lucasmansilla / acrn_chest_x-ray_ia Goto Github PK

Dear authors,

Thanks for presenting this great work. I would like to use your model for chest x-ray registration. I wonder if you could provide the mode.pt so that I can directly perform the inference. Or could you please send me your configuration of the dataset JSRT?

Best,

Hello, your code is written very well, but because there is no pre-training model of the autoencoder, when I was training the model, the loss remained at around 0.49 in the end. Is it because of the pre-training model?

Thanks for sharing this excellent code.

When I run test_aenet.py, an error occurred.

After debugging, I found that the size of test_lbs and z in models.py are inconsistent. The size of test_lbs is (50, 256, 256, 1), but the size of z is (800, 64, 64, 1). As show in two pictures.

I would like to ask how to sovle it.

hi, I notice that, in your 'batch_displacement_warp2d' method, you had assumed that the range of vector field (deformation field) is ranged in [0, xlen][0, ylen].
but I review the "VectorCNN" class, this sub-network may produce negative value. so could you please tell me why the vector field range must be [0, xlen][0, ylen] rather than [-xlen, xlen][-ylen, ylen].
thx.

Dear Lucas:

  I want to ask you about how to deal with the remaining two data sets. If possible, I hope you can give me the processed data .

Looking forward to your reply.

I was wondering whether you could provide information which dataset (JSRT, Montgomery or Shenzen) your CLI model is trained on and whether the training parameters were the ones published in the paper. Thanks!

Hello, I used the predicted displacement field to directly transform the segmentation mask during the test work. However, the test indicators calculated in this way seem to be not very accurate. I would like to ask how to calculate these indicators.

Thank you for your code, @lucasmansilla.
Please guide us, which file do we run firstly? Thank you.

Thank you.