AI for Eye Cancer Prediction

  • Research type

    Research Study

  • Full title

    Development of Artificial Intelligence Techniques for the Prediction of Eye Cancer

  • IRAS ID

    285441

  • Contact name

    Heinrich Heinmann

  • Contact email

    Heinrich.Heimann@liverpoolft.nhs.uk

  • Sponsor organisation

    Liverpool University Hospitals NHS Foundation Trust

  • Duration of Study in the UK

    1 years, 11 months, 30 days

  • Research summary

    We will use artificial intelligence (AI) to assess and monitor lesions in the back of the eye. Choroidal naevi are lesions visible only on detailed examination by an eye professional or retinal camera. These benign lesions are comparable to moles of the skin, and usually do not cause any problems. Choroidal naevi can be found in 5% in the population over 50 years of age, but the majority are never detected. It may be assumed that several million people are affected by choroidal naevi in the UK, comparable to the number of patients with diabetes.

    Choroidal naevi can very rarely transform into a type of eye cancer called choroidal melanoma, potentially leading to a loss in vision, loss of the eye, and death in about 30% of patients. Only 1 in 8,000 naevi will transform into a melanoma. Because of this risk, lifelong examinations are recommended to monitor for progression. Due to the greatly increased use of retinal cameras by High Street Opticians, the number of detected naevi has risen sharply in recent years. This has led to significantly more referrals to secondary care ophthalmic services and specialist oncology services, involving expense, travel, and sometimes extreme anxiety on the patients’ side. Choroidal naevi can be classified by several features based on their appearance and size. Of the well-established six risk factors defining the risk for progression to a melanoma, four can be identified with modern retinal imaging methods available at modern High Street Optician practices.

    Over the past years, in Liverpool we have developed AI tools to identify patients requiring treatment for common retinal diseases, age-related macular degeneration and diabetic retinopathy, based on retinal images without the need for a medical examination.

    Our aim is to develop our existing AI tool to assess the risk factors for progression of a choroidal naevus into a melanoma and to monitor lesions for growth or change over time based on existing clinical datasets from our service.

  • REC name

    London - Camden & Kings Cross Research Ethics Committee

  • REC reference

    20/LO/1126

  • Date of REC Opinion

    22 Oct 2020

  • REC opinion

    Favourable Opinion