2.1 Study population
This study conducted a comprehensive analysis of patients diagnosed with carotid plaques causing more than 50% luminal stenosis. Participants were selected based on their receipt of Doppler ultrasonography, magnetic resonance imaging (MRI) of the carotid arteries, and brain MRI within a one-week window following the onset of symptoms. The study period spanned from January 2019 to December 2023, and a thorough collection of baseline demographic and clinical data was systematically implemented. This study was performed in line with the principles of the World Medical Association Declaration of Helsinki. The study protocol was reviewed and approved by the Institutional Review Board of Shenzhen University General Hospital (Approval No. SUGHKYLL2022070601).
Inclusion Criteria:
1)High-resolution MRI images devoid of motion and flow artifacts, deemed suitable for computational fluid dynamics (CFD) modeling.
2)Comprehensive follow-up data spanning 1 to 3 years post-enrollment.
3)Diagnosis of anterior circulation acute cerebral infarction confirmed by cranial diffusion-weighted imaging (DWI).
Exclusion Criteria:
1)Cardioembolic cerebral infarction attributable to recent myocardial infarction, atrial fibrillation, valvular heart disease, dilated cardiomyopathy, sick sinus syndrome, acute endocarditis, patent foramen ovale, or severe cardiac dysfunction (ejection fraction ≤ 30%).
2)Conditions such as aneurysms, vasculitis, or other pathologies impacting the carotid arteries.
3)Presence of cerebral hemorrhage, transient ischemic attack, cerebral vascular malformations, other intracranial pathologies, peripheral arterial disease, immunological disorders, hematological conditions, or metabolic disorders.
Informed consent
was obtained from all participants or their legally authorized representatives prior to their inclusion in the study. The research was conducted in compliance with the ethical standards of the World Medical Association Declaration of Helsinki.
2.2 Clinical variables
Demographic and clinical data, including essential risk factors for cerebral infarction (hypertension, hyperlipidemia, diabetes, and smoking status), were recorded upon admission. Laboratory parameters such as triglycerides (TG), total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), uric acid (UA), fasting blood glucose (FBG), and homocysteine (HCY) were also collected.
2.3 Collection of follow-up information
Follow-up data spanning 1 to 3 years were collected for each patient. The primary endpoint was the occurrence of new clinical symptoms with continuous neurological impairment, confirmed as anterior circulation acute cerebral infarction by DWI. Follow-up assessments were conducted every six months ± 2 weeks for up to 36 months post-baseline. Data on survival, endpoint events, time of occurrence, paroxysmal atrial fibrillation during follow-up, and other vascular events were collected. Two experienced radiologists independently evaluated endpoint events based on DWI.
2.4 Brain MRI protocol
MRI was performed using a 3.0 Tesla General Electric Company (GE Discovery MR750) with a standard 8-channel head coil (Wk401 Jiangyin Wankang Medical Technology).Before the scanning procedure, patients were instructed to follow specific precautions, including wearing specialized earplugs to protect against the noise associated with the MRI machine. They were positioned supine and advised to keep their heads still to minimize motion artifacts that could affect the image quality.
The MRI protocol included the following sequences:
Diffusion-Weighted Imaging (DWI): Parameters were set with a repetition time (TR) of 7000 ms, echo time (TE) of 75 ms, field of view (FOV) of 22 cm, matrix size of 98×140, slice thickness of 5.0 mm, and interslice spacing of 1.5 mm.
Fluid-Attenuated Inversion Recovery (FLAIR): With TR of 12,000 ms, TE of 87 ms, inversion time of 2800 ms, FOV of 22 cm, matrix size of 226×384, slice thickness of 5.0 mm, and interslice spacing of 1.5 mm.
T2-Weighted Imaging(T2WI): TR of 617 ms, TE of 14 ms, FOV of 22 cm, matrix size of 224×320, slice thickness of 5.0 mm, and interslice spacing of 1.5 mm.
T1-Weighted Imaging(T1WI): TR of 500 ms, TE of 5 ms, FOV of 22 cm, matrix size of 256×256, slice thickness of 5.0 mm, and interslice spacing of 1.5 mm.
Three-Dimensional Time-of-Flight Magnetic Resonance Angiography (3D-TOF-MRA): TR of 22 ms, TE of 3.69 ms, FOV of 200 cm, matrix size of 215×320, slice thickness of 0.50 mm, and a negative interslice spacing of 6.0 mm to ensure contiguous slices.
2.5 Definition of acute cerebral infarction and evaluation of cerebral small vascular disease (CSVD) on Brain MRI.
Acute cerebral infarction was diagnosed based on clinical symptoms within 48 hours of onset and confirmed by characteristic MRI findings: hypointense on T1-weighted imaging (T1WI), hyperintense on T2-weighted imaging (T2WI), high signal on DWI, and low apparent diffusion coefficient (ADC) indicative of restricted diffusion.
Patients were categorized into acute cerebral infarction(ACI) and non- acute cerebral infarction(NACI) groups based on clinical presentations and MRI results.
Imaging features of CSVD were evaluated following the STandards for ReportIng Vascular changes on Euroimaging (STRIVE) criteria[34]. White matter lesions, including periventricular hyperintensity (PVH) and deep subcortical white matter hyperintensity (DSWMH), were graded according to the Fazekas scale (1–3)[35]. Lacunar infarcts were classified by their anatomical location: thalamus, medial capsule, corona radiata, basal ganglia, and brainstem[34]. Cerebral microbleeds (CMBs) were categorized as deep, lobar, or infratentorial[34]. Enlarged perivascular space (EPVS) was quantified in the basal ganglia and centrum semiovale, with categories ranging from 1 (0–10 EPVS) to 3 (26 or more) [36].
The total Cerebral Small Vessel Disease (CSVD) score, ranging from 0 to 4, was calculated to assess CSVD severity on MRI. It encompasses lacunar infarcts, cerebral microbleeds (CMBs) distribution, Fazekas-graded white matter changes (PVH and DSWMH), and the extent of enlarged perivascular space (EPVS).
2.6 carotid plaque MRI protocol
Carotid plaque imaging was conducted on all patients using a 3T General Electric Company (GE Discovery MR750) scanner equipped with an 8-channel carotid surface coil (Wk401 Jiangyin Wankang Medical Technology). The protocol included the following sequences:
3D Phase-Contrast Magnetic Resonance Angiography (3D PC MRA): Parameters were set with a repetition time (TR) of 15.0 ms, echo time (TE) of 3.7 ms, field of view (FOV) of 32 cm × 32 cm, and a slice thickness of 1.8 mm.
3D CUBE T1-Weighted Imaging (3D CUBE T1WI): TR/TE of 575/15 ms, an echo train length (ETL) of 24, a slice thickness of 2 mm, and an FOV of 32 cm × 32 cm.
3D CUBE T2-Weighted Imaging (3D CUBE T2WI) : TR/TE of 3500/120 ms, flip angle of 150°, a slice thickness of 2 mm, and an FOV of 32 cm × 32 cm.
Post-acquisition, the raw data were transferred to a workstation and reconstructed into cross-sectional images with a slice thickness of 0.625 mm using Multiple Planar Volume Reconstruction (MPVR) techniques.
2.7 Duplex ultrasonography protocol
Extracranial carotid artery ultrasound was conducted on all patients using a GE LOGIQ E7 ultrasound system. During the examination, patients were positioned supine. The peak systolic velocity (PSV) and peak diastolic velocity (PDV) within the common carotid artery (CCA) were measured in centimeters per second (cm/s) and recorded for subsequent post-processing analysis. These measurements are critical for the assessment of blood flow dynamics in the carotid arteries.
Carotid plaques were identified as focal areas of increased echogenicity with a minimal intima-media thickness (IMT) of 1.2 mm or greater, consistent with the established criteria for carotid atherosclerotic disease[37, 38]. The identification of these plaques is essential for the evaluation of carotid stenosis and the assessment of stroke risk.
2.8 Computational blood flow modeling
We employed computational fluid dynamics (CFD) for the analysis of blood flow in carotid arteries using the commercial software ANSYS ICEM (ANSYS, Inc., USA). The morphological technique of open and close reconstruction was applied to reduce noise inherent in magnetic resonance angiography (MRA) images, facilitating precise extraction of the carotid arterial tree, plaque identification, and segmentation of the lumen boundary [39]. The blood was modeled as an incompressible Newtonian fluid, and the arterial wall was assumed to be rigid with a no-slip condition. Inflow boundary conditions were defined using the peak systolic (PSV) and diastolic velocities (PDV) of the common carotid artery (CCA). The outflow was modeled using an open boundary condition. Hemodynamic parameters within the stenotic lesions were calculated for three distinct regions: upstream, downstream, and core, based on finite element analysis specific to the patient's carotid artery geometry. These parameters included wall shear stress (WSS) in the upstream (WSSup), downstream (WSSdown), and core (WSScore) regions, as well as pressure (Pup, Pdown, Pcore) and blood flow velocity (Vup, Vdown, Vcore).
To elucidate local hemodynamic characteristics within the carotid plaques, each lesion was further divided into segments: upstream, minimal lumen area (MLA) extending 3 mm in length, and downstream[19]. Hemodynamic quantities were computed for these segments and then averaged to provide a comprehensive assessment of the local flow dynamics.
2.9 Analysis of carotid plaque characteristics
Carotid plaques were characterized as intraluminal lesions with an area greater than or equal to 1 mm², clearly demarcated from the vessel lumen and adjacent tissues[40]. Luminal stenosis was quantified on Maximum Intensity Projection (MIP) reconstructed Phase Contrast Magnetic Resonance Angiography (PC MRA) images, while plaque and vascular areas were measured on 3D CUBE T1-Weighted Imaging (3D CUBE T1WI) sequences. Two experienced radiologists, blinded to clinical data and computational fluid dynamics (CFD) results, independently assessed and documented the following characteristics: the presence of high-risk plaques was determined based on signal intensity patterns, overall morphology, plaque burden, and remodeling index, as well as the angle of the vessel at the internal and external carotid artery junction. Vulnerable plaque components, such as lipid-rich necrosis and hemorrhage, were identified by their mild to marked hyperintense signals on MRI, potentially accompanied by ulceration on the plaque surface. In contrast, stable components, including calcification and fibrous tissue, exhibited hypointense to isointense signals[33]. Plaque burden was calculated as the ratio of the plaque area to the total vascular area. The remodeling index was derived from the ratio of the maximal vascular diameter, encompassing both plaque and lumen at the lesion site, to the average normal lumen diameter at the proximal and distal reference segments of the plaque[19]. A remodeling index of 1.10 or higher was indicative of positive remodeling[19]. In cases of discordance between the radiologists' assessments, a consensus was reached through joint discussion.
2.10 Statistical analysis
The statistical analysis was conducted utilizing IBM SPSS Statistics (version 24.0) and Python (version 3.11.4) software packages. The normality of the data distribution was assessed employing the Kolmogorov-Smirnov test. Continuous variables exhibiting normal distribution were reported as mean ± standard deviation (SD), whereas non-normally distributed continuous variables were depicted using median and interquartile range (IQR). Categorical data were presented in terms of frequency counts and percentages. Group comparisons for continuous variables were performed using the independent samples t-test for normally distributed data and the Mann-Whitney U test for non-normally distributed data. Categorical variables were compared between groups using the chi-square test or Fisher's exact test, as appropriate. To assess the diagnostic efficacy of hemodynamic parameters—wall shear stress (WSS), pressure (P), and velocity (V)—in differentiating between patients with acute cerebral infarction and those without, Receiver Operating Characteristic (ROC) curve analysis was conducted. The Area Under the Curve (AUC) was determined to quantify the overall discriminatory capability of these parameters, with AUC values approaching 1 indicating superior diagnostic accuracy. Furthermore, Python was employed to compute additional diagnostic metrics, including positive and negative likelihood ratios, precision, F1 score, diagnostic odds ratio, and Matthews correlation coefficient for the hemodynamic parameters.
All statistical tests were two-tailed, and a p-value of less than 0.05 was considered to denote statistical significance.