eduzhai > Applied Sciences > Engineering >

Robust Retinal Vessel Segmentation from a Data Augmentation Perspective

  • king
  • (0) Download
  • 20210505
  • Save

... pages left unread,continue reading

Document pages: 10 pages

Abstract: Retinal vessel segmentation is a fundamental step in screening, diagnosis,and treatment of various cardiovascular and ophthalmic diseases. Robustness isone of the most critical requirements for practical utilization, since the testimages may be captured using different fundus cameras, or be affected byvarious pathological changes. We investigate this problem from a dataaugmentation perspective, with the merits of no additional training data orinference time. In this paper, we propose two new data augmentation modules,namely, channel-wise random Gamma correction and channel-wise random vesselaugmentation. Given a training color fundus image, the former applies randomgamma correction on each color channel of the entire image, while the latterintentionally enhances or decreases only the fine-grained blood vessel regionsusing morphological transformations. With the additional training samplesgenerated by applying these two modules sequentially, a model could learn moreinvariant and discriminating features against both global and localdisturbances. Experimental results on both real-world and synthetic datasetsdemonstrate that our method can improve the performance and robustness of aclassic convolutional neural network architecture. Source codes are availablethis https URL

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×