Post-reconstruction filtering of SPECT images by variable conductance diffusion

The authors investigate the 3-D post-reconstruction filtering of single-photon emission computed tomographic (SPECT) images with variable conductance diffusion (VCD) filters. The idea behind VCD is to smooth the noisy image without blurring the edges. In order to get a better estimate of the conductance function used in VCD, the gradient estimated from the Gaussian blurred image, instead of the image itself, is used. The authors also propose a new method of spatially varying the parameter k of the conductance function which permits smoothing within the region but not across the edges. Although the new method does not outperform the traditional one, further refinement may prove it to be useful. Experiments show that the VCD filters remove considerable noise from the noisy images and maintain the edge information at the same time