OBJECTIVE: To investigate a novel deep-learning network that synthesizes virtual contrast-enhanced T1-weighted (vceT1w) magnetic resonance images (MRI) from multimodality contrast-free MR images for nasopharyngeal carcinoma (NPC) patients.
METHODS: This paper presents a retrospective analysis of multi-parametric MRI, with and without contrast enhancement by gadolinium-based contrast agents (GBCAs), obtained from 64 biopsy-proven NPC patients treated at XXXX. A multimodality-guided synergistic network (MMgSN-Net) was developed to leverage complementary information between contrast-free T1-weighted and T2-weighted MRI for vceT1w MRI synthesis. 35 patients were randomly selected for model training, whereas 29 patients were employed for model testing. The synthetic images generated from MMgSN-Net were quantitatively evaluated against real GBCA-enhanced T1w MR images using a series of statistical evaluating metrics, which include mean absolute error (MAE), mean squared error (MSE), structural similarity index (SSIM) and peak signal-to-noise ratio (PSNR). Qualitative visual assessment between the real and synthetic MRI was also performed. Effectiveness of our MMgSN-Net was compared with three state-of-the-art deep-learning networks, including U-Net, CycleGAN, and Hi-Net, both quantitatively and qualitatively. Further, a Turing test was carried out by seven board-certified radiation oncologists from four hospitals for assessing authenticity of the synthesized vceT1w MR images against the real GBCA-enhanced T1w MRI.
RESULTS: Results from the quantitative evaluations demonstrated that our MMgSN-Net outperformed U-Net, CycleGAN and Hi-Net, yielding the top-ranked scores in averaged MAE (44.50 ± 13.01), MSE (9193.22 ± 5405.00), SSIM (0.887 ± 0.042), and PSNR (33.17 ± 2.14). Further, the mean accuracy of the seven readers in the Turing tests was determined to be 49.43%, equivalent to random guessing (i.e., 50%) in distinguishing between real GBCA-enhanced T1-weighted and synthetic vceT1w MRI. Qualitative evaluation indicated that MMgSN-Net gave the best approximation to the ground-truth images, particularly in visualization of tumor-to-muscle interface and the intra-tumor texture information.
CONCLUSIONS: Our MMgSN-Net was capable of synthesizing highly realistic vceT1w MRI that outperformed the three comparing state-of-the-art networks.