Medical Image Fusion Based a Densely Connected Convolutional Networks

Image fusion method based on deep learning has achieved great success. In order to obtain better performance, researchers keep deepening the network structure, but it also brings greater computational burden. Compared with the traditional convolutional structure, Dense Convolutional Network (DenseNet) achieves a good balance between performance and computational complexity. Therefore, in this paper, we present a novel medical image fusion method based on DenseNet, which achieves feature reuse by connecting features over channels, and enables the algorithm to achieve better performance than traditional networks with fewer parameters and calculation costs. The experimental results show that, on the premise that the convolution structure layers of the proposed method is far less than that of the state-of-the-art deep learning method, the fused image details are still well preserved, and the objective assessment indices are better than that of the traditional transformation domain methods and close to the state-of-the-art deep learning method.