EARLY DIAGNOSIS OF ISCHEMIC STROKE: THE CONTRIBUTION OF ARTIFICIAL INTELLIGENCE IN TOMOGRAPHY

Abstract

Introduction: The increasing use of imaging tests, due to greater availability, has significantly increased the demand for radiologists, increasing the response time in the analysis of exams and reducing the accuracy of diagnoses. AI optimizes this flow, speeding up image analysis and identifying altered exams for priority evaluation by the radiologist, generating automated reports, which facilitates decision-making. In neurology, computed tomography is essential for the diagnosis of stroke due to its accessibility and sensitivity. Ischemic stroke is prevalent and impactful in Brazil, with a high mortality and disability rate. Treatments such as thrombolysis and mechanical thrombectomy, applied within a therapeutic window, have shown greater efficacy, highlighting the importance of rapid and accurate diagnoses. Objective: To investigate and analyze the benefits of AI in the early detection of ischemic stroke and the benefits of immediate treatment with thrombolysis and thrombectomy. Materials and Methods: A search was carried out in the PubMed and VHL Regional Portal databases from 2019 to 2024, in English. The selection was conducted using the descriptors “Artificial Intelligence”; “stroke; “tomography” and “detection”. Of the 111 articles surveyed, when applying Qualis CAPES (A1, A2, A3, A4, B1 and B2) and excluding duplicates, 49 were selected and 28 were consistent with the inclusion criteria. Results: Most studies highlight AI as an effective tool in the automated detection of ischemic cerebral infarctions, with the potential to accelerate the early diagnosis of stroke and improve clinical management. AI, especially with deep learning, reduces interpersonal variability and interpretation errors, in addition to offering advantages in diagnostic accuracy, including non-contrast tomography. However, widespread clinical adoption faces challenges, such as the need for standardized protocols and the limitation of some methods in identifying occlusions in smaller vessels. Conclusion: Given the great potential for early diagnosis of ischemic stroke with the use of artificial intelligence, there was a clear gain in the optimization of clinical management. However, the challenges raise the need for the implementation of clearer guidelines, so that the demand for analysis of existing algorithms occur with the evolution of AI, along with the standardization of the method to be used before its effective clinical use. Despite this, with the detailed advancement of technology, AI has the potential to positively transform treatment. In this way, this tool changes the social scenario in a broadly positive way, which reduces the disability of patients affected by stroke and improves the morbidity and mortality rates of the disease.