Variational Model on Wavelet Domain for PET/CT Image Fusion

The objective of the study is to find a new algorithm for PET/CT image fusion. A variational model was used based on the wavelet transform. Firstly, PET and CT images were decomposed using wavelet transform. Then, images in approximate channel and detail channel were fused according to the two proposed assumption. Finally, decomposed images were synthesized to form fused image. Compared with the results form MATLAB wavelet fusion toolbox, the experimental results showed that the new variational image fusion model could provide more accurate result for target location in radiotherapy planning. According to experiments, the new algorithm can reach good results and meet requirement of clinical demands.