Closed Seconight closed 1 year ago
Hi, thank you for your interest towards our work! I have tried the scenario of randomly sampled 20 cases for validation and the scenario of last 20 subjects (with last 20 subject numbers). Both demonstrates similar results and I would recommend to use the scenario of last 20 subjects.
For computing the Dice score, I have another python file to compute it and will upload to the GitHub afterwards.
I got it! Thank you for your reply and work!
Hello! Thank you for your great work in 3DUX-Net! And I have a few questions: I've seen the
dataset_AMOS.json
and there's 180 cases in Training and 20 cases in Test, and in the paper the ratio is 160:20:20. Are the 20 cases in the validation dataset randomly selected from the training set? Or are there other methods of splitting the training dataset? And, I didn't find find the code for calculating the mean dice of each category and the total in thetest_seg.py
. Could you please provide specific code for calculating the metrics in the paper? Thank you for your reply~