Sex differences in quantitative fundus parameters measured by artificial intelligence: a cross-sectional community study

The human eye is often called a window to the soul, but for doctors and researchers, it is increasingly viewed as a window to the rest of the body. The retina, the light-sensitive layer at the back of the eye, is the only place in the body where doctors can see blood vessels and nerves without making an incision. Because these tiny structures are so sensitive to the body's overall health, changes in their shape or size can signal issues with the heart, blood pressure, or metabolism long before other symptoms appear. For years, scientists have used artificial intelligence to measure these features with extreme precision, turning a simple photograph of the eye into a detailed map of hundreds of specific numbers. However, a fundamental question has lingered: do these maps look different depending on whether the person is a man or a woman? While we know that men and women have different body sizes and hormonal profiles, it has been unclear how much of the eye's structure is shaped by these biological differences versus other health conditions like diabetes or high blood pressure. Understanding these natural variations is crucial; if a computer program is trained to spot disease but doesn't know what a "normal" eye looks like for a man versus a woman, it might mistake a healthy difference for a sign of sickness.

To answer this, a large team of researchers in Ganzhou, China, turned to a massive community health screening program. They gathered data from over 21,000 people, capturing more than 32,000 high-quality photographs of retinas. The goal was not to find sick eyes, but to understand the baseline differences between healthy men and women. The team used a sophisticated artificial intelligence system called EVisionAI to analyze each image. Instead of a human doctor squinting at a screen, the software measured 49 distinct features, ranging from the width of blood vessels and the shape of the optic nerve head to the density of the tissue surrounding the nerve. The researchers then compared these measurements between men and women, carefully adjusting for age and twenty different systemic health conditions, such as high cholesterol, kidney function, and liver health, to ensure that any differences they found were truly linked to sex and not just to a specific disease.

The results revealed that sex is a powerful factor in the architecture of the eye. After accounting for age and health conditions, the researchers found significant differences in 41 out of the 49 features they measured. The most consistent differences appeared in the optic disc, the area where the optic nerve enters the eye. Men tended to have optic discs that were more horizontally stretched and rounder, with a larger cup—the central depression in the nerve—relative to the size of the disc itself. In contrast, women's discs were slightly more vertically oriented. These differences were not minor; they were consistent across the vast majority of the measurements. For instance, the ratio of the vertical height to the horizontal width of the disc was notably lower in men, and the distance from the center of the disc to its edge was also different.

The study also uncovered distinct patterns in the blood vessels and the surrounding tissue. Men generally had a lower ratio of small arteries to veins, primarily because their arteries were narrower on average. They also showed a higher density of blood vessels in the area immediately surrounding the optic nerve, but a lower density in the central part of the retina near the macula, which is responsible for sharp, central vision. Furthermore, the tissue surrounding the optic nerve, known as peripapillary atrophy, was larger and more pronounced in men. This area, which can sometimes look like a pale ring around the nerve, was taller and wider in men than in women. Interestingly, the researchers found that these differences remained largely the same even after they adjusted for twenty different health conditions. This suggests that the structural differences between male and female eyes are deeply rooted in biology and are not simply a side effect of men and women having different rates of high blood pressure or diabetes.

One of the most intriguing findings involved a condition called prostatic hyperplasia, an enlargement of the prostate gland that only affects men. The researchers looked at whether men with this condition had different eye features compared to men without it. They found that men with prostate enlargement had smaller distances from the center of their optic nerve to its edge and lower blood vessel density in certain areas. This suggests that even within the male population, specific health conditions can subtly alter the eye's landscape. However, for the vast majority of the other health conditions studied, such as high blood sugar or kidney issues, the differences between men and women remained stable. The study did not find that these common diseases erased or reversed the natural sex differences in the eye.

Despite the clarity of these findings, the researchers are careful to note what they cannot yet say. Because the study was a snapshot in time, it cannot prove that these differences cause specific health outcomes or that they are entirely independent of the size of the eyeball itself. Men, on average, tend to have longer eyeballs than women, and this physical size difference likely influences the shape of the optic nerve and the density of the retinal tissue. Without measuring the exact length of every participant's eye, the study describes the differences as they appear in this specific group, rather than as a universal rule of biology. Nevertheless, the work provides a vital reference map. It shows that when artificial intelligence is used to screen for eye diseases like glaucoma or to assess cardiovascular risk, it must be calibrated to understand what is normal for men and what is normal for women. Ignoring these natural variations could lead to misdiagnoses, where a healthy man's eye is flagged as abnormal simply because the computer expects it to look like a woman's. By establishing these baseline differences, the study helps ensure that the future of eye care is as precise and personalized as the technology that makes it possible.