In classifying chest X-rays (CXR), machine learning, particularly transfer learning, has been widely implemented and has demonstrated an excellent range of accuracy. In this study, ResNet-50, ResNet-101, and VGG16, three popular transfer learning models, were compared in a CXR classification task. This study used a dataset containing CXRs of 3616 COVID- 19-positive cases, CXRs of 1345 viral pneumonia cases, CXRs of 6012 lung opacity (non-Covid lung infection) cases, as well as CXRs of 10,192 normal individuals and corresponding lung masks. The study used Keras and TensorFlow to import the pre-trained models and compare them using the exact same training sets, testing sets, and validation sets. The ResNet-50 achieves a classification accuracy of 90%, the ResNet-101 achieves a classification accuracy of 91%, and the VGG16 achieves a classification accuracy of 89%. This research reached the conclusion that ResNet-101 performs better in such image multiclassification tasks. We speculate that this is because ResNet-101 was trained based on residual learning which is easy to optimize.
The usage of X-rays for classifying and diagnosing pneumonia has shown an excellent range of exactness and accuracy. This paper presents a binary-classification model that diagnoses patients who may have pneumonia by inputting their x-ray images and introduces the concepts used to develop that model. Keras and TensorFlow libraries are used in this analysis to produce a convolutional neural network model. The training data set which are used to train the model contains 5216 samples which represent 5216 different patients with either pneumonia x-ray image or normal x-ray image. The testing data set contains 624 samples which show how well the model generalizes on the new dataset. The model produces a classification accuracy of 75% on the testing set.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.