eduzhai > Applied Sciences > Engineering >

Efficient OCT Image Segmentation Using Neural Architecture Search

  • king
  • (0) Download
  • 20210505
  • Save

... pages left unread,continue reading

Document pages: 5 pages

Abstract: In this work, we propose a Neural Architecture Search (NAS) for retinal layersegmentation in Optical Coherence Tomography (OCT) scans. We incorporate theUnet architecture in the NAS framework as its backbone for the segmentation ofthe retinal layers in our collected and pre-processed OCT image dataset. At thepre-processing stage, we conduct super resolution and image processingtechniques on the raw OCT scans to improve the quality of the raw images. Forour search strategy, different primitive operations are suggested to find thedown- & up-sampling cell blocks, and the binary gate method is applied to makethe search strategy practical for the task in hand. We empirically evaluatedour method on our in-house OCT dataset. The experimental results demonstratethat the self-adapting NAS-Unet architecture substantially outperformed thecompetitive human-designed architecture by achieving 95.4 in mean Intersectionover Union metric and 78.7 in Dice similarity coefficient.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×