Quantitative analysis of corneal nerve fibers in type 2 diabetics with and without diabetic peripheral neuropathy: Comparison of manual and automated assessments

Abstract

AimsTo examine and compare fully-automated and manually measured corneal nerve fiber parameters in type 2 diabetes mellitus (T2DM) patients with and without diabetic peripheral neuropathy (DPN). MethodsA total of 128 T2DM subjects and 24 healthy controls underwent neuropathy assessment and bilateral corneal confocal microscopy (CCM). Five representative nerve fiber images were selected for each participant and analyzed manually and with fully-automated software. Corneal nerve fiber length (CNFL), branch density (CNBD), and fiber density (CNFD) were examined. ResultsManual and full-automated methods for the whole cohort were significantly positive correlated for CNFL, CNBD and CNFD (r = 0.818, 0.845, 0.457, all P < 0.001). Analysis of agreement between the two measurements using Bland-Altman method showed a bias of 2.05 mm/mm2 (95% limits of agreement: −2.03 mm/mm2, 6.13 mm/mm2), 1.62 no./mm2 (95% limits of agreement: −17.92 no./mm2, 21.17 no./mm2), and 16.0 no./mm2 (95% limits of agreement: −0.14 no./mm2, 32.14 no./mm2) for CNFL, CNBD and CNFD respectively. A progressive decrease in manual and full-automated CNFL, CNBD and CNFD accompanied with the occurrence of DPN, The fully-automated method slightly underestimated corneal nerve fiber parameters. ConclusionsThis study demonstrated strong correlations between manual and fully-automated CNFL and CNBD, but not CNFD. Fully-automated corneal nerve fiber parameter quantification may be a fast, objective way to detect DPN.