Automated Grading System for Evaluation of Superficial Punctate Keratitis Associated With Dry Eye

Abstract

An objective, automated analysis of corneal staining provides a quality assurance tool to be used to substantiate clinical grading of key corneal staining endpoints in multicentered clinical trials of dry eye.

Size of the Lesions of Superficial Punctate Keratitis in Dry Eye Syndrome Observed With a Slit Lamp

“…Objective corneal staining was defined as the ratio of the number of pixels identified as corneal erosions to the total number of pixels in the cornea. 13 Similarly, in 2015, Rodriguez et al 14 developed an automated method to detect and quantify . Epifluorescence microscopic image of the epithelium of a fixed human cornea (0.5% paraformaldehyde) after immunolabeling with cytokeratin K3/12 (green cytoplasm) and counterstaining with ethidium (red nuclei).…”

Assessing ocular bulbar redness: a comparison of methods

“…Within the field of ophthalmic imaging, there have been efforts towards developing software analysis tools to objectively quantify important clinical signs. For anterior eye signs, software exists to automatically grade various clinical parameters, including meibomian gland dropout, 31 superficial punctate keratitis 32 and non-invasive tear-breakup-time. 33 The attraction of reliable automatic clinical grading systems, particularly for research applications, is There is a 95% probability that at least 95% of the differences in the population lie outside the limits of 1.28 AE 0.29 units (0.99 and 1.57) and inside the limits of 1.28 AE 0.52 units (0.76 and 1.80).…”

The SUSTech-SYSU dataset for automatically segmenting and classifying corneal ulcers

“…Manual segmentation is considered as being the most reliable but highly time-consuming and of large inter-and intra-rater variability. During the past decade, various semi- [8][9][10] and fully-automated [11][12][13][14] techniques have been developed for segmenting corneal ulcers. Many of these segmentation algorithms employ machine learning techniques which require large training datasets.…”