Transfer Learning and Feature Extraction of Chest X-ray Images for Deep Convolutional Neural Network (CNN)- based Pneumonia Detection

Pneumonia is a lung inflammation that mostly affects the tiny air sacs known as alveoli. A productive or dry cough, chest discomfort, a fever, and breathing difficulties are typical symptoms. The disorder can range in severity. The most prevalent causes of pneumonia are infections with viruses or ba...

Full description

Saved in:
Bibliographic Details
Published in2023 4th IEEE Global Conference for Advancement in Technology (GCAT) pp. 1 - 7
Main Authors Gill, Kanwarpartap Singh, Anand, Vatsala, Gupta, Rupesh
Format Conference Proceeding
LanguageEnglish
Published IEEE 06.10.2023
Subjects
Artificial intelligence
Classification
CNN
Confusion matrix
Convolutional neural networks
Data models
Deep Learning
Lung
Model Training
Pneumonia Disease
Pulmonary diseases
Sequential Model
Task analysis
Transfer learning
Online AccessGet full text

Cover

More Information
Summary:Pneumonia is a lung inflammation that mostly affects the tiny air sacs known as alveoli. A productive or dry cough, chest discomfort, a fever, and breathing difficulties are typical symptoms. The disorder can range in severity. The most prevalent causes of pneumonia are infections with viruses or bacteria; other microbes, certain drugs, or illnesses including autoimmune disorders are less frequently to blame. Cystic fibrosis, chronic obstructive pulmonary disease (COPD), asthma, diabetes, heart failure, a history of smoking, having a defective cough reflex, such as after a stroke, and having a weakened immune system are risk factors. The physical exam and symptoms are frequently used to make a diagnosis. Blood tests, sputum culture, and chest X-rays can all support the diagnosis. The location of the infection may be used to categorise the illness, such as community, hospital, or healthcare-associated pneumonia. Convolutional neural networks (CNNs) are used in this study to diagnose pneumonia. Deep learning is frequently used in the medical industry. CNNs, like the one suggested in this study, are well suited for image-based tasks like medical image analysis because they can automatically learn relevant features from images and make predictions based on those features. This outcome clearly predicts the accuracy of the proposed CNN, which is 92.6 percent, when exploratory data analysis is done.
DOI:10.1109/GCAT59970.2023.10353421