mdciri / vox2vox Goto Github PK

Hello Sir
As I see in your paper, there is Hausdorff Distance as one of the training metrics. But, when I ran the code, I couldn't find the Hausdorff distance. Could I know the code to calculate Hausdorff Distance Sir? Because I still get errors in my own code.
Thank you in advance Sir

history['train'] and history['valid'] both has different indentation in trainv2v file could please me know whether it is supposed to be this way or there is some error because I am facing indentation error in that lines

Hello please explain more the data, How it should be saved and the dirctory of the images and the masks and how to distinguish between each 3D sample?

in utils.py line 111:
X_aug, y_aug = combine_aug(X_aug, y_aug)
shown error as below.
combine_aug() missing 1 required positional argument: 'do'

please help to fix the error, thanks.

hello, could you please share your pretrained weights?

I think if you decided to make your code public, you may have the responsibility to maintain it well, instead of just putting all your manuscript-code online. Otherwise you dont't have to do so.

Hello sir, what is this 40 in the imshape?

Hello, you have a 3npy file in which those are pre-processed and converted to NumPy.
could you please share your preprocessing method and the structure of NumPy files?

Hello, I am unable to understand the contents of class_weights.npy file. My dataset is 3d mr images having 7 classes to classify of heart region . I know something similar you have mentioned in #2 but where to mention the names of my labels and size etc. could you please explain it in detail, thanks.

in models.py line 55 :
x = encoder_step(x, Nfilter_start*np.power(2,d), ks, False)
shown error as below.
encoder_step() takes from 2 to 3 positional arguments but 4 were given

I don't sure to modify anything.

Hello,
Thanks for sharing your code. I'm working on the BRats2018 data set. I ran your code 220 epochs. But I couldn't find the same results as you. My Test results:
Evaluation 57/57 (100.00%) [0.75832042 0.47969846 0.46253171]
[34.80947115 25.43625307 22.91565778].

Unlike your code, I manually separated the datasets.
and class weights:
class_weights = [0.25646058, 59.627304, 16.495895, 42.784115]
class_weights = numpy.array (class_weights)

I couldn't figure out where I made a mistake. I would be very happy if you could help me with this subject.

Hello Sir,
I want to ask, is it possible to implement Vox2Vox for 2D medical images? Because I have the final year project to segment meningioma using Vox2Vox. But the data I get from the hospital is 2D.

If it is possible to implement Vox2Vox for 2D medical images, which part do I have to modify sir?
Sorry If I ask too much, because I'm a beginner in this field.
Thank you, Sir.

Can I run this architecture on MyoPS 2020 dataset which have 6 class segmentation?

Hello Sir
Actually, I tried Vox2Vox for my brain MRI dataset which has only one channel. I've already changed the number of channels, but I always got this error. I am still confused, how could it be? Thank you, sir.

Epoch 1/200
WARNING:tensorflow:Model was constructed with shape (None, 128, 128, 128, 1) for input KerasTensor(type_spec=TensorSpec(shape=(None, 128, 128, 128, 1), dtype=tf.float32, name='input_image'), name='input_image', description="created by layer 'input_image'"), but it was called on an input with incompatible shape (4, 128, 128, 128, 2).
Traceback (most recent call last):
File "/content/drive/MyDrive/Fitria/vox2vox_satu sequence/main.py", line 76, in
h = fit(train_gen, valid_gen, alpha, n_epochs)
File "/content/drive/MyDrive/Fitria/vox2vox_satu sequence/train_v2v.py", line 75, in fit
losses = train_step(Xb, yb, alpha)
File "/usr/local/lib/python3.7/dist-packages/tensorflow/python/util/traceback_utils.py", line 153, in error_handler
raise e.with_traceback(filtered_tb) from None
File "/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py", line 1147, in autograph_handler
raise e.ag_error_metadata.to_exception(e)
ValueError: in user code:

File "/content/drive/MyDrive/Fitria/vox2vox_satu sequence/train_v2v.py", line 27, in train_step  *
    disc_real_output = D([image, target], training=True)
File "/usr/local/lib/python3.7/dist-packages/keras/utils/traceback_utils.py", line 67, in error_handler  **
    raise e.with_traceback(filtered_tb) from None
File "/usr/local/lib/python3.7/dist-packages/keras/engine/input_spec.py", line 249, in assert_input_compatibility
    f'Input {input_index} of layer "{layer_name}" is '

ValueError: Exception encountered when calling layer "Discriminator" (type Functional).

Input 0 of layer "conv3d_8" is incompatible with the layer: expected axis -1 of input shape to have value 2, but received input with shape (4, 128, 128, 128, 3)

Call arguments received:
  • inputs=['tf.Tensor(shape=(4, 128, 128, 128, 1), dtype=float64)', 'tf.Tensor(shape=(4, 128, 128, 128, 2), dtype=float32)']
  • training=True
  • mask=None

How to make test prediction from the trained model? Thank you