An ICA Based Noise Reduction for PET Reconstructed Images
Xian‐Hua Han, Yen‐Wei Chen, Keishi Kitamura, Akihiro Ishikawa, Yoshihiro H. Inoue, Kouichi Shibata, Yukio Mishina, Yoshihiro Mukuta · 2007
The reduction of noise in medical images is an important issue. In this paper, we propose a new ICA-based filter for reduction of noise in reconstruction domain. In the proposed filter, the reconstructed 3D PET images(X-Y plane-slice domain or X-Z plane) are firstly transformed to ICA domain, and then, the components of noise information are removed by a soft thresholding (shrinkage). In this study, the choice of ICA basis function trained from noisy reconstructed images in different plane is considered. The de-noised results with different ICA basis functions and conventional denoising method (wavelet shrinkage and Gaussian filter) are given for comparison. Experimental results show that the reconstructed images of ICA-based denoised images are much clearer and have much better contrast than those with wavelet-domain filters.