the objective characterisation of ultrasonic carotid plaque … · 2017-01-14 · carotid plaque...

Eur J Vasc Endovasc Surg 16, 223-230 (1998)

The Objective Characterisation of Ultrasonic Carotid Plaque Features

T. Elatrozy*, A. Nicolaides, T. Tegos and M. Griffin

Irvine Laboratory for Cardiovascular Investigation and Research, Academic Surgical Unit, Imperial College School of Medicine, St. Mary's Hospital, London, U.K.

Objective: To determine the influence of ultrasonic carotid plaque morphology on the incidence of ipsilateral hemispheric symptoms (IHS). Design: Cross-sectional study. Materials: A consecutive series of 80 patients (96 plaques) with more than 50% ICA stenosis was studied. Methods: B mode ultrasonic images were captured and transferred to a computer on magneto-optic disk and standardised using linear scaling so that adventitia would have a grey scale median (GSM) value of 185-I95 and blood 0-5. The GSM and the percentage of echolucent pixels (PEP) in plaques were determined to measure echodensity. Homogeneity, entropy, and contrast were also determined to measure spatial distribution (heterogeneity) of grey shades in each plaque. Each measurement was correlated to presence or absence of IHS. Results: Twenty-~'ve plaques were associated with IHS and 71 plaques were asymptomatic. In symptomatic plaques the mean of GSM was 23 and the mean of PEP was 70%, compared to 38 and 55% respectively in asymptomatic plaques (p=O.02; Wilcoxon test). Sixty per cent of symptomatic plaques were associated with a homogeneity, entropy, and contrast values of >0.2, <2.95, <150 respectively as compared to 40% in asymptomatic plaques. Multiple regression analysis revealed that the GSM and the PEP were the most signi~cant variables (p =0.001) that are related to presence or absence of IHS. Conclusion: This study indicates that computer aided analysis of ultrasonic B mode features of carotid plaques could identify a potentially high-risk subgroup (patients with IHS). A GSM less than 40 or PEP greater than 50% is a good predictor of IHS related to carotid plaques. The fact that these measurements are operator independent and performed after image standardisation should encourage their use in multicentre clinical trials where different operators and equipment are used.

Key Words: Carotid plaque; Characterisation; Ultrasonic.

Introduction

It has been suggested that ultrasonic carotid plaque morphology may be as important as the percentage diameter stenosis in selecting patients for carotid endarterectomy. 1 Few prospective studies have suggested that an echolucent lesion is more likely to be associated with stroke than an echogenic one. 2'3 The problem has been that the definition and the classification of plaque morphology using ultrasonic varied greatly. 4~7

Recently our group has shown that computer assisted methods of plaque characterisation using digital image processing can provide measurements that may be as important as the degree of stenosis in predicting clinical outcome. 8 A grey scale median (GSM) less than

* Please address all correspondence to: T. S. Elatrozy, St. Mary's Hospital Qeqm-wing, Irvine Laboratory, London-Paddington-Praed Street W2 1NY.

32 (scale 0-255; 0 =black, 255 =white) of the pixels of digitised images of plaques was associated with a 5- fold increase in the incidence of silent infarcts on CT brain scans, providing the same duplex scanner and settings were used. 9

A new method has been developed that allows image standardisation so that measurements of GSM of the same plaque obtained by different scanners or operators are comparable and reproducible with a coefficient of variation of 4.7%J ° The aim of this study was to determine the relationship between ultrasonic plaque morphology as measured by objective computer assisted methods in standardised images and the incidence of ipsilateral hemispheric cerebrovascular symptoms.

Patients and Methods

Eighty consecutive patients with haemodynamically significant carotid bifurcation disease (more than 50%

1078-5884/98/090223+08512.00/0 © 1998 W.B. Saunders Company Ltd.

224 1". Elatrozy et al,

diameter stenosis in the bulb and/or proximal ICA) were admitted to the study. The degree of stenosis was estimated according to the velocity ratio PSVica/ EDVcca 11 and PSVica/PSVcca. 12

All patients have been evaluated carefully by a senior neurologist to determine whether they were symptomatic or asymptomatic. Patients with atrial fibrillation, heart failure, valvular lesions as well as previous cerebral haemorrhage or symptoms of verte- brobasilar insufficiency were excluded.

Duplex settings

Ultrasound scans were performed using an ATL (model HD13000; Advanced Technology Laboratories, Seattle, U.S.A.) duplex scanner with a 7-4 MHz high- resolution multi-frequency probe. B mode scan settings were adjusted so that the maximum dynamic range was used with a linear post-processing curve. Frame rate and persistence were set high. The position of the probe was adjusted so that the ultrasonic beam was at right angles to the arterial wall. The region of interest (plaque, far and near walls and the nearby surrounding tissues) was selected and magnified using the high definition zoom facility. All scan settings were saved in the duplex memory so that they could be instantly selected at the beginning of every examination. The TGC curve was adjusted (gently sloping) to produce uniform intensity of echoes on the screen, but it was vertical in the lumen of the artery where attenuation in blood was minimal so that the echogenicity of the far wall was visually the same as that of the near wall. The overall gain was increased until the appearance of the plaque was judged to be optimal and noise appeared within the lumen. It was then decreased so that at least some areas in the lumen appeared to be free of noise (black).

B mode images of carotid plaques were captured through the direct access link facility in the duplex on a magneto-optic disk. Image standardisation and analysis were performed using Adobe Photoshop TM

Version 3.0 and MATLAB based algorithms (The Math- Works, Inc. Version 4.2c.1).

Using the histogram facility in the program, the median value of the grey levels of all the pixels (grey scale median; GSM) was obtained.

3. Similarly part of the adventitia was selected. Meas- urements of GSM were made at the brightest part of the adventitia on the same arterial wall as the plaque.

4. Image standardisation. Algebraic scaling of the whole image was performed using the "curves" facility of the software. This was linear and based on the two reference points: blood and adventitia. The scale was adjusted so that the grey value of the blood would be in the region of 0-5 and that of the adventitia in the region of 185-195. Thus the grey values of all pixels would change as defined by this new linear scale.

The reproducibility of this method in our de- partment (four observers; two duplex scanners and 23 carotid images) has been previously reported (CV; 0.047 Reliability coefficient 0.99). l°

5. Analysis of echoic features in plaques. In standardised images plaques were outlined by the mouse of a personal computer (PC) and the following measurements were obtained:

Grey Scale Median (GSM): defined as the median of overall grey shades of the pixels in the plaque.

Total percentage of echolucent pixels: defined as the percentage of pixels with grey levels below 40 (see Fig. 1).

Homogeneity: defined as the degree of local homogeneity of grey values. A homogenous plaque indicates that all pixels in the selected region have similar grey levels. Values of homogeneity range between 0 (for very non-homogenous plaques) to 1 (for perfectly homogenous plaques).

Entropy: defined as the degree of disorder or dis- similarity of grey values; a largest value is obtained when the plaque is non-homogenous and has a non- recognisable pattern.

Contrast: defined as a measure of the variability of grey scale difference and hence coarseness of texture. A large value of contrast indicates large variation in grey values of the pixels.

Protocol of image standardisation and analysis

1. The colour information in the image was omitted so that all the processing and analysis were performed on images in grey mode.

2. An area in the blood (free of noise) was selected.

Statistical Me thods

Different measurements made in this paper were not normally distributed and were represented as pro- portions and mean values. Plaques were classified into two groups: symptomatic and asymptomatic. The distribution of the values for each feature measured

Eur J Vasc Endovasc Surg Vol 16, September 1998

Carotid Plaque Characterisation 225

Fig. 1. B mode ultrasonic image of carotid artery plaque magnified using the high definition zoom. The histogram of grey shades in the plaque is shown. The grey scale median (GSM) is seven and percentage of echolucent pixels (percentage of grey levels between 0 and 40) is 80.25%.

is presented in the form of a column scatter to allow visual inspection of the overlap between symptomatic and asymptomatic groups. A non-parametric, M a n n - Whitney, Wilcoxon test was used to give the prob- ability whether the distribution is different between the two groups. Visual inspection of column scatter distributions al lowed us to draw arbitrary cut-off lines that have the highest prevalence of symptomatic plaques to one side. Symptomatic and asymptomatic plaques were subdivided into two subgroups for each feature measured on the basis whether its value fit above or below the cut-off line and Fisher's exact test was used to compare these subgroups. A stepwise progressive multivariate analysis with the symptomatic status of carotid plaques as the independent variable was per formed to determine which feature measured is most predictive of IHS. It should be noted that statistical significance in this paper should take into account the fact that this s tudy is a cross-sectional one where asymptomatic plaques at one stage in the

natural history of the disease would have features in common with symptomatic plaques and therefore an overlap exists. Also with measurements like GSM a small difference between two plaques of say five grey levels on a grey scale of 255 shades of grey may not be clinically important. This is very important when interpret ing the results.

Results

Ninety-six plaques from 80 patients were studied (71 asymptomatic, 25symptomatic). Stenosis in the range of 70-99% was found in 68 plaques. In the remaining 28 stenosis was in the range of 50-69%.

In plaques that were associated with greater than 70% stenosis, 49/68 (72.1%) were asymptomatic while 19/68 (27.9%) were associated with hemispheric symptoms (transient ischaemic attacks (TIAs) in five;


226 1". Elatrozy et al.

110

~ 100

~ 90

.~ 80

O ~ 70

~ 60

o 50

r~ ~ 4o

• 80

~ 2o

0 0 o

0

• 0 O0 0000

@ 000 0 0 0 0 0

_ 0 0 0 8

• 000 • O0

................................ ~)- ...........

@@ 0 0

@@@@ 0

@@• 00000~ • 0000

• 00000 ° e o o

00 0 0 00@0 000 @ 0

@@ 000000

Fig. 2. Grey scale median in symptomatic and asymptomatic plaques. (p=0.02 Wilcoxon Mann-Whitney test). (@) Symptomatic; (O) asymptomatic.

100

r~ 90

.~ 80 S

v 70

5o

4o

~ ao

~ 20

£ lO

• O 0 0 0

• 0 @ 0~0

• 0 o

O0 @@@ 0

,, °8~°

0 0 O0 o

• 0

O0 O0 @@

............ ~ ................................

• 0 • 0000

_ O00

00~00 • ~000

~000 @ 000

O00 0 o

O

0 o 0

Fig. 3. Percentage of echolucent pixels (pixels with grey levels below 40) in symptomatic and asymptomatic plaques. (p = 0.01 Wilcoxon Mann-Whitney test). (@) Symptomatic; (O) asymptomatic.

amaurosis fugax (AF) in two; TIAs and AF in one; stroke with good recovery in 11).

In plaques that were associated with less than 70% stenosis, 22/28 (78.5%) were asymptomatic while 6/28 (21.4%) were associated with hemispheric symptoms (TIAs in one; AF in one; TIAs and AF in two; stroke with good recovery in two).

The mean diameter stenosis for the symptomatic plaques was 78 ± 14 and for asymptomatic plaques was 74 + 14. The difference was not statistically significant (Wilcoxon test p = 0.2).

Measurements of echodensity (GSM and PEP)

The distribution of GSM and PEP values of plaques which are associated with and without symptoms are shown in Figs 2 and 3 respectively.

Twenty-one out of twenty-five (84%) plaques had a GSM less than 40 and PEP above 50%, while only 4/ 25 (16%) had a GSM greater than 40 and PEP below 50% (Table 1).

In asymptomatic plaques, there was no statistically significant difference between the incidence of plaques with GSM less or greater than 40 or PEP below or

above 50%. However, the mean of GSM values in symptomatic plaques (23+17) was significantly different from asymptomatic plaques (37.9 +26; p = 0.02 Wilcoxon text).

The mean of PEP values in symptomatic plaques (70 + 20) was significantly different from asymptomatic plaques (55 ± 25; p = 0.01 Wilcoxon test).

Figure 4 is a scattergram, which demonstrates the relationship between GSM and PEP in symptomatic and asymptomatic plaques. It appeared that plaques with a GSM less than 40 and PEP greater than 50% are more likely to be symptomatic (sensitivity 85%; specificity 51%; relative risk 3.4 ± 0.5 s.m with 95% CI 1.27-9.1).

Measurements of heterogeneity (entropy, homogeneity, and contrast

There was no statistically significant association between homogeneity or contrast and the incidence of symptomatic or asymptomatic plaques (Figs 5 and 6, Table 1). However, entropy was shown to be more significantly associated with symptomatic status. Using an arbitrary cut-off line at 2.9 (Fig. 7), 60% of



Table 1. The association between computer assisted measurements of the grey level characteristics in B mode images of carotid plaques and related ipsilateral hemispheric symptoms (IHS).

Cut-off value Symptomatic Asymptomatic Fisher's (25 plaques) (69 plaques) exact test

<40 21 (84%) 36 (52%) Grey scale median

Percentage of echolucent pixels Homogeneity

Contrast

Entropy

>40 4 (16%) 33 (48%) 0.02 >50% 20 (80%) 36 (52%) <50% 5 (20%) 33 (48%) 0.02

>0.2 14 (56%) 29 (42%) <0.2 11 (44%) 40 (58%) 0.13 <150 15 (60%) 34 (49%) >150 10 (40%) 35 (51%) o.4 <2.9 15 (60%) 24 (35%) >2.9 I0 (40%) 45 (65%) 0.03

90 6"

o

b.O

e ° , e 0

i00

@ O

80 , o.O . o 0 • 0

co oll, I I

7 0 - o • ', I I I

6 0 - o e e ~, • 0 0 l m

%oi 50 ................................ i ..................................................................

• o

40 -- ° d :~ ° co %00

S ° o ° o 3 0 -

2 0 -

1 0 -

0 @ 0 0 O 0

0 O

O

O

I I r [ 0 20 40 60 80 100 120

Grey scale median (GSM) of grey levels of pixels in plaques

Fig. 4. The relationship between grey scale median and the total percentage of echolucent pixels in symptomatic and asymptomatic plaques: the relative risk of symptoms (for the GSM _<40 and PEP >50% against GSM >40 and TPEP <50%) is relative risk 3.4_+0.5 s.~. with 95% CI 1.27-9.1. (O) Symptomatic; (O) asymptomatic.

symptomatic plaques were below that line while only 40% were above the line indicating that symptomatic plaques tend to have a lower entropy value and therefore they are considered less heterogeneous. On the other hand in asymptomatic plaques 65% were found to be above the line (with entropy value greater than 2.9) while only 35% were below that line (i.e. with entropy value lesser than 2.9) indicating that asymptomatic plaques tend to be more homogenous (Chi

squared 3.8 p=0.05; Fisher's exact 2 p=0.04). Mul- tivariate regression analysis using the above measurements together (Table 2) in relation to presence or absence of IHS taken as an independent variable revealed that the GSM and total percentage of echolucent pixels are the only significant variables that could predict plaques with IHS. It seems likely that the finding of a GSM less than 40 and a PEP greater than 50% characterises symptomatic plaques.


2 2 8 T . E l a t r o z y et al.

0.50

0.45

0.40

'H 0.35

0.30

0.25

0.20

o9 = 0.15

o 0.10

0.05

0 . 0 0

• O0

0 o -- 0

• 0

0•00 08 -- @••

0 0000 0000

"" og o O 0

. . . . . . . . . i : ~ " . . . . . . . . . . . . . . %~R8 . . . . . . . . . . . . • •" 000080000 • 8 0 0 0

_ •@ 0 0 0 0 0 8

.. 8 8 8 0 UO00

0

Fig. 5. Homogeneity in symptomatic and asymptomatic plaques. (p=0.04 Wilcoxon Mann-Whitney test). (@) Symptomatic; (0) asymptomatic.

-K

o9

¢.~ o

o9

o9

o9

o

o9

500

450

400

350

300

250

200

150

!oo

5O

0

O0

0 0 0 0

- • o8 • 0

0 o 0

0

@@ 080 - - 0000

0 • @ 0 0 0 O 0

. . . . . . . . . . . . . . . . . . . . . . . . . . . . ~O00-Q . . . . . . . . . . .

• ~8O9o 00 -- O U

• O 0

• ~ 000080000 -- 0000

O0 _ 00

Fig. 6. Contrast in symptomatic and asymptomatic plaques. (p = 0.4 Wilcoxon Mann-Whitney test). (@) Symptomatic; (O) asymptomatic.

;m

o9 H

o9

o9

09

3.9

3.7

3.5

3.3

3.1

2.9

2.7

2.5

2.3

2.1

1.9

1.7

-- 0

0

0 -- • @ 000

0 0 • oOo~ o

- - • • uO;::5

08o0

.-. 0O0o°%00 ~8o~

• 0 0 8 - . . . . . . . . . . . , i . . . . . . . . . . . . . o5O7,o - . . . . . . . . . .

~ OOo~;

- %8 @@

_ • 08 °

• O

• 0

0 0

Fig. 7. Entropy in symptomatic and asymptomatic plaques. (p = 0.05 Wilcoxon Mann-Whitney test). (@) Symptomatic; (O) asympto- rnatic.

Table 2. Multiple regression analysis of computer assisted measurements of the grey level characteristics in B mode images of carotid plaques in relation to their ipsilateral hemispheric symptoms (IHS).

Parameter p value

Grey scale median 0.001" Percentage of echolucent pixels 0.001" Entropy 0.118 Homogeneity 0.194 Contrast 0.880 Degree of stenosis 0.250

* Denotes statistical significance.

Discussion

At present the main criterion available to make a clinical decision whether a patient should be considered for carotid endarterectomy or not, is the degree of stenosis for both symptomatic and asymptomatic patients . ]3-16 This is b ecau se the degree of s tenos i s is the only criterion that has been shown to correlate with stroke in randomised controlled studies, although the presence of ulcerated plaques on angiography also has increased the risk in the North American Symptomatic Carotid Endarterectomy Trial. 17

In this trial it has b e e n d e m o n s t r a t e d that pa t i en t s w i t h severe s tenos i s 70-99% in the in t e rna l carotid artery carry a substantial risk of ipsilateral ischaemic



stroke and that carotid endarterectomy decreased that risk by a factor of six.

in our study there was no statistically significant difference between the mean of the degree of stenosis from symptomatic and asymptomatic groups. How- ever, we have found that the incidence of symptomatic plaques is 27.9% (19/68) when the degree of stenosis is 70% or more and only 21.4% (6/28) when the degree of stenosis is 50-69%. This gives an estimated relative risk of 1.3 (CI 0.6-2) if the degree of stenosis is used as the only criterion that characterises symptomatic plaques. In contrast the incidence of symptomatic plaques was 36% (21/57) in the group with GSM <40 or PEP >50% (i.e. echolucent plaque) while the incidence of symptomatic plaques was only 10% (4/ 37) in the group with GSM >40 or PEP <50% (i.e. echogenic plaque). This gives an estimated relative risk of 3.4 (CI--1.2-9). It appears that the combined criteria of stenosis greater than 70% and a GSM <40 or PEP >50% characterises plaques with IHS.

In this study the incidence of plaques with these combined criteria was 45% (19/42) compared to an incidence of only 9% (5/51) in the absence of one of the above criteria. The estimated relative risk is 4.6 (CI 1.8-11.3).

Considerable work in the last 10 years has identified a number of ultrasonic plaque characteristics that are associated with an increased incidence of hemispheric clinical events (TIAs and/or stroke). These include irregular borders, echolucency and a heterogeneous texture. 4 However, a clear association between plaque ultrasonic characteristics and stroke is lacking. The great variability in the incidence of cerebrovascular events in relation to plaque morphology has resulted from the fact that various classifications have relied on the human eye rather than objective measurements.

Homogenous plaques, for example, can be in- terpreted both as homogenous and totally echolucent or homogenous and totally echogenic. The totally echolucent plaques which correspond to type 1 in Grey- Weale's classification will be of higher risk that the totally echogenic ones that correspond to type 4. On the other hand heterogeneous plaques could be in- terpreted as heterogeneous but predominantly echolucent which correspond to type 2 in Grey-Weale's classification or predominantly echogenic which correspond to type 3. The former is of higher risk than the latter. In this study it is possible that a number can be given so that measurement of echodensity of plaques separate from texture characteristics, in standardised images, can be achieved to compare plaque morphologic characteristics from different centres. What is needed in the future studies is to

provide different cut-off points of such measurements that correlate with plaque stability/instability.

The potential of providing objective and reproducible measurements of echodensity separate from homogeneity is valid for the material included in this study. It needs to be tested prospectively in another cohort of patients and validated again in separate groups of symptomatic and asymptomatic patients. This can be in relation to surrogate end points such as infarcts on CT brain scans or the number of microemboli (high intensity transient signals) in the middle cerebral artery using transcranial Doppler. Fol- lowing such validation, the inter-centre variability of image standardisation and measurement of plaque echodensity should be tested. Finally the methodology should be applied in studies (on carotid plaques) which use the development of stroke as an endpoint. Only when a clear association between plaque echolucency and stroke is demonstrated can clinical decisions be considered. This would have its implication in im- proving selection of patients for carotid endarterectomy with a lower incidence of pre-operative stroke and mortality.

Acknowledgements

The study was supported by a grant from CDER Trust (U.K.). We are grateful to the support of Prof A. O. Mansfield, Mr J. Wolfe, Dr D. J. Thomas and Mr G. Stansby for their support and for allowing us to study their patients.

References

1 JEMU. Consensus concerning the morphology and the risk of carotid plaques. Paris 1996; 176.

2 JOHNSON JM, KENNELLY M, DECESARE D, MORGAN S, SPARROW A. Natural history of asymptomatic carotid plaque. Arch Surg 1985; 120: 1010-1012.

3 BOCK RW, LUSBY RJ. Carotid plaque morphology and in- terpretation of echolucent lesion. In: Labs KH, ed. Diagnostic Vascular Ultrasound. London: Edward Arnold, 1992.

4 WIDDER B, PAULT K, HAIKSPACHER J. Morphological characterisation of carotid artery stenosis by ultrasound duplex scanning. Ultrasound Med Biol 1990; 16: 349-354.

5 JOHNSON JM, KENNELLY M, DECESARE D, MORGAN S, SPARROW A. Natural history of asymptomatic carotid plaque. Arch Surg 1985; 120: 1010-1012.

6 STERPETTI av, SCHULTZ RD, FELDHAUS R. Ultrasonographic features of carotid plaques and the risk of subsequent neurologic deficits. Surgery 1988; 104: 652-660.

7 LANGSFELD M, GRE¥-WEALE AC, LUSBY RJ. The role of plaque morphology and diameter reduction in the development of new symptoms in asymptomatic carotid arteries. J Vasc Surg 1989; 9: 548-557.

8 EL-BARGHHOUTY N, GEROULAKOS G, NICOLAIDES A, AN- DROULAKIS A, BAHAL V. Computer assisted carotid plaque characterisation. Eur J Vasc Endovasc Surg 1995; 9: 389-393.


230 1". Elatrozy et al.

9 EL-BARGHHOUTY N, GEROULAKOS G, NICOLAIDES A, AN- DROULAKIS A, BAHAL V. The identification of the high risk carotid plaque. Eur J Vasc Endovasc Surg 1996; 11: 470-478.

10 ELATROZY TS, ZARKA ZA, GRIFFIN M, TEGOS T, NICOLAIDES A. The way to standardize reporting on carotid plaque echodensity: application of image processing techniques. Presented at the 8th Mediterranean congress of angiology and vascular surgery, Alexandria, 31 May-3 April 1997.

11 NICOLAIDES AN, SHIFRIN EG, BRADBURY A. Angiographic and duplex grading of internal carotid stenosis: can we overcome the confusion? J Endovasc Surg 1996; 3: 1-8.

12 MONETA GL, EDWARDS JM, PAPANICOLAOU G. Screening for asymptomatic internal carotid artery stenosis: duplex criteria for discriminating 60% to 90% stenosis. J Vasc Surg 1995; 21: 989-994.

13 MAYBERG MR, WILSON SE, YATSU F. For the Veteran Affairs Cooperative Studies Program 309 Trialist Group. Carotid endarterectomy and prevention of cerebral ischaemia in symptomatic carotid stenosis. JAMA 1991; 266: 3289-3294.

14 GASECK~ AP, HACHINSKI VC, MANDEL T. Endarterectomy for carotid stenosis. Review of the European and North American Symptomatic Carotid Surgery Trials. NebrMedJ 1992; 77:121-123.

15 CONSENSUS GROUP. Consensus statement on the management of patients with asymptomatic atherosclerotic carotid bifurcation lesions: international angiology. Int AngioI 1995; 14: 5-17.

16 EUROPEAN CAROTID SURGERY TRIALISTS' COLLABORATIVE GROUP. Endarterectomy for moderate symptomatic carotid stenosis. In- terim results from the MRC European Carotid Surgery Trial. Lancet 1996; 347: 1592-1593.

17 NORTH AMERICAN SYMPTOMATIC CAROTID ENDARTERECTOMY TRIAL. ELIASZIW M, STREIELER JY, Fox AJ, HACHINSRI VC, FER- GUSON GG, BARNETT HJ. Significance of plaque ulceration in symptomatic patients with high-grade carotid stenosis. Stroke 1994; 25: 304-308.

Accepted 29 April 1998


the objective characterisation of ultrasonic carotid plaque … · 2017-01-14 · carotid plaque...

Documents