Novel objective approach to retrieve quantitative measures of corneal transparency from depth-resolved images

Session Details

Session Title: Ocular Infections & Other Topics

Session Date/Time: Sunday 15/09/2019 | 08:00-10:00

Paper Time: 09:30

Venue: Free Paper Forum: Podium 4

First Author: : K.Irsch FRANCE

Co Author(s): :    R. Bocheux   B. Rivière   P. Pernot   V. Borderie   K. Plamann                 

Abstract Details

Purpose:

To address the unmet need for an objective means to quantify corneal transparency.

Setting:

The study was conducted at the Quinze-Vingts National Eye Hospital, Paris (France).

Methods:

Based on a physical model of stromal light scattering, an algorithm and related software were developed in Matlab (Mathworks, Inc., USA) in conjunction with R (R Core Team, Vienna, Austria) capable of deriving quantitative transparency parameters from depth-resolved corneal images. Specifically, after automatic segmentation and flattening based on signal intensity at the epithelial surface, the average grey value is computed as a function of stromal depth (z). The resultant z-profiles are fitted using a Bayesian method to an exponential function A(z)~exp(-Bz), allowing the extraction of the photon mean-free path, ls=1/B, a major measure of scattering extent and thus transparency.

Results:

Applicable to any depth-resolved corneal imaging modality, our approach was validated by means of ex-vivo full-field optical coherence tomographic microscopy (FF-OCT or FF-OCM) and the proof of concept demonstrated on in-vivo data obtained from standard clinical spectra-domain optical coherence tomography (SD-OCT). Specifically, the computation of the Birge ratio, as a measure of exponential-decay model inadequacy, enabled the estimation of severity of stromal heterogeneity and the differentiation of heterogeneous, pathological tissues from homogeneous tissues. In the latter, homogeneous tissues, the extraction of the photon mean-free path from the rate of exponential decay allows further quantification of the degree of transparency.

Conclusions:

We developed an optical data analysis-based method that may be implemented into existing depth-resolved corneal imaging methods (e.g., slit-lamp biomicroscopy, OCT, confocal microscopy) and enables, via computation of the light attenuation in the stroma, the characterization and standardization of the state of corneal transparency. This approach has the potential to supply the demand for an objective means to quantify corneal transparency in both the eye-bank and clinical setting, replacing current subjective and qualitative means and aiding in the diagnosis, treatment, monitoring of various diseases as well as ameliorating surgical interventions that can directly impact corneal transparency (e.g., refractive surgery, keratoplasty).

Financial Disclosure:

None