calico / AAC_scoring

Map DEXA images from UKBB dataset to abdominal aortic calcification scores
4 stars 2 forks source link

Model_1 releases #26

Closed JoyceYJ01 closed 2 years ago

JoyceYJ01 commented 2 years ago

Hello, I’ve been trying to reproduce some parts of your work. And noticed that the Model 2 pipeline works all right and seems to capture the abdominal aorta region quite accurately (Great!). However, the Model 1 pipeline seems not to produce any results, probably because the trained segmentation and regression model files are not available to public access? (under https://github.com/calico/AAC_scoring/releases, there isn’t anything). Therefore,

  1. may I ask you a favour to release the trained Model 1 weights/parameters?
  2. And would you mind telling me what parameter you used for converting DICOMs to PNGs? (I have downloaded DXA images from UK Biobank, and Is ‘image_scaled = (np.maximum(image,0) / image.max()) * 255.0’ good enough for all Dicoms' converting process?)

Many thanks!

eugenemel commented 2 years ago

Hi, we believe we fixed this issued, could you try again.