Computer analysis of ultrasonic plaque echolucency in identifying high risk carotid bifurcation lesions.

OBJECTIVES: to confirm that plaque echogenicity evaluated by computer analysis, as suggested by preliminary studies, can identify plaques associated with a high incidence of strokes. MATERIALS AND METHODS: a series of 96 patients with carotid stenosis in the range of 50-99% were studied retrospectively (41 with TIAs and 55 asymptomatic). Carotid plaque echogenicity was evaluated using a computerised measurement of the median grey scale value (GSM). All patients had a CT brain scan to determine the presence of infarction in the carotid territory. RESULTS: the incidence of ipsilateral brain CT infarctions was 16% in the asymptomatic and 32% in the symptomatic plaques (p =0.076). It was 20% for <70% stenosis and 25% for >70% stenosis (p =0.52). It was 9% for plaques which had a GSM >50 and 40% in those with GSM <50 (p <0.001) with a relative risk of 4.6 (95% CI 1.8 to 11.6). CONCLUSIONS: the results confirm that computer analysis of plaque echogenicity is better than the degree of stenosis in identifying plaques associated with an increased incidence of CT brain-scan infarction and consequently useful for identifying individuals at high risk of stroke. What is required is a form of image standardisation in order to apply this method to natural history studies with stroke as the endpoint.

Histological verification of computerised carotid plaque characterisation.

The correlation between the computerised measurement of carotid plaque echogenicity on high resolution ultrasound imaging and plaque histology was studied in 52 patients undergoing carotid endarterectomy. Thirteen plaques were from asymptomatic patients, 15 were associated with amaurosis fugax, 10 with transient ischaemic attacks and 14 with stroke. Longitudinal images of the anterior and posterior component of each plaque were obtained by ATL Ultramark-4 Duplex scanner and were transferred to a computer. Using an image analysis program the median of the overall grey scale content of each plaque component was evaluated and used as a measure of echogenicity. Following carotid endarterectomy each plaque specimen was divided into anterior and posterior component and then fixed, oriented, sectioned and stained in the longitudinal plane corresponding to the ultrasound image. Plaque histology sections were then examined by computer morphometric analysis and the percentage surface areas of fibrous tissue, lipid deposits and haemorrhage were calculated. This was then correlated with the grey scale median for each plaque component. Plaques with a high lipid and haemorrhage content as established histologically had a low grey scale median (Spearman correlation r=-0.351, p<0.05) and those with a high fibrous content had a high grey scale median (r=0.411, p<0.001). This study has shown that computerised measurement of carotid plaque echogenicity correlates well with histology and could be used to predict plaque composition, thus identifying high risk plaques with high lipid and haemorrhage content.

The identification of the high risk carotid plaque.

OBJECTIVE: To determine the influence of carotid plaque morphology and severity of stenosis on symptoms of cerebrovascular disease and cerebral infarction. PATIENTS AND METHODS: One hundred and ninety patients with 329 carotid plaques producing 50-99% stenosis were studied. Carotid plaque echogenicity on ultrasonography was evaluated using computerised measurement of the median of the overall grey scale content (GSM). Heterogeneity was evaluated as the difference between the GSMs of the most echogenic and the most echolucent areas within each plaque and expressed as the heterogeneity index (HI). All patients had a CT brain scan and the presence of ipsilateral cerebral infarction noted. RESULTS: Cerebral infarction was more common in symptomatic than asymptomatic plaques (42% vs. 29%, p<0.02) and in echolucent than echogenic plaques (mean GSM of 37.8 vs. 29.7, p<0.01). Plaques with GSM below or equal to 32 were associated with a higher incidence of cerebral infarction as compared to those above this level, this was significant in both symptomatic and asymptomatic plaques. Symptomatic carotid plaque were less heterogenous than asymptomatic plaques. Plaques associated with cerebral infarction were less heterogenous than those not associated with infarction. CONCLUSION: This study has shown that the identification of the high risk carotid plaques, i.e. those associated with a high incidence of cerebral infarction is possible both in symptomatic and asymptomatic patients. The potential of such analysis in the identification of patients with asymptomatic carotid stenosis with high and low risk of stroke should be explored in a natural history study.

Computer-assisted carotid plaque characterisation.

OBJECTIVE: To determine the relationship between plaque echogenicity as measured by computer and the incidence of cerebral brain infarction. PATIENTS AND METHODS: Eighty-seven patients with 148 plaques producing more than 50% internal carotid artery stenosis were studied. Sixty-nine plaques were in asymptomatic patients, 35 were associated with amaurosis fugax, 19 with transient ischaemic attacks and 25 with stroke. All patients had a CT brain scan and the presence of ipsilateral cerebral infarction was noted. Images of the plaques obtained with an ATL Ultramark-4 Duplex scanner (7.5 MHz high resolution probe) were transferred to a computer. Using an image analysis program a histogram for each plaque was obtained with the number of pixels plotted against the grey scale (0-225). The median of the grey scale was used as a measure of echogenicity. RESULTS: Fifty-three (36%) of the 148 plaques were associated with ipsilateral CT brain infarction. Plaques with a grey scale median more than 32 (echogenic) were associated with an incidence of 11% (7/64) CT infarction. In contrast, plaques with grey scale median below or equal to 32 (echolucent) were associated with 55% (46/84) incidence of CT infarction (chi 2 = 30.35, p < 0.001, relative risk = 22, 95% confidence interval from 4.7 to 108). CONCLUSION: This study indicates that computer analysis of carotid plaque can identify high-risk carotid plaques. The potential of such analysis in the identification of asymptomatic high-risk patients should be explored in further studies.