Key points

History of the development of three-dimensional carotid plaque ultrasound

Early applications of ultrasound in the carotid arteries focused on using the Doppler shift to evaluate blood velocity and flow disturbances; our group used implanted ultrasound probes in the carotid arteries to assess effects of antihypertensive drugs on blood velocity and pulsatility, using spectral analysis to evaluate flow patterns. The primary clinical use was for diagnosing carotid stenosis ; we used a primitive device, the Dopscan, to evaluate effects of antihypertensive drugs on turbulence at sites of stenosis. When the author JDS obtained a duplex scanner because it provided spectral analysis to evaluate drug effects on flow disturbances, he realized that it was possible to image and begin to quantify carotid plaque burden. It was Maria DiCicco RVT, in the author’s laboratory, who told JDS that there was software in the scanner that could measure plaque area, and first measured carotid total plaque area (TPA), in 1990. After JDS told Jon Wikstrand about it, Wikstrand’s group published a method in 1992. Then in 1997, TPA was used to study effects of mental stress on atherosclerosis ; Fig. 1 is reproduced from that paper.

Measurement of plaque area. B-mode ultrasound images of atherosclerotic plaques in the extracranial carotid arteries. ( A ) A small plaque with an area of 0.18 cm ² traced by the trackball in the left common carotid artery. ( B ) A large plaque with an area of 0.42 cm ² in the distal common carotid artery.

( From Barnett PA, Spence JD, Manuck SB, et al. Psychological stress and the progression of carotid atherosclerosis. J Hypertension 1997;15:51; with permission.)

In 1994, around the time that Delcker and Diener published their early work on three-dimensional (3D) ultrasound estimation of plaque volume, Dr Aaron Fenster visited JDS at Victoria Hospital in London, Ontario to ask if he would be interested in using the 3D ultrasound system that Fenster had developed for other purposes to measure 3D carotid plaque volume. JDS visited his laboratory soon after. Fig. 2 shows perhaps the first measurement of plaque volume, in his right carotid artery in 1994 on Fenster’s prototype machine. In 1995 JDS moved to University Hospital and the Robarts Research Institute to collaborate with Dr Fenster, and then later Dr Grace Parraga, to develop 3D ultrasound of carotid plaque.

( A, B ) Measurement of carotid plaque volume, 1994. These small plaques totaling 12 mm ³ were measured using disk segmentation.

Topics discussed in this article include measurement of plaque burden, ulceration, echolucency, and plaque texture.

History of the development of three-dimensional carotid plaque ultrasound

Early applications of ultrasound in the carotid arteries focused on using the Doppler shift to evaluate blood velocity and flow disturbances; our group used implanted ultrasound probes in the carotid arteries to assess effects of antihypertensive drugs on blood velocity and pulsatility, using spectral analysis to evaluate flow patterns. The primary clinical use was for diagnosing carotid stenosis ; we used a primitive device, the Dopscan, to evaluate effects of antihypertensive drugs on turbulence at sites of stenosis. When the author JDS obtained a duplex scanner because it provided spectral analysis to evaluate drug effects on flow disturbances, he realized that it was possible to image and begin to quantify carotid plaque burden. It was Maria DiCicco RVT, in the author’s laboratory, who told JDS that there was software in the scanner that could measure plaque area, and first measured carotid total plaque area (TPA), in 1990. After JDS told Jon Wikstrand about it, Wikstrand’s group published a method in 1992. Then in 1997, TPA was used to study effects of mental stress on atherosclerosis ; Fig. 1 is reproduced from that paper.

Measurement of plaque area. B-mode ultrasound images of atherosclerotic plaques in the extracranial carotid arteries. ( A ) A small plaque with an area of 0.18 cm ² traced by the trackball in the left common carotid artery. ( B ) A large plaque with an area of 0.42 cm ² in the distal common carotid artery.

( From Barnett PA, Spence JD, Manuck SB, et al. Psychological stress and the progression of carotid atherosclerosis. J Hypertension 1997;15:51; with permission.)

In 1994, around the time that Delcker and Diener published their early work on three-dimensional (3D) ultrasound estimation of plaque volume, Dr Aaron Fenster visited JDS at Victoria Hospital in London, Ontario to ask if he would be interested in using the 3D ultrasound system that Fenster had developed for other purposes to measure 3D carotid plaque volume. JDS visited his laboratory soon after. Fig. 2 shows perhaps the first measurement of plaque volume, in his right carotid artery in 1994 on Fenster’s prototype machine. In 1995 JDS moved to University Hospital and the Robarts Research Institute to collaborate with Dr Fenster, and then later Dr Grace Parraga, to develop 3D ultrasound of carotid plaque.

( A, B ) Measurement of carotid plaque volume, 1994. These small plaques totaling 12 mm ³ were measured using disk segmentation.

Topics discussed in this article include measurement of plaque burden, ulceration, echolucency, and plaque texture.

Measurement of plaque burden and its progression/regression

Measurement of carotid intima-media thickness (IMT) had begun around 1985, first in monkeys and then in human subjects. Having been taught atherosclerosis by Dr Daria Haust (for many years the Editor of Atherosclerosis , and a Professor of Pathology at University of Western Ontario), JDS understood early on that IMT did not truly represent atherosclerosis, and decided to focus on quantification of plaque burden.

In 2002 his group reported that TPA was a strong predictor of cardiovascular risk among patients attending cardiovascular prevention clinics. After adjusting for age, sex, blood pressure, cholesterol, smoking, diabetes, homocysteine, and treatment of blood pressure and cholesterol, patients in the top quartile of TPA had a 3.4 times higher 5-year risk of stroke, death, or myocardial infarction. By quartile of TPA, the 5-year risks were approximately 5%, 10%, 15%, and 20%. Thus TPA was much stronger than a Framingham risk profile in predicting risk. During the first year of follow-up, approximately half the patients had plaque progression, a quarter had regression, and a quarter was stable. Those with plaque progression had twice the risk of events, after adjustment for the panel of risk factors listed previously. Our findings were borne out in the Tromsø study, a population-based study in Northern Norway of more than 6000 participants who were healthy at baseline. That study showed that TPA, but not IMT in the distal common carotid, predicted myocardial infarction and stroke. Added to risk calculation using scores based on risk factors, TPA significantly improves risk prediction. Subsequently it has become apparent in meta-analyses that IMT is only a weak predictor of cardiovascular events, adding little to a Framingham risk score, and that progression of IMT does not predict events. Brook and colleagues found that TPA significantly changed risk prediction and was more predictive of coronary stenosis than IMT, coronary calcium score, or C-reactive protein. Sillesen and colleagues found that carotid plaque burden (an estimated plaque volume assessed by moving the probe along the carotid) was much more closely correlated with coronary calcium score than IMT.

Measurement of Three-Dimensional Plaque Volume

Perhaps the first description of estimation of plaque volume was in a study by Hennerici and Steinke in 1987. Initial measurements of carotid plaque volume were laborious; the method, called disk segmentation, required acquisition of a series of two-dimensional cross-sectional slices captured by translating the ultrasound probe along the carotid artery with a mechanical device, tracing of the plaque area in serial cross-sections of the artery, stepping through the plaque at intervals of 1 to 2 mm, and summing the slices to obtain plaque volume ( Fig. 3 ). Early work specified the reproducibility, reliability, interslice distance, and other technical issues. Ludwig and colleagues in 2008 and papers in 2014 and 2015 reported good reproducibility, similar to results previously reported by Fenster’s group, and also found that reproducibility was better for large plaques than for small ones.

Procedure for determining plaque volumes from 3D ultrasound (US) images. ( A ) An approximate axis of the vessel is selected in a longitudinal view ( purple line ) and the internal elastic lamina and lumen boundary are outlined ( yellow ). ( B ) Using the surfaces generated by the vessel contours and the 3D US image, the position of the bifurcation (BF; yellow arrow ) is determined and marked. The axis of the vessel is selected based on the bifurcation point, and marked along the branch as far as the plaque can be measured ( purple line ). This axis is used as a reference for distance measurements. ( C ) All plaques within the measurable distance are outlined, different colors being used for each separate plaque to aid in identification. ( D ) Volumes are calculated for each plaque, and surfaces of the vessel wall and plaques are generated to better visualize the plaques in relation to the carotid arteries.

( From Ainsworth CD, Blake CC, Tamayo A, et al. 3D ultrasound measurement of change in carotid plaque volume: a tool for rapid evaluation of new therapies. Stroke 2005;36(9):1906; with permission.)

The disk segmentation method was not only laborious; it required 2 to 3 months of training and practice to learn to do it reliably and achieve certification in its use, and some candidates (approximately a third) were not able to perform it reliably: it seemed that perfectionists struggled too much with deciding where to demarcate boundaries, and others were too careless. These issues were the impetus for development of semiautomated and automated methods that would be less operator-dependent, less laborious, and perhaps more reproducible and reliable.

Early semiautomated methods reduced the number of slices of the plaque that required manual input to set the proximal and distal ends of each plaque, and tracing the contour of slices at the midpoint, and 25% and 75% of the length of the plaque, with automated interpolation of the surface of the remainder of the plaque assuming uniform plaque geometry, adjustable by the operator ( Fig. 4 ).

User input for semiautomated total plaque volume measurement. In the longitudinal view (and with assistance from the axial view, not shown) the user identifies the maximum and minimum z-values representing the end points of the plaque (Min Z and Max Z). The user identifies the midpoint of the plaque (C1) and finally C2 and C3 are identified and generated in the axial view. Uniform plaque geometry between C2 and C3 is assumed and a final volume is generated.

( From Buchanan D, Gyacskov I, Ukwatta E, et al. Semi-automated segmentation of carotid artery plaque volume from three-dimensional ultrasound carotid imaging. Proc SPIE 2012;8317:831701–4; with permission.)

In 2013 an automated method was developed using mechanical movement of the probe along the artery that provided good agreement with manual segmentation by experts ( Fig. 5 ). However, the machine used to translate the ultrasound probe is large and clumsy, and difficult to use in patients with obesity or short necks. A mechanical sweep, obtained by holding the probe in one location and having the angle changed mechanically ( Fig. 6 ), is faster and more convenient. Recently Graebe and colleagues, using the Philips system used by Sillesen and colleagues in the High-Risk Plaque Bioimage Study, showed that the mechanical sweep gave improved reproducibility of plaque volume quantification compared with manual translation of the probe along the artery ( Fig. 7 ).

Automated measurement of plaque volume. Sample results of algorithm and manual segmentations for three patients (one subject in each column). The panels in the rows were obtained at distances of 0, 1, and 2 mm from the carotid bifurcation. Yellow solid contours, algorithm results; red dashed contours, results of expert 1; green dashed contours, results of expert 2.

( From Cheng J, Li H, Xiao F, et al. Fully automatic plaque segmentation in 3-D carotid ultrasound images. Ultrasound Med Biol 2013;39:2440; with permission.)

Measurement of carotid plaque burden by automated software (Philips). Segment of carotid artery with a plaque ( orange ), which is scanned with a linear array transducer as a series of image slices in transverse section ( top ). Each image is analyzed with semiautomated software to quantify plaque area, plaque gray-scale statistics, percent stenosis, and other metrics of interest. Plaque areas from all images in the entire image sequence were summed as “plaque burden.” ( Lower left ) Common carotid artery with no plaque. The blue border represents the lumen/intima border; the red border represents the media-adventitia boundary. ( Lower right ) Common carotid artery with plaque. The red border and blue border are the same as in previous image, but the orange border represents the boundary of the plaque.

( From Sillesen H, Muntendam P, Adourian A, et al. Carotid plaque burden as a measure of subclinical atherosclerosis: comparison with other tests for subclinical arterial disease in the high risk plaque BioImage study. JACC Cardiovasc Imaging 2012;5:683; with permission.)

Freehand scanning versus mechanical sweep. Carotid artery scanning methods. ( A ) The two-dimensional freehand sweeps were acquired using a linear array 9- to 13-MHz transducer that was moved along the neck in one direction for a 10-s acquisition. ( B ) For the 3D mechanical sweeps, the transducer was held still over the carotid bifurcation. When activated, a linear transducer inside the transducer head was mechanically propagated at a constant speed.

( From Graebe M, Entrekin R, Collet-Billon A, et al. Reproducibility of two 3-D ultrasound carotid plaque quantification methods. Ultrasound Med Biol 2014;40:1642; with permission.)