Quantitative rotational thallium-201 tomography for identifying and localizing coronary artery disease.

Abstract

The purpose of this study was to develop and validate a method for quantifying the uptake, redistribution, and washout of thallium-201 (201Tl) obtained with rotational tomography. This method generates maximum count circumferential profiles of the short-axis slices of the left ventricle, translates them into polar coordinate profiles, and displays them as a bullseye plot, which consists of a series of concentric circles with the apex at the center and the base at the periphery. Normal limits were established for the distribution of 201Tl in 36 patients with a low (less than 5%) probability of coronary artery disease (CAD). Forty-five patients who had undergone coronary angiography were used as a pilot group to define criteria for the identification and localization of perfusion defects. The best agreement with the results of angiography was found when abnormal regions of the bullseye were defined as contiguous defects over 2.5 SDs below normal. These criteria were applied prospectively to 210 points (179 points with greater than 50% diameter stenosis and 31 with less than 50%). Visual, quantitative, and combined visual and quantitative analysis were compared for overall detection of disease and for detection of individual vessel involvement. The overall sensitivity for detection of disease by these methods was 97%, 95%, and 95%, respectively. The specificities were 68%, 74%, and 71% respectively. The sensitivity for detection of individual vessel involvement with the bullseye alone was 78% for the left anterior descending artery (LAD), 89% for the right coronary artery (RCA), and 65% for the left circumflex (LCx). For visual analysis, the results were 70%, 88%, and 50%, respectively, while the use of visual and quantitative analysis combined identified 75% of LAD, 87% of RCA, and 55% of LCx lesions. We conclude that quantitative analysis of rotational 201Tl tomographic images is a highly accurate technique for determining the presence and location of CAD.

Ähnliche Arbeiten