Figure 1.
Illustrated overview of the retrospective MC method for dynamic PET/CT dynamic scan.
Table 1.
Relative positions of the Hoffman brain phantom for different CT scans and PET frames.
Figure 2.
Co-registration accuracy evaluated from FDG phantom study.
A. PET to PET co-registration accuracy B. CT to PET co-registration accuracy.
Figure 3.
Comparison of AC with slightly misaligned CT and fully aligned CT in FDG phantom study.
Panel A: the difference map between PET FR4 AC with a 2(CT1) and PET FR1 AC with an aligned CT (CT1) (taken as the reference PET frame). The small misalignment (2 mm) between PET and CT has subtle impact on AC comparing to Panel B.; Panel B: the difference map between PET FR12 AC with an aligned CT (CT12) and the reference. The difference map shows only the noise.
Figure 4.
Artifacts of AC with largely misaligned CTs in FDG phantom study.
Panel A: the difference map between PET FR12 AC with a 10 mm mismatched CT (CT1) and the reference PET image. The misalignment between PET and CT caused significant AC artifacts; Panel B: the difference map between PET Frame12 AC with a 10 mm mismatched CT (CT1) and the same PET frame AC using the physically aligned CT (CT12). The misalignment between PET and CT caused significant AC artifacts. However, the AC matrix did not add extra noise to the PET images.
Figure 5.
Comparison of AC with aligned CT and MC CT in FDG phantom study.
Panel A: the difference map between PET FR12 AC with movement corrected CT1 and the reference image. After MC, there were no noticeable differences, except noise, among the PET frames; Panel B: the difference map between PET FR12 AC with movement corrected CT1 and the same PET frame with the correctly aligned CT (CT12).
Figure 6.
Movement correction for PET/CT misalignment induced by head translation and rotation in FDG phantom study.
Figure 7.
Displacement during dynamic FDDNP PET/CT scan.
Half of the subjects have displacements over 5(after 40 min).
Figure 8.
Improvement of a late FDDNP PET/CT frame image after MC.
Panel A: An FDDNP PET/CT study on a Down syndrome patient with 10 mm movement in z direction in a late frame of a dynamic scan. Panel B: 3-D orthogonal images show the improvement of the PET/CT images with proper AC after MC. All images are displayed with the same color scale and show clearly the quantitative differences in the PET images reconstructed with misaligned and aligned CT images.
Figure 9.
TACs in both medial temporal region and cerebellum (the reference region), and Logan plot showed less fluctuation after MC for a subject who had large head movement during a dynamic FDDNP PET/CT scan.
A. Comparisons of TACs before and after MC in medial temporal region. B. Comparisons of TACs before and after MC in cerebellum. C. The slope of the Logan plot increased by 7.5% after MC for this patient.
Figure 10.
A FDDNP PET/CT scan subject with both head nodding and shaking movement showed reduction in artifacts in the temporal region, reduction in left-right asymmetry and better hemisphere separation after MC.
Panels A and B show different brain slices of the same subject before and after MC.
Figure 11.
Boxplot showing difference of DVR in Regions of interest (ROIs) before and after MC.
A. in absolute value and B. in percentage of absolute value. ROIs include frontal cortex (FRT), lateral temporal lobe (LTL), medial temporal lobe (MTL), parietal cortex (PAR) and posterior cingulate (PCG).
Figure 12.
Improvement of a FDG PET/CT scan of a DS patient after MC in aspects of late frame, Ki image, image derived input function and Patlak Plot.
A. Orthogonal PET/CT images before and after MC of a late frame in the dynamic brain FDG scan: attenuation correction artifact in the right frontal region (pointed at by arrows) diminished after MC (lower row). B. The Ki image: The Ki values appeared to be more uniform (except the affected regions) and globally higher after MC. In this case, the lower input function accounted for the global increase in the Ki values C. Image Derived Input Function: After MC, a 21% drop in the tail AUC during 30–65 minutes was seen. D. Patlak Plot in the medial temporal region before and after MC: The Patlak plot had less fluctuation after MC. The slope of the plot used for calculating Ki increased by 97.9% after MC for this patient.