Using retina imaging to study the eye and protect sight
Nicky van den Berg developed methods to help computers read images of the retina with the same clarity as human experts. These methods can detect early signs of eye disease and support specialists in preventing vision loss.
The retina, which is located at the back of the eye, carries vital clues about a person’s health. By examining it in detail, medical experts can detect signs of conditions such as diabetes, high blood pressure, and even early signs of Alzheimer’s disease. Taking a retinal scan is quick, easy, and non-invasive, but understanding the intricate network of vessels is far from simple. While human experts can quickly recognize where vessels branch or cross, computers often struggle with these complex patterns. Nicky van den Berg set out to change that as part of her PhD research.
Nicky van den Berg’s goal in her PhD research was to help computers in the automatic and accurate analysis of retinal images, therefore helping doctors detect early signs of disease and protect patients’ vision.
She focused on three main objectives: tracing the complex paths of blood vessels, connecting pieces of the network that appear broken or interrupted, and distinguishing arteries from veins.
By combining mathematical models with advanced image processing, Van den Berg developed methods that allow computers to ‘see’ the retinal network almost like a human expert. She defended her PhD thesis on Friday, November 28 at the department of Mathematics and Computer Science.
Mapping the hidden roads of the eye
Blood vessels in the retina twist, turn, and cross over each other like a miniature city map of roads. Following these paths accurately is crucial for understanding a person’s health, but it’s no easy task, especially for computers.
developed fast and smart methods that can trace every winding vessel across the retina, carefully preserving every branch and crossing. Her techniques allow computers to interpret the vascular network clearly, even when on first look it is complex and confusing.
Connecting the dots
Retinal images are rarely perfect. Tiny gaps, shadows, or blurs can make vessels appear broken or incomplete.
Van den Berg created ways for computers to ‘connect the dots’, thus ensuring that the whole blood vessel network was distinguishable in all images. This ensures that important details about the structure of the blood vessel network are not lost and that the retina’s structure is faithfully represented.
Arteries and veins
However, not all blood vessels are the same. Arteries and veins carry different types of information about a person’s health, and mixing them up can lead to mistakes.
Van den Berg methods allow computers to tell arteries and veins apart in a reliable manner, even when the images are noisy or unclear.
Some retinal scans can also be hazy or low in contrast, which makes it hard to see the vessels clearly. Van den Berg developed ways to enhance the images, highlighting vessels and reducing distractions. This gives doctors and computers a sharper, more accurate view of the retina’s complex network.
Catching problems early
Van den Berg’s work combines several key advances, including precise vessel tracing, connected network mapping, artery and vein classification, and image enhancement.
Together, these methods transform retinal scans into powerful tools for early diagnosis. Subtle changes in the blood vessels can be detected sooner, allowing doctors to identify diseases such as diabetes or high blood pressure at an earlier stage. This not only supports timely treatment, but also helps prevent vision loss.
PhD researcher Nicky van den Berg. Photo: Angeline Swinkels
-
Supervisors
Remco Duits, Olga Mula Hernández, Tos Berendschot (external)
Written by
More on Health
Latest news
