Abstract
Purpose :
To present an assessment of image quality of a portable retinal camera for screening for diabetic retinopathy (DR) as part of a tele-screening system at Clinicas del Azucar (CDA), Monterrey, Mexico.
Methods :
DR screening at two CDA clinics was performed using the Pictor Plus camera (Volk Optical Inc, Mentor, OH). A case is composed of at least one image of the macula and one image of the optic disc from each eye. The Image Quality Analyzer (IQA), a software to detect retinal images of insufficient quality for interpretation was developed by VisionQuest. Images are read within 48 hours in the USA by a certified retinal grader for disease and quality according to the International Clinical DR Grading Scale. Images were processed by the IQA and its results compared to the grader’s quality evaluation.
Results :
From May to October, 2016, 387 cases, comprising more than 1,500 images, were acquired. The number of unreadable cases was 61 or 16%. Of these, 29 cases (48%) are due to photographers’ technical errors (focus, alignment, etc.) and the rest (32 or 52%) due to physiological issues (cataracts, small pupils, etc.). Of the readable cases, 255 (66%) had no signs of DR or Mild NPDR, 51 (13%) had moderate NPDR, 20 (5%) had severe NPDR or worse, and 18 cases (5%) were deemed suspect for CSME.
A subset of 744 images were processed by the IQA and their outputs compared to the grader’s quality evaluation. The IQA correctly identified 84% of the images that the grader deemed as inadequate for interpretation. The IQA likewise identified 56% of the images that the grader classified as low quality. Further, the IQA classified as unreadable, only 27% of the images that the grader classified as medium quality, and only 7% of those classified as good quality.
Conclusions :
These results show that it is clinically feasible to use a portable retinal camera, such as the Pictor Plus, as part of a tele-retinal screening system. Portable cameras can have unreadable image rates comparable to table top cameras as reported in the literature, e.g. 11% to 29%. The results of real-time, automatic image quality assessment show that immediate feedback to the photographer, such as suggestions on what steps to take to eliminate artifacts affecting image quality, can further reduce the percentage of unreadable images.The savings by switching to low cost cameras will facilitate the expansion of this service.
This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.