June 2017
Volume 58, Issue 8
Open Access
ARVO Annual Meeting Abstract  |   June 2017
An artificial intelligence method to estimate region of biomechanical weakness in keratoconic corneas
Author Affiliations & Notes
  • Tushar Grover
    Narayana Nethralaya, Bangalore, Karnataka, India
  • Rohit Shetty
    Narayana Nethralaya, Bangalore, Karnataka, India
  • Abhijit Sinha Roy
    Narayana Nethralaya, Bangalore, Karnataka, India
  • Footnotes
    Commercial Relationships   Tushar Grover, None; Rohit Shetty, None; Abhijit Sinha Roy, Cleveland Clinic Innovations (P)
  • Footnotes
    Support  None
Investigative Ophthalmology & Visual Science June 2017, Vol.58, 3551. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Tushar Grover, Rohit Shetty, Abhijit Sinha Roy; An artificial intelligence method to estimate region of biomechanical weakness in keratoconic corneas. Invest. Ophthalmol. Vis. Sci. 2017;58(8):3551.

      Download citation file:

      © ARVO (1962-2015); The Authors (2016-present)

  • Supplements

Purpose : This study aimed to develop a spatial model of keratoconic biomechanical properties and to develop an artificial intelligence method to estimate the shape and size of the affected zone using the model

Methods : A 3-D geometrical model of the cornea was constructed using tomography (OCULUS Optikgerate Gmbh, Germany). The model included the epithelium and stroma. Transverse fiber dependent material model was used (Sinha Roy et al., J Mech Behav Biomed Mat, 2015). To model KC, it was assumed that: (a) the disease caused local biomechanical degeneration; (b) the strength of the stroma in KC was determined by the residual collagen network that was unaffected by the disease. Thus, the material properties were multiplied by a linear "factor" that varied spatially over the cornea to model localized steepening (Sinha Roy et al., IOVS, 2011). The model was used to compute the shape and size of the region of localized weakening in progressive keratoconus using artificial intelligence (AI) and finite element method (FEM). By evaluating the "factor" at several locations in the stroma, the AI automatically adjusted the factor at each location so as to minimize the difference between the measured anterior corneal curvature of the progressed state of KC and the FEM estimate of the same.

Results : Figure 1 shows the outcomes of the model, when applied to a progressive KC case. The patient progressed to KC two years after the 1st measurement (year 2010) and continued thereafter (2013 and later). The eye had normal topography at 1st visit. Figure 1A shows the axial curvature measured in 2015. Figure 1B shows the model outcome at the same time point. Figure 1C, D and E shows the AI predicted spatially varying map of the "factor" for years 2013, 2014 and 2015, respectively. A "factor" of 1 (Figure 1C, D and E) indicated no change in material properties. For e.g., if at a given location in the cornea, the computed factor was 0.5 and the modulus of the cornea in the unaffected region was 1 MPa, then the modulus of the cornea at that location due to diseases progression was 0.5 MPa (1 x 0.5).

Conclusions : An algorithmic approach using conventional tomography was developed to estimate the region of biomechanical weakening in KC cornea. The method showed the irregular shape and size of the region of biomechanical weakening, which may be used to plan corneal crosslinking procedures.

This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.



This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.