Artificial intelligence in diagnostic radiology: current status, evidence base, and implementation prospects

Abstract

Objective. To systematize current trends in the application of AI in diagnostic radiology, to analyze the evidence base, clinical efficacy, and limitations of real implementation, as well as to assess regulatory, ethical, and organizational and economic factors. Materials and methods. A descriptive analytical review of publications indexed in the PubMed/MEDLINE and RSCI databases for the period of 2020–2026 was conducted, analyzing the primary directions of AI application in radiology. Results. This review presents an analysis of the current state of AI technology application in diagnostic radiology. The main directions of the clinical use of AI systems are examined: detection and classification of paroplasms, quantitative assessment of focal lesions, medical image reconstruction, and generation of structured reports. The evidence base of AI solutions efficacy was analyzed taking into account external validation results. Regulatory requirements, ethical issues, and organizational aspects of integrating AI into radiological practice were discussed. Special attention was paid to the limitations of existing studies and the future prospects for the development of this field. Conclusion. AI systems demonstrate significant potential for enhancing the quality and efficiency of diagnostic radiology; however, their clinical value is determined not only by accuracy metrics, but also by their ability to perform reliably in real-world clinical settings, and their compliance with regulatory requirements.