Hi, I wanna use the ANTsPyNet, but there are always some questions about tensorflow environment

[FreshmanMa]

only requires tensorlfow-version? only uisng under mac or linux? I don't understand, sorry, I'm freshman.

[cookpa]

I don't know about Windows. We recently got ANTsPy to build on Windows, but I don't know if ANTsPyNet will work.

On Mac or Linux, you can try installing with pip:

pip install antspynet

or via setup.py

git clone https://github.com/ANTsX/ANTsPyNet
cd ANTsPyNet
python setup.py install

If you have docker, you can get ANTsPyNet here

https://hub.docker.com/r/cookpa/antspynet/tags

On my Mac, I have

tensorflow==2.3.1 tensorflow-estimator==2.3.0 tensorflow-probability==0.11.1

If you're new to Python, I recommend learning about virtual environments and miniconda, because that will help you create an environment just for ANTsPyNet, where you can control the version of dependencies.

[FreshmanMa]

I don't know about Windows. We recently got ANTsPy to build on Windows, but I don't know if ANTsPyNet will work.

On Mac or Linux, you can try installing with pip:

pip install antspynet

or via setup.py

git clone https://github.com/ANTsX/ANTsPyNet
cd ANTsPyNet
python setup.py install

If you have docker, you can get ANTsPyNet here

https://hub.docker.com/r/cookpa/antspynet/tags

On my Mac, I have

tensorflow==2.3.1 tensorflow-estimator==2.3.0 tensorflow-probability==0.11.1

If you're new to Python, I recommend learning about virtual environments and miniconda, because that will help you create an environment just for ANTsPyNet, where you can control the version of dependencies.

First, thanks to your reply, and the tensorflow need a gpu version ?

[cookpa]

To use the GPU, it needs to be tensorflow-gpu, yes.

[FreshmanMa]

To use the GPU, it needs to be tensorflow-gpu, yes.

Ok, thanks, I'll have a try now.

[FreshmanMa]

Sorry, there is a question I don't know how to deal with. I wanna use antspynet.utilities.deep_atropos() to segment one processed HCP data into Six-tissue segmentation. I rent a server, I use the following codepip install antspynet -i https://pypi.tuna.tsinghua.edu.cn/simple, to install antspynet, and wana use a demo with deep_atropos, but I can't do it . and using python setup.py install ,the error is Reading https://pypi.org/simple/matplotlib/ error: The read operation timed out the code： import ants import antspynet file_path = './dataset/test_GT/531536.nii.gz image = ants.image_read(file_path) segmentation = antspynet.utilities.deep_atropos(image) segmentation_result = segmentation['segmentation'] ants.image_write(segmentation_result, "./dataset/segmentation/531536.nii.gz") but there is error Exception: URL fetch failure on https://figshare.com/ndownloader/files/34821874: 403 -- Forbidden What should I do to resovle it? and if I use the code above, can get the segmentation.nii.gz data ? Thanks!

To use the GPU, it needs to be tensorflow-gpu, yes.

[cookpa]

It seems your server is not allowed to download from Figshare. I don't know how to solve that. If you can download by other means (eg from another machine), here's how you can add it to your ANTsPyNet:

1. mkdir ~/.keras

2. Copy this text into ~/.keras/keras.json

{
    "floatx": "float32",
    "epsilon": 1e-07,
    "backend": "tensorflow",
    "image_data_format": "channels_last"
}

3. mkdir ~/.keras/ANTsXNet

4. Copy the files (eg, https://figshare.com/ndownloader/files/34821874 downloads as brainExtractionRobustT1.h5) to ~/.keras/ANTsXNet

[FreshmanMa]

It seems your server is not allowed to download from Figshare. I don't know how to solve that. If you can download by other means (eg from another machine), here's how you can add it to your ANTsPyNet:

1. mkdir ~/.keras
2. Copy this text into ~/.keras/keras.json

{
    "floatx": "float32",
    "epsilon": 1e-07,
    "backend": "tensorflow",
    "image_data_format": "channels_last"
}

3. mkdir ~/.keras/ANTsXNet
4. Copy the files (eg, https://figshare.com/ndownloader/files/34821874 downloads as brainExtractionRobustT1.h5) to ~/.keras/ANTsXNet

Sorry，I don't know why it did not work. So today, I try use docker, and I had dodocker pull cookpa/antspynet:latest and docker run -it cookpa/antspynet, into a python environment, How can I run a .py script and upload my mri.nii.gz data？and the docker will use the gpu of my pc? I wanna get this kind of segmentation data in picture.

It's hard for me now. Your help will be very useful for me, thanks a lot.

[FreshmanMa]

It seems your server is not allowed to download from Figshare. I don't know how to solve that. If you can download by other means (eg from another machine), here's how you can add it to your ANTsPyNet:

1. mkdir ~/.keras
2. Copy this text into ~/.keras/keras.json

{
    "floatx": "float32",
    "epsilon": 1e-07,
    "backend": "tensorflow",
    "image_data_format": "channels_last"
}

3. mkdir ~/.keras/ANTsXNet
4. Copy the files (eg, https://figshare.com/ndownloader/files/34821874 downloads as brainExtractionRobustT1.h5) to ~/.keras/ANTsXNet

It seems your server is not allowed to download from Figshare. I don't know how to solve that. If you can download by other means (eg from another machine), here's how you can add it to your ANTsPyNet:

1. mkdir ~/.keras
2. Copy this text into ~/.keras/keras.json

{
    "floatx": "float32",
    "epsilon": 1e-07,
    "backend": "tensorflow",
    "image_data_format": "channels_last"
}

3. mkdir ~/.keras/ANTsXNet
4. Copy the files (eg, https://figshare.com/ndownloader/files/34821874 downloads as brainExtractionRobustT1.h5) to ~/.keras/ANTsXNet

Sorry，I don't know why it did not work. So today, I try use docker, and I had dodocker pull cookpa/antspynet:latest and docker run -it cookpa/antspynet, into a python environment, How can I run a .py script and upload my mri.nii.gz data？and the docker will use the gpu of my pc? I wanna get this kind of segmentation data in picture. It's hard for me now. Your help will be very useful for me, thanks a lot.

Can you give me some advice? Thanks

[FreshmanMa]

Sorry, I encounter a new promle. using docker provided by, but can't use GPU. What should I do?

[cookpa]

I didn't build that container with CUDA, because I don't have a machine that supports it. But deep Atropos will run quite quickly without it.

[FreshmanMa]

I didn't build that container with CUDA, because I don't have a machine that supports it. But deep Atropos will run quite quickly without it.

All right, Thanks! Can give me some advice to get the MRI after segmentation, like this picture

[cookpa]

I didn't build that container with CUDA, because I don't have a machine that supports it. But deep Atropos will run quite quickly without it.

All right, Thanks! Can give me some advice to get the MRI after segmentation, like this picture

https://antsx.github.io/ANTsPyNet/docs/build/html/utilities.html#antspynet.utilities.deep_atropos