Abstract
Purpose:
The evaluation of vital staining in the dry eye subject is of primary importance in clinical trials. While the most prevalent staining methodology may be corneal fluoresecien staining, lissamine staining is a valuable tool for the evaluation of conjunctival tissue. This abstract extends previous work on software detection of corneal staining using fluorescein, where pixel intensity variations alone are required for enumerating corneal superficial punctate keratitis (SPK). Analysis of lissamine images, however, entails several additional complications. For example, variations in image hue, edge analysis, and shape analysis must be calculated to detect staining blobs and remove false positives due to dye pooling in tissue folds. We apply this method to a sample dry eye population.
Methods:
High resolution digital images were obtained of the nasal, temporal, and inferior conjunctival tissue using lissamine dye in a population of seven mild to moderate dry eye subjects. Ten images were selected and graded using the Ora Calibra™ dry eye staining scale (0-4) by a clinician based on image quality and representing a maximal range of staining severity. Each image was processed using software with the number of isolated stained regions, and their respective surface areas were calculated. Clinical grade was compared to the log of the number of staining areas, as well as the log of the total stained area.
Results:
The mean number of isolated stained regions was 73.7, and the mean staining area (as a percent of surface area) was 0.3049. Correlation of log region number with clinical grade was 87% (91% versus log region area).
Conclusions:
A software approach to conjunctival grading can efficiently and accurately extract clinical information from a lissamine image. The relative importance of staining area may be due to the differing specificity of lissamine vital staining versus fluorescein.
Keywords: 486 cornea: tears/tear film/dry eye •
479 cornea: clinical science