Hi

Thanks a lot for sharing your work!

will you be making the model public, i am facing some issues running the code without the model.
Line 244 in model_vnet3d.py /log/segmeation/model/Vnet3d.pd-100000

Thanks in advance.

I think whenever the stridez >= 1 or stridewidth >= 1 or strideheight >= 1，wo should generate more than one patch. In line 25, I think it should be
if stridez >= 1 or stridewidth >= 1 or strideheight >= 1:

In this file "corsedata3dprepare.py"
line 21-23
stridewidth = (width - block_width) // numberxy
strideheight = (height - block_height) // numberxy
stridez = (imagez - blockz) // numberz

I think it should be
stridewidth = (width - block_width) // numberxy + 1
strideheight = (height - block_height) // numberxy + 1
stridez = (imagez - blockz) // numberz + 1

How can ı download

I set up my directory to mimic your folder tree for downloading the images. However, the images are not downloading into the folders. Instead they are being saves outside of the designated folder. For example, if the folder is not in the directory, it creates a folder and doesn't save the images in the sub-folder.

    proImage = "D:\Data\KiTS\kits19process\Image\\"
    proMask = "D:\Data\KiTS\kits19process\Mask\\"
    protumorImage = "D:\Data\KiTS\kits19tumorprocess\Image\\"
    protumorMask = "D:\Data\KiTS\kits19tumorprocess\Mask\\"

Hi

I have followed your steps but i am not able to reproduce the results.at epoch 20 i get the accuracy of 44 but then drops to 13 in the next epoch and then picks up to 34.

Do you have any suggestion?

also wanted to confirm the time it take for you to train to 10 epochs, for me it takes less than a minute with a gpu, is it the same for you? i am running train_vnet3d.py

Thanks

hello,
Is there a difference between Kidney Segmentation and Kidney Tumor Segmentation?
Why does Kidney Tumor Segmentation have no results?

我想请问1.2 Preprocess Kidney Tumor中第一步 tumordata2dprepare.py文件里的这个路径（35-38行）是如何设置的？我在开源的文件和数据集中没有找到Image和Mask文件夹（小白一枚，请见谅）
[图片]

Thanks for your contribution.
I noticed that the fifth point in ‘Preprocess’ is split trainSegmentation.csv into training set and test set, but I can't find subset.py in your code.
What should I do, or can you update 'How to Use'?

他的输入路径srcpath = "E:\junqiangchen\data\kits19\kits19process\Image\"是 corsedata2dprepare.py的输出吗

Nice coding! Is there a way to set a numerical value for subset?

help me
data2prepare.py code executes json error ?

I cannot find subset.py in the whole project but there's a step to use subset.py for splitting into train and test set

How can I tell which original image is the result of pressing the results during training?

Hi, I'd like to use you your project as a baseline for experiments in my master thesis. Would that be possible, and if yes could you add a license? Maybe a citation template too :)

Hi junqiangchen,

Thanks for this repo. Could you please tell why did you truncate the image intensities in
'load_itkfilewithtrucation' function in corsedata2dprepare.py file? After truncating, the intensity values are rescaled to 0 and 255, so what was the reason for truncating before with 300 and -200?

Hi

Is there a way to predict on a complete nii.gz case and produce a complete nii.gz file as a segmentation output ?

how would i go about doing this?

Thanks

Hi

Could you please tell me for how many epochs you trained your model ? and what's the training accuracy you were able to achieve? I have trained it for 100 epochs but there is lot of fluctuation in the accuracy and it is not exceeding 50% is that like normal?

also i am running train vnet3d_train.py

Hi,

while i run your codes, data3dprepare.py, it reminds me that there are no such directory or file named /kits19segementation/200_12.bmp

while in your code, you write
trainImage = "E:\junqiangchen\data\kits19\kits19segmentation\Image"

I would appreciate it if you could please tell me how to generate this file.

I run the project on Linux.

hr_samples[0, 0:blockz, 0:block_width, 0:block_height] = image[0:rangz, 0:rangwidth, 0:rangheight]
hr_mask_samples[0, 0:blockz, 0:block_width, 0:block_height] = mask[0:rangz, 0:rangwidth, 0:rangheight]
您好，请问这里是将原图像的左上角128*128的地方返回吗，这样的话我得到的mask全部为0？

Can you show an example on how you run the subset.py? I'm unsuccessful after several attempts.

I'm so close to getting your code to data2dprocess.py to work. However, I keep running to the runtimeerror at the end of download.

RuntimeErrorTraceback (most recent call last)
<ipython-input-10-c6b6b1f0d94c> in <module>()
    193 
    194 
--> 195 proKitsdata()

<ipython-input-10-c6b6b1f0d94c> in proKitsdata()
    146         file_image = kits_subset_path + image_name
    147         # 1 load itk image and truncate value with upper and lower
--> 148         src = load_itkfilewithtrucation(file_image, 300, -200)
    149         mask_path = kits_subset_path + mask_name
    150         seg = sitk.ReadImage(mask_path, sitk.sitkUInt8)

<ipython-input-10-c6b6b1f0d94c> in load_itkfilewithtrucation(filename, upper, lower)
     77     """
     78     # 1,tructed outside of liver value
---> 79     srcitkimage = sitk.Cast(sitk.ReadImage(filename), sitk.sitkFloat32)
     80     srcitkimagearray = sitk.GetArrayFromImage(srcitkimage)
     81     srcitkimagearray[srcitkimagearray > upper] = upper

/usr/local/lib/python2.7/dist-packages/SimpleITK/SimpleITK.pyc in ReadImage(*args)
   8874 
   8875     """
-> 8876     return _SimpleITK.ReadImage(*args)
   8877 class ImageViewer(_object):
   8878     """

RuntimeError: Exception thrown in SimpleITK ReadImage: /tmp/SimpleITK/Code/IO/src/sitkImageReaderBase.cxx:99:
sitk::ERROR: The file "data/kits19/imaging.nii.gz" does not exist.

Any idea on how to solve.