Share：
Share this content in WeChat
X
[Chinese] [PDF] 31 1
• Clinical Article •
Diagnostic efficacy and efficiency of an AI-assisted combined-sequence MRI post-processing protocol in acute ischemic stroke
HUANG Yan  SUN Shuke  TAN Yingxun  WEI Lishan  YA Hanhua 

DOI:10.12015/issn.1674-8034.2026.02.008.


[Abstract] Objective To evaluate the diagnostic efficacy and processing efficiency of an artificial intelligence (AI) assisted, combined-sequence magnetic resonance imaging (MRI) post-processing protocol that integrates conventional sequences with perfusion-weighted imaging (PWI) in patients with acute ischemic stroke (AIS).Materials and Methods This prospective study enrolled 200 patients with AIS who underwent MRI at the First People's Hospital of Hechi between June 2023 and June 2025. The imaging protocol comprised conventional sequences [T1-weighted imaging (T1WI), T2-weighted imaging (T2WI), diffusion-weighted imaging (DWI), fluid-attenuated inversion recovery (FLAIR), magnetic resonance angiography (MRA)] and perfusion-weighted imaging (PWI) sequences [arterial spin labeling (ASL), dynamic susceptibility contrast (DSC)]. Two senior radiologists independently scored image quality on a 5-point Likert scale in a double-blind fashion; inter-observer agreement was assessed using weighted Kappa, and intra-observer reproducibility was evaluated with the intra-class correlation coefficient (ICC). Using the PWI data, participants were allocated to either manual post-processing (PWI group) or AI-assisted post-processing (PWI+AI group). Both approaches involved delineating regions of interest (ROIs) and quantifying cerebral blood flow (CBF) and cerebral blood volume (CBV). Diagnostic performance was evaluated by receiver operating characteristic (ROC) curve analysis, with areas under the curve (AUCs) compared using DeLong's test. Post-processing time was recorded and compared between the two groups.Results Subjective image quality scores for all sequences in the PWI+AI group were ≥ 3 (clinically acceptable), with a higher proportion of 5-point scores (76.0% to 77.5%) than the PWI group (all P < 0.05). Inter-observer agreement was good (weighted Kappa: 0.754 to 0.826; 95% CI: 0.715 to 0.855), and intra-observer reproducibility was excellent (ICC = 0.82, P < 0.001). CBF and CBV values in the PWI+AI group were lower than in the PWI group for both ASL and DSC (all P < 0.001). The PWI+AI group showed superior diagnostic performance (CBF AUC = 0.815, 95% CI: 0.751 to 0.894; CBV AUC = 0.826, 95% CI: 0.765 to 0.912) compared to the PWI group (CBF AUC = 0.674; CBV AUC = 0.681; both P < 0.05). Post-processing time in the PWI + AI group was reduced by 86.2% compared to the PWI group [(2.1 ± 0.6) min vs. (15.2 ± 3.5) min, P < 0.001].Conclusions The AI-assisted MRI combined-sequence post-processing protocol improves diagnostic performance and evaluation efficiency in AIS, providing reliable imaging support for emergency care and clinical decision-making.
[Keywords] acute ischemic stroke；magnetic resonance imaging；perfusion-weighted imaging；artificial intelligence；image post-processing；diagnostic performance；evaluation efficiency

HUANG Yan1   SUN Shuke1   TAN Yingxun1   WEI Lishan1   YA Hanhua2*  

1 Department of Radiology, the First People's Hospital of Hechi, Hechi 546300, China

2 Department of Neurology, the First People's Hospital of Hechi, Hechi 546300, China

Corresponding author: YA H H, E-mail: kjk0362@163.com

Conflicts of interest   None.

Received  2025-10-23
Accepted  2026-01-28
DOI: 10.12015/issn.1674-8034.2026.02.008
DOI:10.12015/issn.1674-8034.2026.02.008.

[1]
Chinese Society of Neurology, Chinese Stroke Society. Chinese guidelines for diagnosis and treatment of acute ischemic stroke 2023[J]. Chin J Neurol, 2024, 57(6): 523-559. DOI: 10.3760/cma.j.cn113694-20240410-00221.
[2]
RYU W S, HONG K S, JEONG S W, et al. Association of ischemic stroke onset time with presenting severity, acute progression, and long-term outcome: a cohort study[J/OL]. PLoS Med, 2022, 19(2): e1003910 [2025-10-22]. https://pubmed.ncbi.nlm.nih.gov/35120123/. DOI: 10.1371/journal.pmed.1003910.
[3]
GBD 2019 Stroke Collaborators. Global, regional, and national burden of stroke and its risk factors, 1990-2019: a systematic analysis for the Global Burden of Disease Study 2019[J]. Lancet Neurol, 2021, 20(10): 795-820. DOI: 10.1016/S1474-4422(21)00252-0.
[4]
HE M L, ZHANG S, LIU X, et al. Global, regional, and national epidemiology of ischemic stroke in young adults, 1990-2021[J/OL]. J Neurol, 2025, 272(5): 354 [2025-10-22]. https://pubmed.ncbi.nlm.nih.gov/40261367/. DOI: 10.1007/s00415-025-13082-4.
[5]
JOHANSEN M C. The future of ischemic stroke diagnosis and a review of underrecognized ischemic stroke etiologies[J]. Neurotherapeutics, 2023, 20(3): 613-623. DOI: 10.1007/s13311-023-01383-3.
[6]
QIN Y L, YU H, CHEN Y Q. Research progress of multimodal MRI in the diagnosis and prognosis assessment of acute ischemic stroke[J]. Chin J Magn Reson Imaging, 2022, 13(8): 112-116. DOI: 10.12015/issn.1674-8034.2022.08.025.
[7]
LI Z Y, ZHANG X M, CAO X H, et al. Magnetic resonance perfusion-weighted imaging signatures of cerebral microcirculation in patients with acute ischemic stroke[J]. Guangxi Med J, 2021, 43(18): 2154-2156. DOI: 10.11675/j.issn.0253-4304.2021.18.02.
[8]
LUO H D, LIU Y Y Q, ZHANG H N, et al. Optimization of multi-parameter MRI with flexible design sequences in the saddle area based on artificial intelligence-assisted compressed sensing technology[J]. Chin J Magn Reson Imaging, 2024, 15(12): 150-156. DOI: 10.12015/issn.1674-8034.2024.12.022.
[9]
YANG Z C, ZHAN Y, SHI N N, et al. Comparative use of artificial intelligence-assisted compressed sensing and parallel imaging for shoulder magnetic resonance imaging[J]. Chin J Magn Reson Imaging, 2024, 15(8): 166-171. DOI: 10.12015/issn.1674-8034.2024.08.025.
[10]
JIANG B, PHAM N, VAN STAALDUINEN E K, et al. Deep learning applications in imaging of acute ischemic stroke: a systematic review and narrative summary[J/OL]. Radiology, 2025, 315(1): e240775 [2025-10-22]. https://pubmed.ncbi.nlm.nih.gov/40197098/. DOI: 10.1148/radiol.240775.
[11]
National Health Commission Emergency Medicine Quality Control Center, Chinese Medical Doctor Association Emergency Physician Branch, World Federation of Chinese Medicine Societies Emergency Professional Committee, et al. Expert consensus on emergency diagnosis and treatment of acute ischemic stroke in China[J]. Chin J Crit Care Med, 2018, 38(4): 281-287. DOI: 10.3969/jz.issn.1002-1949.2018.04.001.
[12]
Chinese Society of Radiology Quality Management and Safety Management Group, Chinese Society of Radiology Magnetic Resonance Group. Expert consensus on standardized application of arterial spin labeling brain perfusion MRI technology[J]. Chin J Radiol, 2016, 50(11): 817-824. DOI: 10.3760/cma.j.issn.1005-1201.2016.11.003.
[13]
OLTHOF S C, WEILAND E, BENKERT T, et al. Optimizing image quality with high-resolution, deep-learning-based diffusion-weighted imaging in breast cancer patients at 1.5 T[J/OL]. Diagnostics (Basel), 2024, 14(16): 1742 [2025-10-22]. https://pubmed.ncbi.nlm.nih.gov/39202230/. DOI: 10.3390/diagnostics14161742.
[14]
YALÇIN S, VURAL H. Brain stroke classification and segmentation using encoder-decoder based deep convolutional neural networks[J/OL]. Comput Biol Med, 2022, 149: 105941 [2025-10-22]. https://pubmed.ncbi.nlm.nih.gov/36055156/. DOI: 10.1016/j.compbiomed.2022.105941.
[15]
GUI H Y, DU J X, ZHANG J J, et al. Evaluation value of 3D-ASL and 4D-TRANCE technique for grading of acute ischemic stroke[J]. Chin Comput Med Imag, 2025, 31(1): 1-7. DOI: 10.19627/j.cnki.cn31-1700/th.2025.01.011.
[16]
ZHOU Y X, JIA S Q, NIE S D. Application of computer-aided diagnosis for cerebrovascular diseases[J]. Chin J Med Phys, 2023, 40(9): 1167-1173. DOI: 10.3969/j.issn.1005-202X.2023.09.018.
[17]
HE Q M, LI G M, JIANG M E, et al. Predicting a favorable (mRS 0-2) or unfavorable (mRS 3-6) stroke outcome by arterial spin labeling and amide proton transfer imaging in post-thrombolysis stroke patients[J/OL]. J Pers Med, 2023, 13(2): 248 [2025-10-22]. https://pubmed.ncbi.nlm.nih.gov/36836482/. DOI: 10.3390/jpm13020248.
[18]
NING L, HEI H H. The value of apparent diffusion coefficient and arterial spin labeling imaging in predicting bleeding transformation after vascular recanalization therapy in acute ischemic stroke[J]. Shanxi Med J, 2022, 51(4): 400-402. DOI: 10.3969/j.issn.0253-9926.2022.04.010.
[19]
WOLFF L, SU J H, VAN LOON D, et al. Inter-rater reliability for assessing intracranial collaterals in patients with acute ischemic stroke: comparing 29 raters and an artificial intelligence-based software[J]. Neuroradiology, 2022, 64(12): 2277-2284. DOI: 10.1007/s00234-022-02984-z.
[20]
WEIL E L, NOSEWORTHY P A, LOPEZ C L, et al. Artificial intelligence-enabled electrocardiogram for atrial fibrillation identifies cognitive decline risk and cerebral infarcts[J]. Mayo Clin Proc, 2022, 97(5): 871-880. DOI: 10.1016/j.mayocp.2022.01.026.
[21]
SUN F X, ZHANG Z, ZHANG L. Comparison of 3D-arterial spin labeling and digital subtraction angiography in the diagnosis of occlusion and collateral circulation in patients with acute ischemic stroke[J]. J Chin Physician, 2023, 25(1): 81-85, 91. DOI: 10.3760/cma.j.cn431274-20220322-00217.
[22]
KUANG Y, CHEN Y, LIU R. Clinical value of 3D-arterial spin labeling for the evaluation of collateral circulation in patients with acute cerebral infarction[J]. Chin J Integr Med Cardio/cerebrovascular Dis, 2023, 21(19): 3616-3620. DOI: 10.12102/j.issn.1672-1349.2023.19.027.
[23]
BIAN X B, ZHANG D K, LÜ J H, et al. The clinical application value of multi-modality MRI technique in acute ischemic stroke[J]. China Med Equip, 2020, 17(3): 57-61. DOI: 10.3969/J.ISSN.1672-8270.2020.03.015.
[24]
ZHOU J, WANG K L, ZHANG X. Clinical analysis of multimodal MRI in the evaluation of collateral circulation in acute ischemic stroke[J]. Chin J CT MRI, 2022, 20(5): 26-29. DOI: 10.3969/j.issn.1672-5131.2022.05.009.
[25]
LI B, LIU G P. Magnetic resonance imaging image segmentation under artificial intelligence neural network for evaluation of the effect of butyphthalide combined with edaravone on neurological function in patients with acute cerebral infarction[J/OL]. Front Neurorobot, 2021, 15: 719145 [2025-10-22]. https://pubmed.ncbi.nlm.nih.gov/34658830/. DOI: 10.3389/fnbot.2021.719145.
[26]
TEICHMANN L J, KHALIL A A, VILLRINGER K, et al. Evaluation of Siemens Healthineers' StrokeSegApp for automated diffusion and perfusion lesion segmentation in patients with ischemic stroke[J/OL]. Front Neurol, 2025, 16: 1518477 [2025-10-22]. https://pubmed.ncbi.nlm.nih.gov/39926019/. DOI: 10.3389/fneur.2025.1518477.
[27]
NAEL K, YOO B, SALAMON N, et al. Acute ischemic stroke: MR imaging-based paradigms[J]. Neuroimaging Clin N Am, 2021, 31(2): 177-192. DOI: 10.1016/j.nic.2021.01.002.
[28]
Chinese Medical Doctor Association Radiologist Branch. Chinese expert recommendations on the standardization of study design of deep learning-based automatic imaging detection in cerebrovascular disease[J]. Chin J Radiol, 2023, 57(1): 17-26. DOI: 10.3760/cma.j.cn112149-20211225-01146.
[29]
GUBSKIY I L, NAMESTNIKOVA D D, CHERKASHOVA E A, et al. Comparative assessment of manual segmentation of cerebral infarction lesions in experimental animals based on magnetic resonance imaging using artificial intelligence[J]. Bull Exp Biol Med, 2025, 178(4): 514-519. DOI: 10.1007/s10517-025-06366-2.
[30]
LI S F, HUANG X, LI D X, et al. A clinical study of multimodal MRI combined with estroke artificial intelligence to evaluate arterial thrombectomy in late time window AIS[J]. Chin J CT MRI, 2024, 22(7): 14-17. DOI: 10.3969/j.issn.1672-5131.2024.07.005.
[31]
SOUN J E, CHOW D S, NAGAMINE M, et al. Artificial intelligence and acute stroke imaging[J]. AJNR Am J Neuroradiol, 2021, 42(1): 2-11. DOI: 10.3174/ajnr.A6883.
[32]
TANG C S, HU C C, SUN J D, et al. Deep learning-based medical images analysis evolved from convolution to graph convolution[J]. J Image Graph, 2021, 26(9): 2078-2093. DOI: 10.11834/jig.200666.
[33]
SHI Z, ZHANG L J. Artificial intelligence in cerebrovascular disease imaging[J]. Chin J Radiol, 2021, 55(8): 793-798. DOI: 10.3760/cma.j.cn112149-20210616-00564.
[34]
HAN X W, LI M, ZHANG B. Application of artificial intelligence in neuroimaging of stroke[J]. Med J Peking Union Med Coll Hosp, 2021, 12(5): 749-754. DOI: 10.12290/xhyxzz.2021-0491.
[35]
WANG H, LI J, ZHANG Y. Research progress of artificial intelligence-assisted magnetic resonance perfusion sequence in evaluating ischemic penumbra of AIS patients[J]. Chin J Magn Reson Imaging, 2025, 16(3): 145-148. DOI: 10.12015/issn.1674-8034.2025.03.023.
[36]
LAN W, AI P Y, XU Q. Deep-learning-based MRI in the diagnosis of cerebral infarction and its correlation with the neutrophil to lymphocyte ratio[J]. Ann Palliat Med, 2021, 10(11): 11370-11381. DOI: 10.21037/apm-21-1786.
[37]
ZHANG L F, LI Y R, BIAN L, et al. Cognitive impairment of patient with neurological cerebrovascular disease using the artificial intelligence technology guided by MRI[J/OL]. Front Public Health, 2022, 9: 813641 [2025-10-22]. https://pubmed.ncbi.nlm.nih.gov/35310781/. DOI: 10.3389/fpubh.2021.813641.
[38]
SUN S S, LIU D Y, ZHOU Y F, et al. Longitudinal real world correlation study of blood pressure and novel features of cerebral magnetic resonance angiography by artificial intelligence analysis on elderly cognitive impairment[J/OL]. Front Aging Neurosci, 2023, 15: 1121152 [2025-10-22]. https://pubmed.ncbi.nlm.nih.gov/36819723/. DOI: 10.3389/fnagi.2023.1121152.
[39]
QU Y C, WANG X C. Research progress of perfusion imaging in the treatment and prognosis assessment of ischemic stroke[J]. Chin J Magn Reson Imaging, 2025, 16(2): 130-134. DOI: 10.12015/issn.1674-8034.2025.02.021.
[40]
RYU W S, SCHELLINGERHOUT D, PARK J, et al. Deep learning-based automatic segmentation of cerebral infarcts on diffusion MRI[J/OL]. Sci Rep, 2025, 15: 13214 [2025-10-22]. https://pubmed.ncbi.nlm.nih.gov/40240396/. DOI: 10.1038/s41598-025-91032-w.

PREV Associations between plasma chitinase-3-like protein 1 levels and white matter microstructure in patients with amnestic mild cognitive impairment: A DTI study
NEXT Research on the correlation among white matter hyperintensities, cerebral blood flow, and collateral circulation in patients with unilateral middle cerebral artery atherosclerotic stenosis
  


LINK


Tel & Fax: +8610-67113815    E-mail: editor@cjmri.cn