Doubts on AAPM dataset Preparation

I have doubts about the publicly available dataset released as part of

the 2016 NIH-AAPM-Mayo Clinic Low Dose CT Grand Challenge.

  1. I have observed most of the LDCT denoising research done using this dataset selects the patient cases of slice thickness 1mm. However, when I check the DICOM header almost all of the abdomen data were recorded under 5mm slice thickness. Then how to get 1mm slice thickness samples?

  2. The papers cited in this dataset said that the normal-dose abdominal CT images have been acquired at 120 kV and 200 effective mAs. However, according to the Dicom header the effective mAs is 500. Then how to get 200mAs data?

  3. When downloading data, we get RDCT and LDCT cases. It said the LDCT has been obtained by simulating Poisson noise. Also, it is quoter does. But I cannot see the effective mAs as 50mAs in any abdomen case. How to get this?

  4. When training the NN, which image file format do we need to use? Either DICOM or other (png)?

1 Like
Category icons made by Freepik from