DOIONLINE 
DOIONLINE Publish In |
International Journal of Advance Computational Engineering and Networking (IJACEN)-IJACEN |
 Journal Home Volume Issue |
||||||||
Issue |
Volume-9,Issue-10 ( Oct, 2021 ) | |||||||||
Paper Title |
Automatic Diagnosis of Pneumonia using Convolution Neural Networks and VGG-16 | |||||||||
Author Name |
Riya Verma, Neha Gupta | |||||||||
Affilition |
B.Tech, Department of Information Technology, JSSATE, Dr. A.P.J. Abdul Kalam Technical University, Uttar Pradesh, Lucknow, India Assistant Professor, Information Technology, JSSATE, Dr. A.P.J. Abdul Kalam Technical University, Uttar Pradesh, Lucknow, India | |||||||||
Pages |
6-11 | |||||||||
Abstract |
Inflammatory lung disease Pneumonia affects the small air sacs in the lungs called Alveoli. It is caused by a bacteria Streptococcus pneumoniae. Symptomsgenerally includes some combination of productive or dry cough, chest pains, difficulty in breathing and fever. In India, where Covid-19 is so prominent,In India, community-acquired pneumonia (CAP) is a major risk factor for mortality, accounting for one out of every three fatalities, according to the World Health Organization (WHO). The diagnosis as well as prognosis of Pneumonia generally requires the demonstration a chest imaging radiograph ofa patient suffering from a clinically comparable condition (eg: fever, dyspnoea, cough, and sputum production). Thus, by Developing an automated method for detecting pneumonia might be advantageous for rapid illness treatment, especially in distant regions. By using the DL models, we can analyse various chest radiographs. Nowadays, Convolution Neural network has gained much of the attention in disease classification. Moreover, the features which were trained in the initial stages of CNN model are useful in the classification task. A pre-trained DenseNet-201 is used to obtain a good accuracy. A total of 5219 chest Fig 1: Pneumonia Infected Lung X-ray pictures, including pneumonia, were obtained. Prior to the deep transfer learning classification task, images from conventional chest radiographs were pre-processed and trained on. 91 percent of the normal and pneumonia pictures can be correctly classified. As a result, the suggested research could be beneficial in pneumonia detection and can also be used in various private sectors for the Patients with pneumonia could be screened correctly. Keywords - Normal And Pneumonia Lungs; Pneumonia; Chest Radiographs; Deep Learning; DenseNet-201; Image Processing; Transfer Learning | |||||||||
| View Paper | ||||||||||
