Abstract
Purpose :
Hand-held cameras are marketed for eye screenings. These devices are often low cost and portable which should improve disease screening. But how do these cameras compare to each other? We sought to assess the utility of 3 hand-held cameras, the PanOptic (Welch Allyn, Skaneateles Falls, NY), Portable Retinal Imaging System (PRIS) (D-Eye, Padova, Italy) and Pictor Plus (Volk, Mentor, OH) to screen for glaucoma.
Methods :
84 patients from ages 18-60 were recruited from Temple Ophthalmology. Exclusion criteria were cataracts, dementia, unable to converse in English, incapable of having photos, or blind in the right eye. 1 image of the right eye was captured with the PanOptic, the PRIS, and the Pictor Plus in that order. For the Pictor Plus, a color image and a red-free image were obtained. A photo montage of the 4 images was created so that all 4 images could be seen at once. 2 reviewers independently evaluated the montages to determine which images were usable for glaucoma screening based on image clarity and optic nerve/nerve fiber layer visibility.
Results :
For the PanOptic, interpreter 1 deemed 77.4% of the images as usable while interpreter 2 deemed 32.1% of images as usable. For the PRIS, 66.7% and 33.3% of the images were deemed usable. For the Volk Pictor Plus, 88.1% and 83.3% of the color images were deemed usable, and for the red-free images, 88.1% and 75.0%. The Cochran Q test detected heterogeneity in the frequency of usefulness for the 4 sets of images in detecting glaucoma regardless of interpreter (Q=93.3 and 26.5, p<.0001 for both). Using the McNemar test, there was a significant difference in frequency of usable images for screening between the PanOptic and the Pictor Plus (p=0.049 to p<0.0001, depending on interpreter and Pictor Plus color vs red-free), and a significant difference between the PRIS and the Pictor Plus (p<0.0001 for all cases). The degree of concordance between interpreters for PanOptic and PRIS was minimal, κ-coefficient=0.203, 95% CI=[0.085, 0.323] and κ-coefficient=0.316, 95% CI=[0.173, 0.459]. The degree of concordance for Pictor Plus color and red-free images was good, κ-coefficient=0.787, 95% CI=[0.610, 0.965] and κ-coefficient=0.6100, 95% CI=[0.4097, 0.8104].
Conclusions :
There is an advantage to using images from the Pictor Plus, whether color or red-free, over images from the PanOptic and PRIS when acquiring optic nerve photos.
This is a 2020 ARVO Annual Meeting abstract.