Abstract
Purpose: :
To identify ageing changes in the human retinal tissue.
Methods: :
Sixty-eight eyes from 41 healthy volunteers were noninvansively imaged using the high-definition optical coherence tomography Cirrus HD-OCT (Carl Zeiss Meditec, Dublin, CA, USA) to collect data from the eye fundus. Because OCT is sensitive to changes in refractive index, changes over time in the refractive index may translate into changes in the reflectivity observed in OCT data between the inner limiting membrane and retinal pigment epithelium. OCT intensity distribution was analyzed both in the linear and logarithmic spaces and fits of known mathematical functions used to compute a set of 51 features describing the data. Functions were designed to account for differences in signal intensity across the eye rather than depending on absolute reflectivity. Data features were grouped by the age of respective eye in group A (age < 40 yrs, n=33, 29.5±4.3 yrs), group B (40 ≤ age < 60 yrs, n=22, 50.1±6.2 yrs) and group C (age ≥ 60 yrs, n=13, 67.5±6.5 yrs). A support vector machine (SVM) classification algorithm was used for the classification of eyes in one of the groups and the leave-one-out approach was used for the validation.
Results: :
Using the entire set of eyes available, 33, 22 and 13, respectively group A, B and C, eyes were correctly classified (overall) in 79.4% of the cases. Using a subset of 13 eyes in each group, group A (age < 40 yrs, 28.7±4.6 yrs), group B (40 ≤ age < 60 yrs, 50.0±6.2 yrs) and group C (age ≥ 60 yrs, 67.5±6.5 yrs), the overall correct classification is of 84.6% of the eyes.
Conclusions: :
The results suggest the presence of signs of the ageing process within the human retina that can be noninvasively detected by the HD-OCT. These findings open novel perspectives of using the eye as a window to changes in age-related neuronal diseases using well known low-cost technologies as HD-OCT.
Clinical Trial: :
http://www.clinicaltrials.gov NCT01220804
Keywords: retina • aging • image processing