June 2013
Volume 54, Issue 15
Free
ARVO Annual Meeting Abstract  |   June 2013
Predicting standard automated perimetry (SAP) sensitivities from spectral domain optical coherence tomography (SD-OCT) retinal nerve fiber layer thickness (RNFLT) using temporally correlated data
Author Affiliations & Notes
  • Lisha Deng
    Discoveries in Sight, Devers Eye Institute, Portland, OR
  • Deborah Goren
    Discoveries in Sight, Devers Eye Institute, Portland, OR
  • Manoj Pathak
    Discoveries in Sight, Devers Eye Institute, Portland, OR
  • Shaban Demirel
    Discoveries in Sight, Devers Eye Institute, Portland, OR
  • Stuart Gardiner
    Discoveries in Sight, Devers Eye Institute, Portland, OR
  • Footnotes
    Commercial Relationships Lisha Deng, None; Deborah Goren, None; Manoj Pathak, None; Shaban Demirel, Carl Zeiss Meditec (F), Heidelberg Engineering (R), Heidelberg Engineering (F); Stuart Gardiner, Allergan (R)
  • Footnotes
    Support None
Investigative Ophthalmology & Visual Science June 2013, Vol.54, 1891. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Lisha Deng, Deborah Goren, Manoj Pathak, Shaban Demirel, Stuart Gardiner; Predicting standard automated perimetry (SAP) sensitivities from spectral domain optical coherence tomography (SD-OCT) retinal nerve fiber layer thickness (RNFLT) using temporally correlated data. Invest. Ophthalmol. Vis. Sci. 2013;54(15):1891.

      Download citation file:


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

      ×
  • Supplements
Abstract
 
Purpose
 

Cross sectional analyses have been employed to study the association between SAP sensitivities and SD-OCT RNFLT. Longitudinal datasets have only become available recently. Such data provide large sample sizes due to repeat visits per subject while reducing the patient recruitment burden. If correlations between visits are accounted for, this method allows more powerful data analysis, which can be used to assess the structure-function relation.

 
Methods
 

328 eyes of 164 subjects with mild to moderate glaucoma or risk factors for its development were tested for a minimum of 5 visits at six month intervals. SD-OCT circle sweeps (TSNIT Average: Mean 88.02μm, 95% confidence limits 111.52μm - 52.0μm) and SAP 24-2 SITA visual fields (MD: Mean -0.43dB, 95% confidence limits -9.19dB - 2.47dB) were performed. Sensitivity at each of the 52 non-blindspot locations was modeled with a two-level (subject, eye within subject) linear mixed effects model including exponential temporal auto-correlation that accounted for correlations across visits (i.e. over time) and between fellow eyes of an individual. Retinal nerve fibre layer thicknesses (RNFLT) at eight 45° sectors were used to predict visual field sensitivity. Backwards elimination was used to identify the strongest SD-OCT sector predictors of sensitivity for each location by stepwise removal of sectors with negative coefficients or non-significant p-values (p>0.05, one-tailed).

 
Results
 

For all visual field locations two-level linear mixed effects models provided good fits to the data. All upper hemifield visual field locations resulted in significant positive coefficients (p<0.05) from inferior and/or inferonasal SD-OCT sectors. Lower hemifield locations were noisier. 25 of the 26 inferior hemifield locations resulted in significant positive coefficients (p<0.05) for the superonasal sector.

 
Conclusions
 

Visual field sensitivities can be predicted using two-level linear mixed effects models that account for correlations between eyes and temporal auto-correlation within sequences. As expected, upper hemifield locations were modeled best by inferior and inferonasal RNFL sectors, whereas lower hemifield locations were modeled best by the superonasal sector.

 
 
Significant sector predictors of visual field sensitivities
 
Significant sector predictors of visual field sensitivities
 
Keywords: 642 perimetry • 610 nerve fiber layer • 629 optic nerve  
×
×

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.

×