The code of image classification on COVID dataset using pytorch on INFORMS 2020 QSR Data Challenge on COVID dataset. We use an ensemble model consisting of Densenet 121 and Residual Attention model. We first split 15% of the data into validation which is not used in the training process. And we select the model which has the highest validation accuracy. Densenet 121 is pretrained on ImageNet, and Residual Attention model is pretrained on Cifar-10. In training, we separately train these two pretrained models in an end-to-end manner. Then we extract features from the last 2nd layer, and perform another classifier on the learned concatenated features by these two models on the whole training dataset. Here we use SVM with random gaussian kernels.
- Python3, Scikit-learn, torch (Please refer to requirement.txt)
The data for this Data Challenge is selected from an open-source data set on COVID-19 CT images. The raw data have been divided into two subsets: training and test sets. The training dataset is provided to participants to develop their models. The training dataset consists of 251 COVID-19 and 292 non-COVID-19 CT images. In addition to the images, meta-information (e.g., patient information, severity, image caption) is provided in a spreadsheet. The details of the original dataset can be found in Zhao et al. (2020).
We extended this work by building a large lung CT scan dataset for COVID-19 curating data from 7 public datasets. The dataset and the dataset description are available in the following links: https://www.kaggle.com/maedemaftouni/large-covid19-ct-slice-dataset https://github.com/maftouni/Curated_Covid_CT.git
The training data is saved in data/training. If you want to use your own data, just replace everything in data/training. It contains two folders where one is COVID images, and another Non-Covid images. The test data should be put in data/test.
The performance might be a little different due to different performance of a certain seed on different devices.
data_prep.pyto train DenseNet121 model:
python Model_densenet121.pyto train residual_attention model:
python Model_residual_attention.pyto train the ensemble model:
python Model_Ensemble.py
Sample classification results
Attention can be viewed, broadly, as a tool to focus the most on the most informative parts of the image:
Here we evaluate the performance of our best model on the training data.
Confusion Matrix
predict Covid predict Non-Covid
Covid 247 4
Non-Covid 2 290
Accuracy
Accuracy: 98.9%
Version 1.0
If you use this code for your research, please cite:
Maftouni, M., Law, A.C, Shen, B., Zhou, Y., Yazdi, N., and Kong, Z.J. “A Robust Ensemble-Deep Learning Model for COVID-19 Diagnosis based on an Integrated CT Scan Images Database,” Proceedings of the 2021 Industrial and Systems Engineering Conference, Virtual Conference, May 22-25, 2021.
You can find the paper here.
- "https://pytorch.org/tutorials/beginner/transfer_learning_tutorial.html"
- "https://pytorch.org/tutorials/beginner/finetuning_torchvision_models_tutorial.html"
- "https://www.learnopencv.com/image-classification-using-transfer-learning-in-pytorch/"
- "https://towardsdatascience.com/https-medium-com-dinber19-take-a-deeper-look-at-your-pytorch-model-with-the-new-tensorboard-built-in-513969cf6a72"
- "https://www.aiworkbox.com/lessons/how-to-define-a-convolutional-layer-in-pytorch#lesson-transcript-section"
- "https://medium.com/udacity-pytorch-challengers/ideas-on-how-to-fine-tune-a-pre-trained-model-in-pytorch-184c47185a20"
- "https://github.com/FrancescoSaverioZuppichini/Pytorch-how-and-when-to-use-Module-Sequential-ModuleList-and-ModuleDict"
- "https://www.kaggle.com/c/understanding_cloud_organization/discussion/112582"
- "https://stackoverflow.com/questions/53290306/confusion-matrix-and-test-accuracy-for-pytorch-transfer-learning-tutorial"
- "https://github.com/tengshaofeng/ResidualAttentionNetwork-pytorch"
- "https://github.com/UCSD-AI4H/COVID-CT"
- "https://arxiv.org/abs/2007.14846"
- "https://arxiv.org/pdf/2003.13865.pdf"