Transfer Learning Based Fusion of CT and MR Images

This paper presents a method for fusing MR and CT images to create a high-contrast fused image that integrates complementary information from both input images. The proposed approach employs wavelet transform to decompose the CT and MR images. Next, features are extracted from the decomposed images by applying transfer learning from the VGG19 model. The fusion of these decomposed images is performed based on the extracted features and the fusion rules outlined in this study. An inverse wavelet transformation is then applied to the fused decomposed images, resulting in a final fused image that encompasses information from both the CT and MR images. The effectiveness of the proposed method is evaluated and compared with five other advanced fusion techniques through both visual assessment and objective measures.