June 2022
Volume 63, Issue 7
Open Access
ARVO Annual Meeting Abstract  |   June 2022
Comparison of segmentation methods for measuring collagen fiber density in confocal reflectance images of the vitreous
Author Affiliations & Notes
  • Eileen Hwang
    Ophthalmology, University of Utah Health John A Moran Eye Center, Salt Lake City, Utah, United States
  • Denise Morgan
    Ophthalmology, University of Utah Health John A Moran Eye Center, Salt Lake City, Utah, United States
  • Mary Elizabeth Hartnett
    Ophthalmology, University of Utah Health John A Moran Eye Center, Salt Lake City, Utah, United States
  • Brittany Coats
    Mechanical Engineering, University of Utah, Salt Lake City, Utah, United States
  • Footnotes
    Commercial Relationships   Eileen Hwang None; Denise Morgan None; Mary Elizabeth Hartnett None; Brittany Coats None
  • Footnotes
    Support  Knights Templar Eye Foundation, NIH Core Grant EY014800, Research to Prevent Blindness, University of Utah 1U4U Innovation Funding, R01EY015130, R01EY017011
Investigative Ophthalmology & Visual Science June 2022, Vol.63, 4105 – F0069. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Eileen Hwang, Denise Morgan, Mary Elizabeth Hartnett, Brittany Coats; Comparison of segmentation methods for measuring collagen fiber density in confocal reflectance images of the vitreous. Invest. Ophthalmol. Vis. Sci. 2022;63(7):4105 – F0069.

      Download citation file:


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

      ×
  • Supplements
Abstract

Purpose : Spatial variation in vitreous collagen fiber density may play a role in vitreous aging. Methods for quantifying fiber density across the eye are not well developed. Our objective was to evaluate the performance of different image segmentation methods for detecting variation in vitreous collagen fiber density in confocal reflectance images.

Methods : Four right fresh young adult porcine eyes were sectioned in the sagittal plane at the nasal limbus. Two-dimensional 70um x 70um confocal reflectance images were obtained every 1mm along an anterior-posterior axis originating at the nasal limbus. Three methods were used to quantify fiber density: 1) manual fiber counting by a single grader; 2) automated fiber counting using the Fiji ridge detection plugin; 3) automated pixel thresholding using the Fiji moments algorithm. Pearson correlation was used to compare methods. Simple linear regression was used to evaluate anterior-posterior density trends.

Results : Confocal reflectance images of porcine vitreous demonstrated a network of linear fibers with a diameter of 1.18 ± 0.51um (mean ± SD). Manual fiber counting yielded an average density of 0.019 ± 0.006 fibers/um2, whereas automated fiber counting yielded an average density of 0.032 ± 0.014 fibers/um2. Automated pixel thresholding resulted in an average of 23 ± 9% of pixels occupied by signal. There were strong correlations between manual fiber counting and automated fiber counting (R-squared=0.80, p<0.0001) and between automated fiber counting and automated pixel thresholding (R-squared=0.85, p<0.0001). All three methods demonstrated a significant trend of decreasing density from anterior to posterior. The density was found to vary by 7.1% per mm (95%CI 4.0-10.2%; p<0.0001) by manual fiber counting and 8.2% per mm (95%CI 3.9-12.6%; p=0.0005) by automated fiber counting, and 7.1% per mm (95%CI 3.3-10.9%; p=0.0006) by automated pixel thresholding.

Conclusions : High correlation was observed between three methods for quantifying fiber density in vitreous confocal reflectance images. By all three methods, collagen fiber density was greater anteriorly, near the vitreous base, which is consistent with prior research. These results support the use of these methods for assessing relative differences in vitreous collagen fiber density.

This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.

 

 

×
×

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.

×