Fractal Dimension of Vitreoretinal Amyloid: Mathematical Characterization of Transthyretin Amyloid Deposition

Inside the human body, certain proteins can sometimes misfold, clumping together into rigid, thread-like structures that the body cannot break down. When this happens, these clumps, known as amyloid, can build up in organs and tissues, causing damage over time. One specific type of this condition, called hereditary transthyretin amyloidosis, affects the eyes as well as the heart and nerves. Even when doctors successfully treat the disease in the rest of the body using modern medicines, the eye can remain a separate, active site where these protein clumps continue to form. This happens because the eye has its own internal machinery for making the protein, allowing the disease to progress locally even when the systemic treatment is working well. For doctors, this creates a difficult challenge: they need a way to see exactly how much damage is happening inside the eye and how fast it is growing, but current methods rely on looking at images and describing what they see with words like "cloudy" or "spiky," which are not precise enough to track subtle changes over time.

A team of researchers at Wills Eye Hospital set out to solve this problem by applying a mathematical concept called fractal dimension to the images of the eye. In simple terms, fractal dimension is a way to measure how rough or complex a shape is. If you look at a perfectly smooth line, it has a low complexity score. If you look at a jagged, craggy coastline or a rough mountain range, the score is much higher because the shape fills up space in a more complicated, irregular way. The researchers suspected that the clumps of amyloid protein growing on the surface of the retina would look much rougher and more complex than the smooth, healthy surface of a normal eye. By turning the eye images into digital maps and counting how many tiny boxes were needed to cover the edges of these protein clumps, they could assign a single number to the roughness of the deposit, turning a vague visual impression into a precise measurement.

To test this idea, the team studied a 77-year-old woman who had the genetic form of the disease and was experiencing worsening vision in both eyes. They used a high-resolution camera that takes cross-sectional slices of the eye, similar to an ultrasound but using light instead of sound, to create detailed maps of the retina. They also gathered similar images from three healthy people to serve as a baseline for comparison. Using custom computer programs, the researchers traced the very top edge of the amyloid deposits in thousands of these eye slices. They created two types of digital outlines for each image: one that traced just the jagged top surface of the clumps to measure their roughness, and another that filled in the entire area of the clump to measure how much space they occupied. They then calculated the complexity score for every single slice.

The results were striking. The healthy eyes showed a very smooth surface, with complexity scores that stayed within a narrow, low range. In contrast, every single slice from the patient's eyes showed a much higher complexity score, with no overlap between the healthy and the diseased eyes. The roughness of the amyloid deposits was so distinct that the computer could clearly separate the sick eyes from the healthy ones without any human guessing. The researchers found that the right eye, which had undergone surgery to remove the vitreous gel five years prior, had even rougher deposits than the left eye, which had surgery more recently. This suggested that the timing of the surgery and the remaining environment inside the eye might influence how the protein clumps arrange themselves. Furthermore, the deposits were not uniform; they were rougher at the top of the eye and became slightly smoother as you moved down toward the bottom, hinting that gravity might play a role in how these heavy protein clumps settle and grow.

The study also looked at how the disease changed over time by reviewing older images of the patient's right eye taken over several years. The data showed that the roughness of the deposits was steadily increasing, rising by a small but measurable amount each year. This confirmed that the mathematical method could detect the slow, progressive nature of the disease, even when the changes were too subtle for the naked eye to notice in a standard exam. In the left eye, which had been surgically cleared more recently, the trend was not significant, likely because the surgery had reset the environment. To ensure their computer method was accurate, the team compared their results with a different, established software program and found that the two agreed almost perfectly, confirming that their new approach was reliable.

This work suggests that measuring the geometric complexity of eye deposits could become a powerful tool for doctors. Instead of relying on subjective descriptions, physicians could use this number to track whether a treatment is working, to see if the disease is speeding up or slowing down, and to compare how different patients respond to therapy. While this initial study focused on just one patient, the clear separation between the healthy and diseased eyes indicates that this method has real potential. As treatments for the systemic form of the disease improve and people live longer, the eye may remain the final frontier where the disease persists. Having a precise, objective way to measure what is happening inside that small, dark compartment could help doctors protect vision in ways that were not previously possible.