June 2017
Volume 58, Issue 8
Open Access
ARVO Annual Meeting Abstract  |   June 2017
Comparison of automated and expert human grading of diabetic retinopathy using smartphone-based retinal photography
Author Affiliations & Notes
  • Tyson Kim
    Ophthalmology and Visual Sciences, University of Michigan Kellogg Eye Center, Ann Arbor, Michigan, United States
  • Patrick Li
    Ophthalmology and Visual Sciences, University of Michigan Kellogg Eye Center, Ann Arbor, Michigan, United States
  • Leslie M Niziol
    Ophthalmology and Visual Sciences, University of Michigan Kellogg Eye Center, Ann Arbor, Michigan, United States
  • Malavika Bhaskaranand
    Eyenuk Inc, Woodland Hills, California, United States
  • Sandeep Bhat
    Eyenuk Inc, Woodland Hills, California, United States
  • Chaithanya Ramachandra
    Eyenuk Inc, Woodland Hills, California, United States
  • Kaushal Solanki
    Eyenuk Inc, Woodland Hills, California, United States
  • Jose R Davila
    Ophthalmology and Visual Sciences, University of Michigan Kellogg Eye Center, Ann Arbor, Michigan, United States
  • Frankie Myers
    Bioengineering, University of California Berkeley, Berkeley, California, United States
  • Clay Reber
    Bioengineering, University of California Berkeley, Berkeley, California, United States
  • David C Musch
    Ophthalmology and Visual Sciences, University of Michigan Kellogg Eye Center, Ann Arbor, Michigan, United States
  • Todd P Margolis
    Ophthalmology and Visual Sciences, Washington University School of Medicine, St. Louis, Missouri, United States
  • Daniel Fletcher
    Bioengineering, University of California Berkeley, Berkeley, California, United States
  • Maria A Woodward
    Ophthalmology and Visual Sciences, University of Michigan Kellogg Eye Center, Ann Arbor, Michigan, United States
  • Yannis Mantas Paulus
    Ophthalmology and Visual Sciences, University of Michigan Kellogg Eye Center, Ann Arbor, Michigan, United States
  • Footnotes
    Commercial Relationships   Tyson Kim, CellScope Retina (P); Patrick Li, None; Leslie Niziol, None; Malavika Bhaskaranand, Eyenuk Inc (E), Eyenuk Inc (P); Sandeep Bhat, Eyenuk Inc (E), Eyenuk Inc (P); Chaithanya Ramachandra, Eyenuk Inc (E), Eyenuk Inc (P); Kaushal Solanki, Eyenuk (E), Eyenuk (P); Jose Davila, None; Frankie Myers, CellScope Retina (P); Clay Reber, CellScope Retina (P); David Musch, None; Todd Margolis, CellScope (P); Daniel Fletcher, CellScope Retina (P); Maria Woodward, None; Yannis Paulus, None
  • Footnotes
    Support  University of Michigan Center for Entrepreneurship Dean’s Engineering Translational Prototype Research Fund; University of Michigan Translational Research and Commercialization for Life Sciences Grant # N021025, and University of Michigan Department of Ophthalmology and Visual Sciences department support
Investigative Ophthalmology & Visual Science June 2017, Vol.58, 659. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to Subscribers Only
      Sign In or Create an Account ×
    • Get Citation

      Tyson Kim, Patrick Li, Leslie M Niziol, Malavika Bhaskaranand, Sandeep Bhat, Chaithanya Ramachandra, Kaushal Solanki, Jose R Davila, Frankie Myers, Clay Reber, David C Musch, Todd P Margolis, Daniel Fletcher, Maria A Woodward, Yannis Mantas Paulus; Comparison of automated and expert human grading of diabetic retinopathy using smartphone-based retinal photography. Invest. Ophthalmol. Vis. Sci. 2017;58(8):659.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Purpose : Diabetic retinopathy (DR) is the leading cause of vision loss and blindness in working-age adults. Retinal photography is a well-validated screening tool for DR but is expensive with limited portability. Smartphone-based photography addresses these limitations. We combine smartphone-based retinal photography with automated image analysis to detect referral-warranted DR (RWDR).

Methods : A mydriatic smartphone-based retinal camera (Cellscope Retina) was used to image diabetic patients at the University of Michigan Kellogg Eye Center Retina Clinic. Images were analyzed with cloud-based EyeApp software to generate screening recommendations (refer/no refer) based on presence of moderate non-proliferative DR or higher and/or markers for clinically significant macular edema (CSME). Images were independently evaluated by two masked readers for severity of DR and/or presence of CSME, and similarly categorized as refer/no refer. Results from EyeApp and masked readers were compared against clinical diagnosis made with slit-lamp biomicroscopy to determine sensitivity and specificity, at both eye-level and patient-level (RWDR defined at the patient-level if present in at least 1 eye of a subject).

Results : 72 patients (144 eyes) were imaged. RWDR was present in 101 eyes (77.1%) and absent in 30 eyes (22.9%) by gold standard clinical diagnosis. For detecting RWDR at the eye-level, EyeApp had a sensitivity of 77.4% and specificity of 70.4%; grader 1 had a sensitivity of 93.9% and specificity of 51.9%; grader 2 had a sensitivity of 88.8% and specificity of 63.0%. At the patient-level, RWDR was present in 55 subjects (76.4%) and absent in 12 subjects (16.7%). For detecting RWDR at patient-level, EyeApp had a sensitivity of 93.9% and specificity of 75.0%; grader 1 had a sensitivity of 98.1% and specificity of 41.7%; and grader 2 had a sensitivity of 96.0% and specificity of 41.7% (Figure 1).

Conclusions : CellScope Retina combined with EyeApp software achieves reasonable sensitivity and specificity in detection of RWDR at the person-level, with lower sensitivity but higher specificity than human graders. As the patient population studied had high prevalence of DR, additional study of a more typical screening population of diabetics in the community is needed.

This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.

 

Sensitivity & specificity for referral-warranted diabetic retinopathy from CellScope Retina images using EyeApp and expert human grading

Sensitivity & specificity for referral-warranted diabetic retinopathy from CellScope Retina images using EyeApp and expert human grading

×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×