We have developed and validated a MHDgN-Net for gadolinium-free contrast-enhanced MRI (GFCE-MRI) synthesis in patients with nasopharyngeal carcinoma (NPC). The developed MHDgN-Net was featured with high generalizability. We first modelled the MHDgN-Net using three hospital datasets to improve the diversity of training samples. Then, the external hospital data was matched to the distribution of training dataset by EDM. Compared to traditional models, the proposed MHDgN-Net can accurately enhance tumor and significantly improve the quality of GFCE-MRI when applying to external hospital data. This technique holds great potential in providing a generalizable gadolinium-free tumor enhancement alternative on data from other hospitals.
This abstract and the presentation materials are available to members only; a login is required.