Purpose
To investigate the retinal screening capabilities of an automated fundus camera
Methods
Prospective study with patients of any race, age, sex, and pathology. A comprehensive ophthalmological examination was performed including auto refraction, best corrected/pinhole visual acuity, slit lamp evaluation, retinal biomicroscopy, intraocular pressure, and if necessary fluorescein angiography, visual field testing, and optic disc topographical imaging. The multifield protocol of the DRS (CenterVue Spa, Italy) fundus camera was used to acquire images of the central, peripheral nasal, peripheral temporal, superior, and inferior retina with a 20 sec delay in between acquisitions. Both eyes were tested with a random selection of the initial tested eye. Undilated and dilated images were acquired while flash intensity was set at a lower level than default to ease the comfort of the patients. Images were graded in a blind masked fashion as unsatisfactory, somewhat satisfactory, satisfactory, and very satisfactory as well as normal, suspect, abnormal, and ungradable by a retinal specialist. A pairwise analysis was performed to compare the undilated and dilated images, while the collective morphological grading outcome of each eye was compared with the ophthalmological examination
Results
In total, 94 patients with mean age 56 ± 16.28 years, mean spherical equivalent 0.10 ± 2.22 D, and mean VA 0.16 ± 0.20 logMAR were included. As expected, dilated images were graded higher than undilated images (0%, 12.7%, 55.5%, and 31.8% vs 17.8%, 40.5%, 25.8%, and 15.9% for unsatisfactory, somewhat satisfactory, satisfactory, and very satisfactory). Likewise, 46.1%, 27.5%, 17.8% and 8.6% of undilated images and 59.8%, 17.6%, 22.5%, and 0% of dilated images were graded normal, suspect, abnormal, and ungradable, respectively. The pairwise morphology grading agreement between undilated and dilated images was 74.6%. When considering suspect and abnormal as one group the corresponding percentage was 92.9%. Central field exhibited the highest agreement followed by the temporal field. The agreement of the collective morphological grading of each eye with the retinal abnormalities clinical examination was higher than 94% for both cases
Conclusions
Retinal screening with the automated fundus camera shows promising results. Further analysis is required to stratify its importance with respect to the various retinal abnormalities