about liver dataset

[zzx332]

Thanks for sharing your work in this field! I have some questions about these words in the paper "Raw scans are resampled into 128 × 128 × 128 voxels after cropping unnecessary area around the target object. For liver CT scans, a simple threshold-based algorithm is applied to find a rough liver bounding box for cropping."

1.What should I do to crop and resample？ Sorry I am not familar with image processing, what kind of tools should I use? 2.During training，how do I select two input image （fixed image and moving image）from dataset （e.g LiTs 130 CT scans，512x512x75）？ 3.If I were to use a liver dataset, what preprocessing steps would I need to perform？

[zsyzzsoft]

If you use our provided datasets, there is nothing else to do --- they are already preprocessed. If you want to make your own dataset, a complete preprocessing pipeline for liver can be found in demo.py, which takes as input two raw CT scans for a forward pass.

[zzx332]

Thank you for your reply,i will check out demo.py. I have another question. How do I choose an image pair from mydataset for training，just random？

[zsyzzsoft]

You can prepocess them into a .h5 file like our provided dataset, then the training script would choose random image pairs for training by default.

[zzx332]

Thank you very much for your reply.