The photographer’s learning curve: Image quality over time in non-mydriatic fundus photographs taken with a portable camera

Abstract

Purpose : Portable fundus cameras offer promise to expand ophthalmic services to underserved areas, but successful task shifting from physicians to healthcare workers (HCWs) depends on the ability of HCWs to attain high quality images with these portable devices. We performed a cross-sectional analysis of fundus image quality (IQ) over time with a handheld non-mydriatic fundus camera used to aid in the diagnosis of vision-threatening diabetic retinopathy.

Methods : A HCW at an Aravind Eye Hospital (Pondicherry, India) obtained three undilated fundus images (macular, nasal, and superotemporal fields) from 274 eyes of 155 participants over 13 months using the Smartscope (Smartscope, Optomed, Finland). The HCW had prior experience as an ophthalmic photographer and underwent a brief training period with the Smartscope (75 eyes) prior to obtaining images for the study. Images were de-identified, ordered randomly, and graded for IQ by two masked retina specialists. Images were graded on a scale of 1 to 3 (1, excellent; 2, acceptable; 3, ungradable). Images were then ordered sequentially and divided into quintiles (162-165 images/quintile) to assess IQ and ungradable rate over time. Wilcoxon Rank-Sum and Chi-square tests were used to compare performance between quintiles.

Results : 1644 IQ grades from 822 images were analyzed. Overall, both graders reported one or more gradable images in 88.0% of the eyes. When evaluating all three fields per eye, the rate of non-gradable images was 17.3% in the 1st quintile, worsened to 29.3% in the 3rd quintile, and improved to 9.9% by the 5th quintile (p<0.001). Mean IQ in the 5th quintile was 1.57 (SD=0.67), which was significantly better than mean IQ in the 1st (M=1.71, SD=0.74, p=0.024), 2nd (M=1.79, SD=0.76, p<0.001), or 3rd (M=1.96, SD=0.79, p<0.001) quintiles.

Conclusions : The HCW in our study developed a 90.1% success rate at producing gradable images by the final quintile of photographs after photographing approximately 290 eyes with the Smartscope. These data demonstrate that with sufficient training, HCWs can likely obtain high quality images with a portable non-mydriatic fundus camera. However, it requires a reasonably high patient volume to give the HCW consistent practice to gain these photography skills. Future research should investigate novel portable fundus imaging options that require less HCW skill to attain high quality images.

This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.