PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Stroke. Author manuscript; available in PMC 2012 July 1.
Published in final edited form as:
PMCID: PMC3169166
NIHMSID: NIHMS295118

Associations of edge detected and manual traced common carotid intima-media thickness (IMT) measurements with Framingham risk factors: the Multi-Ethnic Study of Atherosclerosis

Abstract

Background

Carotid intima-media thickness (IMT) is a marker of cardiovascular disease derived from ultrasound images of the carotid artery. In most outcome studies, human readers identify and trace the key IMT interfaces. We evaluate an alternate approach using automated edge detection.

Methods

We study a subset of 5640 participants with an average age 61.7 years (48% men) of the Multi-Ethnic Study of Atherosclerosis composed of whites, Chinese, Hispanic and African-Americans that are part of the MESA IMT progression study. Manual tracing IMT (mt_IMT) and edge-detected IMT (ed_IMT) measurements of the far wall of the common carotid artery (CCA) served as outcome variables for multivariable linear regression models using Framingham cardiovascular risk factors and ethnicity as independent predictors.

Results

Measurements of mt_IMT was obtainable in 99.9% (5633/5640) and of ed_IMT in 98.9% (5579/5640) of individuals. Average ed_IMT was 0.19 mm larger than mt_IMT. Inter-reader systematic differences (bias) in IMT measurements were apparent for mt_IMT but not ed_IMT. Based on complete data on 5538 individuals, associations of IMT with risk factors were stronger (p < 0.0001) for mt_IMT (model r2: 19.5%) than ed_IMT (model r2: 18.5%).

Conclusion

We conclude that this edge-detection process generates IMT values equivalent to manually traced ones since it preserves key associations with cardiovascular risk factors. It also decreases inter-reader bias, potentially making it applicable for use in cardiovascular risk assessment.

Indexing: Ultrasonography, Risk Factors, Carotid Arteries, Carotid Intima Media Thickness

Introduction

Carotid intima-media thickness (IMT) is a marker of cardiovascular disease13 with measurements mostly done by readers who either place calipers at selected points4 or trace continuous lines along the lumen-intima and media-adventitia wall interfaces of the artery wall5. The distance between these the lumen-intima and media-adventitia interfaces defines IMT. An alternate approach is to use automated edge detectors to identify these interfaces611.

Two consensus groups have proposed the use of edge detection for IMT measurements12, 13. IMT measurements derived from hand tracings and from edge detectors and their associations with risk factors have not been compared in any large population study. We have designed one that uses cost-functions for gradients and echodensity11 and applied it to ultrasound images acquired on a subset of a large multi-ethnic cohort large multi-ethnic cohort, the Multi-Ethnic Study of Atherosclerosis (MESA).

We compare edge detected to manual traced IMT measurements in the MESA IMT progression study, their respective associations with cardiovascular risk factors and the effect on inter-reader differences.

Materials and Methods

Population

MESA recruited and examined a multiethnic population of 6814 men and women aged 45–84 with no history of clinical cardiovascular disease14 between July 2000 and August 2002. The MESA cohort is a multi-ethnic cohort including white, African-American, Hispanic-American, and Chinese participants. Participants were excluded if they had physician diagnosis of heart attack, stroke, transient ischemic attack, heart failure, angina, atrial fibrillation, history of any cardiovascular procedure, weight above 300 lbs, pregnancy, or any medical conditions that would prevent long-term participation. MESA protocols and all studies described herein have been approved by the Institutional Review Boards of all collaborating institutions. The participants studied underwent carotid artery imaging at the baseline visit.

Risk factors and anthropomorphic variables

The risk factors used in this paper are derived from the updated Framingham Risk Score as presented by D’Agostino et al.15: age, gender, smoking and diabetes status, systolic blood pressure, total and HDL cholesterol to which treatment of hypertension has been added.

Age, gender, ethnicity, and medical history were self-reported. Current smoking was defined as self-report of a cigarette in the last 30 days. Resting blood pressure (BP) was measured three times in the seated position using a Dinamap model Pro 100 automated oscillometric sphygmomanometer (Critikon, Tampa, Florida). The average of the last two measurements was used in analyses.

Lipid levels were measured after a twelve-hour fast. Diabetes mellitus was based on self-reported physician diagnosis, use of insulin and/or oral hypoglycemic agent, or fasting glucose ≥126 mg/dL. Total cholesterol was measured using a cholesterol oxidase method (Roche Diagnostics), and HDL after precipitation of non-HDL cholesterol with magnesium/dextran.

Carotid artery measures

Participants were examined supine with the head rotated 45° towards the left side. Imaging was done in the plane parallel to the neck with the jugular vein lying immediately above the common carotid artery (or at 45 degrees from the vertical if the internal jugular vein is not present). Images of the right common carotid artery were centered 10 to 15 mm below (caudad to) the right common carotid artery bulb. A matrix array probe (M12L, General Electric, Milwalkee, WI) was used, with the frequency set at 13 MHz, two focal zones and the frame rate was set at 32 frames-per-second. A super-VHS videotape recording was then made for 20 seconds. Images were digitized at 30 frames-per-second and end-diastolic images (smallest diameter of the artery) were captured. Although the theoretical resolution of the ultrasound at 13 MHz is 0.07 mm, it might be as low as 0.24 mm taking into consideration the number of cycles in the transmitted ultrasound pulses.

IMT was measured over a length of approximately 10 mm starting 5 mm to 10 mm below (caudad to) the right common carotid artery bulb and excluding any carotid artery plaque. Trained readers traced the key two interfaces to obtain manual tracings (mt) on 19” monitors in a low light environment. The readers activated the edge detector after completing their tracings (Please see http://stroke.ahajournals.org). The edge detector did not use the location of the manually traced interfaces but operated on the same arterial segment. The edge detection algorithm weighed gradients between pixels and pixel density values through a dynamic programming process that minimizes cost-functions11. The readers had the option of modifying the cost-function coefficients if the edge detector failed to track an interface. The readers were not given the option of tracing any start points or editing the line tracings. Manual traced (mt) and edge detected (ed) line tracings were processed by the same algorithm16 in order to obtain mt_IMT) and ed_IMT measurements.

Reproducibility was assessed by blinded replicate re-reads by 2 readers of a set of 114 studies, 66 for one reader and 48 for a second reader, both compared to a third reader. The third reader had intra-reader evaluation of reproducibility in a set of 18 individuals. The readers performed the measurements after blindly selecting images from the 20 second video loop.

Statistical analyses

The mean (and standard deviation) values of continuous variables, mt_IMT, ed_IMT are presented. The distribution of dichotomous variables is also shown as % in each group.

A Bland-Altman plot was generated for the paired mt_IMT and ed_IMT measurements. The mean differences and standard deviation between replicate mt_IMT and ed_IMT reading were computed for each reader combination. Analyses were performed using JMP 7.0.2 (SAS Institute, Cary, NC).

Multivariable regression models were fit with mt_IMT and ed_IMT as respective outcomes and the component risk factors of the Framingham Risk Score as predictors. Additionally the models were adjusted for ethnicity. Regression coefficients and partial contributions to model R-square were calculated for each predictor. Overall R-square values were computed and compared using asymptotic testing procedure for correlated correlations. These analyses were run using SAS version 9.1 (Cary, NC) and p-values below 0.05 were considered statistically significant.

Results

There were 5633 individuals with mt_IMT values (mean 0.678 ± 0.190 mm) as compared to 5579 with ed_IMT values (mean 0.867 ± 0.226 mm). We were able to obtain paired IMT values in 5574 individuals. Restricting the analyses to all individuals with complete risk factor data, there were 5538 individuals with a mean age of 61.9 years. Demographics are shown in Table 1. The mean difference between measurements was −0.191 ± 0.15 mm with edge detected IMT values being larger as shown on a Bland-Altman plot (Figure 1).

Figure 1
Bland-Altman plot summarizing the differences between manual traced IMT and edge detected IMT values. Edge detected IMT values are subtracted from manual traced IMT.
Table 1
Basic demographics on all individuals with mt_IMT, ed_IMT values and risk factors

Results of paired replicate studies (Table 2) show that the standard deviations of paired measurements between readers (variance) are lower for manual traced IMT measurements than for edge detected IMT measurements. Measurements made by different readers were similar when the edge detector was used whereas differences between readers were apparent when manual tracings were used. These findings are displayed graphically in figures 2a and 2b. Intra-reader measurements showed a similar effect (Table 2).

Figure 2Figure 2
a Paired differences between readers for manual traced IMT values. Readers 1 and 2 (inter-reader) are compared to reader 3.
Table 2
Reproducibility of IMT measurements based on the difference between replicate readings between readers using either manual traced interfaces or edge detected interfaces.

Table 3 summarizes the strength of the associations between risk factors and the carotid IMT measurements made by both methods. Risk factors account for slightly more (p < 0.0001) of the variability of mt_IMT (19.5%) than for ed_IMT (18.5%).

Table 3
Multivariable linear regression models looking at the associations between risk factors and IMT measurements derived from manual tracings (mt_IMT) and those derived from edge detected values (ed_IMT).

Based on the partial correlations, associations of mt_IMT with age, gender, total cholesterol, and smoking were qualitatively stronger than for ed_IMT. Diabetes and HDL cholesterol had qualitatively stronger associations with ed_IMT than mt_IMT.

Discussion

Edge detected IMT measurements of the common carotid artery far wall can be consistently obtained in a large cross-sectional sample of the population. Edge detected IMT measurements have strong associations with cardiovascular risk factors, similar but slightly weaker than manual traced IMT measurements. Contrary to prior publications, we have not found that edge detected IMT measurements are more reproducible than manual measurements8, 11, 1719 although we show that they decrease inter-reader differences.

Carotid IMT measurements have been proposed as a measure of cardiovascular risk and a means of possibly identifying individuals in need of pharmacotherapy or life style interventions20. Edge detection offers the advantage of obtaining IMT measurements in a standardized fashion so that they can be compared against normative or calibrated values.

We show that IMT measurements made with an edge detector preserve key associations with cardiovascular risk factors (Table 3) while decreasing reader bias (Table 2). As such, they could be substituted for manual traced measurements generating normative data for IMT risk assessment.

We used an algorithm based on dynamic programming that resembles one developed by Wendelhag et al11. This algorithm processes data based on pixel intensity and gradients and is different than edge detector algorithms based on polynomial fitting of intensity curves perpendicular to interfaces or to algorithms using template matching21. The results presented in this paper are specific to our implementation of a specific edge detector and do not apply to other edge detectors.

We were able to obtain automated IMT measurements in 99% (5574) of the 5633 individuals with manual determined IMT values. Wendelhag et al.11 reported that readers identified by hand a new start point for their edge detection process in 17% of cases. Our readers had the option to slightly alter the weighting factors used for edge detection. Pre-selected values were used in 93% of cases for the media-adventitia interface and 83% of cases for the lumen-intima interface (Please see http://stroke.ahajournals.org). Despite these adjustments, Figure 1 shows outliers in individuals with low IMT values and where the algorithm failed (63 cases or 1.1% of individuals). We believe that this algorithm failure depends on the thickness of the media layer since the algorithm requires a minimum number of pixels between the lumen-intima and media-adventitia interfaces. Increasing image size (scaling in pixels/mm) could circumvent this limitation. We have included these 63 cases since they did not substantially alter our findings. By protocol, the same image was measured in order to reduce the variability inherent in image selection. This might bias our results in favor of manual tracings since selecting a different image might have improved edge detection, reduced variability and increased the predictive value of risk factors.

While edge detectors have been used in clinical trials6, associations between risk factors and edge detected IMT measurements have not been studied. In our review of the literature, we have found studies with small groups of subjects where the reproducibility of edge detector data was evaluated or where edge detected IMT values were compared to manual tracings8, 17, 2226. These studies did not evaluate the associations of risk factors with common carotid IMT measurements.

The larger IMT values measured with the edge detector may be due to the mathematical process used to derive edges and the relative thickness of the intima and adventitia (Figure 3). The mathematical location of these edges tends to be different than the line perceived by the human eye. For example, a human reader would tend to trace a line on the lumen-intima interface while the edge detector would place the edge above this line and cause an overestimation of 0.056 to 0.11 mm (one to two pixels at an image scale of 180 pixels/cm). The weighing function in our algorithm would also tend to “pull” the estimated interface towards the denser pixels in the adventitia, thereby further increasing the estimated IMT. The IMT differences between readers (Table 2) are 0.09 to 0.14 mm, lower than the difference between manual traced and edge detected IMT values (0.19 mm). Edge detected mean IMT values show lack of inter-reader differences (Figures 2a and 2b). Contrary to previous studies, the variability of edge detected IMT was slightly higher than for manual tracing8, 11. This may represent an image quality issue since we accepted images from six centers rather than from one laboratory.

Figure 3
Carotid artery far-wall interface differences. The lumen-intima interface (top arrow) is thinner than the media-adventitia interface(bottom arrow). This is in part due to the greater thickness of the adventitia.

Our observations apply to this ultrasound device with its presets for image texture and scale. This may limit the general applicability of our observations.

We conclude that edge detected IMT measurements can be substituted for manual traced IMT measurements in cross-sectional IMT studies. As implemented, this specific edge detector gives IMT measurements comparable to manual traced measurements in a large multi-ethnic patient population. Edge detected IMT values might be better suited for cardiovascular risk assessment since they do not seem to have the significant inter-reader differences seen with manually traced IMT measurements.

Supplementary Material

Acknowledgments

Acknowledgements and Funding:

The authors would like to thank the investigators, the staff, and the participants of the MESA study for their valuable contributions. This research was supported by NIH contracts N01-HC-95159 through N01-HC-95165 and N01-HC-95167 as well as R01 HL069003 and R01 HL081352.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

A full list of participating MESA investigators and institutions can be found at http://www.mesa-nhlbi.org.

Disclosures:

DH O’Leary owns stock in Medpace, Inc. MJ Pencina serves on a DSMB for Abbott.

References

1. Bots ML, Hoes AW, Koudstaal PJ, Hofman A, Grobbee DE. Common carotid intima-media thickness and risk of stroke and myocardial infarction: the Rotterdam Study. Circulation. 1997;96:1432–1437. [PubMed]
2. Chambless LE, Heiss G, Folsom AR, Rosamond W, Szklo M, Sharrett AR, Clegg LX. Association of coronary heart disease incidence with carotid arterial wall thickness and major risk factors: the Atherosclerosis Risk in Communities (ARIC) Study, 1987–1993. American Journal of Epidemiology. 1997;146:483–494. [PubMed]
3. OLeary DH, Polak JF, Kronmal RA, Manolio TA, Burke GL, Wolfson SK., Jr Carotid-artery intima and media thickness as a risk factor for myocardial infarction and stroke in older adults. Cardiovascular Health Study Collaborative Research Group. New England Journal of Medicine. 1999;340:14–22. [PubMed]
4. Anonymous. Atherosclerosis Risk In Communities Manual 6B Ultrasound Assessment Version 1.0. ARIC Coordinating Center Chapel Hill NC. 1989:68. Version 1B.
5. OLeary DH, Polak JF, Wolfson SK, Jr, Bond MG, Bommer W, Sheth S, Psaty BM, Sharrett AR, Manolio TA. Use of sonography to evaluate carotid atherosclerosis in the elderly. The Cardiovascular Health Study. CHS Collaborative Research Group. Stroke. 1991;22:1155–1163. [PubMed]
6. Hedblad B, Wikstrand J, Janzon L, Wedel H, Berglund G. Low-dose metoprolol CR/XL and fluvastatin slow progression of carotid intima-media thickness: Main results from the Beta-Blocker Cholesterol-Lowering Asymptomatic Plaque Study (BCAPS) Circulation. 2001;103:1721–1726. [PubMed]
7. Hodis HN, Mack WJ, LaBree L, Selzer RH, Liu C, Alaupovic P, Kwong-Fu H, Azen SP. Reduction in carotid arterial wall thickness using lovastatin and dietary therapy: a randomized controlled clinical trial. Annals of Internal Medicine. 1996;124:548–556. [PubMed]
8. Schmidt C, Wendelhag I. How can the variability in ultrasound measurement of intima-media thickness be reduced? Studies of interobserver variability in carotid and femoral arteries. Clinical Physiology. 1999;19:45–55. [PubMed]
9. Selzer RH, Hodis HN, Kwong-Fu H, Mack WJ, Lee PL, Liu CR, Liu CH. Evaluation of computerized edge tracking for quantifying intima-media thickness of the common carotid artery from B-mode ultrasound images. Atherosclerosis. 1994;111:1–11. [PubMed]
10. Touboul PJ, Prati P, Scarabin PY, Adrai V, Thibout E, Ducimetiere P. Use of monitoring software to improve the measurement of carotid wall thickness by B-mode imaging. Journal of Hypertension - Supplement. 1992;10:S37–S41. [PubMed]
11. Wendelhag I, Liang Q, Gustavsson T, Wikstrand J. A new automated computerized analyzing system simplifies readings and reduces the variability in ultrasound measurement of intima-media thickness. Stroke. 1997;28:2195–2200. [PubMed]
12. Stein JH, Korcarz CE, Hurst RT, Lonn E, Kendall CB, Mohler ER, Najjar SS, Rembold CM, Post WS. American Society of Echocardiography Carotid Intima-Media Thickness Task F. Use of carotid ultrasound to identify subclinical vascular disease and evaluate cardiovascular disease risk: a consensus statement from the American Society of Echocardiography Carotid Intima-Media Thickness Task Force. Endorsed by the Society for Vascular Medicine. Journal of the American Society of Echocardiography. 2008;21:93–111. [PubMed]
13. Touboul PJ, Hennerici MG, Meairs S, Adams H, Amarenco P, Bornstein N, Csiba L, Desvarieux M, Ebrahim S, Fatar M, Hernandez Hernandez R, Jaff M, Kownator S, Prati P, Rundek T, Sitzer M, Schminke U, Tardif JC, Taylor A, Vicaut E, Woo KS, Zannad F, Zureik M. Mannheim carotid intima-media thickness consensus (2004–2006). An update on behalf of the Advisory Board of the 3rd and 4th Watching the Risk Symposium, 13th and 15th European Stroke Conferences, Mannheim, Germany, 2004, and Brussels, Belgium, 2006; Cerebrovascular Diseases; 2007. pp. 75–80. [PubMed]
14. Bild DE, Bluemke DA, Burke GL, Detrano R, Diez Roux AV, Folsom AR, Greenland P, Jacob DR, Jr, Kronmal R, Liu K, Nelson JC, O'Leary D, Saad MF, Shea S, Szklo M, Tracy RP. Multi-ethnic study of atherosclerosis: objectives and design. American Journal of Epidemiology. 2002;156:871–881. [PubMed]
15. D'Agostino RB, Sr, Vasan RS, Pencina MJ, Wolf PA, Cobain M, Massaro JM, Kannel WB, D'Agostino RB, Sr, Vasan RS, Pencina MJ, Wolf PA, Cobain M, Massaro JM, Kannel WB. General cardiovascular risk profile for use in primary care: the Framingham Heart Study. Circulation. 2008;117:743–753. [PubMed]
16. Polak JF, O'Leary DH. Algorithms for detection of early atherosclerosis on high resolution ultrasound images [Abstract] 17th International Congress of Radiology. 1989:875. Abstract no. 2936.
17. Faita F, Gemignani V, Bianchini E, Giannarelli C, Ghiadoni L, Demi M. Real-time measurement system for evaluation of the carotid intima-media thickness with a robust edge operator. Journal of Ultrasound in Medicine. 2008;27:1353–1361. [PubMed]
18. Stein JH, Korcarz CE, Mays ME, Douglas PS, Palta M, Zhang H, Lecaire T, Paine D, Gustafson D, Fan L. A semiautomated ultrasound border detection program that facilitates clinical measurement of ultrasound carotid intima-media thickness. Journal of the American Society of Echocardiography. 2005;18:244–251. [PubMed]
19. Touboul PJ, Vicaut E, Labreuche J, Belliard JP, Cohen S, Kownator S, Pithois-Merli I. Paroi Arterielle et Risque Cardiovasculaire Study I. Design, baseline characteristics and carotid intima-media thickness reproducibility in the PARC study. Cerebrovascular Diseases. 2005;19:57–63. [PubMed]
20. Naghavi M, Falk E, Hecht HS, Jamieson MJ, Kaul S, Berman D, Fayad Z, Budoff MJ, Rumberger J, Naqvi TZ, Shaw LJ, Faergeman O, Cohn J, Bahr R, Koenig W, Demirovic J, Arking D, Herrera VLM, Badimon J, Goldstein JA, Rudy Y, Airaksinen J, Schwartz RS, Riley WA, Mendes RA, Douglas P, Shah PK, Force ST. From vulnerable plaque to vulnerable patient--Part III: Executive summary of the Screening for Heart Attack Prevention and Education (SHAPE) Task Force report. American Journal of Cardiology. 2006;98:2H–15H. [see comment]. [PubMed]
21. Molinari F, Zeng G, Suri JS. A state of the art review on intima-media thickness (IMT) measurement and wall segmentation techniques for carotid ultrasound. Computer Methods & Programs in Biomedicine. 2010;100:201–221. [PubMed]
22. Cheng D-C, Jiang X. Detections of arterial wall in sonographic artery images using dual dynamic programming. IEEE Transactions on Information Technology in Biomedicine. 2008;12:792–799. [PubMed]
23. Dwyer JH, Sun P, Kwong-Fu H, Dwyer KM, Selzer RH. Automated intima-media thickness: the Los Angeles Atherosclerosis Study. Ultrasound in Medicine & Biology. 1998;24:981–987. [PubMed]
24. Graf S, Gariepy J, Massonneau M, Armentano RL, Mansour S, Barra JG, Simon A, Levenson J. Experimental and clinical validation of arterial diameter waveform and intimal media thickness obtained from B-mode ultrasound image processing. Ultrasound in Medicine & Biology. 1999;25:1353–1363. [PubMed]
25. Holdfeldt P, Viberg M, Gustavsson T. A new method based on dynamic programming for boundary detection in ultrasound image sequences; Conference Proceedings: Annual International Conference of the IEEE Engineering in Medicine & Biology Society; 2008. pp. 3072–3074. [PubMed]
26. Vermeersch SJ, Rietzschel ER, De Buyzere ML, Van Bortel LM, D'Asseler Y, Gillebert TC, Verdonck PR, Segers P. Validation of a new automated IMT measurement algorithm. Journal of Human Hypertension. 2007;21:976–978. [PubMed]