Abstract
Purpose :
Examination of the ganglion cell layer (GCL) or complex (GCC) augments our ability to detect early glaucoma with retinal nerve fibre layer (RNFL) scanning. However the variety of patterns in GCL loss, their location and whether inner plexiform layer (IPL) and RNFL should be included in segmentation algorithms is not clear. We compared 2 OCT devices (Cirrus - Carl Zeiss and RS3000 Advance - Nidek) that use different segmentation techniques, and examined the effect of a larger macular scan size.
Methods :
120 early to moderate glaucoma cases were prospectively enrolled. One eye was analysed, if both had glaucoma the eye with the lesser defect was chosen. Diagnosis was based on stereoscopic optic disc assessment for thinning of neuroretinal rim +/- visual field loss. Humphrey visual field MD mean was -2.5 +/-3.5dB, range +1 to -13.2. Patients were scanned with GCC and peripapillary RNFL scans in random order, same session. The Nidek GCC segmentation was applied to both a 6mm and 9mm ring. Only scans >7/10 quality included. Registration errors, decentration or other identified pathology such as epiretinal membranes were excluded. Detection rates for different scan types were determined using their database software, and patterns of GCC/RNFL loss were assessed qualitatively.
Results :
Of 120 eyes, 92 met the criteria for high quality on all scans. Several qualitatively distinct patterns of loss were more clearly demonstrated on the combined GCL/IPL/RNFL complex (Nidek) which we described as focal arcuate, broad arcuate, double arcuate, diffuse and ring-shaped patterns. When using the GCL/IPL complex only (Cirrus) the patterns were more commonly only ring shaped or diffuse. Using one borderline flagged sector as the minimum criterion, of 92 eyes 73(79%) were abnormal on RNFL, 63(68%) on Cirrus GCL, 81(88%) on Nidek 6mm and 82(89%) on the 9mm GCL complex. Only 55(60%) were abnormal on all 4 scan types. The wider 9mm scan detected 3 extra cases of perimacular arcuates that were missed by both the 6mm and RNFL, providing a small advantage.
Conclusions :
Including RNFL in the segmentation aided detection rates and provided recognizable patterns that reflected both RNFL loss in some cases and GCL loss in others. These might represent different phenotypes of glaucomatous damage. The Nidek analyzes more macular segments which could also explain the improved detection rate.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.