June 2015
Volume 56, Issue 7
Free
ARVO Annual Meeting Abstract  |   June 2015
Automated Segmentation and Quantification in SD-OCT Images to Predict Future Geographic Atrophy Involvement
Author Affiliations & Notes
  • Sijie Niu
    Computer Science and Technology, Nanjing University of Science and Technology, Nanjing, China
    Radiology, Stanford University School of Medicine, Stanford, CA
  • Luis De Sisternes
    Radiology, Stanford University School of Medicine, Stanford, CA
  • Qiang Chen
    Computer Science and Technology, Nanjing University of Science and Technology, Nanjing, China
  • Theodore Leng
    Byers Eye Institute at Stanford, Stanford University School of Medicine, Palo Alto, CA
  • Daniel L Rubin
    Radiology, Stanford University School of Medicine, Stanford, CA
    Medicine (Biomedical Informatics), Stanford University School of Medicine, Stanford, CA
  • Footnotes
    Commercial Relationships Sijie Niu, None; Luis De Sisternes, None; Qiang Chen, None; Theodore Leng, None; Daniel Rubin, None
  • Footnotes
    Support None
Investigative Ophthalmology & Visual Science June 2015, Vol.56, 2839. doi:https://doi.org/
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Sijie Niu, Luis De Sisternes, Qiang Chen, Theodore Leng, Daniel L Rubin; Automated Segmentation and Quantification in SD-OCT Images to Predict Future Geographic Atrophy Involvement. Invest. Ophthalmol. Vis. Sci. 2015;56(7 ):2839. doi: https://doi.org/.

      Download citation file:


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

      ×
  • Supplements
Abstract
 
Purpose
 

Predicting future expansion of Geographic Atrophy (GA) in advanced non-exudative age-related macular degeneration (AMD) is currently based on manual measurement of the current GA region and presents limited performance. We analyzed a series of novel automatically extracted features in SD-OCT images as predictors of GA appearance and expansion in topographic maps. The extracted features include those measured directly from the current affected region and from automatically detected candidate regions of future GA appearance or expansion.

 
Methods
 

GA expansion was analyzed in 18 eyes from 12 advanced non-exudative AMD patients (mean age at baseline 75.81 years (SD 2.97), 56.25% female). For each patient, SD-OCT images were collected at baseline and at a mean follow-up at 2.9 years (SD 0.1 years) from baseline. GA regions were automatically segmented from the images using novel image analysis methods. Candidate regions of future GA involvement were automatically outlined in the baseline images by detecting regions of photoreceptor loss and evolving drusen. We generated topographic maps displaying the relationship of these candidate regions with actual GA expansion from baseline to follow-up for each patient. We extracted quantitative features from the baseline images describing the area of current GA, as well as the area and location of possible GA expansion within the macula. We analyzed the correlation of each of these features with the actual measured expansion rate of GA and its rate of involvement in different topographic macula regions.

 
Results
 

Good correlation was observed qualitatively between regions marked at baseline as candidate locations of future GA involvement and actual GA expansion from baseline to follow-up. A significant correlation (p<0.05) with GA expansion rate was observed for GA area measured at baseline (cc=0.52) and a higher correlation was observed when considering the candidate regions of future expansion (cc=0.56). The location of these candidate regions also showed significant correlation with localized future GA involvement.

 
Conclusions
 

The automated methods showed potential for identiflying regions of future GA involvement in patients diagnosed with advanced non-exudative AMD. Likewise, the extracted features describing future GA involvement showed potential in predicting GA expansion, which may be useful to quantify progression of GA in AMD.

 
×
×

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.

×