Esophageal cancer is the sixth leading cause of cancer-related mortality worldwide. Barium esophagram is an inexpensive, noninvasive, and widely available procedure. However, the variation in tumor size, the interference of surrounding tissues, and esophageal deformation make tumor detection very challenging. This paper proposes a novel algorithm for detecting esophageal cancer, consisting of a detection network and a classification network. A deformable residual bottleneck block is proposed to replace the residual bottleneck block in ResNet50 to sample the deformed esophagus adaptively. The improved ResNet50 is used as the backbone network of the detection network, which is combined with FPN network to predict tumors of various sizes. Then, regions of interest detected from each image of the barium esophagram by the detection network are used as the input of the classification network. The attention module CBAM is introduced into the classification network to enhance the significance of the esophageal region, and improve the classification accuracy. The algorithm is evaluated on 40 positive (1166 images) and 53 negative cases (1547 images), the accuracy of the algorithm on positive and negative cases are 100% (40/40) and 84.90% (45/53) respectively. The experiment demonstrates that our proposed method yields promising results with the barium esophagram dataset.
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.