• Automatic Ocular Disease Identification System

Screening and automatic identification of diabetic retinopathy, glaucoma and cataract using cutting edge ophthalmic instruments and next generation image processing technology

Founded in 1965, Ohira (trading as jomt) is one of the leading manufacturers and suppliers of medical and ophthalmic equipment in Japan. In 2017 Ohira partnered with Effective Solutions to boost their ophthalmic product range sold to over 60 countries, with superior data processing capabilities and configurable machine learning modules of our Ananke IoT platform to automate ocular disease identification.

The three-year project is funded by a grant from Niigata Prefecture Government and guided by Niigata University.

IT CHALLENGE

Integrating IoT capabilities, image processing, machine learning and algorithms to automate the identification of eye disease.

Ohira’s team have deep expertise in ophthalmology — expertise it has nurtured throughout its fifty-year history. To realise the promise of IoT technologies, however, its operations team needed to pair its ophthalmic expertise with IoT skills and knowledge. And it needed to do this without diluting its resources or diverting focus from its core manufacturing priorities.

Ohira also needed accuracy and reliability to safeguard the reputation it had built for over fifty years: it wanted the ability to leverage analysis in real-time to automate detection of diabetic retinopathy, glaucoma and cataract from the fundus photographs taken by its ophthalmic instruments.

SOLUTION

Screening and automatic identification of diabetic retinopathy, glaucoma and cataract using cutting edge ophthalmic instruments and next generation image processing technology

Ohira’s first step was to find an ophthalmic research partner to ringfence the science behind the project and complement its ophthalmic instrument technology with IoT capabilities. Its search led it to Niigata University and Effective Solutions, a provider of real-time data analytics and configurable machine learning solutions.

The second step was to secure grant funding from the Niigata Prefecture Government for ophthalmic and IoT research.

The first phase features:

  • Raw data input
  • Image pre-processing
  • Segmentation of patient lesion
  • Training database
  • Feature extraction
  • Classification
  • Quantitative scoring of disease severity

RESULTS

Into the 3rd month of the three-year project the team had made big strides towards the end solution

Effective solutions collaborated with the Ohira’s team to develop the algorithms to identify the optic disk and blood vessels accurately. Work is underway by the Niigata University research team to prove the concept by analysing image databases.

RESEARCH OPPORTUNITIES

Effective Solutions is looking to partner with world class research organisations and universities for following research opportunities:

  • New algorithm development to identify different eye diseases
  • Automate eye disease identification process using AI
  • Eliminate human intervention for diabetic retinopathy and glaucoma diagnosis, using AI