Camera-based on-line short cessation of breathing detection

Ilde Lorato, Sander Stuijk, Mohammed Meftah, Wim Verkruijsse, Gerard De Haan

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

3 Citations (Scopus)
31 Downloads (Pure)

Abstract

Apnea detection is extremely important in neonatal settings because hypoxia can lead to permanent impairment. Short cessations of breathing are very common in infants and could be used for example for the prediction of longer apneas. The aim of this study is to investigate the accuracy of our on-line cessation of breathing detector. Signals obtained through camera-based respiration monitoring were analyzed in five infants with 91 annotated cessations of breathing. The method proposed is based on the comparison of short-term and long-term standard deviations allowing the detection of sudden amplitude reduction in the signal with a low latency. A new strategy able to detect short cessations of breathing on-line was successfully validated yielding an average accuracy of 93%.

Original languageEnglish
Title of host publicationProceedings - 2019 International Conference on Computer Vision Workshop, ICCVW 2019
PublisherInstitute of Electrical and Electronics Engineers
Pages1656-1663
Number of pages8
ISBN (Electronic)9781728150239
DOIs
Publication statusPublished - 28 Oct 2019
EventICCV 2019 Workshops: The IEEE International Conference on Computer Vision (ICCV) Workshops - The 2nd International Workshop on Computer Vision for Physiological Measurement (CVPM) - Seoul, Korea, Republic of
Duration: 27 Oct 20192 Nov 2019

Conference

ConferenceICCV 2019 Workshops
CountryKorea, Republic of
CitySeoul
Period27/10/192/11/19

Keywords

  • Apnea
  • Apnea detection
  • Camera
  • Cessation of breathing
  • Infants
  • NICU
  • On line detection
  • Respiration
  • Short apnea
  • Video

Fingerprint Dive into the research topics of 'Camera-based on-line short cessation of breathing detection'. Together they form a unique fingerprint.

Cite this