Effect of attenuation correction on lesion detectability in FDG PET of breast cancer

The aim of this study was to compare the visual analysis of attenuation-cor rected and noncorrected F-18-fluoro-2-deoxy-D-glucose (FDG) PET images in p atients with primary or metastatic breast cancer using standardized film do cumentation and to evaluate the influence of attenuation correction on lesi on detectability. Methods: Standard FDG PET of the breasts and of the axill ary regions was performed on 28 women with breast cancer. Transmission scan s were acquired for attenuation correction after administration of FDG. Tra nsverse and coronal slices and maximum intensity projections both with and without attenuation correction were documented in a standardized manner on film. Noncorrected images were displayed with an upper threshold of five ti mes the mean activity in normal lung tissue. Attenuation-corrected images w ere documented with an upper threshold of a standardized uptake value of fi ve. Two independent nuclear medicine physicians, who were unaware of the re sults of clinical investigation, other imaging modalities and histopatholog ic findings, interpreted the images visually, noncorrected images first. Re sults: One hundred eighty-four of 189 lesions in 28 of 28 patients were fou nd on attenuation-corrected and noncorrected images. Seventeen lesions were found in the breasts of 12 patients, In 18 patients, 31 axillary lesions w ere round. Moreover, 141 lesions representing distant metastases were detec ted in 18 patients. Attenuation-corrected images showed the same lesions in all patients but 2, in whom 5 of 189 small pulmonary lesions (2.6%) were n ot detected. Iterative reconstruction did not improve detectability of thes e lesions on attenuation-corrected images. These lesions were confirmed by CT, which revealed diameters of <1 cm. Conclusion: Attenuation correction b y transmission measurement after injection may impair lesion detectability in PET for staging of breast cancer patients. When using the image modaliti es described, noncorrected PET images should be considered in image analysi s.