| Abstract [eng] |
Rationale of the study. Artificial intelligence is currently applied across many fields of medicine, one of which is radiology. In radiology, artificial intelligence can be used for the analysis of radiological images and the identification of pathological changes. For this reason, artificial intelligence is becoming progressively more important in the field of radiology. However, many specialists still question the reliability of artificial intelligence; therefore, it is essential to clearly understand its underlying principles, advantages, disadvantages, and future prospects. This allows for a better assessment of the potential of artificial intelligence and a more critical evaluation of its results. Aim of the study. To review the latest literature on artificial intelligence, its application in radiology, as well as its associated advantages, disadvantages, and future prospects. Objectives: 1. To review the structural components and operating principles of artificial intelligence and evaluate its application in radiology. 2. To evaluate the purposes for which artificial intelligence can be used in radiology and the advantages it provides. 3. To evaluate the disadvantages, challenges, and limiting factors encountered in the application of artificial intelligence in radiology. 4. To review the potential applications and future prospects of artificial intelligence in radiology. Methods and materials. Analysis of scientific literature in the „PubMed“ and „ScienceDirect“ databases. The review included 64 articles published in English between 2016 and 2026. Results. Artificial intelligence is a specialized algorithm developed using computer systems. Artificial intelligence in radiology is widely applied for the analysis of radiological images. This technology can enhance the efficiency of radiologists’ work, enable automated triaging of radiological examinations, improve diagnostic accuracy, and support the prediction of disease progression and treatment outcomes. Furthermore, it has been demonstrated that the implementation of artificial intelligence may reduce patients’ exposure to ionizing radiation while simultaneously improving the quality of radiological images. In addition, artificial intelligence can be utilized for the synthesis of radiological images and in the education and training of radiology residents. Despite these advantages, artificial intelligence also has several limitations. Evidence suggests that artificial intelligence does not always correctly perform the segmentation of anatomical and pathological structures. In addition, the development and application of artificial intelligence systems are often hindered by a lack of high-quality data, due to limited availability, data protection concerns, and potential data bias. Moreover, the operating principles of artificial intelligence are not yet fully understood; therefore, some specialists are skeptical about its integration into clinical practice. Nevertheless, artificial intelligence demonstrates significant potential in radiology and is expected to become an important assistant to radiologists in the future. Conclusions. Artificial intelligence technologies, including machine learning, deep learning, and large language models, are becoming an important part of radiology. Artificial intelligence helps to improve the evaluation of radiological images and support diagnostic decision-making; it has significant potential to automate complex tasks and reduce the workload of radiologists. Despite its wide range of applications, its effective use is limited by technical, ethical, and legal challenges. In the future, broader use of multimodal systems is expected; however, at present, artificial intelligence is considered a supportive tool rather than a replacement for radiologists. Keywords. Artificial intelligence; radiology; radiological images; machine learning; deep learning; large language models. |
