Cascade pulse coupled neural network for multimodal medical image fusion

This paper proposed a novel cascade pulse coupled neural network (CPCNN) with two-layer structure, which is used to multimodal medical image fusion. The first layer of CPCNN contains m single-channel PCNNs, which is used to calculate weighted coefficients for the second layer of CPCNN. The second layer of CPCNN is a multi-channel PCNN, which is used to fuse the source images. The proposed CPCNN model exploits the advantages of both the single-channel PCNN and multi-channel PCNN to obtain better fusion results. It retains the same fusing speed as multi-channel PCNN and achieves a better result similar to the single-channel PCNN and wavelet transform based method. Experimental results showed the better performance of CPCNN in both visual effect and objective evaluation criteria.