Image preprocessing

[ginobilinie]

Image preprocessing

[ginobilinie]

@wangken1994 Thanks for your interest. Personally, I donot involve in the data preprocessing, but i have asked the guy who did this work, and I'll let you know if he answered me.

Thanks.

[ginobilinie]

@wangken1994 We use flirt dof=6 for registration. We use rigid registration here and donot use non-rigid registration.

[wangken1994]

Thank you ,Is the 3T image of each individual registered to the 7T image? Or register data from all training sets to a standard space such as ICBM152? Is the input of the 2D model a slice in one direction? Such as axial。Is the input to the 3D model the .nii data for the entire brain? I am a newbie, thank you very much for answering my question.

[ginobilinie]

The 3T image of each individual is registered to the corresponding 7T image.

[wangken1994]

@ginobilinie Thank you very much.

[wangken1994]

Hellow,
What dose dFA ，dSeg ，step mean in extract23DPatch4MultiModalImg.py？

[wangken1994]

What is the data size of MR and CT data？ 15319350？

[ginobilinie]

dFA is one of the input modality, dSeg is the output Modality, step size is the stride during patch extraction in a sliding window manner.
MR/CT size: 256x256x(120-320)

[wangken1994]

hellow

Why MRI is [5,64,64],ct[1,64,64]?

[wangken1994]

Both of them should be [1,64,64] 。 Did I get it wrong？

[ginobilinie]

@wangken1994 I'd like to make use 5 slices of source modality to predict the corresponding middle slice of the target slice. That's why I set 5x64x64->1x64x64. Actually, if your gpu allows, you can set up large patches, like 5x240x240->1x240x240