Hi,
Thank you for your outstanding work and sharing the code. I have some certain clarifications and I will appreciate your response,
I am working on 3D CT images where each CT has a dimension of 512 x 512 x N , N is number of axial slices. As part of image pre-processing as was written in your paper
a. For 3D images. for each slice, I need to resize the image dimension to 256 x 256 x 3, correct?
b. Does that mean even for 3D images and for inference, segmentation (and dice) was performed slice-by-slice and then averaged to get the volumetric dice?
c. I show one of the example data that you provided. For AbomenCT-1K and CT_Liver, file names are following,
"casenum_00371_sliceid_50.npy", "casenum_00944_sliceid_422.npy". I am assuming you just uploaded only one slice for a particular patient.
Thank you for your response.
Hi,
Thank you for your outstanding work and sharing the code. I have some certain clarifications and I will appreciate your response,
I am working on 3D CT images where each CT has a dimension of 512 x 512 x N , N is number of axial slices. As part of image pre-processing as was written in your paper
a. For 3D images. for each slice, I need to resize the image dimension to 256 x 256 x 3, correct?
b. Does that mean even for 3D images and for inference, segmentation (and dice) was performed slice-by-slice and then averaged to get the volumetric dice?
c. I show one of the example data that you provided. For AbomenCT-1K and CT_Liver, file names are following,
"casenum_00371_sliceid_50.npy", "casenum_00944_sliceid_422.npy". I am assuming you just uploaded only one slice for a particular patient.
Thank you for your response.