Comparison of Retinal Nerve Fiber Layer Imaging by Spectral Domain Optical Coherence Tomography and Scanning Laser Ophthalmoscopy

Abstract

To compare the area and the angular width of localized retinal nerve fiber layer (RNFL) defects imaged by confocal scanning laser ophthalmoscopy (CSLO) and optical coherence tomography (OCT) and to evaluate their agreement. Cross-sectional study. Fifty-one eyes of 41 glaucoma patients. Sixty-one distinctive, localized RNFL defects (17 superior and 44 inferior RNFL defects) detected in RNFL photographs imaged by a CSLO were identified. These patients underwent RNFL imaging with a spectral-domain OCT. The RNFL thickness deviation maps (50×50 pixels) generated by the OCT revealed the locations of abnormal RNFL thicknesses with abnormal pixels denoted in red (RNFL thickness less than the lower 99% normal distribution) or yellow (RNFL thickness less than the lower 95% normal distribution). The RNFL thickness deviation maps were aligned and overlaid with the corresponding CSLO RNFL photographs. The area and the angular width of RNFL defects from the corresponding retinal regions in the CSLO RNFL photographs and the OCT RNFL thickness maps were measured and compared. Their agreement was analyzed with the Bland-Altman plot. The area and the angular width of RNFL defects and the agreement of RNFL defects measurements between OCT images and CSLO RNFL photographs. The area and the angular width of RNFL defects measured with the CSLO RNFL photographs were 1.11 ± 0.57 mm² and 23.80 ± 10.38°, respectively, which were significantly smaller than those measured by the OCT RNFL thickness deviation map when abnormal RNFL thickness was defined as less than the lower 95% centile ranges (2.27 ± 0.92 mm² and 74.16 ± 28.74°, respectively; both P < 0.001). When abnormal RNFL thickness was defined as less than the lower 99% centile ranges, a significant difference in angular width (42.11 ± 22.19°; P<0.001), but not in area (1.19 ± 0.68 mm²; P = 0.444) was found between the 2 imaging methods. Bland-Altman plots revealed that a larger RNFL defect was associated with a greater difference in angular width between OCT and CSLO RNFL photography measurements. The agreement of RNFL defect measurements between CSLO RNFL photography and OCT was poor. The OCT RNFL thickness deviation map could reveal additional RNFL abnormalities not detectable by CSLO RNFL photography. Proprietary or commercial disclosure may be found after the references.