Improved tracking of corneal immune cell dynamics using in vivo confocal microscopy

Abstract

In vivo confocal microscopy (IVCM) is a widely used technique for imaging the cornea of the eye with a confocal scanning light ophthalmoscope. Cellular resolution and high contrast are achieved without invasive procedures, suiting the study of living humans. However, acquiring useful image data can be challenging due to the incessant motion of the eye, such that images are typically limited by noise and a restricted field of view. These factors affect the degree to which the same cells can be identified and tracked over time. To redress these shortcomings, here we present a data acquisition protocol together with the details of a free, open-source software package written in Matlab. The software package automatically registers and processes IVCM videos to significantly improve contrast, resolution, and field of view. The software also registers scans acquired at progressive time intervals from the same tissue region, producing a time-lapsed video to facilitate visualization and quantification of individual cell dynamics (e.g., motility and dendrite probing). With minimal user intervention, to date, this protocol has been employed to both cross-sectionally and longitudinally assess the dynamics of immune cells in the human corneal epithelium and stroma, using a technique termed functional in vivo confocal microscopy (Fun-IVCM) in 68 eyes from 68 participants. Using the custom software, registration of ‘sequence scan’ data was successful in 97% of videos acquired from the corneal epithelium and 93% for the corneal stroma. Creation of time-lapsed videos, in which the averages from single videos were registered across time points, was successful in 93% of image series for the epithelium and 75% of image series for the stroma. The reduced success rate for the stroma occurred due to practical difficulties in finding the same tissue between time points, rather than due to errors in image registration. We also present preliminary results showing that the protocol is well suited to in vivo cellular imaging in the retina with adaptive optics scanning laser ophthalmoscopy (AOSLO). Overall, the approach described here substantially improves the efficiency and consistency of time-lapsed video creation to enable non-invasive study of cell dynamics across diverse tissues in the living eye.