James K. H. Goh et al.
Singapore Eye Research Institute, Singapore National Eye Center, National University of Singapore
Confocal Scanning Laser Ophthalmoscopy (cSLO), based on green and infrared laser technology, allows retinal imaging through undilated pupils and media opacities. We conducted a cross-sectional study to evaluate the performance of a cSLO compared with digital retinal camera for non-mydriatic diabetic retinopathy (DR) assessment.
190 participants with diabetes were consecutively recruited from the Singapore Indian Eye Study, a population-based survey of eye diseases in Indian adults aged above 46 years, residing in Singapore. Images were taken from all participants’ central (ETDRS field 1) and nasal field of both eyes, first using a cSLO (EasyScan, i-Optics, the Netherlands) before pharmacological pupil dilation, and then with a standard digital retinal camera (Canon CR-DGi, Canon, Japan) pre-dilation (for first 102 subjects only) and post-dilation.
Two masked and trained graders reviewed all images for gradability, and DR severity level using the modified Airlie House classification system (no DR, mild nonproliferative DR [NPDR], moderate NPDR or worse). The sensitivity, specificity, positive/negative predictive values (PPV/ NPV), and 95% confidence intervals were calculated to assess the diagnostic performance of cSLO and undilated fundus photography. Cohen’s weighted Kappa was used to calculate the concordance of DR severity.
The mean age of the 190 participants was 63.6 years (standard deviation 9.04) and 52.1% were male. 67 participants (35.3%) had DR (48 mild NPDR, 19 moderate NPDR or worse). Comparing 202 eyes from 102 subjects, we found that gradability of cSLO images was significantly higher than undilated fundus photography (89.1% vs 53.0%, p<0.01). After excluding ungradable images, we compared 330 eyes from 174 subjects using dilated 2-field color fundus photography as reference standard: cSLO had a sensitivity of 41.9% (31.4-52.3), specificity of 98.4% (96.8-100), PPV of 90.0% (80.7-99.3) and NPV of 82.8% (78.4-87.1). We found no significant difference in the diagnostic performance between cSLO and undilated fundus photography (all p-values >0.10). The concordance of DR severity assessment between dilated fundus photography and cSLO was 0.510 (0.427–0.593).
cSLO offers a diagnostic performance equivalent to undilated fundus photography, with superior gradability of images. Therefore, cSLO is a potential tool for non-mydriatic DR screening in primary care setting.