Fig 1.
Overview of the workflow of reconstruction of PET images using different attenuation correction modes.
Fig 2.
Example of a subtraction image of the MR-based μ-map and the registered CT-based μ-map of a patient in which the LACs from the lung and the bone tissue were replaced by the LACs from the MR- μ-map in axial view (A), sagittal view (B) and coronal view (C). The histogram analysis of all patients shows that most of the voxels within the patient from the subtraction images provide HU of around 0 (>90% of the voxels show a deviation of more than 10% from the maximum value (0.1cm-1)) which stands for the high precision of the performed registration.
Fig 3.
Example of ROI placements in the basal lung parts.
Left hand side: ROIs superposed with coregistered μ-map from CT. Right hand side: ROIs superposed with the correlating PET.
Table 1.
Overview of mean values and SD of measured Hounsfield units (HU) of CTs from PET/CT in different lung regions.
A gradient from anterior to posterior can be seen in both contrast-enhanced and non-contrast-enhanced CTs, pronounced between the middle and posterior regions. Values of contrast-enhanced CTs and non-contrast-enhanced CTs as well as the differences between the anterior and posterior regions did not differ significantly.
Table 2.
SUVmean including standard deviations in different regions and different PET images.
P-values describe the significance of differences between the anterior and posterior lung regions (p-value a.-p.) in the respective PET.
Fig 4.
Typical example of an axial view of the right lung of a patient to demonstrate the effect of the replacement of the CT-based LACs of lung tissue (μ-map CT) with the LAC from the MR-based segmentation method from the Biograph mMR (μ-map CT_MRLUNG).
To show the differences of LACs, a subtraction map is given in the upper row on the right-hand side (μ-map CT_MRLUNG—CT); as one can see, the subtraction of μ-map CT from μ-map CT_MRLUNG leads to an underestimation of LACs in the posterior regions and an overestimation in the anterior regions (given in LACs). This results in higher SUVs in PET_CTAC as compared to PET_CTAC_MRLUNG in the posterior regions, or, in other words, an underestimation of SUVs in the posterior regions of PET_CTAC_MRLUNG as shown in the subtraction image PET CTAC_MRLUNG–CTAC.
Table 3.
Overview of relative differences in % between the differently reconstructed PETs and regions.
Note the differences of SUV-underestimation in the posterior lung parts while the overestimations in the anterior regions do not change significantly. P-Values describe the significance of differences between the particular lung regions in the different PET images. The SUVs of PETCTAC_MRLUNG_NOBONE and PETMRAC images did not differ significantly in any region.