Abstract
69
Objectives MRI has high sensitivity for detecting pelvic disease, with the tradeoff of decreased specificity. We examined the effect of fused 18F-FDG PET with MRI on pelvic lesion characterization.
Methods IRB approval was obtained for this retrospective study. Our institution’s electronic medical record was queried for whole body PET and pelvic MRI studies performed within 2 months of each other and between Jan 2007 and June 2011. 250 pairs of PET and MRI studies were identified in 250 patients. Each pair of PET and MRI reports was reviewed for discordant findings, defined as any pelvic lesion identified on either PET or MRI but not on both. Each discordant lesion was evaluated on fused PET/MRI images for qualitative FDG-avidity, SUV, lesion size, MRI contrast enhancement, and presence of restricted diffusion by two radiologists in a joint readout. The PET and MRI datasets were registered manually using strict anatomic criteria on a Syngo TrueD workstation (Siemens, Malvern, PA). Specificity of MRI and fused PET/MRI was calculated on the basis of pathology reports and/or at least 6 months of clinical and radiological follow-up.
Results A total of 57 discordant lesions were identified, including 37 osseous lesions, 12 lymph nodes, 7 ovarian masses, 3 colonic findings, and 2 uterine findings. Eight findings were detected by FDG PET only (14%), and 49 lesions were detected by MRI only (86%). Fused images improved concordance in three lesions. Two lesions were false negatives on PET (Figure 1). Fused PET/MRI helped localize a third lesion, a focus of PET uptake, to an external iliac lymph node on MRI. Among this cohort of lesions, the specificity of MRI alone versus fused PET/MRI was 39% and 94% respectively.
Conclusions Integrated FDG-PET and MRI with software fusion can increase the concordance rate of pelvic PET and MRI findings and can increase the specificity of MRI. Additional data are needed to confirm the statistical significance of these preliminary findings