June 2013
Volume 54, Issue 15
Free
ARVO Annual Meeting Abstract  |   June 2013
Detection and Quantification of Autofluorescence Abnormalities in Patients with Neovascular Macular Degeneration using a Fully Automated Image Analysis Algorithm
Author Affiliations & Notes
  • Kathryn Pepple
    Ophthalmology, Duke Eye Center, Durham, NC
  • Qing Nie
    Ophthalmology, Duke Eye Center, Durham, NC
  • Sally Ong
    Ophthalmology, Duke Eye Center, Durham, NC
  • Scott Cousins
    Ophthalmology, Duke Eye Center, Durham, NC
  • Footnotes
    Commercial Relationships Kathryn Pepple, None; Qing Nie, None; Sally Ong, None; Scott Cousins, Alcon (F), Alcon (C), Heidelberg Engineering (C), Narrow River (C), Nordic Biotech (C), PanOptica (C), Pfizer (C), Salutaris Medical Devices (C), Sanofi-Fovea (C), Valeant Ophthalmics (C), Imagen Biotech (I)
  • Footnotes
    Support None
Investigative Ophthalmology & Visual Science June 2013, Vol.54, 4089. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Kathryn Pepple, Qing Nie, Sally Ong, Scott Cousins; Detection and Quantification of Autofluorescence Abnormalities in Patients with Neovascular Macular Degeneration using a Fully Automated Image Analysis Algorithm. Invest. Ophthalmol. Vis. Sci. 2013;54(15):4089.

      Download citation file:


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

      ×
  • Supplements
Abstract
 
Purpose
 

To develop a fully automated fundus autofluorescence (AF) image analysis algorithm and evaluate the occurrence of retinal pigment epithelium abnormalities (RPE) in patients with neovascular macular degeneration.

 
Methods
 

A retrospective chart review was performed to identify patients with NVAMD that had fundus AF images obtained at presentation and at follow up at least one year later. A custom Matlab algorithm was designed to allow fully automated segmentation and quantification of hypo-AF and hyper-AF. The computer segmentation results were compared against the boundaries defined by three expert graders who performed manual segmentation. The baseline area of abnormal AF was compared to the follow up area and a rate of change was calculated.

 
Results
 

22 eyes of 16 patients with NVAMD were identified. The computer algorithm successfully identified areas of hyper-AF and hypo-AF in all eyes. The area identified by the computer algorithm had good agreement with the area identified by the human graders. In all eyes at baseline, abnormal hypo-AF was present, and on average, the area increased in size by the final follow up. The average size of the hyper- AF area also increased from baseline, but less than the area of hypo-AF. An example of the output from the automated software is included below.

 
Conclusions
 

Image analysis of fundus AF using a fully automated algorithm can provide rapid, objective information about RPE pathology in patients with NVAMD. This software has the potential to allow for monitoring of RPE changes to detect disease progression and in response to medical therapy.

 
 
The area of hypo-AF is outlined in red. This area overlaps the region of normal foveal hypo-AF, but also includes an additional area of hypo-AF present over a choroidal neovascular (CNV) lesion. Areas of hyper-AF are outlined in blue. The areas of hypo-AF and hyper-AF increase over 3 years.
 
The area of hypo-AF is outlined in red. This area overlaps the region of normal foveal hypo-AF, but also includes an additional area of hypo-AF present over a choroidal neovascular (CNV) lesion. Areas of hyper-AF are outlined in blue. The areas of hypo-AF and hyper-AF increase over 3 years.
 
Keywords: 412 age-related macular degeneration • 550 imaging/image analysis: clinical • 688 retina  
×
×

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.

×