April 2014
Volume 55, Issue 13
Free
ARVO Annual Meeting Abstract  |   April 2014
Automated Segmentation and Quantification of Retinal Layers in Patients with Hydroxychloroquine Toxicity
Author Affiliations & Notes
  • Luis de Sisternes
    Radiology, Stanford University, Stanford, CA
  • Michael F Marmor
    Ophthalmology, Byers Eye Institute at Stanford, Palo Alto, CA
  • Theodore Leng
    Ophthalmology, Byers Eye Institute at Stanford, Palo Alto, CA
  • Daniel Rubin
    Radiology, Stanford University, Stanford, CA
  • Footnotes
    Commercial Relationships Luis de Sisternes, None; Michael Marmor, None; Theodore Leng, None; Daniel Rubin, None
  • Footnotes
    Support None
Investigative Ophthalmology & Visual Science April 2014, Vol.55, 4799. doi:https://doi.org/
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Luis de Sisternes, Michael F Marmor, Theodore Leng, Daniel Rubin; Automated Segmentation and Quantification of Retinal Layers in Patients with Hydroxychloroquine Toxicity. Invest. Ophthalmol. Vis. Sci. 2014;55(13):4799. doi: https://doi.org/.

      Download citation file:


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

      ×
  • Supplements
Abstract
 
Purpose
 

To demonstrate a new automated method quantifying regional changes in retinal layers in spectral-domain optical coherence tomography (SD-OCT), to enhance detection of hydroxycholoroquine (HCQ) toxicity and its progression.

 
Methods
 

We studied clinical SD-OCT scans recorded on a Zeiss (Cirrus) instrument from 8 normal eyes, and 15 eyes previously classified as early, moderate or severe HCQ toxicity. We used gradient statistics to produce an initial estimation of 9 retinal layers visible on SD-OCT and a weighted median filtering iterative process to refine their location. The segmentation of layers was validated ophthalmologically by a retina specialist (MFM). Quantitative features describing 5 mean thickness measurements (nerve fiber layer, outer nuclear layer, inner, outer, and whole retina) were automatically generated for 9 macula regions (similar to ETDRS regions), together with en face maps showing regional thinning. Loss of the ellipsoid zone (EZ, also called IS/OS line) was recognized where signal strength of the line was too low to be located, and quantitative features were extracted describing its morphology.

 
Results
 

Total retina thickness measurements were very similar (<5% difference) between our method and the Cirrus instrument. In normal eyes, the automated retinal layer outlines were accurately located, and en face maps showed no focal defects. Retinopathy patients showed progressive thinning of the outer retina, beginning in parafoveal areas but extending with more severe disease into the fovea and peripheral regions of the posterior retina. Small regions of thinning could be visualized more readily in the generated en face maps (showing the whole macula) than in routine cross-sectional scans. Our method could also show the morphology of EZ line damage as retinopathy increases.

 
Conclusions
 

We have developed methods to automatically measure a set of SD-OCT imaging features that may be helpful for recognizing and tracking HCQ toxicity with more precision than is presently possible. Preliminary results in patients with HCQ toxicity show a promising ability to follow changes in the EZ line, and to separate inner and outer retinal damage. Refinements may allow monitoring of patients for foveal cone damage and visual loss.

 
 
Inner and outer retina en face maps of a normal eye and a HCQ eye with no visible retinopathy or RPE change, but showing a full ring of photoreceptor damage.
 
Inner and outer retina en face maps of a normal eye and a HCQ eye with no visible retinopathy or RPE change, but showing a full ring of photoreceptor damage.
 
Keywords: 503 drug toxicity/drug effects • 549 image processing • 542 grouping and segmentation  
×
×

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.

×