Table 2.
Numerical results of different methods on brain and pelvis sCT images.
| Method | Brain | Pelvis | ||||
|---|---|---|---|---|---|---|
| MAE (HU) | PSNR (dB) | NCC | MAE (HU) | PSNR (dB) | NCC | |
| RF‐based | 69.8 ± 15.2 | 24.41 ± 1.71 | 0.955 ± 0.002 | 69.7 ± 19.7 | 24.25 ± 2.20 | 0.893 ± 0.026 |
| GAN‐based | 66.9 ± 15.6 | 25.10 ± 2.02 | 0.937 ± 0.021 | 74.7 ± 20.0 | 22.08 ± 2.7 | 0.877 ± 0.053 |
| 2D Cycle GAN‐based | 59.0 ± 11.9 | 25.75 ± 1.81 | 0.953 ± 0.009 | 65.4 ± 18.6 | 23.45 ± 2.97 | 0.903 ± 0.037 |
| The proposed | 55.7 ± 9.4 | 26.59 ± 2.27 | 0.963 ± 0.008 | 50.8 ± 15.5 | 24.45 ± 2.64 | 0.929 ± 0.028 |
GAN, Generative adversarial networks; HU, Hounsfield units; MAE, mean absolute error; NCC, normalized cross correlation; PSNR, peak signal‐to‐noise ratio; CT, synthetic computed tomography.