In this project, we were asked to experiment with a real-world dataset, and develop an advanced ML based Classifier that can scan chest XRays and COVID19 cases. After performing the required tasks on a dataset of my choice, herein lies my final report.
Keywords: Machine Learning | Classification | Artificial Intelligence | CNN
Corona virus disease (COVID-19) is an infectious disease caused by a newly discovered corona virus.Most people infected with the COVID-19 virus will experience mild to moderate respiratory illness and recover without requiring special treatment. Older people, and those with underlying medical problems like cardiovascular disease, diabetes, chronic respiratory disease, and cancer are more likely to develop serious illness.
The best way to prevent and slow down transmission is be well informed about the COVID- 19 virus, the disease it causes and how it spreads. Protect our self and others from infection by washing hands or using an alcohol based rub frequently and not touching your face. The COVID-19 virus spreads primarily through droplets of saliva or discharge from the nose when an infected person coughs or sneezes, so it’s important that we also practice respiratory etiquette (for example, by coughing into a flexed elbow). At this time, there are no specific vaccines or treatments for COVID-19. However, there are many ongoing clinical trials evaluating potential treatments.
Artificial intelligence applied to the medical domain can have very real consequences.
We have gathered all the images from the positive COVID images.We ended up with 140 images of positive COVID X26 rays with view-’PA’ out of 350 total images and same 140
number of images were taken from the Kaggle chest X rays
for negative COVID images. Total number of images in each
folder before training are as shown in table 1.
The data set is available at:
Some Important terminologies which are used in algorithm:
- CNN: A convolutional neural network (CNN) is a type of artificial neural network used in image recognition and processing that is specifically designed to process pixel data.
- Relu: The output of ReLu is the maximum value between zero and the input value. An output is equal to zero when theinput value is negative and the input value when the input is positive.
- Max pooling layer: Max pooling is a pooling operation that selects the maximum element from the region of the feature map covered by the filter.
- Convolutional layer: a convolutional neural network is a class of deep neural networks, most commonly applied to analyzing visual imagery. Binary Cross-Entropy Loss: It is a Sigmoid activation plus a Cross-Entropy loss. Unlike Softmax loss it is inde- pendent for each vector component (class), meaning that the loss computed for every CNN output vector component is not affected by other component values.
- Adam optimizer: Adam is an optimization algorithm that can be used instead of the classical stochastic gradient descent procedure to update network weights iterative based in training data.
- Keras Fit Generator: It in python can be used to train our machine learning and deep learning models.
- Confusion Matrix: A confusion matrix is a table that
is often used to describe the performance of a classification
model (or "classifier") on a set of test data for which the true
values are known.
Steps involved in Algorithm: - Pre-process the given input X ray images and convert them to RGB channel ordering to make them ready for our CNN.
- Initialize data augmentation generator object where each image is first normalized by dividing each image by 255 so that its pixels are now between 0 and 1. After that, rotation of image is done.
- We instantiate the CNN model with nodes of size 32,64,64,128,128 in each CNN layer respectively. Filters of size (3*3) are used to detect the patterns in images.
- First, apply Relu then add the Max pooling layer after each convolutional layer and a drop-out of 0.25. To compile the network, use the Binary cross-entropy loss and Adam optimizer.
- To start CNN training process, we make a call to Keras fit generator method, while passing in chest X-ray data via data augmentation object.
- For evaluation, first make predictions on the testing set and grab the prediction indices, then generate and print out a confusion matrix report and save it to disk.
From Figure 1. We have found that COVID-19 detector is obtaining approximately 96% accuracy on our test dataset, alongwith 100% sensitivity and 98% specificity implying that:
- Of patients that do have COVID-19 (i.e., true positives), we could accurately identify them 100% of time.
- Of patients that do not have COVID-19 (i.e., true negatives), we could accurately identify them only 98% of the time.
Refer Table 2. On testing Model on real life data, we have found that it predicting the same level which was shown by actual level. Refer Figure 2 to check the result.
COVID-19 tests are currently hard to come by there are simply not enough of them and they cannot be manufactured fast enough, which is causing panic. Given that there are limited COVID-19 testing kits, we need to rely on other diagnosis measures. Since COVID-19 attacks the epithelial cells that line our respiratory tract, we can use X-rays to analyze the health of a patient’s lungs. It could be possible to use X-rays to test for COVID-19 without the dedicated test kits. A drawback is that X-ray analysis requires a radiology expert and takes significant time —which is precious when people are sick around the world. Therefore, this CNN based automated model will be helpful that time.
- S Das, Prediction of covid-19 disease progression in india : Under the effect of national lock- 136 down (2020). 137
- C Hill, SciPy. (Cambridge University Press), (2016). 138
- M Chinazzi, et al., The effect of travel restrictions on the spread of the 2019 novel coronavirus 139 (covid-19) outbreak. Science 368, 395–400 (2020).
Table 1. Number of images in each folder before training
| Image Type | Count |
|---|---|
| True-Positive images | 115 |
| True-Negative images | 115 |
| False-Positive images | 25 |
| False-Negative images | 25 |
Table 2. Accuracy of Model
| Data Category | Accuracy in % |
|---|---|
| Training Accuracy (CNN) | 95.22 |
| Test Accuracy (CNN) 96.00 | 115 |
Fig 1. Result after 10 epochs of validation steps of 2 and batch size of 32.
Fig 2. Prediction by Model.