Automated ROP Diagnostic System based on Comparisons and U-Net Segmentation

Peng Tian, Jennifer Dy, Deniz Erdogmus, Susan Ostmo, J. Peter Peter Campbell, Michael F.F. Chiang, Stratis Ioannidis

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Retinopathy of Prematurity (ROP) is a disease affecting premature infants and may lead to childhood blindness. Due to lack of trained ophthalmologists, developing a fully automated ROP diagnostic system can significantly benefit the infants affected by ROP. Based on manually segmented features, previous work produces severity scores for ROP with excellent prediction accuracy. However, when automated segmentation employed, a significant accuracy drop is observed. In this paper, we show that U-Net segmentation, which is automated, comes at no accuracy loss.

Original languageEnglish (US)
Title of host publication14th ACM International Conference on PErvasive Technologies Related to Assistive Environments, PETRA 2021
PublisherAssociation for Computing Machinery
Pages33-36
Number of pages4
ISBN (Electronic)9781450387927
DOIs
StatePublished - Jun 29 2021
Event14th ACM International Conference on PErvasive Technologies Related to Assistive Environments, PETRA 2021 - Virtual, Online, Greece
Duration: Jun 29 2021Jul 1 2021

Publication series

NameACM International Conference Proceeding Series

Conference

Conference14th ACM International Conference on PErvasive Technologies Related to Assistive Environments, PETRA 2021
Country/TerritoryGreece
CityVirtual, Online
Period6/29/217/1/21

Keywords

  • learning from comparisons
  • Retinopathy of Prematurity
  • U-Net

ASJC Scopus subject areas

  • Human-Computer Interaction
  • Computer Networks and Communications
  • Computer Vision and Pattern Recognition
  • Software

Cite this