Edge-detected common carotid artery intima-media thickness and incident coronary heart disease in the multi-ethnic study of atherosclerosis

Two-stage artificial intelligence model for jointly measurement of atherosclerotic wall thickness and plaque burden in carotid ultrasound: A screening tool for cardiovascular/stroke risk assessment

MOTIVATION

The early screening of cardiovascular diseases (CVD) can lead to effective treatment. Thus, accurate and reliable atherosclerotic carotid wall detection and plaque measurements are crucial. Current measurement methods are time-consuming and do not utilize the power of knowledge-based paradigms such as artificial intelligence (AI). We present an AI-based methodology for the joint automated detection and measurement of wall thickness and carotid plaque (CP) in the form of carotid intima-media thickness (cIMT) and total plaque area (TPA), a class of AtheroEdge™ system (AtheroPoint™, CA, USA).

METHOD

The novel system consists of two stages, and each stage comprises an independent deep learning (DL) model. In Stage I, the first DL model segregates the common carotid artery (CCA) patches from ultrasound (US) images into the rectangular wall and non-wall patches. The characterized wall patches are integrated to form the region of interest (ROI), which is then fed into Stage II. In Stage II, the second DL model segments the far wall region. Lumen-intima (LI) and media-adventitial (MA) boundaries are then extracted from the wall region, which is then used for cIMT and PA measurement.

RESULTS

Using the database of 250 carotid scans, the cIMT error using the AI model is 0.0935±0.0637 mm, which is lower than those of all previous methods. The PA error is found to be 2.7939±2.3702 mm². The system's correlation coefficient (CC) between AI and ground truth (GT) values for cIMT is 0.99 (p < 0.0001), which is higher compared with the CC of 0.96 (p < 0.0001) shown by the earlier DL method. The CC for PA between AI and GT values is 0.89 (p < 0.0001).

CONCLUSION

A novel AI-based strategy was applied to carotid US images for the joint detection of carotid wall thickness (cWT) and plaque area (PA), followed by cIMT and PA measurement. This AI-based strategy shows improved performance using the patch technique compared with previous methods using full carotid scans.

Natural History of Atherosclerosis and Abdominal Aortic Intima-Media Thickness: Rationale, Evidence, and Best Practice for Detection of Atherosclerosis in the Young. J Clin Med. 2019;8. [PMID: 31408952 PMCID: PMC6723244 DOI: 10.3390/jcm8081201]

Atherosclerosis underlies most myocardial infarctions and ischemic strokes. The timing of onset and the rate of progression of atherosclerosis differ between individuals and among arterial sites. Physical manifestations of atherosclerosis may begin in early life, particularly in the abdominal aorta. Measurement of the abdominal aortic intima-media thickness by external ultrasound is a non-invasive methodology for quantifying the extent and severity of early atherosclerosis in children, adolescents, and young adults. This review provides an evidence-based rationale for the assessment of abdominal aortic intima-media thickness-particularly as an age-appropriate methodology for studying the natural history of atherosclerosis in the young in comparison to other methodologies-establishes best practice methods for assessing abdominal aortic intima-media thickness, and identifies key gaps in the literature, including those that will identify the clinical relevance of this measure.

Associations of edge-detected and manual-traced common carotid artery intima-media thickness with incident peripheral artery disease: The Multi-Ethnic Study of Atherosclerosis

Common carotid artery (CCA) intima-media thickness (IMT) is associated with coronary heart disease and can be measured on ultrasound images either by hand or with an automated edge detector. The association of CCA IMT with incident peripheral artery disease (PAD) is poorly studied. We studied 5467 participants of the Multi-Ethnic Study of Atherosclerosis composed of non-Hispanic white, Chinese, Hispanic, and African American participants with a mean age of 61.9 years (47.8% men). Framingham Risk Factors, manual-traced IMT (mt-IMT), and edge-detected IMT (ed-IMT) were entered into multivariable Cox proportional hazards models with incident PAD as the outcome. There were 87 events during a median follow-up of 12.2 years. In fully adjusted models and expressing the hazard ratios (HR) as an increment in SD values, both mt-IMT and ed-IMT were significantly associated with incident PAD: HR 1.36 (95% CI: 1.15, 1.61) and 1.29 (95% CI: 1.04, 1.60), respectively. We conclude that ed- and mt-CCA IMT measurements are associated with incident PAD. ClinicalTrials.gov Identifier: NCT00063440.

Risk factors and utility of maximum carotid intima-media thickness as a surrogate marker for coronary artery stenosis

Background

This study investigated the factors associated with coronary artery stenosis in outpatients. Furthermore, the usefulness of maximum carotid intima–media thickness (maximum-IMT) as a surrogate marker of coronary artery stenosis was evaluated.

Methods

We conducted a single-center retrospective study. A total of 601 outpatients (338 males; 263 females; mean age, 69.8±10.0 years) who underwent coronary computed tomography angiography between April 2006 and March 2012 were analyzed. The associations between coronary artery stenosis (≥75%) as determined by coronary computed tomography angiography and clinical and laboratory parameters were evaluated by multivariate logistic regression. Validation of maximum-IMT as measured by ultrasonography as a surrogate marker of coronary artery stenosis was analyzed by receiver operating characteristic (ROC) curve analysis.

Results

The estimated glomerular filtration rate (eGFR: mL/min/1.73 m²) (odds ratio [OR] 0.985, p<0.01), diabetes mellitus (OR 1.98, p<0.05), and maximum-IMT (mm) (OR 1.76, p<0.01) were significantly associated with coronary artery stenosis (≥75%). In analysis of each group categorized by identified factors, such as renal impairment (eGFR <60 mL/min/1.73 m²) and diabetes mellitus, the ROC curve of maximum-IMT was significant in the group of patients with diabetes mellitus without renal impairment (p<0.01) (cutoff value of maximum-IMT, 2.0 mm; sensitivity, 0.74; and specificity, 0.54) but not in other groups.

Conclusion

Renal impairment, diabetes mellitus, and increased maximum-IMT may be significant risk factors of coronary artery stenosis. Maximum-IMT as measured by ultrasonography may be a useful surrogate marker for coronary artery stenosis in patients with diabetes mellitus without renal impairment but not in other patients.

HIV related stigma, perceived social support and risk of premature atherosclerosis in South Asians

Objective

This study sought to determine the association between social support and stigma experienced by HIV-positive patients and presence of subclinical cardiovascular disease.

Methods

We implemented a cross sectional study in 67 HIV-positive patients and 52 controls from a community health care center in central India. The participants underwent an in-depth survey and a clinical and laboratory assessment of cardiovascular risk. Carotid-intimal thickness (CIMT) was used as a marker of subclinical cardiovascular disease.

Results

On comparing the HIV and age and sex-matched control population, HIV patients had lower body weight (P=<0.001), and lower systolic blood pressures (P = 0.002). Despite the lack of higher cardiac risk factor prevalence and lower lipid abnormalities, HIV patients had higher right, left and average CIMT values than controls (P < 0.001 for all). HIV patients also showed higher prevalence of abnormal CIMT (≥ 0.9 mm) than controls (32% vs. 0%, P < 0.001). HIV patients with increased CIMT (n = 37) in comparison with those with normal CIMT (n = 30) were more frequently males (P = 0.023), had higher systolic blood pressures (P = 0.002), lower CD4 counts (P = 0.033) and experienced higher enacted stigma (P = 0.044). On multivariable stepwise logistic regression, systolic blood pressure (odds ratio:1.06, P = 0.002) and stigma score > 25^th percentile value (odds ratio:3.84, P = 0.037) were independent predictors of the abnormal CIMT.

Conclusions

HIV-positive patients from central India have a higher prevalence of abnormal CIMT as a marker of subclinical cardiovascular disease than the general population. This predisposition to increased cardiovascular risk may be related to complex interactions between HIV disease and stigma-related healthcare inequalities.

Cardiovascular markers of inflammation and serum lipid levels in HIV-infected patients with undetectable viremia

Patients successfully treated for HIV infection still have an increased risk for cardiovascular morbidity and mortality, which might be related not only to traditional risks, but also to inflammation and dyslipidemia. We examined the relationship of serum lipid levels with plasma biomarkers of inflammation using a composite inflammatory burden score (IBS) based on individual (>75^th percentile) measurements from the following seven markers: CD40L, tPA, MCP-1, IL-8, IL-6, hCRP and P-selectin. IBS was categorized as 0 (none of the biomarkers >75^th percentile), 1, 2 and 3 or more scores. Correlations between the IBS and lipid parameters were examined by ordered logistic regression proportional odds models to estimate the odds of more elevated biomarkers. 181 male patients with undetectable HIV-viremia were included into the study. In the multivariate model, a one-unit increase (mmol/L) of total cholesterol and triglycerides was associated with a 1.41-fold (95% CI, 1.13–1.76) and 1.37-fold (95% CI, 1.18–1.60) increased odds of having a greater IBS, respectively. Those with an IBS score ≥1 compared to none had 2.14 (95% CI, 1.43–3.20) higher odds of having a one-unit increased total cholesterol/HDL-cholesterol ratio. In successfully treated HIV-infected persons dyslipidemia was associated with inflammation.

Carotid Intima-Media Thickness as Surrogate for and Predictor of CVD

Carotid artery intima-media thickness (IMT) is a noninvasive measurement of the artery wall thickness, inclusive of atherosclerotic plaque, obtained using ultrasound imaging. In the MESA (Multi-Ethnic Study of Atherosclerosis) study, IMT measurements are used as a surrogate for subclinical cardiovascular disease and as a variable predictive of cardiovascular events. IMT measurements of the common carotid artery are available in more than 99% of the MESA population and are predictive of cardiovascular events. More importantly, IMT and plaque thickness measurements made in the internal carotid artery and carotid bulb are also available in more than 98% of the population and are also strongly predictive of cardiovascular events. This article reviews the techniques used to obtain the MESA IMT values, compares them to those made in other epidemiological studies, and summarizes how they have been used in the MESA study as both surrogates for and predictors of cardiovascular disease.

Post-processing reproducibility of the structural characteristics of the common carotid artery in a Flemish population

Introduction

Common carotid artery (CCA) intima-media thickness (IMT), lumen diameter, and maximum plaque thickness were assessed on ultrasound images. The objective of the study was to evaluate the intra- and inter-reader reproducibility of the measurements following a standardised protocol.

Methods

Two readers performed the off-line measurements on B-mode ultrasound images of the distal CCA, in a randomly selected subset (n = 60) from a Flemish population cohort (FLEMENGHO). We calculated the coefficient of variation, the interclass correlation coefficient (ICC) and reproducibility according to the Bland–Altman method.

Results

The intra-reader bias for the measurements of left and right side CCA IMT were −0.003 ± 0.04 mm (p = 0.55) and 0.01 ± 0.04 mm (p = 0.03), respectively. The intra-reader bias of the lumen diameter was −0.04 ± 0.25 mm (p = 0.27) for the left and 0.02 ± 0.22 mm (p = 0.45) for the right side. The measurements for the maximum plaque thickness showed no intra-reader differences with bias 0.07 ± 0.2 mm (p = 0.26) for the left and −0.03 ± 0.2 mm (p = 0.55) for the right side. The inter-reader analysis showed good reproducibility for the left and right side CCA IMT with bias 0.004 ± 0.06 mm (p = 0.57) and −0.008 ± 0.05 mm (p = 0.19), respectively, but the lumen diameter measurements showed inter-reader differences, with bias 0.17 ± 0.27 mm (p < 0.0001) for the left and 0.10 ± 0.21 mm (p = 0.0006) for the right side. The inter-reader bias for the maximum plaque thickness were 0.07 ± 0.2 mm (p = 0.21) and −0.1 ± 0.4 mm (p = 0.26) for the left and right side, respectively.

Conclusion

The results demonstrated a reliable reproducibility of carotid wall structural measurements, allowing for an adequate further analysis of the entire population cohort.

•

The reproducibility of carotid arterial characteristics was evaluated.

•

A sample of participants was chosen from a population cohort.

•

The readers followed a standardised protocol, using commercially available software.

•

The reproducibility of all characteristics was satisfactory.

Carotid plaque-thickness and common carotid IMT show additive value in cardiovascular risk prediction and reclassification

Background and aims

Carotid plaque size and the mean common carotid intima-media thickness measured in plaque-free areas (PF CC-IMT_mean) have been identified as predictors of vascular events (VEs), but their complementarity in risk prediction and stratification is still unresolved. The aim of this study was to evaluate the independence of carotid plaque thickness and PF CC-IMT_mean in cardiovascular risk prediction and risk stratification.

Methods

The IMPROVE-study is a European cohort (n = 3703), where the thickness of the largest plaque detected in the whole carotid tree was indexed as cIMT_max. PF CC-IMT_mean was also assessed. Hazard Ratios (HR) comparing the top quartiles of cIMT_max and PF CC-IMT_meanversus their respective 1–3 quartiles were calculated using Cox regression.

Results

After a 36.2-month follow-up, there were 215 VEs (125 coronary, 73 cerebral and 17 peripheral). Both cIMT_max and PF CC-IMT_mean were mutually independent predictors of combined-VEs, after adjustment for center, age, sex, risk factors and pharmacological treatment [HR (95% CI) = 1.98 (1.47, 2.67) and 1.68 (1.23, 2.29), respectively]. Both variables were independent predictors of cerebrovascular events (ischemic stroke, transient ischemic attack), while only cIMT_max was an independent predictor of coronary events (myocardial infarction, sudden cardiac death, angina pectoris, angioplasty, coronary bypass grafting). In reclassification analyses, PF CC-IMT_mean significantly adds to a model including both Framingham Risk Factors and cIMT_max (Integrated Discrimination Improvement; IDI = 0.009; p = 0.0001) and vice-versa (IDI = 0.02; p < 0.0001).

Conclusions

cIMT_max and PF CC-IMT_mean are independent predictors of VEs, and as such, they should be used as additive rather than alternative variables in models for cardiovascular risk prediction and reclassification.

•

Taken by themselves, both cIMT_max and PF CC-IMT_mean are associated with the incidence of cardiovascular disease.

•

Complementarity of cIMT_max and PF CC-IMT_mean in cardiovascular risk prediction/reclassification is still under debate.

•

In our study, both cIMT_max and PF CC-IMT_mean were mutually independent predictors of cerebral and coronary vascular events.

•

In reclassification analyses, PF CC-IMT_mean adds to a model including Framingham Risk Factors and cIMT_max and vice-versa.

•

cIMT_max and PF CC-IMT_mean should be used as additive rather than alternative variables in risk prediction/reclassification.

Association between carotid atherosclerosis and different subtypes of hypertension in adult populations: A multiethnic study in Xinjiang, China

Background

Ethnic differences in non-invasive measurements of carotid atherosclerosis are being increasingly reported, but the association between carotid atherosclerosis and different subtypes of hypertension in adult populations is not fully understood in different ethnicities. We aimed to investigate the association of carotid atherosclerosis with different subtypes of hypertension in different ethnicities in Xinjiang, a northwestern province in China.

Methods

A total of 14,618 participants (5,757 Hans, 4,767 Uygurs, and 4,094 Kazakhs) from 26 villages of seven cities in Xinjiang were randomly selected from the Cardiovascular Risk Survey conducted during 2007 and 2010. A standard questionnaire, a physical examination and biochemical tests were employed.

Results

The mean common carotid intima-media thickness (CIMT) for the 14,618 participants was 0.86±0.003 mm. The CIMT gradually increased with age. Men (0.92±0.005 mm) had a higher CIMT than women (0.81±0.004 mm). The Uygur participants (0.82±0.006 mm) had a lower CIMT than the Han (0.88±0.005 mm) and Kazakh participants (0.88±0.005 mm). The overall prevalences of carotid intimal thickening and carotid plaques were 12.4% and 9.7%, respectively. The prevalence of CIMT varied for the different subtypes of hypertension. Multivariate logistic regression analysis showed different risk factors for abnormal CIMT in different ethnicities. The associations between abnormal CIMT and the different subtypes of hypertension within different ethnic backgrounds were also different. The risk factors for abnormal CIMT included systolic-diastolic hypertension (SDH) in Han participants (OR: 1.323, 95% CI: 1.100–1.590), SDH (OR: 1.426, 95% CI: 1.160–1.753) and isolated-systolic hypertension (ISH) (OR: 1.844, 95% CI: 1.470–2.313) in Uygur participants, and isolated-diastolic hypertension (IDH) (OR: 1.536, 95% CI: 1.170–2.016) in Kazakh participants.

Conclusion

There was an ethnic difference in the prevalence of abnormal CIMT in Xinjiang, a northwestern province in China. The associations between abnormal CIMT and the subtypes of hypertension varied among the different ethnic groups. Among the studied populations, Han participants with SDH, Uygur participants with SDH and ISH, and Kazakh with IDH were more likely to suffer carotid atherosclerosis than those with other subtypes of hypertension. Participants with different ethnic backgrounds had different sets of risk factors for abnormal CIMT.

Current status of carotid ultrasound in atherosclerosis

Cardiovascular disease (CVD) primarily caused by atherosclerosis is a major cause of death and disability in developed countries. Sonographic carotid intima-media thickness (CIMT) is widely studied as a surrogate marker for detecting subclinical atherosclerosis for risk prediction and disease progress to guide medical intervention. However, there is no standardized CIMT measurement methodology in clinical studies resulting in inconsistent findings, thereby undermining the clinical value of CIMT. Increasing evidences show that CIMT alone has weak predictive value for CVD while CIMT including plaque presence consistently improves the predictive power. Quantification of plaque burden further enhances the predictive power beyond plaque presence. Sonographic carotid plaque characteristics have been found to be predictive of cerebral ischaemic events. With advances in ultrasound technology, enhanced assessment of carotid plaques is feasible to detect high-risk/vulnerable plaques, and provide risk assessment for ischemic stroke beyond measurement of luminal stenosis.

Radioiodine Therapy Does Not Change the Atherosclerotic Burden of the Carotid Arteries

BACKGROUND AND AIM

Atherosclerosis evolves or accelerates when arteries are exposed to ionizing radiation, both early and late after exposure. Radioiodine therapy of benign thyroid disease exposes the carotid arteries to 4-50 Gy, and may thereby increase the risk of atherosclerosis. Increased risk of cerebrovascular events has been reported after radioiodine therapy. This study aimed to examine whether atherosclerosis develops early or late after radioiodine therapy of benign thyroid disease.

METHOD

Patients treated for benign thyroid disorders (nontoxic goiter, adenoma, and hyperthyroidism) were examined with ultrasound for the main outcome, carotid intima media thickness (CIMT), and for plaque presence (plaque presence only in late damage). Signs of early damage from radioiodine were studied in 39 radioiodine-treated patients, who were examined before treatment and at 1, 3, 6, and 12 months after treatment. Late changes were studied in a cross-sectional case-control design, with radioiodine-treated patients as cases (n = 193) and patients treated with surgery as controls (n = 95). Data were analyzed with repeated measurement for longitudinal data, and with multivariate regression for cross-sectional data. Results were adjusted for age, sex, cholesterol, smoking status, known atherosclerotic disease, and body mass index.

RESULTS

No changes in CIMT were found in the patients followed prospectively for one year after treatment with radioactive iodine for benign thyroid disease (p = 0.58). In the study on late effects, there was no difference in CIMT (p = 0.25) or presence of plaques (p = 0.70) between those treated with radioactive iodine and those treated with surgery (9.8 and 5.6 years since treatment, respectively). Furthermore, the level of thyrotropin (TSH) did not influence these atherosclerosis markers.

CONCLUSION

No early changes in CIMT were detected in patients treated with radioactive iodine for benign thyroid disease. No signs of late effects of radioactive iodine on CIMT or plaque presence were found after 10 years of follow-up. The radiation to the carotid arteries by radioactive iodine therapy for benign thyroid disease may therefore have no or low effect on atherosclerotic burden of the carotid arteries in general.

Validation of the Carotid Intima-Media Thickness Variability: Can Manual Segmentations Be Trusted as Ground Truth?

The common carotid artery intima-media thickness (IMT) is widely accepted and used as an indicator of atherosclerosis. Recent studies, however, have found that the irregularity of the IMT along the carotid artery wall has a stronger correlation with atherosclerosis than the IMT itself. We set out to validate IMT variability (IMTV), a parameter defined to assess IMT irregularities along the wall. In particular, we analyzed whether or not manual segmentations of the lumen-intima and media-adventitia can be considered reliable in calculation of the IMTV parameter. To do this, we used a total of 60 simulated ultrasound images with a priori IMT and IMTV values. The images, simulated using the Fast And Mechanistic Ultrasound Simulation software, presented five different morphologies, four nominal IMT values and three different levels of variability along the carotid artery wall (no variability, small variability and large variability). Three experts traced the lumen-intima (LI) and media-adventitia (MA) profiles, and two automated algorithms were employed to obtain the LI and MA profiles. One expert also re-traced the LI and MA profiles to test intra-reader variability. The average IMTV measurements of the profiles used to simulate the longitudinal B-mode images were 0.002 ± 0.002, 0.149 ± 0.035 and 0.286 ± 0.068 mm for the cases of no variability, small variability and large variability, respectively. The IMTV measurements of one of the automated algorithms were statistically similar (p > 0.05, Wilcoxon signed rank) when considering small and large variability, but non-significant when considering no variability (p < 0.05, Wilcoxon signed rank). The second automated algorithm resulted in statistically similar values in the small variability case. Two readers' manual tracings, however, produced IMTV measurements with a statistically significant difference considering all three variability levels, whereas the third reader found a statistically significant difference in both the no variability and large variability cases. Moreover, the error range between the reader and automatic IMTV values was approximately 0.15 mm, which is on the same order of small IMTV values, indicating that manual and automatic IMTV readings should be not used interchangeably in clinical practice. On the basis of our findings, we conclude that expert manual tracings should not be considered reliable in IMTV measurement and, therefore, should not be trusted as ground truth. On the other hand, our automated algorithm was found to be more reliable, indicating how automated techniques could therefore foster analysis of the carotid artery intima-media thickness irregularity.

An insight into elasticity analysis of common carotid artery using ultrasonography

Elastance is a distinguished marker in diagnosing various arterial diseases as studies have reported carotid artery–related diseases linked with stiffness index (β) values greater than 5. This study was to estimate elasticity of common carotid artery by measuring the diameter during systolic and diastolic phases using pixel tracing of successive frames and blood pressure. The B-mode ultrasonography video containing arterial wall motion was captured and fragmented into image frames. Each pixel on the greyscale image was converted into RGB intensity values. The diameter of the artery as well as the thickness of the wall was measured by tracing the pixel displacements from successive frames during arterial pulsation. The study was conducted on 19 subjects aged 25–40 years. The systolic and diastolic carotid artery lumen diameters and carotid intima-media thickness were calculated as 7.1 ± 0.7, 6.3 ± 0.6 and 0.5 ± 0.05 mm (mean ± standard deviation), respectively. The mean stiffness index (β), Peterson’s modulus and Young’s modulus of elasticity were 5.2 ± 1.1, 69 ± 15 kPa and 453 ± 99 kPa, respectively. The pixel displacements in tunica intima, tunica media and tunica adventitia were not homogeneous, due to varied macro-constituents such as endothelial tissues, smooth muscle cells, elastin lamina, fibrous tissue and micro-constituents such as collagen, fibroblast and elastin. We found that women have smaller arteries, and the stiffness increased during the systolic phase.

CAROTID WALL MEASUREMENT AND ASSESSMENT BASED ON PIXEL-BASED AND LOCAL TEXTURE DESCRIPTORS

Aim of this paper is to develop an automated system for the classification and characterization of carotid wall status and to develop a robust system based on local texture descriptors. A database of 200 longitudinal ultrasound images of carotid artery is used. One-hundred images with Intima-Media Thickness (IMT) value higher than 0.8[Formula: see text]mm are considered as high risk. Six different rectangular pixel neighborhoods were considered: four areas centered on the selected element, with sizes [Formula: see text], [Formula: see text], [Formula: see text], and [Formula: see text] pixels, and two noncentered areas with sizes [Formula: see text] pixels upwards and downwards. We have extracted various texture descriptors (31 based on the co-occurrence gray level matrix, 13 based on the spatial gray level dependence matrix, and 20 based on the gray level run length matrix (GLRLM) from neighborhood. We have used Quick Reduct Algorithm to select 12 most discriminant features from extracted 211 features. Each pixel is then assigned to the vessel lumen, to the intima-media complex, or to the adventitia by using an integrated system of three feed-forward neural networks. The boundaries between the three regions are used to estimate the IMT value. The texture features associated with GLRLM are found to be clinically most significant. We have obtained an overall classification accuracy of 79.5%, sensitivity of 87%, and specificity of 72%. We observed a unique classification pattern between low risk and high risk images: in the latter ones, a considerable number of pixels of the intima–media complex ([Formula: see text]) was classified as belonging to the adventitia. This percentage is statistically higher than that of low risk images ([Formula: see text]; [Formula: see text]). Locally extracted and pixel-based descriptors are able to capture the inner characteristics of the carotid wall. The presence of misclassified pixels in the intima–media complex is associated to higher cardiovascular risk.