Ability of MultiColor scanning laser ophthalmoscope to detect non-glaucomatous retinal nerve fiber layer defects in eyes with retinal diseases

All of the procedures used in this study conformed to the tenets of the Declaration of Helsinki. The procedures used were approved by the Ethics Committee of Kagoshima University Hospital.

This was a retrospective, cross sectional study that was performed at the Kagoshima University Hospital, Kagoshima, Japan. We initially reviewed the medical records of 297 eyes of 176 patients who had color fundus images obtained by a swept source optical coherence tomographic instrument (DRI OCT Triton, Topcon, Japan) and MC-SLO images obtained by Spectralis MultiColor scanning laser ophthalmoscope. The images were recorded on the same day between December 2016 and February 2017. The eyes that had NFLDs that were detected by two experienced ophthalmologists (SS, NK) in at least one of the MC-SLO or color fundus images, i.e., blue, green, red, or merged, of the MC-SLO, or in the color or red-free images were studied. For this examination, we did not use the enlargement function of the instruments and evaluated the original screen of the viewer attached to the instruments. When the interpretation was split between raters, the images were excluded. In addition, to investigate whether non-glaucomatous NFLD were observed in normal eyes in the MC-SLO and CF images, 46 eyes of 23 healthy subjects who did not have any ocular abnormalities in the ophthalmologic examinations were evaluated. The NFLDs appeared as dark slits or bands or diffuse darkness of the retina that passed toward or touched the optic disc for not more than 60° of the circumference [10]. Eyes with an IOP > 21 mmHg and eyes with optic discs that had glaucomatous changes such as generalized or focal enlargement of the cup, superficial splinter hemorrhage, development of vessel overpass were excluded. To include our subjects appropriately, the evaluations of the optic disc was done by two experienced ophthalmologists. Eyes with poor quality in MC-SLO images or color fundus images were also excluded.

The visibility of NFLDs was divided into three grades: grade 0, not visible; grade 1, barely visible; and grade 2, clearly visible and defined as the “NFLD scores”. The NFLDs in the blue, green, red, and merged MC-SLO images were graded and also in the color and red-free images by two graders (HT, HS). In our preliminary experiments, the NFLDs tended to be more detectable in the MC-SLO images than in the CF images. Thus, the order of examination for the NFLDs was always begun with the color fundus images followed by the red-free images and then the MC-SLO images.

The mean NFLD score for each color was statistically compared by Steel-Dwass tests. The intra- and inter-rater agreement of the two graders was assessed by weighted kappa analysis.

The demographics of the patients are presented in Table 1. Thirty-one eyes of 26 patients, 19 men and 7 women, were studied. There were 14 eyes of 10 patients with diabetic retinopathy (DR), 11 eyes of 10 patients with age-related macular degeneration (AMD), 3 eyes of 3 patients with a branch retinal vein occlusion (BRVO), and 3 eyes of 3 patients with an epiretinal membrane/macular hole (ERM/MH). The mean age of the patients was 63.1 ± 11.2 years with a range of 42 to 86 years. The mean refractive error (spherical equivalent) was − 0.72 ± 1.17 diopters, and the mean axial length was 23.1 ± 1.17 mm. The mean best-corrected visual acuity (BCVA) was 0.25 ± 0.37 logMAR units (snellen visual acuity, 20/35). Sixteen eyes were pseudophakic. All patients with DR and BRVO had been treated by retinal photocoagulation. Five eyes with DR and 3 eyes with MH/ERM had undergone vitrectomy prior to this study. In these cases, 3 eyes with DR and 3 eyes with MH/ERM had the inner limiting membrane (ILM) peeled during the vitrectomy.

The kappa index of intra-and inter-rater agreements for each channel of the MC-SLO and CF images were good at more than 0.6 for both (Table 2).

Comparisons of the NFLD scores for each image obtained by MC-SLO are shown in Fig. 1. In the healthy subjects, we studied 13 men and 10 women with an average age of 32.2 years (range: 20–45 years old). The average refractive error was − 1.97 ± 1.73 diopters and axial length was 24.2 ± 2.06. In these normal eyes, we did not detect any non-glaucomatous NFLD in all eyes. Representative case of healthy eye was shown in Fig. 2a.

The non-glaucomatous NFLDs were detected by blue wavelength in all cases and 90.3% of NFLDs were detected by green wavelength and merged images and 6.5% by the red wavelength. The mean NFLD score was 1.58 ± 0.49, 1.13 ± 0.54, 0.06 ± 0.24. and 1.16 ± 0.56 for the blue, green, red wavelengths and merged images, respectively. The NFLD score for the blue wavelength was significantly higher than that for the green and red wavelengths (P < 0.05 and P < 0.01, respectively) but not the merged image. The NFLD score for red was significantly lower than all other images. The mean AL was not significantly correlated with the NFLD score for each wavelength of the MC-SLO images (data not shown).

NFLD were detected in both the fundus photographs and MC-SLO images in 13 eyes of 13 cases. The average number of NFLDs in the MC-SLO images was 1.30 ± 0.63 in the merged images, 1.69 ± 0.85 in the blue SLO images, 1.38 ± 0.63 in the green SLO images, and 0.08 ± 0.27 in the red SLO images. In the color fundus photographs and red-free images, only one NFLD could be confirmed in each case (Fig. 3). There was a statistically significant difference between the number in the blue images of the MC-SLO and the color fundus/RF images. A representative case is shown in Fig. 2b; two NFLDs can be seen in the blue and merged images of MC-SLO, but only one can be barely seen in the fundus photograph/RF.

The RNFL thickness maps determined by DRI-OCT were available for 26 eyes. In 23 eyes, a thinner RNFL corresponding with the NFLD detected by SLO with blue wavelength was found (88%).

Three representative cases are shown in Fig. 2a-d.