Building an AI system to Detect COVID-19 from X-Ray
In my trials to help the World in fighting against the COVID-19 virus, I came up with an experimental model through the deep neural network (artificial intelligence) to discover Covid-19 virus from X-RAY/CT scan of patients. The model is currently trained on the global open source IEEE dataset, with an accuracy of 96% on test data.
Github Repo: [Link]
Goal
The tasks are as follows using chest X-ray or CT (preference for X-ray) as input to predict these tasks:
Healthy vs Pneumonia (prototype already implemented Chester with ~74% AUC, validation study here)
Bacterial vs Viral vs COVID-19 Pneumonia
Survival of patient
Dataset
Currently this git repo is maintaining the COVID-19 Xray dataset: [Link]
Model Statistics and Accuracy
VGG16 model was utilized as base model, the layers were freezed and on top of it some more layers were added. The model was trained on 370 samples of data and tested on 36 samples. The testing accuracy currently reported is 95%.
Confusion Matrix (accuracy 98%, sensitivity 93% and specificity 100%)
Correct Healthy Patient Detection (True Negatives): 13 Incorrect Covid-19 Detection (False Positives): 1 Incorrect Healthy Patient Detection (False Negatives): 0 Correct Covid-19 Detection (True Positives): 28 Total Patietns Diagnosed with Covid-19: 28
Here is the the result of model on one of the samples:
Commentaires