Share:
Share this content in WeChat
X
• Clinical Article •
Development and validation of an MRI radiomics model for the diagnosis of cerebral venous thrombosis
ZHANG Gaoxing  WANG Qianqian  YANG Wenhui  WANG Meijia  ZHANG Xiaoling  TANG Min 

DOI:10.12015/issn.1674-8034.2026.11.008.


[Abstract] Objective To construct a fused model integrating multi-sequence MRI radiomic features, imaging features and clinical data, and to evaluate its diagnostic performance in cerebral venous sinus thrombosis (CVST), so as to provide quantitative references for early non-invasive auxiliary diagnosis of CVST.Materials and Methods A retrospective analysis was conducted on 242 patients clinically suspected of CVST, randomly divided into a training set (169 cases) and a validation set (73 cases) at a 7∶3 ratio. Region of interest (ROI) delineation of the dural venous sinuses was manually performed on T2WI. The resulting ROI masks were then propagated to the co-registered T1WI and Fluid-attenuated inversion recovery (FLAIR) sequences via rigid registration. Pyradiomics was used to extract radiomics features, and highly correlated redundant features were removed based on Spearman correlation coefficients. Ten-fold cross-validation and least absolute shrinkage and selection operator (LASSO) were then employed to screen the remaining features. Meaningful imaging and clinical indicators were screened through univariate and multivariate logistic regression analyses. Single-sequence models and the fused model were constructed based on the LR algorithm. Model performance was evaluated using receiver operating characteristic (ROC) curves and the area under the curve (AUC). The DeLong test was used to compare differences in AUC, and decision curve analysis (DCA) was used to assess the clinical application value of the models.Results The fused model demonstrated the optimal performance, with an AUC of 0.974 [95% confidence interval (CI): 0.955 to 0.992] in the training set and an AUC of 0.946 (95% CI: 0.898 to 0.993) in the validation set. In the validation set, the sensitivity, specificity, accuracy, positive predictive value, and negative predictive value were 89.7%, 79.4%, 84.9%, 0.833, and 0.871, respectively, Brier = 0.095. The DeLong test showed that the AUC of the fused model was superior to that of the single-sequence models (P < 0.05). Decision curve analysis demonstrated that this model provided clinical net benefit.Conclusions The fused model exhibits favorable diagnostic efficacy for CVST.
[Keywords] cerebral venous sinus thrombosis;radiomics;magnetic resonance imaging;logistic models;nomograms

ZHANG Gaoxing1, 2   WANG Qianqian1   YANG Wenhui2   WANG Meijia3   ZHANG Xiaoling4   TANG Min1*  

1 Department of MRI, Shaanxi Provincial People's Hospital, Xi'an 710068, China

2 School of Medical Technology, Shaanxi University of Chinese Medicine, Xianyang 712046, China

3 Graduate School of Xi'an Medical University, Xi'an 710021, China

4 Department of Scientific Research, Provincial People's Hospital, Xi'an 710068, China

Corresponding author: TANG M, E-mail: tm1981tm@163.com

Conflicts of interest   None.

Received  2026-05-27
Accepted  2026-09-29
DOI: 10.12015/issn.1674-8034.2026.11.008
DOI:10.12015/issn.1674-8034.2026.11.008.

[1]
SILVIS S M, DE SOUSA D A, FERRO J M, et al. Cerebral venous thrombosis[J]. Nat Rev Neurol, 2017, 13(9): 555-565. DOI: 10.1038/nrneurol.2017.104.
[2]
NGUYEN V N, DEMETRIOU A N, DALLAS J, et al. Cerebral venous sinus thrombosis[J]. Neurosurg Clin N Am, 2024, 35(3): 343-353. DOI: 10.1016/j.nec.2024.02.006.
[3]
DIX C, HUNT B J. The changing face of cerebral venous sinus thrombosis: emerging new causes and treatments[J]. J Thromb Haemost, 2024, 22(12): 3346-3354. DOI: 10.1016/j.jtha.2024.08.012.
[4]
BAKRADZE E, SHU L Q, HENNINGER N, et al. Delayed diagnosis in cerebral venous thrombosis: associated factors and clinical outcomes[J/OL]. J Am Heart Assoc, 2023, 12(19): e030421 [2026-05-26]. https://pubmed.ncbi.nlm.nih.gov/37753785/. DOI: 10.1161/jaha.123.030421.
[5]
SALEHI OMRAN S, SHU L Q, CHANG A, et al. Timing and predictors of recanalization after anticoagulation in cerebral venous thrombosis[J]. J Stroke, 2023, 25(2): 291-298. DOI: 10.5853/jos.2023.00213.
[6]
ABDULAZIZ S S, HASHIM A R, HASRAT N H, et al. Clinico-epidemiological study on cerebral venous sinus thrombosis: manifestations, complications, risk factors, and diagnostic challenges[J/OL]. Cureus, 2026, 18(1): e102448 [2026-05-26]. https://pubmed.ncbi.nlm.nih.gov/41769570/. DOI: 10.7759/cureus.102448.
[7]
SAPOSNIK G, BUSHNELL C, COUTINHO J M, et al. Diagnosis and management of cerebral venous thrombosis: a scientific statement from the American heart association[J/OL]. Stroke, 2024, 55(3): e77-e90 [2026-05-26]. https://pubmed.ncbi.nlm.nih.gov/38284265/. DOI: 10.1161/STR.0000000000000456.
[8]
KHAN F, SEYAM M, SHARMA N, et al. New horizons for diagnostic pitfalls of cerebral venous thrombosis: clinical utility of a newly developed cerebral venous thrombosis diagnostic score: a case report and literature review[J/OL]. Am J Case Rep, 2021, 22: e932123 [2026-05-26]. https://pubmed.ncbi.nlm.nih.gov/34224551/. DOI: 10.12659/AJCR.932123.
[9]
ZEDDE M, PASCARELLA R. Non-thrombotic filling defects in cerebral veins and sinuses: when normal structures mimic a disease[J/OL]. Neurol Int, 2025, 17(1): 9 [2026-05-26]. https://pubmed.ncbi.nlm.nih.gov/39852773/. DOI: 10.3390/neurolint17010009.
[10]
SADIK J C, JIANU D C, SADIK R, et al. Imaging of cerebral venous thrombosis[J/OL]. Life, 2022, 12(8): 1215 [2026-05-26]. https://www.mdpi.com/2075-1729/12/8/1215. DOI: 10.3390/life12081215.
[11]
ROSA S, FRAGATA I, AGUIAR DE SOUSA D. Update on management of cerebral venous thrombosis[J]. Curr Opin Neurol, 2025, 38(1): 18-28. DOI: 10.1097/WCO.0000000000001329.
[12]
LAMBIN P, WOODRUFF H C, MALI S A, et al. Radiomics Quality Score 2.0: towards radiomics readiness levels and clinical translation for personalized medicine[J]. Nat Rev Clin Oncol, 2025, 22(11): 831-846. DOI: 10.1038/s41571-025-01067-1.
[13]
WEI B Q, LI Y J, ZHANG X Y, et al. Research progress of deep learning combined with radiomics in musculoskeletal diseases[J]. Chin J Magn Reson Imaging, 2026, 17(3): 213-220. DOI: 10.12015/issn.1674-8034.2026.03.031.
[14]
MCCAGUE C, RAMLEE S, REINIUS M, et al. Introduction to radiomics for a clinical audience[J]. Clin Radiol, 2023, 78(2): 83-98. DOI: 10.1016/j.crad.2022.08.149.
[15]
DUMITRIU LAGRANGE D, REYMOND P, BRINA O, et al. Spatial heterogeneity of occlusive thrombus in acute ischemic stroke: a systematic review[J]. J Neuroradiol, 2023, 50(3): 352-360. DOI: 10.1016/j.neurad.2023.01.004.
[16]
LIU Z Y, LIU Z Y, WANG Y, et al. Machine learning-based predictive analysis of energy efficiency factors necessary for the HIFU treatment of adenomyosis[J/OL]. Front Physiol, 2025, 16: 1602866 [2026-05-26]. https://pubmed.ncbi.nlm.nih.gov/40895435/. DOI: 10.3389/fphys.2025.1602866.
[17]
SANTO B A, JANBEH SARAYI S M M, MCCALL A D, et al. Multimodal CT imaging of ischemic stroke thrombi identifies scale-invariant radiomic features that reflect clot biology[J]. J Neurointerv Surg, 2023, 15(12): 1257-1263. DOI: 10.1136/jnis-2022-019967.
[18]
XU Q Q, ZHU Y, ZHANG X, et al. Clinical features and FLAIR radiomics nomogram for predicting functional outcomes after thrombolysis in ischaemic stroke[J/OL]. Front Neurosci, 2023, 17: 1063391 [2026-05-26]. https://pubmed.ncbi.nlm.nih.gov/36908776/. DOI: 10.3389/fnins.2023.1063391.
[19]
LIU L Z, GAO F, LI Y M, et al. An interpretable radiomics-based model using susceptibility-weighted imaging for non-invasive prediction of tertiary lymphoid structures in hepatocellular carcinoma[J/OL]. J Hepatocell Carcinoma, 2025, 12: 2197-2211 [2026-05-26]. https://pubmed.ncbi.nlm.nih.gov/41049157/. DOI: 10.2147/JHC.S551462.
[20]
LI J, ZHOU C R, QU X X, et al. Perilesional dominance: radiomics of multiparametric MRI enhances differentiation of IgG4-Related ophthalmic disease and orbital MALT lymphoma[J/OL]. BMC Med Imaging, 2025, 25(1): 238 [2026-05-26]. https://link.springer.com/article/10.1186/s12880-025-01771-5. DOI: 10.1186/s12880-025-01771-5.
[21]
ZHU Y X, YANG Z Y, LIU Z P, et al. Multimodal fusion of mpMRI radiomics, clinical features, and hematological biomarkers enhances machine learning-based prediction of biochemical recurrence in prostate cancer patients[J/OL]. Chem Biol Drug Des, 2025, 106(6): e70206 [2026-05-26]. https://onlinelibrary.wiley.com/doi/10.1111/cbdd.70206. DOI: 10.1111/cbdd.70206.
[22]
ALMEGREN M O. Cerebral venous thrombosis: a comprehensive narrative review[J]. Brain Circ, 2025, 11(3): 178-186. DOI: 10.4103/bc.bc_158_24.
[23]
ZUO W B, CHEN S C, LIN J S, et al. Cerebral venous thrombosis: current status and challenges[J/OL]. Brain Behav, 2025, 15(9): e70844 [2026-05-26]. https://onlinelibrary.wiley.com/doi/10.1002/brb3.70844. DOI: 10.1002/brb3.70844.
[24]
UDOVENKO A, MAKOGONENKO Y, KOROLOVA D, et al. Formation and elimination of soluble fibrin and D-dimer in the bloodstream[J]. Croat Med J, 2023, 64(6): 421-429. DOI: 10.3325/cmj.2023.64.421.
[25]
Chinese Stroke Association. Expert consensus on lipid management in panvascular diseases (2025 edition)[J].Natl Med J China, 2026, 106(6): 504-22. DOI: 10.3760/cma.j.cn112137-20250804-01959.
[26]
National Health Commission of the People's Republic of China. Guideline for prevention and treatment of cerebrovascular disease(2024 edition)[J]. Chin J Magn Reson Imaging, 2025, 16(1): 1-8. DOI: 10.12015/issn.1674-8034.2025.01.001.
[27]
RANJAN R, KEN-DROR G, SHARMA P. Pathophysiology, diagnosis and management of cerebral venous thrombosis: a comprehensive review[J/OL]. Medicine, 2023, 102(48): e36366 [2026-05-26]. https://pubmed.ncbi.nlm.nih.gov/38050259/. DOI: 10.1097/md.0000000000036366.
[28]
LI H, GUAN X J, LIU L, et al. Cerebral venous outflow insufficiency: a study on symptoms and venous stenosis classification[J/OL]. MedComm, 2026, 7(3): e70609 [2026-05-26]. https://pubmed.ncbi.nlm.nih.gov/41768367/. DOI: 10.1002/mco2.70609.
[29]
CAO Y, CHEN R C, KATZ A J. Why is a small sample size not enough [J]. Oncologist, 2024, 29(9): 761-763. DOI: 10.1093/oncolo/oyae162.
[30]
FIELD T S, FRAGATA I. Current management of cerebral venous thrombosis[J]. Stroke, 2026, 57(7): 2228-2236. DOI: 10.1161/STROKEAHA.125.051340.
[31]
XU W L, GAO L S, LI T, et al. The performance of CT versus MRI in the differential diagnosis of cerebral venous thrombosis[J]. Thromb Haemost, 2018, 118(6): 1067-1077. DOI: 10.1055/s-0038-1642636.
[32]
BOUKERCHE F, BALAKRISHNAN S, KALAPOS P, et al. Detection of cerebral cortical vein thrombosis with high-resolution susceptibility weighted imaging: a comparison with MR venography and standard MR sequences[J]. Neuroradiology, 2023, 65(5): 885-892. DOI: 10.1007/s00234-023-03123-y.
[33]
SHEN J, TAO Z, CHEN W, et al. Malignant isolated cortical vein thrombosis as the initial manifestation of primary antiphospholipid syndrome: lessons on diagnosis and management from a case report[J/OL]. Front Immunol, 2022, 13: 882032 [2026-05-26]. https://pubmed.ncbi.nlm.nih.gov/35547735/. DOI: 10.3389/fimmu.2022.882032.
[34]
WU X Q, YING J B, LI J, et al. A deep learning based radiomics model for differentiating intraparenchymal hematoma induced by cerebral venous thrombosis[J/OL]. Sci Rep, 2025, 16(1): 1423 [2026-05-26]. https://pubmed.ncbi.nlm.nih.gov/41350423/. DOI: 10.1038/s41598-025-31206-8.
[35]
CHANG Y Z, ZHU H Y, SONG Y Q, et al. High-resolution magnetic resonance imaging-based radiomic features aid in selecting endovascular candidates among patients with cerebral venous sinus thrombosis[J/OL]. Thromb J, 2023, 21(1): 116 [2026-05-26]. https://pubmed.ncbi.nlm.nih.gov/37950211/. DOI: 10.1186/s12959-023-00558-4.

PREV Correlation between asymmetric deep medullary veins sign and DTI-ALPS index in patients with acute ischemic stroke
NEXT Application value of APT imaging combined with T2-FLAIR in non-contrast-enhanced diffuse gliomas
  



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