MEW-UNet

This is the official code repository for "MEW-UNet: Multi-axis representation learning in frequency domain for medical image segmentation" [arXiv]

0. Main Environments

• python 3.8
• pytorch 1.8.0
• torchvision 0.9.0

1. Prepare the dataset and our weights.

• Synapse dataset can be found at the repo of TransUnet.

• Our weights of the Synapse dataset can be download here, with a password of ex8x

• After downloading the weights, you could put weights into this file('./our_weights/')

2. Test data folder format

• data
  • Synapse
    • test_vol_h5
      • case0001.npy.h5
      • case0002.npy.h5
      • case0003.npy.h5
      • case0004.npy.h5
      • case0008.npy.h5
      • case0022.npy.h5
      • case0025.npy.h5
      • case0029.npy.h5
      • case0032.npy.h5
      • case0035.npy.h5
      • case0036.npy.h5
      • case0038.npy.h5

3. Test our model.

cd MEW-UNet

python test.py

After testing about 20 mins, you can obtain the results in './test_log/test_log/'