to bridge this gap by providing insights into the decision-making processes of these models [2]. This paper comprehensively examines XAI methods applied to radiological image classification, focusing on chest X-ray datasets and pneumonia detection models [3]. A detailed exploration of model architectures, feature attribution techniques, and evaluation metrics is conducted to understand the role of explainability in medical AI [4]. Furthermore, key challenges in implementing explainability frameworks and future directions for research and clinical adoption are discussed [5]. This study emphasizes the need for integrating XAI into radiology to ensure AI-driven systems are not only accurate but also transparent and trustworthy.