A novel automated segmentation method for retinal layers in OCT images proves retinal degeneration after optic neuritis

Aim

The evaluation of inner retinal layer thickness can serve as a direct biomarker for monitoring the course of inflammatory diseases of the central nervous system such as multiple sclerosis (MS). Using optical coherence tomography (OCT), thinning of the retinal nerve fibre layer and changes in deeper retinal layers have been observed in patients with MS. Here, we first compare a novel method for automated segmentation of OCT images with manual segmentation using two cohorts of patients with MS. Using this method, we also aimed to reproduce previous findings showing retinal degeneration following optic neuritis (ON) in MS.

Methods

Based on a 5×5 expansion of the Prewitt operator to efficiently calculate the gradient of image intensity, we introduce an automated algorithm for the segmentation of intraretinal layers. We evaluated this algorithm by comparison to manually segmented two-dimensional OCT images at the macular level for 125 patients from two separate cohorts of patients with MS. Of these patients, 52 had suffered from unilateral ON+ within 6 months prior to measurement.

Results

When comparing ON+ eyes with ON− eyes, both manual and automated segmentation demonstrated a significant inter-eye thinning in the ganglion cell layer in ON+ eyes. We also observed an increased thickness of the inner nuclear (INL) and the outer segment-retinal pigment epithelium (OS-RPE) layers of ON+ eyes in both cohorts. These findings corroborate previous data, thus demonstrating the validity of our approach.

Conclusions

The algorithm presented here was found to be a valid tool for replacing cumbersome manual segmentation methods in the quantification of inner retinal layers in OCT. The observed increases in thickness of INL and OS-RPE may be attributed to primary retinal inflammation, repair and/or plasticity mechanisms following the immune attack.