System and methods for precision tumor delineation in cancer treatment or diagnosis for subjects with carcinoma based on contrast agent-free, virtual contrast-enhanced mri data

A system for precision tumor delineation in cancer treatment and diagnosis for subjects in need thereof based on synthesis of contrast agent-free, virtual MRI data, the system comprises:

• a data acquisition module acquiring MRI data from at least two medical institutions, the acquired MRI data comprising contrast agent-free T1-weighted (T1w), T2-weighted (T2w) MRI data, and contrast agent-enhanced MRI (CE-MRI) data;
• a data pre-processing module pre-processing the acquired MRI data comprising resampling thereof by standardizing image size of different MRI data acquired from different medical institutions and subjecting the resampled MRI data to normalization or standardization, said normalization or standardization comprising normalization or standardization based on a whole image dataset, a single image data, or image data obtained from a single subject;
• a model training module utilizing the T1w-MRI and T2w-MRI data from a training dataset containing some of the T1w-MRI, T2w-MRI and CE-MRI data after said pre-processing as an input of the neural network while utilizing the CE-MRI data from the training dataset as a learning target of the neural network only for model training and internal evaluation;
• and
• a virtual image generation module utilizing contrast agent-free MRI data from at least one external medical institution for external evaluation prior to generating a virtual contrast-enhanced MRI (VCE-MRI) data, said at least one external medical institution being different from the at least two medical institutions.