eduzhai > Applied Sciences > Engineering >

DeScarGAN Disease-Specific Anomaly Detection with Weak Supervision

  • king
  • (0) Download
  • 20210505
  • Save

... pages left unread,continue reading

Document pages: 13 pages

Abstract: Anomaly detection and localization in medical images is a challenging task,especially when the anomaly exhibits a change of existing structures, e.g.,brain atrophy or changes in the pleural space due to pleural effusions. In thiswork, we present a weakly supervised and detail-preserving method that is ableto detect structural changes of existing anatomical structures. In contrast tostandard anomaly detection methods, our method extracts information about thedisease characteristics from two groups: a group of patients affected by thesame disease and a healthy control group. Together with identity-preservingmechanisms, this enables our method to extract highly disease-specificcharacteristics for a more detailed detection of structural changes. Wedesigned a specific synthetic data set to evaluate and compare our methodagainst state-of-the-art anomaly detection methods. Finally, we show theperformance of our method on chest X-ray images. Our method called DeScarGANoutperforms other anomaly detection methods on the synthetic data set and byvisual inspection on the chest X-ray image data set.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×