PLoS One. 2015 Oct 1;10(10):e0139148. doi: 10.1371/journal.pone.0139148.

Results of Automated Retinal Image Analysis for Detection of Diabetic Retinopathy from the Nakuru Study, Kenya.
 

Hansen MB1*, Abràmoff MD2, Folk JC2, Mathenge W3, Bastawrous A4, Peto T5.
  • 1NIHR Biomedical Research Centre at Moorfields Eye Hospital NHS Foundation Trust and UCL Institute of Ophtalmology, London, United Kingdom; Research Unit of Ophthalmology, University of Southern Denmark, Odense, Denmark.
  • 2Department of Ophthalmology and Visual Sciences, University of Iowa Hospital and Clinics, Iowa City, IA, 52242, United States of America.
  • 3Rwanda International Institute of Ophthalmology, P.O. Box 312, Kigali, Rwanda.
  • 4International Centre for Eye Health, Department of Clinical Research, Faculty of Infectious and Tropical Diseases, London School of Hygiene and Tropical Medicine (LSHTM), London, United Kingdom.
  • 5NIHR Biomedical Research Centre at Moorfields Eye Hospital NHS Foundation Trust and UCL Institute of Ophtalmology, London, United Kingdom.

* Contact: Dr. Morten B. Hansen, Morten.Hansen@moorfields.nhs.uk

 

Abstract

OBJECTIVE: Digital retinal imaging is an established method of screening for diabetic retinopathy (DR). It has been established that currently about 1% of the world’s blind or visually impaired is due to DR. However, the increasing prevalence of diabetes mellitus and DR is creating an increased workload on those with expertise in grading retinal images. Safe and reliable automated analysis of retinal images may support screening services worldwide. This study aimed to compare the Iowa Detection Program (IDP) ability to detect diabetic eye diseases (DED) to human grading carried out at Moorfields Reading Centre on the population of Nakuru Study from Kenya.

PARTICIPANTS: Retinal images were taken from participants of the Nakuru Eye Disease Study in Kenya in 2007/08 (n = 4,381 participants [NW6 Topcon Digital Retinal Camera]).

METHODS: First, human grading was performed for the presence or absence of DR, and for those with DR this was sub-divided in to referable or non-referable DR. The automated IDP software was deployed to identify those with DR and also to categorize the severity of DR.

MAIN OUTCOME MEASURES: The primary outcomes were sensitivity, specificity, and positive and negative predictive value of IDP versus the human grader as reference standard.

RESULTS: Altogether 3,460 participants were included. 113 had DED, giving a prevalence of 3.3% (95% CI, 2.7-3.9%). Sensitivity of the IDP to detect DED as by the human grading was 91.0% (95% CI, 88.0-93.4%). The IDP ability to detect DED gave an AUC of 0.878 (95% CI 0.850-0.905). It showed a negative predictive value of 98%. The IDP missed no vision threatening retinopathy in any patients and none of the false negative cases met criteria for treatment.

CONCLUSIONS: In this epidemiological sample, the IDP’s grading was comparable to that of human graders’. It therefore might be feasible to consider inclusion into usual epidemiological grading.

PMID: 26425849

 

Supplement

Studies from the last decade have shown predictions for the increase in prevalence of diabetes mellitus (DM).  The International Diabetes Federation (IDF) estimates that currently 1 in 11 adults have diabetes equal to 415 million people world wide. Their predictions for 2040 is that 1 in 10 adults equal to 642 million would be living with DM. [1]

People living with DM are at risk of developing co-morbidities to their DM such as diabetic retinopathy (DR) in the eyes. Earlier DR was the leading cause of visual impairment in industrialised countries. This was also the situation in the UK but due to a well organised DR screening programme this is no longer the case. Regular eye screening is the keystone in discovering DR changes in the retina at an early stage. European recommendation for people with DM is an annual control of the retina. The treatment at the earliest stages is tighter blood glucose control, therefor it’s important to interfere before the DR develops into a sight-threatening disease where at this stage either laser or vitrectomy treatment is necessary.

 

With the rapid increase in people living with DM the numbers needed to be eye screened annually is set to increase as well. This increase creates an even bigger burden on health care personnel than the high burden already seen today. Therefor in order to maintain a high quality screening service within the timely recommendation we need to look at new methods to lower time and cost of the services. Our idea was to test whether computer-based algorithm like the Iowa Detection Program (IDP) could function in a screening setting. The idea was the IDP should divide screened participants into two categories; no-DR and moderate to proliferative DR changes based on the retina fundus image. By dividing participants into two categories an ophthalmologist could focus their time only on the ones showing DR changes.  Leaving the healthy participants with no DR changes aside until their next annual eye screening examination.

 

Studies have shown that normally 70% of a population with DM show no signs of DR for some years but still there is put a lot of effort into screening these patients as well. Our hypothesis was if a reliable software could screen out most the the 70% of healthy individuals without human involvement, the screening service would be more cost-effective leaving only patients with DR changes to be seen by ophthalmologist. Our study showed that the IDP did lower the number of patients needed to be seen by 60%. Another advantages of an automated software are its ability to function and keep running every minute of the day. This would give opportunity for analysing fundus images doing the night-time and the ophthalmologist only to see the relevant images with DR changes the following day. This would increase time available for the once in need of treatment and more patients to be treated.

 

Our study showed the IDP to be reliable in screening participant into two group of no-DR and moderate to proliferative DR changes, without missing any sight-threatening DR eye disease. Still the IDP lacks some of the features human graders and ophthalmologist have. The software does not currently diagnose other diseases than DR. Therefor if patients with DM shows no-DR changes in the retina image but shows signs of glaucoma this would not be picked up by the software at the moment. The software is therefor not fully functional for a screening purpose at its current setting. We hope that further testing and new development to the algorithm could implement a more complete screening tool not only for DR but for glaucoma and age-related macular degeneration (AMD) as well. It is our plan that future studies should investigate in implementing these diseases to get the best results in a safe eye screening setting.

 

References:

  1.   IDF diabetes atlas – Home [Internet]. [cited 30 Mar 2016]. Available: http://www.diabetesatlas.org/
  2.   Liew G, Michaelides M, Bunce C. A comparison of the causes of blindness certifications in England and Wales in working age adults (16-64 years), 1999-2000 with 2009-2010. BMJ Open. 2014;4: e004015. doi:10.1136/bmjopen-2013-004015
  3.   Population screening programmes: NHS diabetic eye screening (DES) programme – GOV.UK [Internet]. [cited 30 Mar 2016]. Available: https://www.gov.uk/topic/population-screening-programmes/diabetic-eye

 

 

Multiselect Ultimate Query Plugin by InoPlugs Web Design Vienna | Webdesign Wien and Juwelier SchönmannMultiselect Ultimate Query Plugin by InoPlugs Web Design Vienna | Webdesign Wien and Juwelier Schönmann