search

MIC-DKFZ / MedNeXt

[MICCAI 2023] MedNeXt is a fully ConvNeXt architecture for 3D medical image segmentation.
https://arxiv.org/pdf/2303.09975
Apache License 2.0
345 stars 26 forks source link

How to train MedNeXt #5

Closed lovergos closed 1 year ago

lovergos commented 1 year ago

Dear author, Thank you for your good job，I want to ask you about the training process, and how to train MedNeXt,

saikat-roy commented 1 year ago

Hey @lovergos . I'm sorry but I simply haven't had the time to clean the code for the training yet. I trained it by creating a nnunet v1 trainer - that is, by replacing the architecture inside a standard nnUNetTrainerV2 subclass and using patches of size 128 x 128 x 128

This should not be hard to replicate if you have experience with nnUNet v1. However, I will have code for this training in the near future.

If you have any other questions on the training process, please let me know. Otherwise, keep an eye out for an update to the repo for the training code.

saikat-roy commented 1 year ago

There is code and instructions now to recreate the MICCAI2023 experiments in nnUNet(v1). Please reopen this issue if you have more questions.