Artificial intelligence based customized cross-linking: a novel approach to strengthen the cornea

Session Details

Session Title: Cross-Linking

Session Date/Time: Tuesday 10/10/2017 | 14:00-16:30

Paper Time: 14:38

Venue: Room 4.6

First Author: : P.Kusumghar INDIA

Co Author(s): :    R. Shetty   A. Sinha Roy   K. Sainani              

Abstract Details

Purpose:

To study the refractive and keratometric outcomes of an artificial intelligence (AI) based approach for topography guided customized crosslinking

Setting:

Narayana Netralaya post graduate eye institute and multi-speciality eye hospital

Methods:

15 keratoconic eyes underwent topography-guided crosslinking (Avedro Inc., USA). Three dimensional AI based model was used to calculate shape and size of cone and design concentric treatment zones to deliver differential energy .The topography was used to design a patient specific UV beam, which was centered and had peak intensity at a point determined by AI model. The delivered energy varied from 15 at center to 3 J/cm2 at periphery. The maximum treated diameter was 8 mm. Topography and corneal deformation was measured with Pentacam HR and Corvis-ST. Refractive and keratometric data was measured pre and post (6 months) surgery.

Results:

Mean MRSE, Flat and steep axis keratometry, mean keratometry and maximum keratometry showed reduction 3 months after the surgery. Root mean square (RMS) of anterior lower order aberration, higher order aberration and total aberration did not show a significant difference before and 3 months after the procedure. Similar results were obtained with the RMS of posterior corneal surface.

Conclusions:

A mathematical AI based approach was used to calculate the treatment algorithm for customizable topography guided crosslinking. This method shows similar refractive, keratometric and biomechanical outcomes to those reported earlier in literature. The safety profile of AI based crosslinking is also acceptable with no significant endothelial damage. Further study is needed with varying pre-progression material property parameters.

Financial Disclosure:

NONE