Robustness and classification capabilities of MRI radiomic features in identifying carotid plaque vulnerability.

OBJECTIVES: 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. METHODS: 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. RESULTS: 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. CONCLUSIONS: 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. ADVANCES IN KNOWLEDGE: 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.

Vessel structural stress mediates aortic media degeneration in bicuspid aortopathy: New insights based on patient-specific fluid-structure interaction analysis.

This study aimed to assess the relationship between local mechanical stimuli and regional aortic tissue degeneration using fluid-structure interaction (FSI) analysis in patients with bicuspid aortic valve (BAV) disease. Nine patients underwent ascending aortic replacement were recruited. Tissues were collected to evaluate the pathology features in four regions, greater curvature (GC-region), posterior (P-region), anterior (A-region), and lesser curvature (LC-region). FSI analysis was performed to quantify vessel structural stress (VSS) and flow-induced parameters, including wall shear stress (WSS), oscillatory shear index (OSI), and particle relative residence time (RRT). The correlation between these biomechanical metrics and tissue degeneration was analyzed. Elastin in the medial layer and media thickness were thinnest and the gap between fibers was biggest in the GC-region, followed by the P-region and A-region, while the elastin and media thickness were thickest and the gap smallest in the LC-region. The collagen deposition followed a pattern with the biggest in the GC-region and least in the LC-region. There is a strong negative correlation between mean or peak VSS and elastin thickness in the arterial wall in the GC-region (r = -0.917; p = 0.001 and r = -0.899; p = 0.001), A-region (r = -0.748; p = 0.020 and r = -0.700; p = 0.036) and P-region (r = -0.773; p = 0.014 and r = -0.769; p = 0.015), and between mean VSS and fiber distance in the A-region (r = -0.702, p = 0.035). Moreover, strong negative correlation between mean or peak VSS and media thickness was also observed. No correlation was found between WSS, OSI, and RRT and aortic tissue degeneration in these four regions. These findings indicate that increased VSS correlated with local elastin degradation and aortic media degeneration, implying that it could be a potential biomechanical parameter for a refined risk stratification for patients with BAV.

Identification of high risk clinical and imaging features for intracranial artery dissection using high-resolution cardiovascular magnetic resonance.

BACKGROUND: Intracranial artery dissection (IAD) often causes headache and cerebral vascular ischemic events. The imaging characteristics of IAD remain unclear. This study aims to characterize the appearance of culprit and non-culprit IAD using high-resolution cardiovascular magnetic resonance imaging (hrCMR) and quantify the incremental value of hrCMR in identifying higher risk lesions. METHODS: Imaging data from patients who underwent intervention examination or treatment using digital subtraction angiography (DSA) and hrCMR using a 3 T CMR system within 30 days after the onset of neurological symptoms were collected. The CMR protocol included diffusion-weighted imaging (DWI), black blood T1-, T2- and contrast-enhanced T1-weighted sequences. Lesions were classified as culprit and non-culprit according to imaging findings and patient clinical presentations. Univariate and multivariate analyses were performed to assess the difference between culprit and non-culprit lesions and complementary value of hrCMR in identifying higher risk lesions. RESULTS: In total, 75 patients were included in this study. According to the morphology, lesions could be classified into five types: Type I, classical dissection (n = 50); Type II, fusiform aneurysm (n = 1); Type III, long dissected aneurysm (n = 3); Type IV, dolichoectatic dissecting aneurysm (n = 9) and Type V, saccular aneurysm (n = 12). Regression analyses showed that age and hypertension were both associated with culprit lesions (age: OR, 0.83; 95% CI 0.75-0.92; p < 0.001 and hypertension: OR, 66.62; 95% CI 5.91-751.11; p = 0.001). Hematoma identified by hrCMR was significantly associated with culprit lesions (OR, 16.80; 95% CI 1.01-280.81; p = 0.037). Moreover, 17 cases (16 lesions were judged to be culprit) were diagnosed as IAD but not visible in DSA and 15 were Type I lesion. CONCLUSION: hrCMR is helpful in visualizing and characterizing IAD. It provides a significant complementary value over DSA for the diagnosis of IAD.

Association of Hypertension With Both Occurrence and Outcome of Symptomatic Patients With Mild Intracranial Atherosclerotic Stenosis: A Prospective Higher Resolution Magnetic Resonance Imaging Study.

BACKGROUND: Intracranial atherosclerotic plaque causing mild luminal stenosis might lead to acute ischemic events. However, the difference between culprit and nonculprit lesions is unclear, as are the factors associated with favorable treatment outcomes. PURPOSE: To quantify characteristics of intracranial atherosclerosis with mild luminal stenosis and to identify factors associated with lesion type (culprit or nonculprit) and with clinical outcomes. STUDY TYPE: Prospective POPULATION: 293 patients who had acute stroke with mild luminal stenosis (<50%) in the middle cerebral or basilar artery. FIELD STRENGTH/SEQUENCE: 3.0 T higher resolution magnetic resonance imaging (hrMRI) of intracranial arteries and whole brain MR images. ASSESSMENT: Morphological and compositional analysis of plaques was performed. This included assessment of plaque volume, plaque burden, remodeling ratio, eccentricity, intraplaque hemorrhage, and enhancement ratio. Clinical outcomes were assessed according to the modified Rankin Scale (mRS) at day 90, with a favorable outcome being defined as a 90-day mRS ≤2. STATISTICAL TESTS: The odds ratios (ORs) with 95% confidence intervals (CIs) were calculated by a logistic regression model. RESULTS: Hypertension (OR 5.2; 95% CI 2.6-10.3; P < 0.05) and hrMRI enhancement ratio (OR 2.7; 95% CI 1.4-5.1; P < 0.05) were independently associated with lesion type. Patients without hypertension had significantly more (P < 0.05) favorable outcomes (124/144) than patients with hypertension (97/149). Most hypertensive patients without any previous blood pressure control (54/63) had a favorable outcome. However, these patients were significantly younger (P < 0.05) than those with adequate blood pressure control. After adjusting for all significant characteristics, hypertension duration (OR 1.19; 95% CI 1.09-1.29; P < 0.05), hypertension management (OR 2.49; 95% CI 1.18-5.26; P < 0.05), and enhancement ratio (OR 0.01; 95% CI 0.001-0.157; P < 0.05) were found to be independent high-risk factors for outcome prediction. DATA CONCLUSION: hrMRI provided incremental value over traditional risk factors in identifying higher risk intracranial atherosclerosis with mild luminal stenosis. LEVEL OF EVIDENCE: 2 TECHNICAL EFFICACY: Stage 2.

Quantitative Histogram Analysis on Intracranial Atherosclerotic Plaques: A High-Resolution Magnetic Resonance Imaging Study.

BACKGROUND AND PURPOSE: Intracranial atherosclerosis is one of the main causes of stroke, and high-resolution magnetic resonance imaging provides useful imaging biomarkers related to the risk of ischemic events. This study aims to evaluate differences in histogram features between culprit and nonculprit intracranial atherosclerosis using high-resolution magnetic resonance imaging. METHODS: Two hundred forty-seven patients with intracranial atherosclerosis who underwent high-resolution magnetic resonance imaging sequentially between January 2015 and December 2016 were recruited. Quantitative features, including stenosis, plaque burden, minimum luminal area, intraplaque hemorrhage, enhancement ratio, and dispersion of signal intensity (coefficient of variation), were analyzed based on T2-, T1-, and contrast-enhanced T1-weighted images. Step-wise regression analysis was used to identify key determinates differentiating culprit and nonculprit plaques and to calculate the odds ratios (ORs) with 95% CIs. RESULTS: In total, 190 plaques were identified, of which 88 plaques (37 culprit and 51 nonculprit) were located in the middle cerebral artery and 102 (57 culprit and 45 nonculprit) in the basilar artery. Nearly 90% of culprit lesions had a degree of luminal stenosis of <70%. Multiple logistic regression analyses showed that intraplaque hemorrhage (OR, 16.294 [95% CI, 1.043-254.632]; P=0.047), minimum luminal area (OR, 1.468 [95% CI, 1.032-2.087]; P=0.033), and coefficient of variation (OR, 13.425 [95% CI, 3.987-45.204]; P<0.001) were 3 significant features in defining culprit plaques in middle cerebral artery. The enhancement ratio (OR, 9.476 [95% CI, 1.256-71.464]; P=0.029), intraplaque hemorrhage (OR, 2.847 [95% CI, 0.971-10.203]; P=0.046), and coefficient of variation (OR, 10.068 [95% CI, 2.820-21.343]; P<0.001) were significantly associated with plaque type in basilar artery. Coefficient of variation was a strong independent predictor in defining plaque type for both middle cerebral artery and basilar artery with sensitivity, specificity, and accuracy being 0.79, 0.80, and 0.80, respectively. CONCLUSIONS: Features characterized by high-resolution magnetic resonance imaging provided complementary values over luminal stenosis in defined lesion type for intracranial atherosclerosis; the dispersion of signal intensity in histogram analysis was a particularly effective predictive parameter.