Abstract:
to Lung Cancer, Pneumonia, Tuberculosis, and COVID-19, which have a very terrible effect
on the body, the patient may die and COVID-19 spreads very easily. It is possible to get rid of
these diseases if they are detected as soon as possible. With the help of Artificial Intelligence
detection of these types of diseases will be very easy and quick. AI can detect these types of
diseases automatically and accurately. The use of this type of automated and accurate process in
the medical sciences will be very beneficial in the modern era of science and technology. In our
study, we use Deep Learning to detect 4 major types of lung diseases including- Lung Cancer,
Pneumonia, Tuberculosis and COVID-19. We have used a total of 7,255 patients’ chest X-ray
images as the dataset. The dataset is divided into 5 classes including- Pneumonia, COVID-19,
Tuberculosis, Lung Cancer and Normal. Classify those diseases we used 6 very popular deep
transfer learning models including- ResNet50, VGG16, EfficientNet, VGG19, MobileNet, and
InceptionV3. Among them, the best accuracy has been gained in ResNet50. Based on chest X-ray
images, we obtained a test accuracy of 97.65% and a training accuracy of 99.67% from ResNet50.
We also obtained trained accuracy of 97.44%, 97.23%, 96.82%, 94.33% and 90.73% respectively
in VGG16, Efficient Net, VGG19, Mobile Net and Inception-V3.
