search

icon-lab / ResViT

Official Implementation of ResViT: Residual Vision Transformers for Multi-modal Medical Image Synthesis
Other
127 stars 26 forks source link

Hello,I have a question #4

Closed dontknow123456 closed 2 years ago

dontknow123456 commented 2 years ago

In the first issue of icon-lab/Resvit，you gave this image image I know that the left is source and the right is target. Could you please tell me which one is t1，t2 or flair ？

dontknow123456 commented 2 years ago

Hi，professor. For the many to one task, could you please tell me that source and target image are still lying side by side? For example, this? Brats18_CBICA_ALU_1_t1_t2_flair_98

onat-dalmaz commented 2 years ago

Hello, For many-to-one tasks, the source modalities should be embedded in the channels of the source image. For example: 2491 In this one, T1 and T2-weighted images are embedded into red and green channels of the source image, whereas the target FLAIR image is embedded into green channel of the target image.

dontknow123456 commented 2 years ago

thanks a lot！