seungjunlee96 / emergency-triage-of-brain-computed-tomography-via-anomaly-detection-with-a-deep-generative-model Goto Github PK

I am encountering a RuntimeError with the message "memory format option is only supported by strided tensors" in my PyTorch code. This issue seems related to memory layout and occurs during the execution of the projector.py script.

ubuntu@c0f2685c32d4:/workspace/easiofy$ python projector.py --query_save_path=./demos
Setting up Perceptual loss from device [0]
/opt/conda/lib/python3.8/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
warnings.warn(
/opt/conda/lib/python3.8/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or None for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing weights=VGG16_Weights.IMAGENET1K_V1. You can also use weights=VGG16_Weights.DEFAULT to get the most up-to-date weights.
warnings.warn(msg)
Loading model from: /workspace/easiofy/lpips/weights/v0.1/vgg.pth

test_results
[INFO] query images = ./demos/tumor_edema
/opt/conda/lib/python3.8/site-packages/torch/utils/data/dataloader.py:558: UserWarning: This DataLoader will create 12 worker processes in total. Our suggested max number of worker in current system is 8, which is smaller than what this DataLoader is going to create. Please be aware that excessive worker creation might get DataLoader running slow or even freeze, lower the worker number to avoid potential slowness/freeze if necessary.
warnings.warn(_create_warning_msg(
Traceback (most recent call last):
File "projector.py", line 297, in
project(args)
File "projector.py", line 169, in project
latent_e = encoder(img_reals)
File "/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1511, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1520, in _call_impl
return forward_call(*args, **kwargs)
File "/workspace/easiofy/model.py", line 717, in forward
out = self.convs(input)
File "/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1511, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1520, in _call_impl
return forward_call(*args, **kwargs)
File "/opt/conda/lib/python3.8/site-packages/torch/nn/modules/container.py", line 217, in forward
input = module(input)
File "/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1511, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1520, in _call_impl
return forward_call(*args, **kwargs)
File "/opt/conda/lib/python3.8/site-packages/torch/nn/modules/container.py", line 217, in forward
input = module(input)
File "/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1511, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1520, in _call_impl
return forward_call(*args, **kwargs)
File "/workspace/easiofy/op/fused_act.py", line 83, in forward
return fused_leaky_relu(input, self.bias, self.negative_slope, self.scale)
File "/workspace/easiofy/op/fused_act.py", line 97, in fused_leaky_relu
return FusedLeakyReLUFunction.apply(input, bias, negative_slope, scale)
File "/opt/conda/lib/python3.8/site-packages/torch/autograd/function.py", line 553, in apply
return super().apply(*args, **kwargs) # type: ignore[misc]
File "/workspace/easiofy/op/fused_act.py", line 56, in forward
out = fused.fused_bias_act(input, bias, empty, 3, 0, negative_slope, scale)
RuntimeError: memory format option is only supported by strided tensors.

Hi,

First of all a very commendable work done here in the triage. I have some doubts about the input png demo images. As far as I can see, the png is made by combining three grey level dicoms at different HU levels. Can you pls elaborate at what Hu levels are those at.

Also, it would be helpful if you could elaborate on the preprocessing of dicom data for the custom dataset creation

We are currently trying to utilise the pretrained model for the detection of anomalies in CT scans. We have a set of sample CT scan images ready for experimentation. However, we are facing uncertainty regarding the creation of 'bet.npy' files associated with these images.

Could someone kindly provide guidance or instructions on how to generate the 'bet.npy' files for our CT scan dataset? Any assistance or insights would be greatly appreciated.

Thank you in advance for your support.

Pretrained model checkpoint link has expired，share it again? thank you

Dear seungjunlee,

I hope this message finds you well. I am writing to inform you that we have successfully run your model on our custom dataset. Excitedly, we have obtained some preliminary results and would greatly appreciate your expertise in reviewing them.

In an effort to contribute to the advancement of the model's performance and utility, we have meticulously prepared and conducted experiments with our dataset. We believe that sharing our findings with you could help refine the model's capabilities and potentially uncover areas for improvement.

To facilitate this collaboration, we have attached the results of our experiments for your review. We kindly request your insightful feedback on the following aspects:

Accuracy: Are our results in line with your expectations for this model? If not, could you please guide us on potential reasons for discrepancies and how we might enhance accuracy?

Performance Metrics: Have we appropriately evaluated the model's performance metrics? Are there additional metrics we should consider to provide a comprehensive assessment?

Data Preprocessing: Could any improvements be made to our data preprocessing methods to better align with the model's requirements and optimize performance?

Model Configuration: Are there specific adjustments or fine-tuning techniques we should consider to enhance the model's compatibility with our dataset?

General Recommendations: Based on our findings, do you have any general recommendations or insights that could aid us in further refining our experiments and achieving better results?

Your expertise and guidance are invaluable to us, and we eagerly await your feedback. Please feel free to reach out to us with any questions or clarifications you may have regarding our experiments or the attached results.

Thank you very much for considering our findings and for your continued dedication to advancing the field.

Warm regards,

Anant Paliwal
[email protected]
study ID_aaf114858f patient ID_5f3cc378 subarachnoid.zip

Description:
I am facing challenges with anomaly detection on my custom dataset. The anomaly detection algorithm marks the entire area in red, suggesting a problem. It seems to be indicating issues across the entire image.

Concerns:

• The red marking covers the entire area, not specific regions.
• CT scan involves a patient with subarachnoid.

Attached Files:

• [ CT scan of a patient]
• [DCM to PNG conversion]
• [Nifty files]
• [bet.npy file]

Contact Information:

• Email: [[email protected]]
  study ID_aaf114858f patient ID_5f3cc378 subarachnoid.zip

Additional Note:
I have also emailed you the datasets for your reference.

@seungjunlee96, could you kindly review and provide guidance on resolving this issue? Your assistance is highly appreciated.

I look forward to your response. Thank you!

Hello,

Thank you for your response.

We attempted to utilize the Brain CT Brain Extraction Tool available at CT_BET, as employed in your study. However, we encountered difficulties and were unable to achieve success. Is there an alternative method for generating 'bet.npy' files?

Thank you

Dear SeungjunLee,

I hope this message finds you well. First and foremost, I want to express my sincere gratitude for your prompt response to my GitHub issue and for taking the initiative to fix the problem by updating the Docker file in the project. Your efforts are highly appreciated, and it's great to see such an active and supportive community around this project.

I am currently trying to run the project on my system, which is equipped with an Nvidia RTX 4070 Ti Super graphics card with 16GB GPU RAM. However, I'm encountering a "CUDA out of memory" error. I understand that my GPU has less memory compared to the 4x Titan RTX 8000 GPUs mentioned in the project documentation.

Given this constraint, I am reaching out to seek your guidance on how to adapt the project to run on my GPU with the available resources. Are there specific parameters or configurations that I can adjust to accommodate the lower GPU memory?

Your expertise and insights would be invaluable in helping me overcome this challenge, and I appreciate any guidance or recommendations you can provide. Thank you once again for your dedication to the project and for your assistance in resolving these issues.

Best regards,

Anant Paliwal
anantpal07