International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 09 Issue: 05 | May 2022
p-ISSN: 2395-0072
www.irjet.net
PNEUMONIA DIAGNOSIS USING CHEST X-RAY IMAGES AND CNN M S Naga Sathyashree1, Lavanya2, Manvitha B Patil3, Mounika V4, Dr.Kiran Kumari Patil5 1,2,3,4 Student,
School of computer science and engineering, Reva University, Karnataka, India and Professor of UIIC, Reva University, Karnataka, India ---------------------------------------------------------------------***--------------------------------------------------------------------5Director
Abstract - A pneumonia diagnosis system was developed
radiologist X-ray of the chest to detect pneumonia symptoms radiographs. The radiologist expert usually requires other information from the patient, such as the detailed medical record and phlegm condition [5]. It would be advantageous if an automated classification system can be developed to assist medical advisors or radiologists in the diagnosis of pneumonia.
using convolutional neural network (CNN) based feature extraction. InceptionV3 CNN was used to perform feature extraction from chest X-ray images. The extracted feature was used to train three classification algorithm models to predict the cases of pneumonia from the Kaggle dataset. The three models are Support Vector Machines, Neural Networks, and KNearest Neighbour The confusion matrix and performance evaluation were presented to represent the sensitivity, accuracy, precision, and specificity of each of the models. Results show that . The sensitivity of the Neural Network model was 84.1 percent, followed by support vector machines (83.5 percent) and the K-Nearest Neighbour Algorithm (83.5 percent) (83.3 percent ). The Support vector machines model obtained the highest AUC of all the classification models, at 93.1 percent.
Since X-ray radiographs are essentially images, CNN can be used to extract features VGG-16 and DenseNet-169, are examples of XCeption.are some of the CNN models that have been utilised for image identification of pneumonia [2, 3]. Rules-based, Bayesian network, Fuzzy C-means method, To predict, support vector machines, Nave Bayers, K-Nearest Neighbor, random forest, and other types of classifiers could be employed. The case of pneumonia.and Decision Tree [2, 8, 9]. Chapman and co-workers reported that the decision tree attained a precision of 85%, followed by rules-based (80%) and Bayesian network (72%), for the identification of pneumonia [8]. However, most of the papers focused on the feature extractions and did not evaluate the performance on different classifiers [9, 10].
Key Words: convolutional neural network, K-Nearest Neighbour, InceptionV3 CNN, X-ray Images, Neural Networks.
1. INTRODUCTION
1.1Related work
Pneumonia is a common illness for the childhood community ranging from bacteria to viral pneumonia or the combination of both [1]. Pneumonia is life-threatening and one of the primary causes of excessive child mortality rates in rural settings. According to the World Health Organization (WHO), pneumonia is responsible for one-third of all infant fatalities in India [2].
J. Zhou and W. Ge proposed A common, deadly, but preventable consequence of an Stroke-associated pneumonia (SAP) (AIS) is a kind of acute ischemic stroke. Identifying people who are most likely to develop SAP is crucial. as soon as possible. On the other hand, Previous clinical prediction methods have not been widely used. practise. As a result, we set out to use machine learning (ML) techniques to create a model that may predict SAP in Chinese AIS patients .Although challenging to implement, the XGBoost model, which comprises six common traits, can ISAN and PNA scores do not accurately predict SAP in Chinese AIS patients.
The presence of an aberrant area known as lung opacity, which looks opaque due to the attenuation of the x-ray beam in comparison to the surrounding tissues is required for the diagnosis of pneumonia [3]. Traditional X-ray chest radiography, Magnetic resonance imaging (MRI) and computerised tomography (CT) scan (MRI)are all options for detecting pneumonia [4, 5]. Among these methods, X-ray chest radiography is the most economical option compared to other imaging diagnostics for pneumonia detection [6].
There is currently no equipment available for early diagnosis of pneumonia caused by using a ventilator, according to Chung-Hung Shih and Yu-Hsuan Liao (VAP). As a result, he recommends employing an offline gas detection device to track the development of pneumonia metabolites and to identify them early. The new method collects breath samples from VAP patients using a e-nose with a low-cost microarray is simple to connect to an ICU mechanical ventilation system. However, this is the standard approach of implementing apps.
However, X-ray chest radiography is inferior in diagnosing pneumonia, especially for patients below five years old. This is due to the subtle differences in terms of scale, shape, intensity, and textures, which complicates the diagnosis [7]. Besides, other illness such as lung scarring, and congestive heart failure could also be misidentified as pneumonia [2]. Therefore, pneumonia diagnosis requires a skillful
© 2022, IRJET
|
Impact Factor value: 7.529
|
ISO 9001:2008 Certified Journal
|
Page 1793