June 2017
Volume 58, Issue 8
Open Access
ARVO Annual Meeting Abstract  |   June 2017
Automatic analysis of retinal oximetry images
Author Affiliations & Notes
  • Sveinn Hakon Hardarson
    Institute of Physiology, University of Iceland, Reykjavik, Iceland
  • Robert Arnar Karlsson
    Oxymap ehf., Reykjavik, Iceland
  • Olof Birna Olafsdottir
    Ophthalmology, University of Iceland, Reykjavik, Iceland
  • Thorunn S Eliasdottir
    Ophthalmology, University of Iceland, Reykjavik, Iceland
  • Toke Bek
    Aarhus University Hospital, Aarhus, Denmark
  • Einar Stefansson
    Ophthalmology, University of Iceland, Reykjavik, Iceland
  • Footnotes
    Commercial Relationships   Sveinn Hardarson, Oxymap ehf. (C), Oxymap ehf. (I), Oxymap ehf. (P); Robert Karlsson, Oxymap ehf. (I), Oxymap ehf. (E), Oxymap ehf. (P); Olof Olafsdottir, None; Thorunn Eliasdottir, None; Toke Bek, None; Einar Stefansson, Oxymap ehf. (I), Oxymap ehf. (P), Oxymap ehf. (S)
  • Footnotes
    Support  Icelandic Research Council
Investigative Ophthalmology & Visual Science June 2017, Vol.58, 681. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Sveinn Hakon Hardarson, Robert Arnar Karlsson, Olof Birna Olafsdottir, Thorunn S Eliasdottir, Toke Bek, Einar Stefansson; Automatic analysis of retinal oximetry images. Invest. Ophthalmol. Vis. Sci. 2017;58(8):681.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Purpose : Oxygen saturation in retinal blood vessels can be measured from fundus images, taken at two different wavelengths. The purpose of the study was to develop and test software to automate and improve analysis of oximetry images.

Methods : The Oxymap T1 oximeter takes two images at the same time, one with light at a wavelength that is sensitive to changes in oxygen saturation (600nm) and one at an insensitive wavelength (570nm). Oximetry images were analysed with manual and automatic approaches. The manual analysis requires that the user selects vessel segments for analysis according to a detailed protocol. The automatic analysis software selects measurement points and automatically calculates median saturation values for each eye.
For testing of repeatability and sensitivity, images were taken of 22 healthy individuals. These images were also used to calibrate the automatic method. Additional testing was performed on oximetry images of 54 diabetic patients.

Results : Standard devation of repeated manual measurements of the same healthy eye was 0.96 percentage points for arterioles and 1.95 percentage points for venules. The corresponding values for automatic measurements were 1.17 (arterioles) and 1.61 percentage points (venules). According to the automatic method, oxygen saturation in retinal arterioles was 92.6±3.1% during breathing of room air and 94.8±2.7% during pure oxygen breathing (p<0.0001). Corresponding values for venules were 58.8±5.2% (normoxia) and 78.2±7.6% (hyperoxia, p<0.0001). The table shows results for 54 diabetic patients, analyzed with the automatic software.

Arteriolar saturation in the PDR group was significantly higher than saturation in the other two groups (p<0.05). For venules, differences between all pairs of groups were statistically significant.

Conclusions : Automatic analysis of retinal oximetry images is faster and less subjective than manual analysis. Repeatability and sensitivity of the automatic method have been established. The automatic analysis can be used for oximetry on diabetic patients, where it yields similar results as have previously been found with the manual approach.

This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.

 

Table. Oxygen saturation (mean±SD) measured with automatic software. DM=Diabetes mellitus. NPDR=Non-proliferative diabetic retinopathy. PDR= Proliferative diabetic retinopathy.

Table. Oxygen saturation (mean±SD) measured with automatic software. DM=Diabetes mellitus. NPDR=Non-proliferative diabetic retinopathy. PDR= Proliferative diabetic retinopathy.

 

Figure. Automatic classification into arterioles and venules. Current software requires minor manual correction of classification.

Figure. Automatic classification into arterioles and venules. Current software requires minor manual correction of classification.

×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×