key: cord-0682361-do8o5q2t authors: Thirukrishna, J. T.; Krishna, Sanda Reddy Sai; Shashank, Policherla; Srikanth, S.; Raghu, V. title: Survey on Diagnosing CORONA VIRUS from Radiography Chest X-ray Images Using Convolutional Neural Networks date: 2022-01-08 journal: Wirel Pers Commun DOI: 10.1007/s11277-022-09463-x sha: 3b34988f886b06469b86cc828ed5ad9a591fbf5e doc_id: 682361 cord_uid: do8o5q2t Corona Virus continues to harms its effects on the people lives across the globe. The screening of infected persons has to be identified is a vital step because it is a fast and low-cost way. Certain above mentioned things can be recognized by chest X-ray images that plays a significant role and also used for examining in detection of CORONA VIRUS(COVID-19). Here radiological chest X-rays are easily available with low cost only. In this survey paper, Convolutional Neural Network(CNN) based solution that will benefit in detection of the Covid-19 positive patients using radiography chest X-Ray images. To test the efficiency of the solution, using data sets of publicly available X-Ray images of Corona virus positive cases and negative cases. Images of positive Corona Virus patients and pictures of healthy person images are divided into testing images and trainable images. The solution which are providing the good results with classification accuracy within the test set-up. Then GUI based application supports for medical examination areas. This GUI application can be used on any computer and performed by any medical examiner or technician to determine Corona Virus positive patients using radiography X-ray images. The result will be precisely obtaining the Covid-19 Patient analysis through the chest X-ray images and also results may be retrieve within a few seconds. COVID-19 is an infectious and fast spreading deadly virus and it was spreading all over the globe. The World Health Organization declared COVID-19 as a pandemic disease on March 11th 2020.The announcement of the pandemic also starts the panic of the increasing the spread of CORONA VIRUS [1, 2] . It is illustrated as a global safety emergency of its time and it has spread everywhere across all different countries. Government of varied nations are imposed different limitations and restrictions such as flight limitations, lockdown, social distancing and spreading awareness of the consciousness about cleanliness. But the Virus was spread at a high speed all over the world. For the infected people, the virus was directly attacked on the lungs. There are some assumptions that old or elder people with other diseases such as diabetes, Blood pressure etc., will be infected easily and it may affect their health deeply. In early stages, there is no correct and proper medical diagnosis for COVID-19. About 78,115,053 positive cases are found across all countries in the world until 24th DEC 2020, where 1,717,640 deaths and 54,890,244 recovered cases were found [3] . In order to prevent this virus, the sick patient has to be screened with proper medical diagnosis. At early stages the detection was done by testing kits manually using a technique called Reverse Transcription Polymerase Chain Response (RT-PCR) test on respiratory tracts [4] . The procedure which was used earlier was used to detect the disease. However, the testing method was manual, complicated, lack of equipment, and time-taking procedure with a normal positive success rate. The symptoms of the COVID-19 virus are having emphysema causing fever, whooping cough and breathing failure. Most of the CORONA VIRUS cases have identical similar spots on radiography chest X-ray photographs, those identical spots can be easily identified by comparing with other positive patient cases. Even though normal lungs X-ray images may serve early broadcast of infected cases, the X-rays of differing viral cases of pneumonia are comparably which may protrude with various other contagious and erythrogenic. Hence, it is hard for radiologist to identify corona virus from other different types of virus [5] . The complications of coronavirus are like viral infection and it can sometimes cause an incorrect separate within the current conditions. Hence, a wrong treatment can cause a non-corona viral infection is wrongly decided as supportable of getting CORONA VIRUS and during this process, giving in treatment with high price, and risk of implementing a positive CORONA VIRUS patient. Presenting, a lot of medical difficulties like brain disease detection, any many other detection, are using Artificial Intelligence (AI) based solutions [6] . For image classification using the Deep Learning techniques therefore it can reveal images with high quality. In Convolutional Neural Network has been displayed incredibly useful in learning and extraction, thus widely considered and approved by many research persons and groups. Convolutional Neural Network is used to increase image quality in dim light images from a very speed endoscopic and was put in to differentiate the thought of respiratory lungs through images, the finale of pneumonia by means of chest X-ray images [7] . According to transfer learning concept in Deep Learning was used for the detection of pathology utilizing trained ImageNet designs. Due to panic situation the testing of CORONA VIRUS testing is present a tough task due to the unfeasible of the diagnosis system. Due to the less harness of CORONA VIRUS testing kits, we need to look upon various diagnosis procedures. Since CORONA VIRUS present on the cells called epithelial cells that presents on the lungs. We are going to use X-rays images to find the presence of cells on an infected lung [8] . The examiners are using radiography X-ray images to research pneumonia and many other lungs related diseases. In this present world many hospitals having their own Radiography X-ray imagining machines. So check patient X-ray images instead of using testing kits, whether it is infected with COVID-19 or not [9] . The drawback of the radiography examiner can't able to diagnose many patient X-ray images very fast and correctly. Hence, developing an automated analysis application will save medical field person's precious time. Today, many are describing deep-learning techniques are the best for the image classification [10] . In efforts for regulating spreading of corona virus, an outsized percent of suspicious cases need to be examined for correct medication and quarantine. Pathogenic research government facility testing provides highest accuracy outcome, even though sometimes it predicts wrong or negative results [11] . Fast & accurate techniques are badly required to overcome this pandemics situation. During this pandemic situation creating model that diagnosis Corona gives more advantages to us for following social distance, as Covid virus attacks epithelial cells that are present in respiratory tracks of lungs, creating model that identifies these cells and predict the users affected with positive or not. Here the model would extract the features like identifying these cells, so giving results even faster than pathogenic test So by increasing the chances of saving life's and time to control the disease by predicting the person results faster than before. After doing survey concluded on using Deep Convolutional Neural Network (DCNN), a model that mainly focus on classifying radiography X-ray images by using classification techniques of Deep Learning. As the project main motive to save life's, accuracy takes important role for doing this, so by adding more X-ray images for training the model and performing more iterations on the model, the Deep Convolutional Neural Network(DCNN) accuracy are often improved more for the model [12] . The process of identifying and detecting COVID virus has become more importance allaround the world for some months. Covid virus has taken the first place for spreading so fast that has become hard to control [13] . Covid has become so hard for detecting as the person are not showing symptoms immediately. Thus it is more important to find new methods to differentiate the Covid positive people with normal people to eliminate the possibility. Artificial Learning can be used to examine a person for COVID-19 as an alternative to traditional time-consuming and expensive methods [14] . Even though there are many papers on Covid virus, this paper is focused on detecting Covid virus using Artificial Learning classification techniques using X-ray pictures and predict the people is positive to Covid virus or not. Several research areas have implemented Artificial Intelligence. One of the most advantages of AI is that they are often implemented during a trained model to classify unseen images. In this study, Artificial Intelligence was used to detect whether a patient is positive for Corona-virus by analysing their lungs X-ray pictures. Artificial intelligence can also be used to predict the status of person like he is positive to corona or not by using existing evidence. Thus, predicting possibilities within the immediate future can help authorities to adopt the required measures [15] . Concept is to get idea of techniques that are used to diagnosis the corona virus and the second concept is to forecast the number of cases that can come in upcoming days. The paper also suggests that existing models are delicate and unpredictable COVID-19 Diagnosis Using Deep Learning, the advantages of Machine Learning (ML) are increasing quickly in various fields such as malware detection, mobile malware detection, medicine, and knowledge retrieval. Deep-learning algorithms enable computational models composed of multiple processing layers to find out data representation through several abstraction layers. They trained a computer model to perform classification tasks directly from pictures. According to LeCun et al., deep-learning models feature high accuracies and may improve human output in certain instances [16] . X-ray machines use light or radio waves as radiation to look at the affected parts of the body due to cancers, lung diseases, bone dislocations, and injuries. Meanwhile, CT scans are used as sophisticated X-ray machines to look at the soft structures of active body parts for better views of the particular soft tissues and organs. The advantages of using X-rays over CT scans are that X-rays are quicker, safer, simpler, and less harmful than CT scans. The proposed a Convolutional Neural Network-based model to identify Covid patients using 450 X-ray images, in which 250 images belong to Covid patients and the 200 images belong to healthy people. He applied this concept in 3 Convolutional Neural Network models:-Residual Network-50, Residual inception v-3, and inception Convolutional Neural Network using five-fold cross-validation and submitted the report that Residual Network-50 had the only detection accuracy (98%) [17] . Extracting the attributes by using Deep Convolutional Neural Network algorithm from chest X-ray images and classified images as either infected or healthy using a SVM [18] . They collected two datasets the first dataset contains the collection of 25 infected patient's images and 25 non-infected patient's images while the other dataset contains X-ray images of 133 infected patients and 133 non-infected patients. They applied separate feature extractions on each dataset using various models and achieved a 94.38% accuracy with ResNet-50 and SVM. Furthermore, Hemdan et al. put forward a framework, called Covidx-net, which will assist radiologists in diagnosing Covid patients using X-ray. They evaluated their framework employing a collection of data of fifty X-ray images divided into two classes: 25 Covid-positive person images and 25 Covid-negative person images. The images used were resized to 224 × 224 pixels. The COVIDX-Net framework employs 7 deep learning models such as MobileNet. ResNet-v2. The authors trained model outcome indicate that the VGG19 and DenseNet models delivered comparable execution with an F-score of 91% for COVID-19 cases. In addition, an arrangement that uses multi-level threshold and an SVM to identify Covid persons by the help of using X-ray images. Their model was implemented by using 50images (20 healthy and 30 Covid infected) with a resolution of 512 × 512 pixels. Their arrangement achieved a performance of 94.32%, accuracy of 99.64% and specificity of 96.13.7% [19, 20] . To validate the proposed method, we require two types of chest related X-ray images they are common X-ray image and the other one is Corona affected patient X-ray image. While chest X-ray images of common category had been collected from a GitHub or from Kaggle dataset which contains some images selected from Chest X-ray dataset. Granting them in a notable number of infected COVID-19 patients universally, but chest x-ray images that are accessible online are not mostly significant and dispersed. Kaggle chest X-ray data is a far-fetched popular database containing chest X-ray images of normal or healthy, viral, and bacterial-pneumonia. Positive and mistrust CORONA VIRUS images were acquired in open available resources. Lungs X-ray images for regular and effected with pneumonia were used from this gathering to generate the up to date database collection. This model aims to organize a given chest X-ray image into common or COVID-19 category which contains few various stages gathering, pre-processing, feature selection, feature extraction, training. The detailed information of each stage has been in the following sections. The first stage is gathering, in this process we can collect the overall x-ray images in which it consists of both Corona and non-Corona x-ray images. Pre-processing refers to all transformation of the image before it is fed to the machine, training a convolutional Neural networks on the images. The Techniques Provided in Data Pre-processing. Data Cleansing. Cleaning "dirty" data. Real-world data tend to be incomplete, inconsistent and noisy. Data Integration, combining data from multiple sources, Data Transformation. Constructing data cube, Data Reduction. Reducing representation of data set. Data which tends to be incomplete leads to inconsistency and noise that affects the remaining part of the data containing x-ray attributes. Data cleaning can be adopted to resolve these issues. A selection algorithm can be seen for presenting new characteristics subsets, along with an approximation measure which tells the different detail subsets. Feature selection is used to simplify the models to make them users to be interpreted, and used to enhanced generalization by decreasing over fitting, avoid the curse of dimensionality. Feature extraction is also involved in minimizing the amount of available sources needed to describe a huge set. One of the major problems, while performing or analyse the complex data is the problem arise from the amount of variables involved in it. By examine of huge amount of the variables we required a huge amount of memory study power, and it also cause a sorting algorithm of over fitting samples and observe poorly to latest samples [20] (Figs. 1, 2) . Figure 3 depicts that CNN is very efficient algorithm which is used for image processing and pattern recognition. It has some features such as simple structure, less training parameters and adaptable. To training this model we required to indicate input training data source, required data transformation instructions, name of the information allocate that data to be anticipated. The evaluation parameters for X-ray images based on the analysis of segmented lungs area, enhanced difference of lung area, and mined image of abnormal tissues. The given input dataset X-ray images applied for identified and classified based on the convolutional neural networks. The large volume of Chest X-ray dataset that is four gigabytes of data sets from KAG-GLE applied into the Tensorflow software that will process as an input given dataset and training the dataset then classifying the images based on the above mentioned parameters and then perform test through manual testing and finally produces the output. From the above discussions, it is evident that recent advances have been made in the diagnosis of COVID-19 corona virus detection it lacks the early diagnostic tools. Even though there are several methods achieved noticeable advancements with high sensitivity or less false positive. There are many challenges to be addressed, to overcome all these challenges we are proposing Deep Convolutional Neural Networks method. The reason to choose CNN is that it can extract the spatial from the data using kernels, which other networks are not capable of. The proposed method uses D-CNN for the detection of COVID-19 based on the chest radiography X-ray images. A DCNN is collection of numerous fully connected and threshold layers, followed by different layers that determines the result. Funding This work was performed by above mentioned authors. It is a survey paper. There is no funding applicable for this work. Hence 'Not applicable' for this work. Conflict of interest All authors have participated in (a) conception and design, or analysis and interpretation of the data; (b) drafting the article or revising it critically for important intellectual content; and (c) approval of the final version. This manuscript has not been submitted to, nor is under review at, another journal or other publishing venue.. The authors have no affiliation with any organization with a direct or indirect financial interest in the subject matter discussed in the manuscript. conducted by Tamil Nadu Cricket Association (TNCA) and also played Zonal, Inter-zonal cricket matches conducted by Anna University. Engineering, Dayananda Sagar Academy of Technology and Management, Bangalore, India. He was attended more workshops. He was completed Architecting with google compute engine online course from Coursera. Policherla Shashank UG scholar, final year B.E in Information Science and Engineering, Dayananda Sagar Academy of Technology and Management, Bangalore, India. He was completed Architecting with google compute engine online course from Coursera. S. Srikanth UG scholar, B.E in Information Science and Engineering, Dayananda Sagar Academy of Technology and Management, Bangalore, India. He was completed Architecting with google compute engine online course from Coursera. DL-CRC: Deep learning-based chest radiograph classification for COVID-19 detection: A novel approach Early detection of COVID19 by deep learning transfer Deep learning based chest X-Ray image as a diagnostic tool for COVID-19 Covid-19 detection using CNN transfer learning from X-ray Images Identification of images of COVID-19 from Chest X-rays using deep learning: Comparing COGNEX VisionPro deep learning 1.0 software with open source convolutional neural networks ECOVNet: An ensemble of deep convolutional neural networks based on efficientnet to detect COVID-19 from chest X-rays Deep convolutional approaches for the analysis of Covid-19 using chest X-Ray images from portable devices A novel approach for QoS enhancement with revision scheme using SeDSR protocol in wireless sensor networks Automatic detection of COVID-19 infection using chest X-ray images through transfer learning Classification of covid-19 from chest x-ray images using deep convolutional neural networks. medRxiv Classification of COVID-19 in chest X-ray images using DeTraC deep convolutional neural network Revamp energy efficiency in homogeneous wireless sensor networks using optimized radio energy algorithm and power-aware distance source routing protocol Diagnosis of COVID-19 from X-rays using combined CNN-RNN architecture with transfer learning Pneumonia and COVID-19 detection using convolutional neural networks Finding covid-19 from chest x-rays using deep learning on a small dataset A lightweight deep learning model for COVID-19 detection Covid-19 detection using artificial intelligence Automated deep transfer learning-based approach for detection of COVID-19 infection in chestX-rays OREA for improving data packet transmission in wireless sensor networks with cloud security mechanism On the detection of covid-19 from chest x-ray images using CNNbased transfer learning He was completed Architecting with google compute engine online course from Coursera Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.