Value of blood-pool subtraction in cardiac indium-111-labeled platelet imaging.

Blood-pool subtraction has been proposed to enhance 111In-labeled platelet imaging of intracardiac thrombi. We tested the accuracy of labeled platelet imaging, with and without blood-pool subtraction, in ten subjects with cardiac thrombi of varying age, eight with endocarditis being treated with antimicrobial therapy and ten normal controls. Imaging was performed early after labeled platelet injection (24 hr or less) and late (48 hr or more). Blood-pool subtraction was carried out. All images were graded subjectively by four experienced, "blinded" readers. Detection accuracy was measured by the sensitivity at three fixed levels of specificity estimated from receiver operator characteristic curve analysis and tested by three-way analysis of variance. Detection accuracy was generally improved on delayed images. Blood-pool subtraction did not improve accuracy. Although blood-pool subtraction increased detection sensitivity, this was offset by decreased specificity. For this population studied, blood-pool subtraction did not improve subjective detection of abnormal platelet deposition by 111In platelet imaging.