Medical Image Fusion Based on Scale-Aware Structure-Preserving Texture Filter and Multi-scale Decomposition

Medical image fusion technology plays an important role in clinical applications as diagnoses. Medical image fusion refers to the fusion of pairs of image information. In this paper, we propose a multi-scale medical image fusion method based on the scale-aware structure-preserving texture filter. Our method performs the multi-scale decomposition of input images (SPECT/MRI) by using the scale-aware structure-preserving texture filter, from which we obtain the base information and feature information content. Correspondingly, we use different fusion strategies in the different layers. For the base information, we consider entropy as the weight and perform a weighted summation. For the feature information, we use a simple and effective fusion rule, namely, a filtering operation is performed on the feature image, and thus obtain the amount of significant information. Then, we calculate the weighted average for the feature information of multiple layers. The subjective and objective experiments demonstrate the superiority of our method over five state-of-the-art fusion methods.