mars11121 / DuDoTrans

This code is an official implementation of "DuDoTrans: Dual-Domain Transformer for Sparse-View CT Reconstruction".
MIT License
26 stars 0 forks source link

如何预处理AAPM数据? #3

Closed Nioolek closed 1 year ago

Nioolek commented 1 year ago

你好,非常感谢您开源了您的相关代码。 想请问一下,下载下来的AAPM数据,只有 projection data 和 image data ,但是网络的输入应该是正弦图?想请问一下如何生成正弦图呢?

mars11121 commented 1 year ago

The input to the network is only the CT image, and the dataloader will process it and generate simulated sinograms.

Nioolek commented 1 year ago

The input to the network is only the CT image, and the dataloader will process it and generate simulated sinograms.

Thank you for your reply. But the image data format of AAPM dataset is .ima, not dcm. I think the code only can read dcm, so should I convert ima to dcm ?

mars11121 commented 1 year ago

The input to the network is only the CT image, and the dataloader will process it and generate simulated sinograms.

Thank you for your reply. But the image data format of AAPM dataset is .ima, not dcm. I think the code only can read dcm, so should I convert ima to dcm ?

The AAPM data format, which I downloaded two years ago, is .dcm. You can also try other datasets whose format is .dcm with the code. Otherwise, you can also rewrite a new dataloader, with the correct normalization.

Nioolek commented 1 year ago

OK. thank you for your reply