COVID Pneumonia Prediction Based on Chest X-Ray Images Using Deep Learning
2022 IEEE International Conference on Communications, ICC 2022
; 2022-May:2580-2585, 2022.
Article
in English
| Scopus | ID: covidwho-2029227
ABSTRACT
COVID which is one of the deadliest Pandemic of this era stuck the entire world which emerged from Wuhan, China in 2019. The pandemic had an extensive impact on unemployment and even deaths. This Pandemic was so new that a lot of medical doctors were involved in research towards diagnosing the chest X-ray images for COVID symptoms. Along with COVID, there have been other complications found like Pneumonia which resulted in the second wave of COVID leading to deaths. There has been good research done by Deep learning researchers in predicting the COVID and also COVID with Pneumonia classification based on Chest X-rays. But the challenge in earlier work is the limited data set which ultimately resulted in higher accuracy. The reason being smaller data set had very fewer number features for training which ultimately resulted in higher accuracy during prediction. So, towards obviating the above-mentioned challenge, we here have collected a fairly larger data set for better prediction. In addition, authors have proposed Convolution Neural Network - Long Short-Term Memory (CNN-LSTM) model by allowing ResNEt-101 as pretrained model for CNN along with other pretrained deep learning models like ResNEt-101, Inception V3, DenseNET-169, and Inception-ResNET V2. In addition to the prediction of chest X-ray images into different classes as COVID, COVID with Pneumonia, Viral Pneumonia, and Healthy, GradCAM has been used for giving a visual explanation for deep learning model resulting in higher accuracy which are ResNET-101, DenseNET-169 and CNN-LSTM. The GradCAM shows the Model built can predict the image perfectly. These would be stored in Cloud for access by doctors for medication. © 2022 IEEE.
Full text:
Available
Collection:
Databases of international organizations
Database:
Scopus
Type of study:
Prognostic study
Language:
English
Journal:
2022 IEEE International Conference on Communications, ICC 2022
Year:
2022
Document Type:
Article
Similar
MEDLINE
...
LILACS
LIS