Evaluation of Artificial Intelligence in Diagnostic Radiology: A Systematic Review

Background of the Study Diagnostic radiology is one of the most important medical specialties in modern healthcare. Radiological imaging techniques such as X-ray, Computed Tomography (CT), Magnetic Resonance Imaging (MRI), Ultrasound, and Nuclear Medicine imaging play a crucial role in the early detection, diagnosis, and monitoring of diseases. These imaging modalities allow clinicians to visualize internal anatomical structures non-invasively, thereby improving diagnostic accuracy and guiding treatment planning (1,2). Over the past two decades, there has been a rapid increase in the use of diagnostic imaging in clinical practice due to technological advancements and the growing demand for accurate disease diagnosis. Modern imaging equipment produces large volumes of high-resolution data, which require careful interpretation by radiologists (3). As imaging examinations increase, radiologists face a significant workload burden, which may contribute to diagnostic fatigue, reporting delays, and potential human error (4). Objective: To systematically evaluate the diagnostic performance, applications, and clinical utility of AI technologies in diagnostic radiology. Methods: A systematic review was conducted following standard guidelines (e.g., PRISMA). Relevant studies were identified through electronic databases such as PubMed, Scopus, and Google Scholar. Inclusion criteria comprised peerreviewed articles focusing on AI applications in diagnostic radiology, including modalities such as X-ray, CT, MRI, and ultrasound. Studies published in English over the last 10 years were included. Data extraction focused on study design, AI techniques used, diagnostic accuracy, sensitivity, specificity, and clinical outcomes. Results (Core Components): The review included multiple studies demonstrating that AI significantly enhances diagnostic performance in radiology. Key findings include: • AI models showed high sensitivity and specificity in detecting diseases such as lung nodules, fractures, tumors, and brain abnormalities. • Deep learning algorithms outperformed traditional image analysis methods in several studies. • AI-assisted diagnosis reduced interpretation time and improved workflow efficiency. • Integration of AI with radiologist interpretation resulted in higher diagnostic accuracy compared to either method alone. • Challenges identified included data variability, lack of standardization, and need for large validated datasets. Conclusion: Artificial Intelligence has shown significant potential in improving diagnostic accuracy, efficiency, and workflow in radiology. While AI cannot replace radiologists, it serves as a powerful supportive tool that enhances clinical decision-making. Further research, standardization, and regulatory frameworks are required for its widespread and safe implementation. Summary: This systematic review highlights the growing importance of AI in diagnostic radiology, emphasizing its benefits in disease detection, workflow optimization, and improved patient care. Despite promising results, challenges remain in validation, ethical considerations, and integration into routine clinical practice.