Throughout  history,  epidemics  and  chronic  diseases  have  claimed  the  lives  of  many  people  and caused major crises that have taken a long time to overcome. The 2019 novel coronavirus (COVID-19) pandemic appeared in Wuhan, China in December 2019 and has become a serious public health problem worldwide. It is an acute resolved disease, but it can also be deadly, with a 2% case fatality  rate. The early and automatic diagnosis of Covid-19  may  be  beneficial  for  timely  referral  of  the  patient  to quarantine, and  monitoring of the spread of the disease.  Some  tests  requiring  significant time to produce results (days), and a projected up to 30% false positive rate, other timely approaches to diagnosis are worthy of investigation. Gold standard testing currently is a RNA-based assay using nasopharyngeal swabs.  Identifying positive COVID-19 cases in early stages helps with isolating the patients as quickly as possible, hence breaking the chain of transition and flattening the epidemic curve.

Medical images analysis is one of the most promising research areas, it provides  facilities  for  diagnosis  and  making  decisions  of  a  number  of  diseases  such  as  MERS, COVID-19. Recently, researchers, specialists, and companies around the world are rolling out deep learning and  image  processing-based  systems  that  can  fastly  process  hundreds  of  X-Ray  and computed tomography (CT) images to accelerate the diagnosis of pneumonia such as SARS, COVID-19, and aid in its containment.    The  false  negative  rate  is  projected  to  be  as  high  as  30%  and  test  results   can  take  some  time  to  obtain. X-ray  machines  are  widely  available  and  provide images  for  diagnosis  quickly.

Consequently, there has been an urgent surge of interest to develop Deep Neural Network (DNN)-based diagnosis solutions, mainly based on Convolutional Neural Networks (CNNs), to facilitate identification of positive COVID-19 cases. CNNs, however, are prone to lose spatial information between image instances and require large datasets.

In work Automatic detection from X-Ray images utilizing Transfer Learning with Convolutional Neural Networks, a dataset of X-Ray images from patients with common pneumonia,  Covid-19,  and  normal  incidents  was  utilized  for  the  automatic detection  of  the  Coronavirus. The aim  of  the  study  is  to  evaluate  the performance  of  Convolutional  Neural  Network  architectures proposed  over  recent  years  for  medical  image  classification. With  transfer  learning,  an  overall accuracy of 97.82% in the detection of Covid-19 is achieved.

In  paper Automated Methods for Detection and Classification Pneumonia based on X-Ray Images Using Deep Learning,  authors  present  a  comparison  of  recent  Deep  Convolutional Neural Network (DCNN) architectures for automatic binary classification of pneumonia images based fined tuned  versions  of  (VGG16,  VGG19,  DenseNet201,  Inception_ResNet_V2,  Inception_V3,  Resnet50, MobileNet_V2 and Xception). As result they can conclude that fine-tuned  version  of  Resnet50,  MobileNet_V2  and  Inception_Resnet_V2  show  highly  satisfactory performance with rate of increase in training and validation accuracy (more than 96% of accuracy).

In  work Diagnosing COVID-19 Pneumonia from X-Ray and CT Images using Deep Learning and Transfer Learning Algorithms,  authors  focus  on  proposing  AI  tools  that  can  be  used  by  radiologists  or  healthcare  professionals  to  diagnose COVID-19  cases  in  a  quick  and  accurate  manner. This  study  aims  to  build  a  comprehensive  dataset  of  X-rays and  CT  scan  images  from  multiple  sources  as  well  as  provides     a  simple  but  an  effective  COVID-19  detection  technique  using deep learning  and  transfer  learning  algorithms. A  simple  convolution  neural  network  (CNN)  and  modified  pre- trained  AlexNet  model  are  applied  on  the  prepared  X-rays  and CT  scan  images  dataset. The  result  of  the  experiments  shows  that the utilized models can provide accuracy up to 98% via pre-trained  network  and  94.1%  accuracy  by  using  the  modified CNN.

FINDING COVID-19 FROM CHEST X-RAYS USING DEEP LEARNING ON A SMALL DATASET  explores  how  useful  chest  X-ray  images can  be  in  diagnosing  COVID-19  disease. A  pre- trained deep convolutional neural network has been tuned on 102 COVID-19 cases and  102  other  pneumonia  cases  in  a  10-fold  cross  validation. On  a  test  set  of  20  unseen  COVID-19  cases  all    were  correctly  classified  and  more  than  95%  of  4,171  other  pneumonia  examples were correctly classified. COVID-CAPS: A CAPSULE NETWORK-BASED FRAMEWORK FOR IDENTIFICATION OF COVID-19 CASES FROM X-RAY IMAGES presents an alternative modeling framework based on Capsule Networks, referred to as the COVID-CAPS, being capable of handling small datasets, which is of significant importance due to sudden and rapid emergence of COVID-19. Initial results based on a dataset of X-ray images show that COVID-CAPS has advantage over previous CNN-based models.