Keywords:
Artificial Intelligence, Neuroradiology brain, CT, CAD, Computer Applications-Detection, diagnosis, Ischaemia / Infarction
Authors:
A. Ayobi, P. Chang, D. Chow, C. Filippi, S. Quenet, M. Tassy, Y. Chaibi
DOI:
10.26044/ecr2023/C-19206
Conclusion
A deep learning AI-based application, CINA-ASPECTS, can identify areas of acute ischemia with high concordance to expert neuroradiologists. The device shows promising results in the automatic detection of regions with EIC and in the computation of the ASPECT score. The results demonstrated that it performed properly and matched with the expert visual assessments, considered as the GT. Such AI-based techniques can be used to accurately determine patients deemed eligible for mechanical thrombectomy or any other treatment, thus, improving clinical workflow and patient outcomes [9-10].
In addition, the conjunctive use of the software as a diagnostic aid significantly improved the accuracy of physicians’ interpretations in a realistic clinical workflow. Similarly, the readers’ standard error decreased when assisted by the software, suggesting that less discrepant and more standardized readings may be expected when using the device. Indeed, the potential clinical benefit of this type of algorithm is directly related to the improvement in radiologists’ accuracy and variability, compared to the conventional use of NCCT images alone. In fact, the characterization of brain tissue abnormalities by means of the ASPECT score is more precise when using CINA-ASPECTS’ outputs, leading to a better patient selection for adequate treatment.
Future studies in larger prospective cohorts are needed to confirm the standardized use of the software. Moreover, an analysis of the impact of the device on patient outcomes may provide more insight into the advantages of this software.