International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 11 Issue: 12 | Dec 2024
p-ISSN: 2395-0072
www.irjet.net
PNEUMONIA DETECTION BASED ON CONVOLUTION NEURAL NETWORK Mr. Basant Rawot1, Dr. Heli Shah2, 1Parul Institute of Technology, Parul University, India.
2HOD of The Department, Parul Institute of Technology, Parul University, India.
----------------------------------------------------------------------------***-------------------------------------------------------------------------Abstract: In the past year, the novel coronavirus has affected over 1 million individuals and resulted in more than 50,000 deaths. COVID19 infection has the potential to progress into pneumonia, a condition that can be identified through a chest x-ray and should be managed appropriately. This study presents a novel approach for automatically detecting COVID-19 infection using chest radiographs. The dataset compiled for this study comprises 480 X-rays obtained from patients diagnosed with coronavirus and from healthy patients. Due to the limited availability of COVID-19 patient images, we will utilize transfer learning principles for this undertaking. We employ various architectures of convolutional neural networks (CNNs) that have been trained on ImageNet. We then modify these networks to function as feature extractors specifically for X-ray images. CNNs are subsequently fused with integrated machine learning methodologies, including Support vector machines (SVM) and LSTM. The results indicate 93.75% accuracy with an F1 score of 93.86%. Hence, the suggested method exhibits efficacy in identifying pneumonia. Keywords: Convolution neural networks, ImageNet, Multilayer perceptron, Support vector machine, X-ray images
1. Introduction A lung air sac infection causes pneumonia. The lung structure is made up of pulmonary alveoli, bronchus, and lobes connected to the main veins. Every part has to work correctly to guarantee an effective exchange of carbon dioxide and oxygen. Along with fluids in their air sacs, pneumonia patients will have alveolar inflammation. There are only a few of the medical examination technologies developed such as CT scans, X-ray and ultrasound as a result of medical technology progress. Pneumonia diagnosis is regarded as best achieved with CT scans. VGG and ResNet networks are utilized to accurately classify lung ultrasound images of pneumonia according to different clinical phases using self-generated LUS datasets. The introduction of deep learning in recent years has resulted in a paradigm shift in medical image analysis. Convolutional neural networks (CNNs) are highly effective at tasks like classifying images, segmenting parts of images, and identifying features within images. Because of these capabilities, CNNs are widely used in creating automated systems for medical imaging. CNNs have shown potential in detecting pneumonia, offering the possibility of quickly and independently identifying lung infections. This advancement could significantly transform patient care. This study focuses on accurately detecting pneumonia in the lungs using chest X-rays, which can be used in the real world by medical practitioners to treat pneumonia. The problem with traditional methods involves lots of time and resource intensive in the diagnosis of pneumonia. Hence, the use of artificial technology such as convolution neural networks solves the problem by simplifying the process of diagnosis of pneumonia. The main objectives of this study are: a) Integrating deep learning techniques for pneumonia detection. b) Using feature extraction and Image classification for diagnosis.
2. Literature Review The research paper "Pneumonia Detection Using CNN-based Feature Extraction" focuses on creating a Convolutional system to effectively diagnose pneumonia in X-rays. It highlighted the global impact of pneumonia as a leading cause of death and the difficulties in diagnosing it accurately. The authors then describe the proposed CNN-based approach for detection and categorization of images. The CNN model was trained on a dataset of 5856 chest X-ray images, consisting of 2,700 normal and
© 2024, IRJET
|
Impact Factor value: 8.315
|
ISO 9001:2008 Certified Journal
|
Page 272