Method for optical coherence elastography of the cornea

Abstract

The material properties of the cornea are important determinants of corneal shape and refractive power. Corneal ectatic diseases, such as keratoconus, are characterized by material property abnormalities, are associated with progressive thinning and distortion of the cornea, and represent a leading indication for corneal transplantation. We describe a corneal elastography technique based on optical coherence tomography (OCT) imaging, in which displacement of intracorneal optical features is tracked with a 2-D cross-correlation algorithm as a step toward nondestructive estimation of local and directional corneal material properties. Phantom experiments are performed to measure the effects of image noise and out-of-plane displacement on effectiveness of displacement tracking and demonstrated accuracy within the tolerance of a micromechanical translation stage. Tissue experiments demonstrate the ability to produce 2-D maps of heterogeneous intracorneal displacement with OCT. The ability of a nondestructive optical method to assess tissue under in situ mechanical conditions with physiologic-range stress levels provides a framework for in vivo quantification of 3-D corneal elastic and viscoelastic resistance, including analogs of shear deformation and Poisson's ratio that may be relevant in the early diagnosis of corneal ectatic disease.

Figures

Figure 1

A Zemax model and measured spot profile of the sample arm scanner used in these experiments.

Figure 2

Diagramatic representation of the setup of the OCT elastographic measurement system.

Figure 3

Optically measured versus imposed mechanical displacement of human corneal tissue by a micrometer positioning stage. The stage was accurate to 2±1 μm according to manufacturer specifications. The upper and lower boundaries represent the positioning tolerence of the stage. The error bars on the measured values indicate the standard deviation of the measured displacement.

Figure 4

Software phantom analysis of the effect of SNR and window size on the displacement algorithm correlation coefficient and displacement error. (a) Correlation values increase and (b) average magnitude of the displacement error decreases with increasing kernel size and increasing SNR. The range of SNR values commonly encountered in OCT systems is well within the limits explored here. (c) Scatter plot of number of displacement errors across all window sizes and all noise levels as a function of correlation coefficient. Each noise-level and window-size group consists of 294 total displacement measures.

Figure 5

Plot showing the effect of a simulated compressive strain on the resultant correlation coefficient.

Figure 6

Effect of spatial sampling density on the correlation coefficient with known out-of-plane displacements. The sampling referenced refers to the separation between lateral samples in the image. Axial sampling was constant. The sample was not displaced in the direction of the image plane. Displacement tracking was only performed in two dimensions.

Figure 7

Examples of 2-D displacement maps obtained during a downward axial displacement of the cornea. The raw OCT image is one of the pair of images representing this displacement increment. Horizontal (lateral) displacements are mapped in the central frame, and vertical (axial) displacements are presented in the lower frame. All displacements are expressed in micrometers with the magnitude indicated by the legend. The images were obtained along the vertical corneal meridian: the left side of all three images represents the inferior portion of the central cornea, and the right side represents the superior portion of the central cornea.

Figure 8

(a) An image of the cornea taken from a displacement image pair with regions of interest demarcated by squares. The cornea is oriented inferior to superior from left to right. The lateral displacements correspond to displacements in the superior-inferior directions of the vertical meridian (b) A vector map taken from the upper left region shown in (a) and representing anterior corneal stroma. The chart depicts cumulative horizontal displacement versus cumulative vertical displacement, both measured with optical coherence optical feature tracking, of the six regions shown in (a). No less than 275 independent data points were used to calculate the average values shown in (c).