ARTIFICIAL INTELLIGENCE IN IDENTIFYING ANATOMICAL STRUCTURES AND SUPPORTING DIAGNOSIS IN ADULT IMAGING EXAMS

Abstract

Introduction: Artificial intelligence (AI) applied to imaging interpretation aims to improve the visualization of anatomical structures, enhance diagnostic accuracy, and reduce human errors. Despite these benefits, challenges remain regarding data privacy, algorithmic bias, and the lack of consolidated guidelines. Objective: To evaluate the accuracy of AI in the visualization of anatomical structures and in providing diagnoses through imaging examinations. Method: An integrative review was conducted in PubMed, ScienceDirect, and LILACS, including articles published in the last five years, in English or Portuguese, with full text available, focusing on the application of AI in imaging exams in adults. The search terms used were: Adult, Adulto, Image Interpretation (Computer-Assisted), Generative Artificial Intelligence, Intelligent Systems, Artificial Intelligence, Diagnostic Imaging, and Diagnóstico por Imagem. After screening 1,299 publications, 11 studies met the inclusion criteria. Results: AI showed potential in several clinical contexts, including image reconstruction with reduced radiation and contrast, advanced lesion detection, segmentation for guided therapies, integration of clinical and anatomical data into predictive models, therapeutic monitoring in chronic diseases, automated exam planning, portable screening for ocular pathologies, cardiac volumetric modeling, prediction of lymph node metastases, segmentation of brain metastases, and applications in telepathology. These approaches improved precision, efficiency, and safety, particularly in settings with limited resources and workforce shortages. Conclusion: AI proved to be a promising tool to enhance diagnostic accuracy, optimize workflows, and expand access to imaging. However, its effective and safe implementation requires adequate infrastructure, professional training, standardized protocols, and ethical regulation.