Change patterns in the corneal sub-basal nerve and corneal aberrations in patients with dry eye disease: An artificial intelligence analysis

Abstract

We aimed to investigate the change patterns in corneal sub-basal nerve morphology and corneal intrinsic aberrations in dry eye disease (DED). Our study included 229 eyes of 155 patients with DED and 40 eyes of 20 healthy control. We used the Oculus keratograph and the ocular surface disease index questionnaire to assess their signs and symptoms. In vivo confocal microscopy was used to observe the corneal sub-basal nerves, corneal endothelial cells, and Langerhans cells (LCs). An artificial intelligence (AI) technique run by the deep learning model generated the sub-basal nerve fibre parameters. Furthermore, we used the Pentacam HR system to measure the corneal intrinsic aberrations and corneal surface regularity indices. DED patients more frequently had increased anterior and total corneal aberrations than controls (P < 0.05). In addition, DED had decreased average density and maximum length of corneal nerve. (Both P < 0.01) The LC number was significantly correlated with maximum length (CC = −0.19, P = 0.01) of the sub-basal nerve fibre. Furthermore, the corneal nerve average density was negatively correlated with IHD, and anterior, posterior, and total corneal aberrations (All P < 0.05) especially the higher-order aberrations. Significant correlations were seen between corneal nerve morphology changes, analysed by AI and corneal intrinsic aberrations, particularly higher-order aberrations.