June 2013
Volume 54, Issue 15
ARVO Annual Meeting Abstract  |   June 2013
Clinical Utility of Thresholding Segmentation in Ultra-Widefield (Optos®) Fundus Autofluorescence Images
Author Affiliations & Notes
  • Sarah MacIver
    Univ of Waterloo Sch of Optom, Waterloo, ON, Canada
  • Jerome Sherman
    Clinical Sciences, SUNY State College of Optometry, New York, NY
    SUNY Eye Institute, State University of New York, New York, NY
  • Natalie Hutchings
    Univ of Waterloo Sch of Optom, Waterloo, ON, Canada
  • Footnotes
    Commercial Relationships Sarah MacIver, None; Jerome Sherman, Optos, Inc. (F), Optos, Inc. (C), Optos, Inc. (R), Annidis (C), Annidis (R), Zeiss (R); Natalie Hutchings, ALCON (F), ALCON (R)
  • Footnotes
    Support None
Investigative Ophthalmology & Visual Science June 2013, Vol.54, 5513. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Sarah MacIver, Jerome Sherman, Natalie Hutchings; Clinical Utility of Thresholding Segmentation in Ultra-Widefield (Optos®) Fundus Autofluorescence Images. Invest. Ophthalmol. Vis. Sci. 2013;54(15):5513.

      Download citation file:

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

  • Supplements

Purpose: Fundus Autofluorescence (AF) is an imaging modality that indicates retinal pigment epithelium (RPE) integrity with hypo-AF signifying RPE loss /atrophy and hyper-AF signifying RPE metabolic stress. This study investigates the clinical utility of thresholding segmentation for identifying areas of hyper/hypoAF in panoramic AF (PAF) images

Methods: A retrospective review was carried out for 20 PAF images of normal (n=2) and diseased (n=18) retinas, including maculopathies, posterior uveitis, and other peripheral retinal disorders. All PAF images were captured with the Optos 200Tx (200° ultra-widefield). Image analysis was carried out using Image J (NIH Image). A circular region of interest (ROI) centered around the macula and free from imaging artifacts was selected, eliminating the far periphery The maximum entropy threshold (MET) and the mean and SD of the grey level in ROI was determined. Four levels of AF were identified: High AF: pixels above MET +1SD, High-Mod AF: pixels above MET, Mod-Low AF: pixels below MET-1SD, and Low AF: pixels below mean gray-1SD. The percentage area of the ROI containing each AF classification of pixels was determined

Results: In normal images the MET and mean threshold were at a similar grey level increment. In diseased eyes, the average grey level for the MET was significantly higher than the average mean threshold (MET=118.8; Mean T= 73.3; paired t-test p<0.0001). The percentage area of the whole image showing mod-high and high AF was smaller for those cases with pathology (Mean [Range]: High AF 5.53% [0.43-14.11] Mod-High AF 15.49% [2.24-36.42]) than in those without pathology (Mean: High AF 20.28% Mod-High AF 49.53%), Cases with pathology had large areas with mod-low AF (Mean 65.10%) than those without pathology (Mean 33.80%). However, the percentage area of each AF classification was dependent on the condition. In diseased eyes, smaller percentage areas of the ROI exhibited moderate AF than in the normal eyes.

Conclusions: Segmentation using the MET and mean grey level offers an objective assessment of fundus AF images. MET and mean threshold were similar in normal eyes and showed large areas of moderate AF. Diseased eyes showed larger areas of High- or Low-AF (condition dependent) than moderate AF. These techniques offer potential for quantitative, longitudinal assessment of AF and, hence, RPE integrity

Keywords: 550 imaging/image analysis: clinical • 549 image processing • 701 retinal pigment epithelium  

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.