Image synthesis method from MRI (Magnetic Resonance Imaging) to cerebral blood volume quantitative image for brain tumor

• Main Authors: XIA YONG, MA HAOJIE, WANG BAO, LIU YINGCHAO, PAN YONGSHENG
• Format: Patent
• Language: Chinese; English
• Published: 29.10.2024
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; PHYSICS

The invention discloses an image synthesis method for brain tumor from MRI to cerebral blood volume quantitative images, comprising: 1) a GAN-based image synthesis algorithm which mainly comprises a GAN network based on feature consistency and a three-dimensional encoder-decoder network with mean absolute error loss; and 2) a 3D (three-dimensional) incremental Encoder-Decoder network, wherein the network mainly comprises an encoder and a decoder facing a CBV (Cone Beam Value) mode for each mode. According to the method, the GAN-based image synthesis algorithm is provided, and the magnetic resonance imaging perfusion graph can be generated from the easily acquired standard MRI sequence, so that the enforceability of the method for distinguishing the tumor progress from the radiation effect by improving the perfusion cerebral blood volume is improved. Meanwhile, the 3D incremental Encoder-Decoder network can learn from a plurality of subjects with only a small number of ASL-CBF paired images, so that the CBV im