rakshith2597/lung-nodule-detection-luna-16

Lung nodule detection- LUNA 16

Lung-nodule-detection-LUNA-16

This Github repository,has the code used as part of my Bachelor's in technology main-project. The purpose of this code is to detect nodules in a CT scan of lung and subsequently to classify them as being benign, malignant.

Abstract:

Abstract—Lung cancer is one of the leading cause for cancer related death in the world. Early detection of the tumor is a crucial part of giving patients the best chance of recovery. However, analysis and cure of lung malignancy have been one of the greatest difficulties faced by humans over the most recent couple of decades. Deep learning gives us to increase the accuracy of the automated initial diagnosis. This project uses an approach that utilizes a network with features of U-Net architecture to classify cancer nodules as benign or malignant with an accuracy of 92.38 and a low percentage of false positives(<10%).

Dataset used: LUNA 16

Trained model and results have not been uploaded in the repo due to its size.

Front-end of the CAD system is in the repo mentioned below. https://github.com/Soumya-Raj/Main-project

Repo organization:

data_prep :- Directory contains the code used to prepare the LUNA16 dataset for training.

train_codes :- Directory contains the code used to train the network.

plots:- Directory contains the scripts to evaluate the network and also the dice plots.

lung-nodule-detection-luna-16's Issues

General Working

Hello there. Thanks for this great work. I just finished a course on Convolutional Neural Networks and want to do a good project to get a better understanding of CNN. So as a beginner, I have no idea about the project. So could you tell me the order in which I should run the files and the basic task the python file does? Thanks in Advance.

What does 'Masks' refer to?

Hello,

My question here might have a few parts that are somehow correlated, it might not make sense to reader but I will try to make it as clear as possible.

1-I have noticed you're using the term detection. This commonly (but not exclusively) refers to detection in the form of bounding boxes. However, your implementation (and Abstract) seems to be referring to only solving the classification problem. Is my understanding correct?

2-You also refer to using 'Masks', as I managed to interpret some of the script, I think you are creating masks of rectangular shape using the LUNA-16 ground-truth. where 0s correspond to background and 1s denote nodules. Then you're training a UNet-like network to perform segmentation (initially) and then interpreting the resulting masks as final classification (or detection in the form of bounding boxes?). So to clarify my questions :
a- If you're using masks as described above, What is the intuition behind these masks (i.e. how are they being created)?
b- what is the final output of the framework, is it in the form of bounding boxes, or is it only the classification per input.

Looking forward for your response.

Many thanks.

Documentation

Hello,

I am very interested in reproducing your work, I am currently working on LUNA-16 nodule detection. I would really appreciate and it would be very helpful if you point out or provide the documentation of your work. This could be wither your paper or thesis (if exists). This would help me and others to further understand the workflow and will assist in reproducing your work.

Looking forward to hearing back from you.

Many thanks.