Improved attenuation correction via appropriate selection of respiratory-correlated PET data

PURPOSE We propose a respiratory-correlated PET data processing method (called "BH-CT-based") based on breath-hold CT acquisition to reduce the smearing effect and improve the attenuation correction. The resulting images are compared with the ungated PET images acquired using a standard, free-breathing clinical protocol. METHODS The BH-CT-based method consisted of a list-mode acquisition with simultaneous respiratory signal recording. An additional breath-hold CT acquisition was also performed in order to define a tissue position from which PET events can be selected. A phantom study featured a 0.5-ml sphere (filled with 18F-fluorodeoxyglucose ((18)F-FDG) solution) pushed onto a rubber balloon (filled with (18)F-FDG solution and iodinated contrast agent). The feasibility of the BH-CT-based method was also assessed in two patients. RESULTS In the phantom study, the contrast-to-noise ratios (CNRs) were -1.6 for the Ungated volume and 5.1 for the BH-CT-based volume. For patients, CNRs were higher for BH-CT-based volumes than those for Ungated volumes (17.3 vs. 6.3 and 7.3 vs. 3.8, for patients 1 and 2, respectively). Bias-variance measurements were performed and yielded bias reduction of 40% with BH-CT-based. CONCLUSION The application of a BH-CT-based method decreases motion bias in PET images. This method resolves issues related to both PET-to-CT misregistration and erroneous attenuation correction and increases lesion detectability.