Table 8 Statistical significance analysis of TDN vs. Other methods.
From: Multimodal medical image fusion combining saliency perception and generative adversarial network
Metric | Method | Mean ± std. dev. | p-value | 95% confidence interval (CI) |
|---|---|---|---|---|
Accuracy | MMIF-NSST | 0.9145 ± 0.008 | 0.0021 | [0.9102, 0.9188] |
MedFusionGAN | 0.9254 ± 0.007 | 0.0016 | [0.9213, 0.9295] | |
GIAE-Net | 0.9358 ± 0.006 | 0.0009 | [0.9324, 0.9392] | |
TDN (proposed) | 0.9457 ± 0.005 | – | [0.9428, 0.9486] | |
Precision | MMIF-NSST | 0.897 ± 0.009 | 0.0032 | [0.8911, 0.9029] |
MedFusionGAN | 0.904 ± 0.007 | 0.0025 | [0.8993, 0.9087] | |
GIAE-Net | 0.918 ± 0.006 | 0.0011 | [0.9144, 0.9216] | |
TDN (proposed) | 0.9274 ± 0.005 | – | [0.9245, 0.9303] | |
Sensitivity | MMIF-NSST | 0.912 ± 0.010 | 0.0045 | [0.9055, 0.9185] |
MedFusionGAN | 0.924 ± 0.008 | 0.0038 | [0.9187, 0.9293] | |
GIAE-Net | 0.928 ± 0.007 | 0.0029 | [0.9234, 0.9326] | |
TDN (proposed) | 0.934 ± 0.006 | – | [0.9305, 0.9375] | |
Specificity | MMIF-NSST | 0.906 ± 0.009 | 0.0052 | [0.9005, 0.9115] |
MedFusionGAN | 0.914 ± 0.008 | 0.0041 | [0.9086, 0.9194] | |
GIAE-Net | 0.927 ± 0.007 | 0.0028 | [0.9225, 0.9315] | |
TDN (proposed) | 0.934 ± 0.006 | – | [0.9305, 0.9375] | |
Computational efficiency | MMIF-NSST | 6.2 min | 5.2 h | 0.48 s |
MedFusionGAN | 7.8 min | 6.8 h | 0.42 s | |
GIAE-Net | 8.4 min | 7.3 h | 0.38 s | |
TDN (proposed) | 5.1 min | 4.3 h | 0.31 s |
