Template based attenuation correction for PET in MR-PET scanners

This work investigates a template based procedure for attenuation correction (TBA) of PET scans acquired in future hybrid MR-PET scanners which will not offer a measured attenuation correction. A previous report of our group described a method (TBA-SPM) how individual attenuation maps can be obtained from an attenuation template which is spatially normalized to the SPM2 standard brain shape. Attenuation maps of females and males obtained from PET transmission scans were used as input for this template. The present study replaces the template referring to SPM2 by a female and a male attenuation template (fAT and mAT), each based on four measured attenuation images (TBA-f&m). The corresponding T1-MR templates (fMR and mMR) were also available. Thus, possible morphological gender-related differences, not considered when the standardized SPM2
¹
brain shape is used, may be taken into account. To examine this approach PET scans of 15 female and 15 male subjects of an ongoing study were attenuation corrected using the templates fAT and mAT. For this purpose and depending on the subject’s gender the fMR or mMR templates were warped onto the individual MR image. The resulting warping matrix was applied to fAT or mAT, respectively. The individualized attenuation maps were used to reconstruct the PET emission data. These were compared to PET images attenuation corrected with the conventional PET based transmission data (PBA). While the relative differences between PBA and TBA=f&m reconstructed images averaged over each group and all regions of interest were 0.57% ± 3.76% for females and −0.59% ± 3.56% for males, the corresponding values obtained with the TBA-SPM method showed an overestimation with similar standard deviations (2.39% ± 3.76% for females and 2.42% ± 3.37% for males). In conclusion, the alternative gender-related template method TBA-f&m gives acceptable results with no significant differences between the genders.