Comparison of retinal pathology visualization in multi-spectral scanning laser imaging

Abstract

Purpose : To compare retinal pathology visualization in multi-spectral confocal scanning laser ophthalmoscope (cSLO) imaging between the Spectralis and Optos devices.

Methods : This cross-sectional study included 42 eyes from 30 patients with common retinal pathologies: age-related macular degeneration (19 eyes), diabetic retinopathy (10 eyes) and epiretinal membrane (ERM, 13 eyes). All patients underwent retinal imaging with a color fundus camera (Topcon TRC-50DX) and two multi-spectral cSLO devices: Spectralis HRA-2 system and Optos P200 system. The Optos image was cropped to a similar size as the Spectralis image to include the macula. Seven retina specialists graded the images in 2 parts. In the first part, the graders were instructed to mark as many retinal pathologies as possible in a given image within a 5 X 5 grid centered on the fovea. The number of marked cells was counted for each pathology. A pathology detection rate was defined as the percent of graders that identified the pathology on a given image. In the second part, the 3 images from each eye were grouped together. The graders were instructed to rank the images according to the quality of retinal pathology visualization.

Results : The overall intergrader agreement of the number of marked cells was similar across all imaging modalities. The average number of marked cells in all eyes was higher in the Spectralis compared with Topcon (1.43 cells, p < 0.0001) and Optos (1.59 cells, p <0.0001). This difference was largest in eyes with ERM. No such difference was found between the Topcon and Optos images (0.16 cells, p = 0.67). The average pathology detection rate for all pathologies was similar across the 3 modalities (Topcon 65%, Spectralis 66%, Optos 66%, all p > 0.05). However, the average ERM detection rate was significantly higher in Spectralis (92%) compared with Topcon (69%, p = 0.001) or Optos (68%, p = 0.0007). Overall, Spectralis was ranked higher than Optos for pathology visualization (67% vs. 50% of the times, respectively), but lower than Topcon (83% of the times).

Conclusions : Larger area of macular involvement and better ERM visualization were noted in the Spectralis images compared with the Topcon and Optos images. The overall detection of retinal pathology was similar across all three imaging modalities.

This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.

 

Part 1 - 5 X 5 grid centered in the fovea for image grading.

 

Part 2 - the 3 images of the same eye for image ranking.