tommy-qichang / asyndgan Goto Github PK

请问AsynDGAN的输入数据如何生成的

请教测试阶段如何计算AJI以及显示实例分割图片

How do I split the training dataset into 10 subsets based on the segmentation volume size?

Can you give me the data format of BRATS2018 divided into 10 categories? I wanted to run your code, but unfortunately it failed. If I know the format of your processed data, I can test my own data. Looking forward to your reply, thank you.

Could you give me a training process? Thank you

Can you please explain how the segmentation process is done? I have generated the synthetic images using AsynDGAN, but I cannot find any code or explanation for the segmentation task.

In your paper, you train a generator that could map the mask to a real-like images, and use these generated images as training samples. You treat the label in datasets as mask and input the model (i'm not sure if my understand is correct), however, how the local entities get label as mask? I mean, how can we get the input of generator if we want to generate large number of images? Does random generate mask as input is make sense?

I have got the latest_net_G.pth, but I always get an error when testing. I used save_syn.py to test. Can you give me some tips?Looking forward to your reply。thanks

Can you explain the preprocessing done on the nuclei dataset? And how should I train and test the code for nuclei dataset.

Hi Tommy,

Thank you for your inspiring work! It looks interesting to me. I was wondering if it was possible that you can provide your pre-trained label-to-image generator based on BraTs 2018 dataset? It would be very much appreciated if possible! Thank you again.

test后的结果包含realA realB fakeB 分别是什么含义？ 如何重建为3维图像呢？

你好，我没有办法获取数据集，您能分享一下吗？

Can you explain the testing part for the BraTS Dataset, if possible can please share the script used for the same.

it seems your work belong to distributed learning. is that so？ it is very interesting. so, how to build this work amongs differents institutions? can you give me some tips?

For the results produced in the paper for nuclei segmentation, which model was used?