Automated evaluation of ASPECTS from brain computerized tomography of patients with acute ischemic stroke

Abstract

Background and Purpose: Precise evaluation of brain computerized tomography (CT) is a crucial step in acute ischemic stroke evaluation. Electronic Alberta Stroke Program Early CT Score (E-ASPECTS) helps in the selection of patients who may be eligible for thrombolysis. This paper seeks to assess the performance of emergency physicians (EPs) in the evaluation of ASPECTS scores with and without the use of E-ASPECTS and to compare their results with neuroradiologists.Methods: A total of 116 patients were selected. Initially, two EPs and two neuroradiologists evaluated the admission nonenhanced CT without E-ASPECTS. Then, after 30 days, they re-evaluated the images using E-ASPECTS. Sensitivity, specificity, Matthew's correlation coefficients (MCC), and receiver operating characteristic curves were generated for analysis before and after the software use.Results: Eps' performances improved when they used E-ASPECTS, with their results closer to those obtained by neuroradiologists. In the initial evaluation, MCC values for the two EPs were -0.01 and 0.04, respectively. After the software assistance, they obtained 0.38 and 0.43, respectively, which was closer to the scores obtained by the neuroradiologists (0.53 and 0.39, respectively).Discussion: This is the first study that has specifically compared neuroradiologists' and EPs' performances before and after using E-ASPECTS. E-ASPECTS assisted and improved the evaluation of the images of patients with acute ischemic stroke. Conclusion:Artificial intelligence in the emergency room may increase the number of patients treated with tissue-type plasminogen activators.

The Utility of Automated ASPECTS in Acute Ischemic Stroke for Intravenous Recombinant Tissue Plasminogen Activator (IV-rtPA) Therapy

“…The automated APECTS scoring for acute ischemic stroke has been reported using several other AI softwares, such as Syngo.via Frontier ASPECT Score Prototype V2 (Siemens Healthineers, Erlangen, Germany), Brainomix e-ASPECTS ® (Brainomix Ltd., Oxford, UK) and RAPID ASPECTS (iSchemaView, Inc., Menlo Park, CA, USA). Although the reading accuracy of these AI systems was shown to be comparable to that of physicians [ 10 , 11 , 12 , 13 , 14 , 15 , 16 , 19 ], these reports were limited to a population of relatively severe and extensive infarction, with the majority of cases including clot retrieval therapy and LVO [ 14 , 15 , 16 ]. In addition, neither physicians nor the AI software had a high reading accuracy for infarcts in the perforating branch region, where relatively mild cases are common [ 19 ].…”

“…As a means of addressing such issues, a head CT image reading software utilizing artificial intelligence (AI) was recently developed, and its reading accuracy of ASPECTS was found to be equivalent to that of stroke specialists in cerebral infarction [ 10 , 11 , 12 , 13 , 14 ]. However, studies have examined the usefulness of the software in determining the indication for endovascular therapy, mainly in cases where the target patient underwent endovascular therapy or in cases of large vessel occlusion (LVO) [ 14 , 15 , 16 ].…”