Purpose
The purpose of this study is to investigate the usefulness of color tone and phase analysis for predicting central choroidal thickness (CCT) and to evaluate the algorithm using normal fundus photographs.
Methods
One hundred fundus photographs of healthy eyes taken between January 2012 and April 2014 at Tokyo University Hospital were used. There were 60 males and 40 females and the average age was 63.7±14.6 (mean±SD). By using in-house built software, 6,000mm ETDRS charts were placed to define subfields in the macular area, and the color tone and phase of each foveal and parafoveal subfield was quantified. The CCT was measured by Spectral-Domain optical coherence tomography (Spectralis, Heidelberg, Germany) with enhanced depth imaging and categorized into thin (~100mm), moderate (101-250mm), and thick (250~ mm). The prediction algorithm was developed with the data of randomly selected 50 eyes and its predictability was confirmed by the data of the other 50 eyes.
Results
The CCT was 227.8±97.7mm (mean±SD). The fundus photographic parameters related to CCT were analyzed with forward stepwise regression analysis, which showed that the CCT was correlated with the quantified red and green color tones of retina excluding the nasal side area. The prediction algorithm calculated with the data of these factors was evaluated in other 50 eyes, which showed that the CCT were precisely predicted in 42 (84%) of the eyes.
Conclusions
The color tone analysis based on our software was limited by its moderate ability to predict choroidal thickness, but is useful to predict central choroidal thickness when OCT data is not available.