Figure 1.
Two MSE plots or parameter selections of the neighbor window (A) and the controlling parameters (B) of KIBF algorithm.
Figure 2.
The 18F-FDG PET simulation settings.
(A) A brain phantom composed of gray matter, white matter and a small tumor; (B) the blood input function and regional time activity curves.
Table 1.
Kinetic parameters used in the 18F-FDG PET simulation.
Figure 3.
MSE plots for parameter selections of the standard deviation for GF algorithm (A) and the controlling parameters of BF algorithm (B).
Figure 4.
The ground truth and the activity images reconstructed by different algorithms at frames #6, #16, and #26.
(A) are the ground truth; (B) are the results from the direct FBP reconstruction; (C) are the results from the FBP images filtered by the GF algorithm ( voxel); (D) are the results from the FBP images filtered by the BF algorithm (
voxel,
); and (E) are the results from the FBP images filtered by the present KIBF algorithm (
voxel,
). All images are with a same display window.
Figure 5.
Three TAC plots from different algorithms for the corresponding voxels located in the gray matter (A), white matter (B) and small tumor regions (C), respectively.
Figure 6.
Box plots of the mean activities with standard deviations in the gray matter, white matter and small tumor regions from different algorithms at (A) the frame #6; (B) the frame #16; and (C) the frame #26.
Figure 7.
PSNRs at each frame of the activity images reconstructed by different algorithms.
Table 2.
The TSNRs of the activity images reconstructed by different algorithms.
Figure 8.
The ground truth and the estimated parametric images from the activity images reconstructed by different algorithms.
(A) is the ground truth; (B) is the result from the direct FBP reconstruction; (C) is the result from the FBP image filtered by the GF algorithm ( voxel); (D) is the result from the FBP image filtered by the BF algorithm (
voxel,
); and (E) is the result from the FBP image filtered by the present KIBF algorithm (
voxel,
). All images are with a same display window.
Figure 9.
Box plots of the mean value of with standard deviations in the gray matter, white matter and small tumor regions from different algorithms.
Figure 10.
The NSD versus Bias tradeoff curves of the influx rate estimated by the GF, BF and KIBF algorithms for different ROIs in the brain phantom.
Figure 11.
The parametric images estimated by different algorithms.
(A) is the result from the direct OSEM reconstruction; (B) is the result from the OSEM image filtered by the GF algorithm ( voxel); (C) is the result from the OSEM image filtered by the BF algorithm (
voxel,
); and (D) the result is from the OSEM image filtered by the KIBF algorithm (
voxel,
). All images are with a same display window.
Figure 12.
Two TAC plots from different algorithms for the corresponding voxels located in the ROI1 (A) and ROI2 (B), respectively.