mr7495 / covid-ct-code Goto Github PK

This notebook doesn't run because of this error. I can't figure out what you meant by img_name here as it has indeces.

NameError Traceback (most recent call last)
in ()
15 full_add=os.path.join(r,file)
16 if 'SR_2' in full_add: #Select only SR_2 folder
---> 17 index1=img_name.index('patient')
18 index2=img_name.index('_SR')
19 if 'covid' in full_add:

NameError: name 'img_name' is not defined

Greetings. Thank you for your exellent work.
I am curious, in your code in COVID_Train&Validation.ipynb, you selected 3 layers from the ResNet50V2 framework ["conv4_block1_preact_relu", "conv5_block1_preact_relu", "post_relu"], what are the reasons of selecting these layers? And if I want to replace the backbone model to InceptionV3, what will be the layers selected in that model?
If you coudl kindly help me out I would be most grateful.

Hi, I needed a help on removing Dicom images from CT scans which is in Dicom format.

Can you help me on this.

Thanks

Dear Mohamad,
First, I thank you for your great job.
I am also interested in the topic Covid-19 CT scan now. I am having very promising models to predict Covid-19 CT scans. So I am trying to expand my experiments via external validity. Can you give me your ENTIRE dataset like folder style, e.g. Covid-19 folder containing only images classified from Covid-19 patients, and Non-Covid-19 folder consiting of scans from healthy subjects?
Best regards.
Linh

Additionally, I am receiving the following error:

This is for the COVID and Validation code, when it gets the predicted class index.

The error is for the following line:
pred_ind=np.argmax(net.predict(np.expand_dims(np.expand_dims(img,axis=0),axis=3))[0])

Do you know what might be the problem? Thanks!

I really like your code and I am trying to run it with the given dataset. I am trying to use my NVIDIA gpu, but whenever I do, it runs into memory problems and raises the following error:

2020-11-22 21:40:39.870589: W tensorflow/core/common_runtime/bfc_allocator.cc:245] Allocator (GPU_0_bfc) ran out of memory trying to allocate 1.09GiB with freed_by_count=0. The caller indicates that this is not a failure, but may mean that there could be performance gains if more memory were available.

Function call stack: keras_scratch_graph at the end.

However, when I reduce the batchsize, the error stops, but I would want to run this code with a larger batchsize. Do you know what may be the problem?