Visual and semiquantitative analysis of 18F-fluorodeoxyglucose positron emission tomography using a partial-ring tomograph without attenuation correction to differentiate benign and malignant pulmonary nodules.
- Additional Document Info
- View All
OBJECTIVE: Many studies have reported the use of attenuation-corrected positron emission tomography with 18F-fluorodeoxyglucose (FDG PET) with full-ring tomographs to differentiate between benign and malignant pulmonary nodules. We sought to evaluate FDG PET using a partial-ring tomograph without attenuation correction. METHODS: A retrospective review of PET images from 77 patients (range 38-84 years of age) with proven benign or malignant pulmonary nodules was undertaken. All images were obtained using a Siemens/CTI ECAT ART tomograph, without attenuation correction, after 185 MBq 18F-FDG was injected. Images were visually graded on a 5-point scale from "definitely malignant" to "definitely benign," and lesion-to-background (LB) ratios were calculated using region of interest analysis. Visual and semiquantitative analyses were compared using receiver operating characteristic analysis. RESULTS: Twenty lesions were benign and 57 were malignant. The mean LB ratio for benign lesions was 1.5 (range 1.0-5.7) and for malignant lesions 5.7 (range 1.2-14.1) (p < 0.001). The area under the ROC curve for LB ratio analysis was 0.95, and for visual analysis 0.91 (p = 0.39). The optimal cut-off ratio with LB ratio analysis was 1.8, giving a sensitivity of 95% and a specificity of 85%. For lesions thought to be "definitely malignant" on visual analysis, the sensitivity was 93% and the specificity 85%. Three proven infective lesions were rated as malignant by both techniques (LB ratio 2.6-5.7). CONCLUSIONS: FDG PET without attenuation correction is accurate for differentiating between benign and malignant lung nodules. Results using simple LB ratios without attenuation correction compare favourably with the published sensitivity and specificity for standard uptake ratios. Visual analysis is equally accurate.
has subject area