“An important research achievement has been made in the field of medical image fusion. In response to the problems of unstable training and insufficient ability to extract semantic information from medical image fusion methods based on generative adversarial networks (GANs), the research team proposes a dual coupled interactive fusion GAN (DCIF-GAN). This innovative solution performs well in integrating PET/CT medical images of lung tumors, effectively improving image quality. DCIF-GAN achieves coupling between the generator and discriminator by designing a dual generator dual discriminator structure, and strengthens interactive fusion through a global self attention mechanism. At the same time, the feature extraction and reconstruction module coupled with CNN Transformer is introduced, which significantly improves the ability to extract the internal local and global feature information of the same modal image. In addition, the application of cross modal interactive fusion module further integrates global interaction information between different modalities. Experimental verification shows that DCIF-GAN has improved the average gradient, spatial frequency, structural similarity, standard deviation, peak signal-to-noise ratio, and information entropy by 1.38%, 0.39%, 29.05%, 30.23%, 0.18%, and 4.63%, respectively, compared to the optimal method among the other four methods. This achievement not only highlights the information of the lesion area, but also makes the fused image structure clearer and the texture details richer. This study provides a new solution for the field of medical image fusion and is expected to open up new directions for the development of computer-aided diagnostic technology.”