Accuracy of whole-body FDG-PET/CT for detecting brain metastases from non-central nervous system tumors
- 235 Downloads
Positron emission tomography (PET) using 18F-fluoro-2-deoxy-d-glucose (FDG) has a limitation in detecting cerebral metastases; however, the feasibility of detection by inline PET/computed tomography (CT) system remains unknown. We evaluated the accuracy of FDG-PET/CT of body imaging protocol for the detection of cerebral metastases when compared with PET alone and CT alone.
Fifty patients underwent whole-body FDG-PET/CT scanning including the brain and contrastenhanced brain MR (magnetic resonance) scan. PET-only, CT-only, and the fused images were interpreted, and the confidence of presence of cerebral metastases was recorded using a five-point grading scale. Area under the receiver-operating characteristic (ROC) curve (Az) was calculated. Differences among the three modalities were tested with the Cochran-Q test, followed by multiple comparisons using the McNemar test with Bonferroni adjustment.
Magnetic resonance imaging revealed 70 cerebral metastatic lesions in 20 patients. Patient-based analysis showed that the sensitivity, specificity, accuracy, and Az of PET-alone interpretation were 45%, 80%, 66%, and 0.6025, respectively, those of CT-alone interpretation were 50%, 97%, 78%, and 0.7158, respectively, and those of fused-image interpretation were 50%, 93%, 76%, and 0.7242, respectively. ROC analysis revealed significant differences among the three interpretation methods (P = 0.0238) and between PET and PET/CT (P = 0.0129). The sensitivity of PET, CT, and fused-image interpretation for detecting 70 lesions was 13%, 20%, and 20%, respectively.
Even with an integrated PET/CT scanner of body imaging protocol, the sensitivity of cerebral metastases remained unsatisfactory. To assess intracranial lesions, MR scanning should still be considered.
KeywordsBrain Metastasis FDG PET CT
Unable to display preview. Download preview PDF.
- 8.Cleves M. Receiver operating characteristic (ROC) analysis. Stata Tech Bull 1999;52:19–33 [reprinted in Stata Tech Bull Reprints, vol. 9, p. 212–29].Google Scholar