Table 7 Technical comparison of multimodal medical image fusion methods.
From: Multimodal medical image fusion combining saliency perception and generative adversarial network
Method | Approach | Key features | Advantages | Limitations |
|---|---|---|---|---|
Proposed (TDN) | Temporal Decomposition + GAN | Saliency perception, generative adversarial learning | High fusion accuracy, reduced noise, robust feature extraction | Computational complexity |
MMIF-NSST36 | NSST-based multimodal fusion | Structural & spectral feature enhancement | Preserves structural details | Less effective in texture preservation |
Jiang et al.21 | Intuitionistic fuzzy sets + Laplacian pyramid | Lightweight, similarity-based fusion | Computationally efficient | Lower performance in complex textures |
GIAE-Net28 | Gradient-intensity oriented network | Gradient-based multimodal fusion | High precision and specificity | High computational cost |
TUFusion24 | Transformer-based universal fusion | CNN + Transformer | Strong generalization, high efficiency | Requires large training data |
AMMNet25 | Attention-based MobileNetV3 | Multiscale convolutional layers | Sharp image texture, lightweight | Lower contrast preservation |
