Share：
Share this content in WeChat
X
[Chinese][PDF]1762
• Clinical Article •
Prognostic value of deep learning models based on dual-center MRI-DWI in predicting outcomes of intravenous thrombolysis for acute ischemic stroke
YANG Huan  WANG Wenxi  ZHANG Jun  YU Yang  WANG Zhanqiu  WU Lei 

Cite this article as: YANG H, WANG W X, ZHANG J, et al. Prognostic value of deep learning models based on dual-center MRI-DWI in predicting outcomes of intravenous thrombolysis for acute ischemic stroke[J]. Chin J Magn Reson Imaging, 2025, 16(1): 95-103. DOI:10.12015/issn.1674-8034.2025.01.015.


[Abstract] Objective To develop a deep learning model based on MRI diffusion-weighted imaging (DWI) and evaluate its ability to predict 90-day outcomes in acute ischemic stroke (AIS) patients undergoing intravenous thrombolysis.Materials and Methods A retrospective analysis was conducted on clinical and imaging data from 677 AIS patients treated with intravenous thrombolysis at two hospitals. MRI-DWI images were collected through picture archiving and communication systems (PACS). A deep neural network was used to extract imaging features. Dataset 1 (Hospital 1) was randomly split into a training set (70%) and a testing set (30%) to develop four models: a clinical features-based machine learning model (Model A), an MRI-DWI radiomics features-based machine learning model (Model B), a deep learning model using MRI-DWI features (Model C), and a combined model integrating clinical and deep learning features (Model D) to predict 90-day outcomes [Patients with a modified Rankin Scale (mRS) score less than 2 at 90 days are considered to have a good prognosis]. Dataset 2 (Hospital 2) was used for external validation. Predictive performance was assessed using the receiver operating characteristic (ROC) curve and area under the curve (AUC).Results The AUCs of Models A, B, and C were 0.705 [95% confidence interval (CI): 0.613 to 0.792], 0.846 (95% CI: 0.777 to 0.906), and 0.877 (95% CI: 0.811 to 0.934), respectively. Model D demonstrated superior predictive performance with an AUC of 0.930 (95% CI: 0.890 to 0.963). External validation showed consistent performance, with AUCs of 0.887 (95% CI: 0.798 to 0.960) for Model C and 0.947 (95% CI: 0.891 to 0.984) for Model D.Conclusions MRI-DWI radiomics features play a crucial role in predicting 90-day outcomes in AIS patients treated with intravenous thrombolysis. Deep learning models outperform traditional machine learning models, and the integration of clinical and deep learning features provides a robust tool for personalized prognosis and treatment planning in AIS.
[Keywords] acute ischemic stroke；prognosis model；machine learning；deep learning；radiomic；magnetic resonance imaging；diffusion-weighted imaging

YANG Huan1   WANG Wenxi2   ZHANG Jun3   YU Yang2   WANG Zhanqiu2   WU Lei4*  

1 Department of Emergency, the First Hospital of Qinhuangdao, Qinhuangdao 066000, China

2 Department of Imaging Center, the First Hospital of Qinhuangdao, Qinhuangdao 066000, China

3 Department of Imaging, Qinhuangdao Funing District People's Hospital, Qinhuangdao 066300, China

4 Department of Stroke Center, the First Hospital of Qinhuangdao, Qinhuangdao 066000, China

Corresponding author: WU L, E-mail: 18903350011@189.cn

Conflicts of interest   None.

Received  2024-07-01
Accepted  2024-12-16
DOI: 10.12015/issn.1674-8034.2025.01.015
Cite this article as: YANG H, WANG W X, ZHANG J, et al. Prognostic value of deep learning models based on dual-center MRI-DWI in predicting outcomes of intravenous thrombolysis for acute ischemic stroke[J]. Chin J Magn Reson Imaging, 2025, 16(1): 95-103. DOI:10.12015/issn.1674-8034.2025.01.015.

[1]
Neurology Branch of Chinese Medical Association, Cerebrovascular Disease Group of Neurology Branch of Chinese Medical Association. Chinese guidelines for diagnosis and treatment of acute ischemic stroke 2018[J]. Chin J Neurol, 2018, 51(9): 666-682. DOI: 10.3760/cma.j.issn.1006-7876.2018.09.004.
[2]
GBD 2016 Causes of Death Collaborators. Global, regional, and national age-sex specific mortality for 264 causes of death, 1980-2016: a systematic analysis for the Global Burden of Disease Study 2016[J]. Lancet, 2017, 390(10100): 1151-1210. DOI: 10.1016/S0140-6736(17)32152-9.
[3]
Chinese Nursing Association Internal Medicine Committee, Xuanwu Hospital, Capital Medical University. Acute ischemic cerebral apoplexy intravenous thrombolysis nursing guide[J]. Chin J Nurs, 2023, 58(1): 10-15. DOI: 10.3761/j.issn.0254-1769.2023.01.001.
[4]
MARKO M, POSEKANY A, SZABO S, et al. Trends of r-tPA (recombinant tissue-type plasminogen activator) treatment and treatment-influencing factors in acute ischemic stroke[J]. Stroke, 2020, 51(4): 1240-1247. DOI: 10.1161/STROKEAHA.119.027921.
[5]
GBD 2015 Neurological Disorders Collaborator Group. Global, regional, and national burden of neurological disorders during 1990-2015: a systematic analysis for the Global Burden of Disease Study 2015[J]. Lancet Neurol, 2017, 16(11): 877-897. DOI: 10.1016/S1474-4422(17)30299-5.
[6]
XU H N, LIANG H, WANG Z P, et al. Application of 3.0T HR-MRI technique in AIS and an analysis on the influence factors of prognosis[J]. China Medical Equipment, 2024, 21(1): 63-68. DOI: 10.3969/j.issn.1672-8270.2024.01.013.
[7]
XIE X H, BAI Q K, ZHAO Z G, et al. Application of fast multimode MRI based emergency assessment of hyperacute stroke thrombolysis[J]. Chinese Computed Medical Imaging, 2014, 20(4): 309-312. DOI: 10.3969/j.issn.1006-5741.2014.04.002.
[8]
YU Y, XIE Y, THAMM T, et al. Use of deep learning to predict final ischemic stroke lesions from initial magnetic resonance imaging[J/OL]. JAMA Netw Open, 2020, 3(3): e200772 [2024-07-01]. https://www.ncbi.nlm.nih.gov/pubmed/32163165. DOI: 10.1001/jamanetworkopen.2020.0772.
[9]
YU H, WANG Z, SUN Y, et al. Prognosis of ischemic stroke predicted by machine learning based on multi-modal MRI radiomics[J/OL]. Front Psychiatry, 2022, 13: 1105496 [2024-07-01]. https://www.ncbi.nlm.nih.gov/pubmed/36699499. DOI: 10.3389/fpsyt.2022.1105496.
[10]
LI Y, LIU Y, HONG Z, et al. Combining machine learning with radiomics features in predicting outcomes after mechanical thrombectomy in patients with acute ischemic stroke[J/OL]. Comput Methods Programs Biomed, 2022, 225: 107093 [2024-07-01]. https://www.ncbi.nlm.nih.gov/pubmed/36055039. DOI: 10.1016/j.cmpb.2022.107093.
[11]
KIANMEHR H, GUO J, LIN Y, et al. A machine learning approach identifies modulators of heart failure hospitalization prevention among patients with type 2 diabetes: A revisit to the ACCORD trial[J/OL]. J Diabetes Complications, 2022, 36(9): 108287 [2024-07-01]. https://www.ncbi.nlm.nih.gov/pubmed/36007486. DOI: 10.1016/j.jdiacomp.2022.108287.
[12]
FATAHZADEH M, GLICK M. Stroke: epidemiology, classification, risk factors, complications, diagnosis, prevention, and medical and dental management[J]. Oral Surg Oral Med Oral Pathol Oral Radiol Endod, 2006, 102(2): 180-191. DOI: 10.1016/j.tripleo.2005.07.031.
[13]
XU J J, WANG Z N, XU X, et al. The study of machine learning based on DWI infarction and cerebrospinal fluid in predicting malignant edema after thrombectomy for acute stroke[J]. Journal of Medical Imaging, 2013, 34(1): 7-10.
[14]
YANG H, LV Z, WANG W, et al. Machine learning models for predicting early neurological deterioration and risk classification of acute ischemic stroke[J/OL]. Clin Appl Thromb Hemost, 2023, 29: 10760296231221738 [2024-07-01]. https://www.ncbi.nlm.nih.gov/pubmed/38115694. DOI: 10.1177/10760296231221738.
[15]
LIU J M, LUO H W, YU P F, et al. Construction and evaluation of a machine-learning-based model for predic-ting healthcare-associated infection in patients with acute ischemic stroke[J]. Chin J Infec Contl, 2023, 22(2): 129-135. DOI: 10.12138/j.iSSN.1671-9638.20233300.
[16]
PETER R, KORFIATIS P, BLEZEK D, et al. A quantitative symmetry-based analysis of hyperacute ischemic stroke lesions in noncontrast computed tomography[J]. Med Phys, 2017, 44(1): 192-199. DOI: 10.1002/mp.12015.
[17]
XU C F, HUANG Y, ZHANG K Z. Acute ischemic stroke patients with onset timewithin 4.5 h identified by machine learning combined with magnetic resonance diffusion weighted imaging[J]. Journal of Jilin University (Medical Science Edition), 2023, 49(6): 1635-1641. DOI: 10.13481/j.1671-587X.20230631.
[18]
LIU Z, HUANG X B, PENG M Y, et al. Construction of a nomogram model for hemorrhagic transformation risk after mechanical thrombectomy in patients with acute stroke[J]. Chin J Magn Reson Imaging, 2022, 13(4): 15-19. DOI: 10.12015/issn.1674-8034.2022.04.003.
[19]
KANAZAWA T, TAKAHASHI S, MINAMI Y, et al. Early prediction of clinical outcomes in patients with aneurysmal subarachnoid hemorrhage using computed tomography texture analysis[J]. J Clin Neurosci, 2020, 71: 144-149. DOI: 10.1016/j.jocn.2019.08.098.
[20]
TANG T Y, JIAO Y, CUI Y, et al. Penumbra-based radiomics signature as prognostic biomarkers for thrombolysis of acute ischemic stroke patients: a multicenter cohort study[J]. J Neurol, 2020, 267(5): 1454-1463. DOI: 10.1007/s00415-020-09713-7.
[21]
LUO X, CHENG Y, WU C, et al. An interpretable machine learning-based prediction model for risk of death for patients with ischemic stroke in intensive care unit[J]. Journal of Southern Medical University, 2023(7): 1241-1247. DOI: 10.12122/j.issn.1673-4254.2023.07.21.
[22]
KOO T K, LI M Y. A guideline of selecting and reporting intraclass correlation coefficients for reliability research[J]. J Chiropr Med, 2016, 15(2): 155-163. DOI: 10.1016/j.jcm.2016.02.012.
[23]
SIMON N, FRIEDMAN J, HASTIE T, et al. Regularization paths for Cox's proportional hazards model via coordinate descent[J]. J Stat Softw, 2011, 39(5): 1-13. DOI: 10.18637/jss.v039.i05.
[24]
JIANG F, JIANG Y, ZHI H, et al. Artificial intelligence in healthcare: past, present and future[J]. Stroke Vasc Neurol, 2017, 2(4): 230-243. DOI: 10.1136/svn-2017-000101.
[25]
KHEMANI B, PATIL S, KOTECHA K, et al. Detecting health misinformation: A comparative analysis of machine learning and graph convolutional networks in classification tasks[J/OL]. MethodsX, 2024, 12: 102737 [2024-07-01]. https://pubmed.ncbi.nlm.nih.gov/38774687/. DOI: 1016/j.mex.2024.102737.
[26]
EGERT J, BROMBACHER E, WARSCHEID B, et al. DIMA: data-driven selection of an imputation algorithm[J]. J Proteome Res, 2021, 20(7): 3489-3496. DOI: 10.1021/acs.jproteome.1c00119.
[27]
MOULTON E, VALABREGUE R, PIOTIN M, et al. Interpretable deep learning for the prognosis of long-term functional outcome post-stroke using acute diffusion weighted imaging[J]. J Cereb Blood Flow Metab, 2023, 43(2): 198-209. DOI: 10.1177/0271678X221129230.
[28]
YE W, TAO Y J, CHEN X C, et al. Deep ensemble evolutionary multi-classification method for predicting prognosis of stroke[J]. Comput Eng Appl, 2023, 59(5): 95-105. DOI: 10.3778/j.iSSN.1002-8331.2211-0077.
[29]
KIM D Y, CHOI K H, KIM J H, et al. Deep learning-based personalised outcome prediction after acute ischaemic stroke[J]. J Neurol Neurosurg Psychiatry, 2023, 94(5): 369-378. DOI: 10.1136/jnnp-2022-330230.
[30]
TURHON M, LI M, KANG H, et al. Development and validation of a deep learning model for prediction of intracranial aneurysm rupture risk based on multi-omics factor[J]. Eur Radiol, 2023, 33(10): 6759-6770. DOI: 10.1007/s00330-023-09672-3.
[31]
LIU Y, YU Y, OUYANG J, et al. Functional outcome prediction in acute ischemic stroke using a fused imaging and clinical deep learning model[J]. Stroke, 2023, 54(9): 2316-2327. DOI: 10.1161/STROKEAHA.123.044072.
[32]
DHAR R. Automated quantitative assessment of cerebral edema after ischemic stroke using CSF volumetrics[J/OL]. Neurosci Lett, 2020, 724: 134879 [2024-07-01]. https://pubmed.ncbi.nlm.nih.gov/32126249/. DOI: 1016/j.neulet.2020.134879.
[33]
JO H, KIM C, GWON D, et al. Combining clinical and imaging data for predicting functional outcomes after acute ischemic stroke: an automated machine learning approach[J/OL]. Sci Rep, 2023, 13(1): 16926 [2024-07-01]. https://www.ncbi.nlm.nih.gov/pubmed/37805568. DOI: 10.1038/s41598-023-44201-8.
[34]
LIU Y, SHAH P, YU Y, et al. A clinical and imaging fused deep learning model matches expert clinician prediction of 90-day stroke outcomes[J]. AJNR Am J Neuroradiol, 2024, 45(4): 406-411. DOI: 10.3174/ajnr.A8140.
[35]
SUN Y Y, ZHOU L Y, GE X L, et al. Research on prognosis and warning indicators of acute ischemic stroke treated with thrombolytic therapy[J]. Chinese Journal of Eme rgency Medicine, 2019, 28(2): 214-218. DOI: 10.3760/cma.j.issn.1671-0282.2019.02.017.
[36]
SUN H, ZHOU F, ZHANG G, et al. A novel nomogram for predicting prognosis after mechanical thrombectomy in patients with acute ischemic stroke[J]. Curr Neurovasc Res, 2021, 18(5): 479-488. DOI: 10.2174/1567202618666211210154739.
[37]
PAPADOPOULOU A, BOUNTOUVI E, KARACHALIOU F E. The molecular basis of calcium and phosphorus inherited metabolic disorders[J/OL]. Genes (Basel), 2021, 12(5): 734 [2024-07-01]. https://www.ncbi.nlm.nih.gov/pubmed/34068220. DOI: 10.3390/genes12050734.
[38]
WANG X R, YUAN L, SHI R, et al. Predictors of coronary artery calcification and its association with cardiovascular events in patients with chronic kidney disease[J]. Ren Fail, 2021, 43(1): 1172-1179. DOI: 10.1080/0886022X.2021.1953529.
[39]
CHEN Y, ZHAO X, WU H. Arterial stiffness: A focus on vascular calcification and its link to bone mineralization[J]. Arterioscler Thromb Vasc Biol, 2020, 40(5): 1078-1093. DOI: 10.1161/ATVBAHA.120.313131.
[40]
DOSHI S M, WISH J B. Past, present, and future of phosphate management[J]. Kidney Int Rep, 2022, 7(4): 688-698. DOI: 10.1016/j.ekir.2022.01.1055.

PREV A preliminary comparative study on the characteristics of resting-state brain functional networks in patients with comorbid insomnia in first-episode and recurrent depression
NEXT Analysis of brain MRI abnormalities in infantile spasms treated with vigabatrin
  


LINK


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