Abstract
Purpose: :
To propose a methodology for building average 3D models or atlases of the human cornea based on topography data (Orbscan II; Bausch & Lomb).
Methods: :
454 normal OD topographies were selected from a pool of 4553 so that only one eye per subject was used, with spherical equivalents within ± 3.00 D from emmetropia and a refractive cylinder < 1.00 D. Ages ranged from 19.3 to 69.3 years with an average (SD) of 39.9 (11.0) years.
Results: :
Methodology consisted in four steps: (1) Data acquisition from both anterior and posterior corneal surfaces in the format of a 101 x 101 grid of z elevations evenly spaced (every 0.1 mm) along the x and y axes, (2) Alignment of the topographies on a unique median BFS (normalization) to reduce the large variability in size between corneas, (3) Map generation to extract local variation information, and (4) Statistics maps including average (variance) and median (percentile) for each point of the grid. To demonstrate the informative potential of this methodology, several atlases were generated as a function of age. Results showed that the anterior corneal surface flattens and gets closer to the anterior BFS by a mean of 0.25±0.76µm per five–year interval (R2=0.68, p=0.0017). A progressive lateral translation of the posterior apex with respect to the anterior apex of 3.2±6.7µm per five–year interval (R2=0.85, p=0.0001) was also observed. Other specific models were constructed to study the effect of laterality (enantiomorphism), gender, and ametropia. Such numerical atlases could also be helpful in the design of algorithms targeting the detection of corneal shape abnormalities, such as keratoconus or previous laser surgery.
Conclusions: :
A new concept of a 3D numerical corneal shape model or atlas is proposed for the characterization of a population, based on corneal topography. This methodology will allow for the first time to compare topographies of populations or to compare a patient to a reference population.
Keywords: topography • computational modeling • image processing