search

qgao21 / CoreDiff

32 stars 3 forks source link

Questions about the mayo2016 dataset #2

Open chienwhen opened 10 months ago

chienwhen commented 10 months ago

How to get the 'mat' format files of the mayo2016 datasets ？I downloaded it and realized that the format of the file is IMA. Can you upload the relevant preprocessing code?

qgao21 commented 10 months ago

As the Mayo 2016 dataset offers only 25% dose data, we employed the LD-CT-simulation method available at LD-CT-simulation to generate simulated ultra-low dose data, stored in the ".mat" format. If you don't require the simulation of ultra-low dose data, you can directly utilize the Pydicom package for reading "DICOM" format data. Afterwards, apply the following code to normalize the data.

MIN_B = -1024
MAX_B = 3072
img = img.astype(np.float32) - 1024
img[img < MIN_B] = MIN_B
img[img > MAX_B] = MAX_B
img = (img - MIN_B) / (MAX_B - MIN_B)
chienwhen commented 10 months ago

Thank you!