Clinical Evaluation of Correction Algorithm for Corvis ST Tonometry

Abstract

 

Purpose

 

To develop and clinically evaluate a correction algorithm to improve the accuracy of intraocular pressure (IOP) measurements made by the non-contact Corvis ST (CVS) tonometer, which considers the combined effects of variations in central corneal thickness (CCT), age and the IOP level itself.

 

Methods

 

Nonlinear numerical simulations based on the finite element method were used to represent corneal behaviour under the effect of IOP and the CVS air pressure. The simulations considered important biomechanical properties including the cornea’s non-uniform thickness, elliptical topography, weak stromal inter-lamellar cohesion, low epithelial and endothelial stiffness, and hyperelastic and hysteretic material behaviour. The simulations represented the CVS procedure on corneas to obtain a correction algorithm based on the values of CCT, age and the measured IOP. The effectiveness of the algorithm in reducing the effects of corneal parameters on CVS-IOP was assessed using three independent clinical databases involving 632, 1047 and 99 participants, respectively and obtained at Smile Eyes Clinics, Germany, Humanitas Clinical and Research Center, Italy and Wenzhou Medical University, China.

 

Results

 

CCT was the dominant stiffness factor affecting CVS-IOP, and both CCT and age had a nonlinear (second order) effect on the IOP measurements. The correction algorithm was successful in significantly reducing the strength of association between CVS-IOP and CCT in the three clinical databases respectively, from 0.031 mmHg/um (R²=0.204) to 0.004 mmHg/um (R²=0.004), from 0.029 mmHg/um (R²=0.095) to 0.009 mmHg/um (R²=0.016) and from 0.033 mmHg/um (R²=0.222) to 0.001 mmHg/um (R²=00), Figs. 1 and 2. The association with age was weak, remaining below 0.06 mmHg/year (R²=0.05) in all three databases and reducing slightly with the correction algorithm.

 

Conclusions

 

The CVS correction algorithm considerably reduced the dependence of CVS-IOP on CCT, addressing an important error source in CVS-IOP measurements. The consistency in performance of the new algorithm over three independent clinical databases indicates the suitability of the algorithm for clinical utility.  

 

Fig 1 Effectiveness of algorithm in reducing association of CVS-IOP with CCT for a clinical database with 632 participants

 

 

Fig 2 Effectiveness of algorithm in reducing association of CVS-IOP with CCT for a clinical database with 1047 participants