For this purpose, in this study, we propose an artificial intelligence-based method has that will speed up the diagnosis of the pandemic by saving labour and expense. In the study, X-Ray images were processed with the most up-to-date deep image processing techniques, and an objective decision support system was created, independent of the doctors’ expertise. Our proposed system can classify x-ray images as Normal, Covid -19, and Viral Pneumonia using pre-trained deep learning networks (AlexNet, MobileNetV2, GoogleNet, ResNet8, and ResNet50) with transfer learning (TL) and K-fold validation (KFCV). The overall accuracies of the networks (AlexNet, MobileNetV2, GoogleNet, ResNet8, and ResNet50) were 98%, 97. 4%, 97. 4%, 97. 4%, 98. 59% respectively. Also, for the average performance metrics values (Sensitivity, Specificity, Precision, F1-score, and Negative predictive value(NPV)). The best results were recorded by ResNet 50+TL+KFCV (97. 9% Sensitivity, 98. 9% Specificity, 97. 9% Precision, 97. 9 % F1-Score, and 98. 9% NPV ). It is easy to diagnose in the advanced stages of the disease. As with most diseases, early diagnosis is important in COVID-19. With the proposed system based on deep learning, a very useful tool in combating the pandemic has been created by determining the disease at an early stage. The proposed system can also be used in areas with a shortage of health personnel such as rural and remote areas.
For this purpose, in this study, we propose an artificial intelligence-based method has that will
speed up
the diagnosis of the pandemic by saving
labour
and expense. In the study, X-Ray images
were processed
with the most up-to-date deep image processing techniques, and an objective decision support
system
was created
, independent of the doctors’ expertise. Our proposed
system
can classify x-ray images as Normal,
Covid
-19, and Viral Pneumonia using pre-trained deep learning networks (
AlexNet
, MobileNetV2,
GoogleNet
, ResNet8, and ResNet50) with transfer learning (TL) and K-fold validation (
KFCV
). The
overall
accuracies of the networks (
AlexNet
, MobileNetV2,
GoogleNet
, ResNet8, and ResNet50) were 98%, 97. 4%, 97. 4%, 97.
4%
, 98. 59%
respectively
.
Also
, for the average performance metrics values (Sensitivity, Specificity, Precision, F1-score, and
Negative
predictive value(NPV)). The best results
were recorded
by
ResNet
50+TL+
KFCV
(97. 9% Sensitivity, 98. 9% Specificity, 97.
9%
Precision, 97. 9 % F1-Score, and 98. 9% NPV
)
. It is easy to diagnose in the advanced stages of the disease. As with most diseases, early diagnosis is
important
in COVID-19. With the proposed
system
based on deep learning, a
very
useful tool in combating the pandemic has
been created
by determining the disease at an early stage. The proposed
system
can
also
be
used
in areas with a shortage of health personnel such as rural and remote areas.