Dynamic curvature topography for evaluating the anterior corneal surface change with Corvis ST

Abstract

Background: The measurement of dynamic parameters, such as the length of applanation and the amplitude of deformation, is significant for evaluating corneal properties. Most of the corneal properties (related to shape) including the anterior corneal curvature and the thickness of cornea can be easily measured using some existing techniques. However, they only provide the static or pseudo-dynamic analysis. Based on Corvis ST images, the dynamic features after corneal boundaries detection and parameter estimation will be helpful for corneal analysis. Material: The study included 40 eyes in normal group (ranging from 19 to 45 years old) and 30 eyes in keratoconus group (ranging from 16 to 40 years old). These eyes were examined by Corvis ST and for each one a sequence of 140 images was obtained. Besides, 11 subjects of each group were also tested by Pentacam. Methods: By analyzing the video from the Corvis ST imaging, the fully dynamic curvature topography is proposed to evaluate the response of the anterior corneal surface to the air puff. The new method not only quantitatively measures the intact variation of anterior corneal surface but also provides an intuitive way to observe the dynamic change of the anterior corneal surface in the whole air stream process. The proposed method consists of three main steps: cornea segmentation, curvature estimation and integrated visualization. An automatic segmentation method based on the combination of prior knowledge with phase symmetry and asymmetry theory is firstly presented to detect the corneal boundaries. The Landau-new method is then used to estimate the anterior corneal surface. The corneal dynamic topography is finally obtained by combining the dynamic parameters with the original Corvis ST video, which is an improvement of the fusion technique proposed by Li et al. Results and conclusion: By comparing the segmentation results with manual method and built-in method of Corvis ST, the accuracy and robustness of our proposed segmentation method is demonstrated. The correctness of the estimated corneal anterior curvatures is also evaluated by comparing it with that of Pentacam which is considered to be able to provide the first-class measurement currently. The dynamic topography may be used to distinguish the dynamic behavior of normal corneas from that of keratoconus.