COVID 19 Severity of Pneumonia Analysis Using Chest X Rays
Narayana Darapaneni, Shweta Ranjane, Uday Shankar Pallavajula Satya, Dr Prashanth, M Harichandan Reddy, Anwesh Reddy Paduri, Aravind Kumar Adhi, Vachaspathi Madabhushanam
Abstract
Purpose: To identify pneumonia location and determine the severity of pneumonia using deep learning network on chest X-ray images Methods: Data from RSNA Pneumonia detection challenge [1] from Kaggle is used for train and test analysis. Identifying images and calculating severity percentage of lung opacity in pneumonia present images by drawing bounding box Results: With 4668 X-ray images trained and tested on 1500 X-ray images, initial model has shown a mean average precision (mAP) of 0.90 on train set and 0.89 on test set. Conclusion: The intention is to leverage on existing studies and develop a better performing and highly accurate deep learning model to calculate severity percentage in a pneumonia present chest x-ray image.