Article Figures & Data
Figures
- FIGURE 1.
(A) Simple image consisting of 2 intensity values. (B) Profile drawn through image.
- FIGURE 2.
(A) Threshold T = 60 set between intensity values of 20 and 100. (B and C) Images after image partitioning using threshold T: P1 (B) and P2 (C).
- FIGURE 3.
Low-pass Butterworth filter applied to partitioned images separately: filtered P1 (A) and filtered P2 (B).
- FIGURE 4.
Filtered images. (A) Filter is directly applied to original image shown in Figure 1. (B) Filter is applied to partitioned images shown in Figure 3, and filtered images are recombined according to their membership maps. (C) Close-up profiles to compare the 2 methods. Solid curve corresponds to direct filtering method, and dashed curve corresponds to proposed method.
- FIGURE 5.
(A) SPECT image of human torso with external marker (arrow). (B) Result from applying low-pass Butterworth filter to SPECT image shown in A. Large Gibbs phenomenon or decaying oscillation is seen around marker. (C) Result from applying proposed partitioned image filtering technique using same low-pass filter. Decaying oscillation around marker is reduced.
- FIGURE 6.
Example of partitioned filtering of patient bone SPECT image: raw filtered backprojection reconstruction (A), direct low-pass Butterworth filtering (B), and proposed partitioned low-pass Butterworth filtering (C). D, E, and F are zoom-in subimages of A, B, and C, respectively.
