search

microsoft / Recursive-Cascaded-Networks

[ICCV 2019] Recursive Cascaded Networks for Unsupervised Medical Image Registration
https://arxiv.org/abs/1907.12353
MIT License
358 stars 86 forks source link

itk::ERROR: Key '0008|103e' does not exist when run demo.py #64

Open Hoodini-Gzj opened 2 years ago

Hoodini-Gzj commented 2 years ago

Thank you for your previous reply. I have encountered several problems recently, and I hope to get your advice again Q1：This error occurs when I run the demo.py with the lung CT data(LIDC-IDRI 512 512 133) :itk::ERROR: Key '0008|103e' does not exist when run demo.py I remember you mentioned the threshold-based algorithm algorithm, is this the cause of the error, or the image dimension, could you tell me how to modify? Q2：The input of RCN network is 128*3, but I want to use different sizes of data, what should I do, modify the network structure or cut the size? Q3：I made some changes to the eval.py and recursive_cascaded_networks.py to visualize the results (1)add 'real_flow', 'warped_moving', 'image_fixed', 'image_moving' to keys (2)The following lines are added at the end of the main function for val_subset in val_subsets:

output_png_dir = 'E:\gzj\Recursive-Cascaded-Networks\png_results_voxel-5' os.makedirs(output_png_dir) for i in range(len(results['real_flow'])): im_flow = RenderFlow(results['real_flow'][i]) skimage.io.imsave(os.path.join(output_png_dir,'flow'+ str(i) +'.png'), im_flow) warped_img = results['warped_moving'][i] show_image(warped_img, os.path.join(output_png_dir,'warped'+ str(i) +'.png')) fixed_img = results['image_fixed'][i] show_image(fixed_img, os.path.join(output_png_dir,'fixed'+ str(i) +'.png')) moving_img = results['image_moving'][i] show_image(moving_img, os.path.join(output_png_dir,'moving'+ str(i) +'.png'))

(3)add 'image_moving': img2, in recursive_cascaded_networks.py I succeeded in getting the visualization results on SLIVER dataset.However, parameter errors occurred in the tests on datasets LiTS and LSPIG. eval.py: error: unrecognized arguments: -v lits and eval.py: error: unrecognized arguments -v lspig --paired

Your suggestions are very important to me and I am looking forward to your reply again.

zsyzzsoft commented 2 years ago
  1. You can modify the code that requests the key in demo.py
  2. The easiest way is to resize or crop.
  3. Please check whether your have modified argument parser.
Hoodini-Gzj commented 2 years ago

Thanks for reply.I resize the scans to 128 128 128,but the resolution is too low,the image is very blurry warped_moving_5 I'm learning your code now, thanks again

  1. You can modify the code that requests the key in demo.py
  2. The easiest way is to resize or crop.
  3. Please check whether your have modified argument parser.