In this study, we applied image processing of combined contrast limited adaptive histogram equalization (CLAHE) and wavelet de-noise processing in Faster R-CNN with the aim of improving the detection accuracy of nodules in images obtained from chest radiographs. The CNN network was selected between VGG-16 and ResNet50 based on the accuracy of image evaluation. Gradient-weighted class activation mapping (Grad-CAM) was used to verify the observation areas that contributed to the network classification. We then verified the detection accuracy for 53 new images of clinically confirmed nodules. In the image evaluation, the detection accuracy was higher with ResNet50. However, verification of the area of interest in the original images by Grad-CAM revealed that 36.0% of the images focused on areas other than lesions. Lesion detection was then attempted using Faster R-CNN in 104 clinical images. When the number of anchors in the verification was set to 30, the highest detection accuracy was 65.5%. Image processing performed with combined CLAHE and wavelet de-noise processing with Faster R-CNN achieved an accuracy of 76.4% for detecting of nodules in images obtained from chest radiographs.
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.