Carotid plaque characterization using digital image processing and its potential in future studies of carotid endarterectomy and angioplasty.
PURPOSE: To corroborate the validity of a computerized methodology for evaluating carotid lesions at risk for stroke based on plaque echogenicity. METHODS: The records of 96 carotid endarterectomy patients (59 men; median age 69.5 years, range 52 to 83) with stenoses > 50% were studied retrospectively. Forty-one patients (43%) had been symptomatic preoperatively. All patients had undergone computed tomography (CT) to detect infarction in the carotid territory and a duplex scan to measure carotid stenosis. Plaque echogenicity was analyzed by computer, expressing the echodensity in terms of the gray scale median (GSM). The incidence of CT-documented cerebral infarction was analyzed in relation to symptomatology, percent stenosis, and echodensity. RESULTS: Symptoms correlated well with CT evidence of brain infarction: 32% of symptomatic patients had a positive CT scan versus 16% for asymptomatic plaques (p = 0.076). The mean GSM value was 56 +/- 14 for plaques associated with negative CT scans and 38 +/- 13 for plaques from patients with positive scans (p < 0.0001). However, there was no difference in the GSM value between plaques with > or < 70% stenosis. Furthermore, the incidence of CT infarction was 40% in the cerebral territory of carotid plaques with a GSM value < 50 and only 9% in those with a GSM > 50 (p < 0.001). CONCLUSIONS: Computerized analysis of plaque echogenicity appears to provide clinically useful data that correlates with the incidence of cerebral infarction and symptoms. This method of analyzing plaque echolucency could be used as a screening tool for carotid stent studies to identify high-risk lesions better suited to conventional surgical treatment.