Predicting submacular haemorrhagen using deep learning

My Topol fellowship problem / project:

A large, acute bleed under the centre of the retina (acute submacular haemorrhage) is a rare but sight devastating complication of age related wear and tear at the centre of the retina (wet age-related macular degeneration (AMD)), which affects approximately  5.4  patients  per million per  annum.  Current treatments are limited to injection (with anti-VEGF), surgery (vitrectomy/tPA/pneumatic displacement), or observation. These provide a limited benefit for improvement of vision. Severe sight impairment can have a serious effect on someone’s well-being, general health and ability to cope with activities of daily living. It also impacts significantly on carers and social services, which may need to help patients when they develop significant vision loss.

We are running a clinical trial (The London Project, under Professor Lyndon DaCruz) reviewing the safety and feasibility of a patch which replaces the layer under the retina which supports the retina with nutrients (retinal pigment epithelium), in restoring vision in patients affected with an acute submacular haemorrhage secondary to wet AMD.

During my time as a Topol Digital fellow I will look to deliver a digital project trying to identify which patients with wet AMD go on to develop an acute submacular haemorrhage using artificial intelligence. We are also reviewing how to improve imaging of the eye during surgery (using intra-operative OCT retinal imaging) using artificial intelligence. A robotic arm which could use this improved imaging, to help a surgeon  deliver  therapies under the retina in the future, is currently  in development.  Our team at Moorfields will analyse millions of images from  over 27 centres  in the UK  to  develop an algorithm  which can predict submacular haemorrhage and then test it on Moorfields data. I will also work with the robotic surgery department at Kings College London, under Dr Christos Bergeles, on the improvement of imaging during surgery inside the eye.

These digital projects go alongside the clinical trial, looking to create a pathway to predict who may need this novel therapy, and deliver this treatment with the assistance of robotic systems in the future. The longer-term aim would the delivery of this therapy for milder forms of wet AMD, which is the leading cause of vision loss for the over 60s in developed countries.

After completing undergraduate training at University of Sheffield, I secured academic foundation programme training, where I undertook basic research in retinal angiogenesis in insulin resistant mice. I went on to work as a clinical research fellow where I undertook research on novel imaging technologies (OCT angiography), as well as running clinical trials, before securing an ophthalmology registrar post in North Thames.

I am currently an Ophthalmology trainee at Moorfields Eye Hospital, who is currently working towards a PhD in the field of vitreoretinal surgery. Ophthalmology for me is attractive as a craft and intellectual speciality, where both surgery and medicine are practiced.

My PhD focuses on prediction and treatment of submacular haemorrhage in wet age-related macular degeneration (AMD) using conventional and robotic systems. The 3 main projects of the PhD are the clinical trial (London Project) for treatment of submacular haemorrhage in wet AMD using a stem-cell derived retinal pigment epithelium (RPE) patch, prediction of submacular haemorrhage in wet AMD using deep learning and assessment of novel robotic systems in assisting vitreoretinal surgery for delivery of subretinal therapeutics.

The field of artificial intelligence holds a lot of promise in the field of the ophthalmology, due to the abundance of imaging data. We are reviewing if deep learning can aid in prediction of sight threatening conditions, for which we are currently testing novel therapeutics.

Outside of work I enjoy learning to code, squash and exploring new cuisines and cultures.