AI based Diabetic Retinopathy Screening System

DRISTI analyses retinal fundus images for Diabetic Retinopathy lesions and directs suspicious cases for detailed ophthalmologist's review. Most of the negative cases can be screened out without ophthalmologist's intervention

Brief Description

Diabetic Retinopathy (DR) is the damage caused to retina by complications of diabetes which can eventually lead to blindness. Presently the ratio of ophthalmologist to general population is 1 : 80,000 which makes it nearly impossible to implement screening programs manually with human expertise alone. DRISTI, using quantitative image analysis algorithms, analyses retinal fundus images for DR lesions and directs DR suspicious cases for detailed ophthalmologist review. DR negative cases are screened out without ophthalmologist intervention.










Features and Technical Specifications

  • Analyses images and automatically classifies retina as either Non-DR or suspicious of DR.

  • Patient prioritization based on automated severity scoring.

  • Only cases having suspicion of DR need ophthalmologist's review.

  • Increase screening efficiency and reduce workload of ophthalmologists by up to 78%.

  • Automated assessment of retinal image quality and real-time operator notification for re-imaging.

  • Detects retinal pathologies like micro-aneurisms, dot, blot, flame haemorrhages, exudates, cotton-wool-spots automatically.

  • Detects optic disc and retinal blood vessels automatically.

  • Intuitive GUI for automated analysis of image, patient or patient batch.

  • Systematic storage of patient details and automated analysis results.

  • Reports automated analysis results.

  • Application in population screening for DR.

  • Can be operated even by a non-ophthalmologist
    Platform required(if any)

  • Fundus camera

  • 40-50 deg field of view

  • Min 3MP color sensor

  • Software for image capture and archival along with patient details

  • Provision to capture image in lossless formats like BMP, TIFF etc.

  • Workstation

  • Core i5, 3 Ghz with 4GB RAM or better configuration

  • Monitor with 1920x1080 resolution

  • Windows 7 64 bit OS


  • Motorized table with height adjustment for mounting the fundus camera

  • Enclosed room with controlled lighting, so as not to interfere with the imaging

 Download Brochure

Contact Details for Techno Commercial Information

Group Head,
Health Informatics & Software Technology Group,
email -