search

Ruyi-Zha / naf_cbct

MIT License
106 stars 20 forks source link

Attenuation synthesis problem #1

Closed Zhentao-Liu closed 1 year ago

Zhentao-Liu commented 1 year ago

It's a good idea to apply implicit function and hash table to reconstruction CBCT volume. However, there is a problem in attenuation synthesis. In your paper, you give the following equation: att While in the code your provide: acc = torch.sum((raw[..., 0] + noise) * dists, dim=-1). I think they are not the same. Would you give some explanation?

Ruyi-Zha commented 1 year ago

Thanks for your interest. There is a preprocessing step for raw X-ray images before we actually use them. We do logarithm operation on raw data such that the intensity becomes a simple sum, which is what the code shows.

puresky123 commented 1 year ago

Hello, may I ask when the raw X-ray images was preprocessed? Has the raw X-ray images in the raw folder you provided been preprocessed

Ruyi-Zha commented 1 year ago

Hi, if I remember correctly, the img.mat files in the raw folder are not preprocessed (normalized), while dataGenerator/generateData.py will preprocess (normalize) the 3D volume and pack it into the *.pickle file.

puresky123 commented 1 year ago

Thanks a lot