To assess how radiomic features may be combined with plaque morphological and compositional features identified by multi-contrast MRI to improve upon conventional risk assessment models in determining culprit carotid artery lesions. Fifty-five patients (mean age: 62.6; 35 males) with bilateral carotid stenosis who experienced transient ischaemic attack (TIA) or stroke were included from the CARE-II multi-centre carotid imaging trial (ClinicalTrials.gov Identifier: NCT02017756). They underwent MRI within 2 weeks of the event. Classification capability in distinguishing culprit lesions was assessed by machine learning. Repeatability and reproducibility of the results were investigated by assessing the robustness of the radiomic features. Radiomics combined with a relatively conventional plaque morphological and compositional metric-based model provided incremental value over a conventional model alone (area under curve [AUC], 0.819 ± 0.002 vs 0.689 ± 0.019, respectively, P = .014). The radiomic model alone also provided value over the conventional model (AUC, 0.805 ± 0.003 vs 0.689 ± 0.019, respectively, P = .031). T2-weighted imaging-based radiomic features had consistently higher robustness and classification capabilities compared with T1-weighted images. Higher-dimensional radiomic features outperformed first-order features. Grey Level Co-occurrence Matrix, Grey Level Dependence Matrix, and Grey Level Size Zone Matrix sub-types were particularly useful in identifying textures which could detect vulnerable lesions. The combination of MRI-based radiomic features and lesion morphological and compositional parameters provided added value to the reference-standard risk assessment for carotid atherosclerosis. This may improve future risk stratification for individuals at risk of major adverse ischaemic cerebrovascular events. The clinical relevance of this work is that it addresses the need for a more comprehensive method of risk assessment for patients at risk of ischaemic stroke, beyond conventional stenosis measurement. This paper shows that in the case of carotid stroke, high-dimensional radiomics features can improve classification capabilities compared with stenosis measurement alone.

The objectives were to develop and evaluate a machine learning model based on a combination of biomechanics and image texture analysis, to improve the detection of high-risk carotid plaques. Sixty-five patients, who underwent high-resolution, multi-contrast, magnetic resonance imaging (MRI) of the carotid artery wall within two weeks of a TIA or stroke, were assessed. The MR images were provided by the CARE-II multi-centre carotid imaging trial (ClinicalTrials.gov Identifier: NCT02017756). Following 3D artery construction, parametric maps of structural stress, wall shear stress, and inward remodeling were computed using a one-way fluid structure interaction (FSI) approach. A radiomics pipeline was developed to derive image texture features from mechanics maps and the MR images. Machine learning models were then developed to distinguish non-culprit and culprit carotid plaques, where culprit plaques were deemed responsible for the symptoms associated with the TIA/stroke. The performance of a combined model, developed from the most predictive features of the mechanics map and MR image models, was compared with mechanics and MRI-based models individually. Mechanics [Accuracy = 0.68, AUC = 0.75 ± 0.03] and MRI-based models [Accuracy = 0.67, AUC = 0.71 ± 0.04] showed greater predictive capabilities for culprit lesions than the measurement of vessel stenosis alone [Accuracy = 0.62, AUC = 0.57 ± 0.04] (p < 0.001). The combined mechano-radiomics model [Accuracy = 0.76, AUC= 0.82 ± 0.04] (p=0.037) showed significant improvement in the prediction of culprit plaques compared with MRI and mechanics map features alone, as well as the clinically conventional measurement of vascular stenosis.