Experience of applying artificial intelligence technologies to ischemic stroke diagnosis based on CT images

Ischemic stroke remains one of the leading causes of mortality and disability worldwide, necessitating improvements in early diagnostic methods. Despite the widespread use of computed tomography (CT), detecting early signs of ischemia remains challenging due to the modality’s limited sensitivity during the initial hours. Artificial intelligence (AI) technologies show potential for enhancing diagnostic accuracy, including the identification of cerebral ischemia signs, though their clinical applicability requires rigorous evaluation.

Aim of the study was to evaluate the diagnostic performance of AI-based services for automated analysis of brain CT scans in detecting acute ischemic lesions within the Moscow Experiment framework.

Material and methods. We conducted a retrospective study of 100 non-contrast brain CT scans (50 with ischemic stroke, 50 normal controls) selected from the Unified Radiological Information Service of Unified Medical Information and Analytical System of the City of Moscow. Diagnosis verification was performed by two independent radiologists, with expert consultation in disputed cases. We assessed the following metrics: sensitivity, specificity, accuracy, and area under the ROC curve (AUC) with 95 % confidence intervals. Scans were processed by three comprehensive AI services specifically designed to detect pathological changes on non-contrast brain CT scans in patients with suspected acute ischemic stroke.

Results. Two of the three AI services demonstrated high diagnostic accuracy: AUC > 87 %, sensitivity ≥ 83 %, specificity ≥ 83 %. The third service showed reduced sensitivity (68%), indicating a risk of missed diagnoses.

Conclusions. AI services can serve as effective decision-support tools in ischemic stroke diagnosis, particularly in time-constrained scenarios or when expert resources are limited. However, the observed variability in performance underscores the necessity for strict validation before clinical implementation.

1. Nikitin A.S., Asratyan S.A., Smirnov D.S., Kemezh Yu.V., Urvantseva O.M. Prevalence of massive ischemic stroke in a regional vascular center. Nevrologicheskiy zhurnal = Neurological Journal. 2017;(1):12–18. [In Russian].

2. Krylov V.V., Nikitin A.S., Dashyan V.G., Arutyunov N.M., Lagoda O.V., Belkin A.A., Vilenskii B.S., Gushcha A.O., Ermakov N.A., Zaitsev A.A. Surgery of massive ischemic stroke. Moscow: GEOTAR-Media; 2016. 136 p. [In Russian].

3. Ignatyeva V.I., Voznyuk I.A., Shamalov N.A., Reznik A.V., Vinitskii A.A., Derkach E.V. Socio-economic burden of stroke in the Russian Federation. Zhurnal nevrologii i psikhiatrii imeni S.S. Korsakova = S.S. Korsakov Journal of Neurology and Psychiatry. 2023;123(8–2):5–15. [In Russian]. doi:10.17116/jnevro20231230825

4. Clinical guidelines: Ischemic stroke and transient ischemic attack. Ministry of Health of the Russian Federation; 2024. 45 p. [In Russian].

5. Vasilev Yu.A., Arzamasov K.M., Kolsanov A.V., Vladzimirskiy A.V., Omelyanskaya O.V., Pestrenin L.D., Nechaev N.B. Experience of using artificial intelligence-based software for 800,000 fluorography studies. Vrach i informatsionnye tekhnologii = Medical Doctor and Information Technologies. 2023;(4):54–65. [In Russian]. doi: 10.25881/18110193_2023_4_54

6. Artyukova Z.R., Kudryavtsev N.D., Petryaykin A.V., Abuladze L.R., Smorchkova A.K., Akhmad E.S., Semenov D.S., Belyaev M.G., Belaya Zh.E., Vladzimirskiy A.V., Vasilev Yu.A. Application of artificial intelligence algorithm for assessing vertebral bone mineral density using computed tomography data. Meditsinskaya vizualizatsiya = Medical Visualization.

7. Morozov S.P., Vladzimirskiy A.V., Shulkin I.M., Ledikhova N.V., Arzamasov K.M., Andreycheno A.E., Logunova T.A., Omelyanskaya O.V., Gusev A.V. Feasibility study of artificial intelligence technologies in radiological diagnostics. Vrach i informatsionnye tekhnologii = Medical Doctor and Information Technologies. 2022;(1):12–29. [In Russian]. doi: 10.25881/18110193_2022_1_12

8. Zelter P.M., Kolsanov A.V., Chaplygin S.S., Pervushkin S.S. Visual and automated assessment of lung lesion volume on computed tomography in COVID-19 pneumonia. Vestnik meditsinskogo instituta «REAVIZ»: reabilitatsiya, vrach i zdorov’ye = Bulletin of the Medical Institute “REAVIZ”: Rehabilitation, Physician and Health. 2020;(6):5–13. [In Russian]. doi: 10.20340/vmi-rvz.2020.6.1

9. Qiu W., Kuang H., Teleg E., Ospel J.M., Sohn S.I., Almekhlafi M., Goyal M., Hill M.D., Demchuk A.M., Menon B.K. Machine learning for detecting early infarction in acute stroke with non-contrast-enhanced CT. Radiology. 2020;294(3):638–644. doi: 10.1148/radiol.2020191193

10. Andropova P.L., Gavrilov P.V., Kolesnikova P.A., Kushner A.V., Vladzimirskiy A.V., Vasilev Yu.A., Trofimova T.N. Diagnostic efficacy of individual automated CT image analysis systems for detecting ischemic stroke in the middle cerebral artery territory. Sibirskiy zhurnal klinicheskoy i eksperimentalnoy meditsiny = Siberian Journal of Clinical and Experimental Medicine. 2023;38(3):194–200. [In Russian]. doi:10.29001/2073-8552-2023-39-3-194-200

11. Andropova P.L., Kolesnikova P.A., Gavrilov P.V., Kokorev O.V., Chernova E.V., Kurnosenko O.A., Galustyan D.K., Boitsova M.G., Trofimova T.N. Impact of an automated system for detecting ischemic changes in the middle cerebral artery territory on clinical decision-making. Meditsinskaya vizualizatsiya = Medical Visualization. 2024;28(3):42–52. [In Russian]. doi:10.24835/1607-0763-1446

12. Artificial intelligence in radiology: Per aspera ad astra. Eds.: Yu.A. Vasilev, A.V. Vladzimirskiy. Moscow: Izdatelskie resheniya, 2025. 491 p.

13. Vladzymyrskiy A.V., Vasilev Yu.A., Arzamasov K.M., Omelyanskaya O.V., Ledikhova N.V., Morozov S.P., Gombolevskiy V.A., Balanyuk E.A., Yermolaev S.O., Zhyvodenko V.S., Idrisov I.M., Kirpichev Yu.S., Logunova T.A., Nuzhdina V.A., Rakovchen V.G., Slepushkina A.V. Computer vision in radiology: the first stage of the Moscow experiment. 2nd ed. Moscow: Izdatel’skiye resheniya, 2023. 388 p. [In Russian]

14. Bobrovskaya T.M., Vasilev Yu.A., Nikitin N.Yu., Vladzimirskiy A.V., Omelyanskaya O.V., Chetverikov S.F., Arzamasov K.M. Sample size estimation for assessing diagnostic accuracy of artificial intelligence-based software in radiology. Sibirskiy zhurnal klinicheskoy i eksperimentalnoy meditsiny = Siberian Journal of Clinical and Experimental Medicine. 2024;39(3):188–198. [In Russian]. doi: 10.29001/2073-8552-2024-39-3-188-198

15. Akbarzadeh M., Sanaie S., Rafsanjani M., Hosseini M.S. Role of imaging in early diagnosis of acute ischemic stroke: A literature review. Egypt J. Neurol. Psychiatry Neurosurg. 2021;57:126. doi:10.1186/s41983-021-00432-y

16. Pchelintseva O.I., Omelyanskaya O.V. On the features of ethical review in studies using artificial intelligence technologies and systems at the Moscow City Government Healthcare Institution “Scientific and Practical Clinical Center for Diagnostics and Telemedicine Technologies of the Moscow City Healthcare Department”. Digital Diagnostics. 2022;3(2):156–161. [In Russian]. doi: 10.17816/DD107983

17. Morozov S.P., Andreychenko A.E., Chetverikov S.F., Vladzimirskiy A.V., Omelyanskaya O.V., Arzamasov K.M., Vasilev Yu.A., Kolsanov A.V., Gusev A.V., Ledikhova N.V. Web tool for ROC analysis of diagnostic test results. Certificate of state registration of computer program No. 2022617324; published 19.04.2022. [In Russian].