Advancing Early Detection of Diabetic Retinopathy in Western Australia

An update from Dr Danuta Sampson, 2024 DRWA Research Grant Recipient

Every year Diabetes Research WA supports research that brings us closer to earlier detection, improved treatment and a healthier future for Western Australians living with diabetes. In 2023, we awarded our 2024 grant to a project that reflects these goals with clarity. Led by physicist and vision science researcher Dr Danuta (Danka) Sampson from the Lions Eye Institute and The University of Western Australia, this work is strengthening our understanding of diabetic retinopathy by harnessing the power of advanced imaging and big data.

Why this research matters

More than 128,000 people in Western Australia live with diabetes. Many more are estimated to have the condition without knowing it. One of the most serious complications is diabetic retinopathy, a progressive eye disease that damages the smallest blood vessels at the back of the eye. These tiny vessels, known as the microvasculature, can begin to change long before a person notices any difference in their vision. Catching these changes early is vital because timely intervention can prevent or slow vision loss.

Despite significant progress in eye imaging, clinicians still face a challenge. Current diagnostic tools are not sensitive enough to reliably detect the earliest signs of microvascular dysfunction, which limits opportunities for early intervention.

This is where Dr Sampson’s project steps in.

Building a harmonised imaging dataset for Western Australia

Modern imaging known as optical coherence tomography angiography (OCTA) has transformed eye care. OCTA allows clinicians and researchers to see the smallest blood vessels in incredible detail without the need for invasive dyes or lengthy procedures. It is quick, safe and has become increasingly common in eye clinics across the country.

However, there is a major barrier to fully realising the potential of OCTA.

Different clinics use different OCTA instruments and these devices do not always produce comparable images. The variations are technical rather than biological, yet they make it impossible to reliably combine data across sites. Without harmonisation, it is difficult to build the large datasets required to discover early biomarkers of diabetic eye disease.

Dr Sampson’s project, Harmonisation of Western Australia multi site OCTA datasets for new biomarker discovery and application in diabetic retinopathy, is designed to overcome this barrier. The aim is to develop and validate a data analysis framework that removes these technical differences so that OCTA images collected across Western Australia can be analysed as one.

Progress so far

The team has made significant progress. In the first half of the project they have:

• Reviewed more than 1,000 OCTA images collected from Lions Eye Institute (Perth) and Lions Outback Vision (Broome)
• Built a dataset of 400 high quality images of Western Australians with and without diabetic retinopathy to test and refine harmonisation models.
• Recruited 70 healthy volunteers, imaging each of them with three different OCTA instruments. This volunteer dataset is crucial for validating the harmonisation approach.
• Begun analysing early differences in vascular parameters across metropolitan and regional WA, including preliminary comparisons between Indigenous and non Indigenous participants.

During her recent presentation at the DRWA Grant Award Breakfast, Dr Sampson shared that the team is now preparing to implement and refine the computer models that will harmonise OCTA images across different devices. These models are based on an established statistical method known as ComBat, which has been used for more than twenty years to harmonise large medical imaging datasets in other fields. Applying this method to OCTA is a significant and innovative step forward.

The early results are promising. The team has already demonstrated that harmonisation reduces instrument related differences in vascular measurements. Once complete, this will allow data from multiple clinics to be pooled into a single large dataset, giving researchers the statistical power needed to identify new early biomarkers of diabetic retinopathy.

What the early findings are beginning to show

Although full analysis is still underway, the preliminary findings have highlighted important patterns.

The team noted meaningful differences in the microvascular parameters of participants from metropolitan Perth compared with those from Broome. They also observed that the average age of participants with diabetic retinopathy differed considerably between the two regions. This has raised new research questions about how diabetes related vessel changes may vary between groups, including Indigenous and non Indigenous Australians.

Dr Sampson explained that understanding these differences is essential, saying that these early observations are an important reminder that diabetic retinopathy does not always follow the same pattern across populations. Larger and harmonised datasets will allow the team to explore this in a more rigorous and meaningful way.

Looking ahead

The next stage of the project will focus on refining the harmonisation models, generating pilot data and deepening the collaboration with the project’s Community and Consumer Involvement group. This group includes people with lived experience of diabetes who help ensure that the research remains grounded, relevant and aligned with community needs.

Dr Sampson shared her ambition for the future. She hopes that OCTA will eventually become a routine part of eye screening for people living with diabetes, especially once the technology is supported by clear diagnostic biomarkers.

A message of thanks

Dr Sampson expressed her gratitude to all collaborators who have contributed data, clinical expertise and volunteer support, including colleagues from Lions Eye Institute, Lions Outback Vision, The University of Western Australia, Fremantle Hospital and University College London. She also acknowledged the ongoing support provided by Diabetes Research WA and the donors who make this work possible.

Her commitment to improving early detection of diabetic retinopathy is clear. By building one of the first harmonised OCTA datasets in Australia, this project is creating new opportunities to detect vision threatening changes earlier and more accurately than ever before.