Alzheimer’s disease can begin changing the brain years before memory problems become obvious. That long silent period is one of the reasons researchers are looking for better ways to detect risk earlier, before a person or family first notices cognitive symptoms.
A recent Mayo Clinic Press article and Tomorrow’s Cure podcast episode explored one promising research direction: using retinal imaging and artificial intelligence to study patterns that may be associated with Alzheimer’s disease and related brain changes. The conversation featured Dr. Oana Dumitrascu of Mayo Clinic and Dr. Yalin Wang of Arizona State University, whose work sits at the intersection of neurology, eye imaging, computer science, and population health.
The idea is compelling because the retina is part of the central nervous system and can be viewed directly and non-invasively. Eye images are already familiar in many clinical settings. If researchers can validate reliable retinal markers of brain health, routine eye imaging could one day help support risk assessment, clinical trial enrollment, and prevention-focused research.
That future is not here yet. Retinal imaging with AI is not currently a routine diagnostic test for Alzheimer’s disease. The work described by Mayo Clinic Press is still a research frontier, and its value will depend on careful validation, diverse data, and appropriate follow-up.
The retina as a research window into the brain
The retina has long interested researchers because it offers a rare view into nervous system and vascular tissue without the need for invasive procedures. In clinical eye care, retinal imaging is already used to evaluate and monitor conditions such as diabetic retinopathy, glaucoma, and other retinal or vascular changes.
In the Mayo Clinic Press transcript, Dr. Dumitrascu explains that the eye is an extension of the brain and that the retina offers a way to observe vascular health and brain-related changes non-invasively. Her work has explored whether features in the retinal vasculature and deposits of proteins associated with Alzheimer’s disease, including amyloid and tau, may relate to brain health.
“My vision is that the patient can have, when they have their regular eye scans with their optometrist every year, that particular eye scan can be coupled with an AI model that is really fast and very well-validated and accurate.”
— Dr. Oana Dumitrascu, Mayo Clinic neurologist
That quote captures the long-term promise. It also points to the standard the field will need to meet. For retinal imaging to become useful in this context, AI models would need to be accurate, explainable, validated across diverse populations, and connected to appropriate medical interpretation.
Looking for signals before symptoms appear
The Mayo Clinic Press episode emphasizes that Alzheimer’s disease may unfold quietly for years before symptoms become noticeable. According to the transcript, nearly 7 million Americans are living with Alzheimer’s disease today, and that number is expected to rise as the population ages.
Dr. Dumitrascu notes that Alzheimer’s disease is often diagnosed too late, when symptoms are already apparent and some therapies may be less applicable. She also explains that Alzheimer’s disease can coexist with cerebrovascular disease, which may affect eligibility for some Alzheimer’s-targeting therapies.
That is why the presymptomatic stage is such an important research focus. At that point, a person may still have normal cognition and no difficulty with daily activities, while biological changes may already be present. If researchers can identify people earlier, they may be able to support prevention-focused clinical trials, monitor disease processes more effectively, and guide future care planning with more information.
Where AI enters the picture
A standard retinal image may contain subtle information that is difficult for the human eye to quantify. Vascular branching, vessel structure, spatial patterns, and small changes in the retina may be too complex or too time-consuming to assess manually at scale.
Dr. Wang describes AI as a way to analyze millions of tiny image features simultaneously. In the transcript, he notes that Alzheimer’s-related retinal changes may be much subtler than findings seen in diabetic retinopathy, which makes computational analysis especially important.
The research team is also studying how AI might improve the usability of retinal images collected in real-world settings. In rural or mobile clinic environments, images may be lower quality than those captured in tightly controlled academic settings. Dr. Wang describes work in Arizona where a portion of retinal images collected in the field were not usable, and AI tools were able to convert some of those images into usable data.
That point matters. A tool that works only under ideal conditions may not help the communities that could benefit most from lower-barrier screening and research access. Scalability depends on performance in the real world.
The promise and limits of routine eye imaging
One reason this research is attracting attention is that eye imaging is more familiar, less invasive, and often more accessible than specialized neurological testing. Many people already see an optometrist or ophthalmologist for routine care, and fundus photography is widely used in eye care settings.
The current diagnostic pathway for Alzheimer’s disease is very different. The Mayo Clinic Press transcript describes a process that often begins when a patient presents with cognitive concerns to a primary care provider or neurologist. Evaluation may include cognitive screening, detailed neuropsychological testing, laboratory testing, brain MRI, and sometimes specialized tests such as cerebrospinal fluid analysis or amyloid PET imaging.
Retinal imaging would not replace those tools. If validated, it may become an additional way to flag risk, support research, or help identify people who may need closer monitoring. That distinction should remain clear in patient-facing communication.
A routine eye scan supported by AI may one day contribute to Alzheimer’s risk assessment. It should not be presented as a standalone Alzheimer’s diagnosis.
Validation is the dividing line
The Mayo Clinic Press conversation returns several times to validation, and for good reason. AI can detect patterns, but researchers must prove that those patterns are clinically meaningful.
Dr. Dumitrascu explains that retinal imaging findings need to be validated against established Alzheimer’s biomarkers such as amyloid PET imaging, cerebrospinal fluid analysis, and emerging blood-based biomarkers. Dr. Wang also emphasizes the need for large, multicenter studies, diverse populations, and validation across different environments.
This is where careful science matters more than excitement. Researchers need to understand whether an AI model is detecting Alzheimer’s-specific changes, vascular disease, aging patterns, image artifacts, or a combination of factors.
The transcript also highlights the importance of explainable AI. A risk score is not enough if clinicians and researchers cannot understand what biological signal the model is using. Dr. Dumitrascu describes her interest in identifying the underlying biomarker that tells the AI model a disease process may be present.
For clinical trust, the system cannot simply be fast. It has to be interpretable, reliable, and clinically grounded.
A possible role in clinical trials
One of the most practical near-term applications may be in research rather than routine diagnosis. If retinal imaging and AI can help identify people at higher risk before symptoms appear, it may support clinical trial recruitment and monitoring.
That matters because prevention-focused Alzheimer’s trials need to identify participants earlier in the disease process. The transcript also discusses the need for monitoring tools that can follow people over time during trials. A non-invasive retinal imaging approach could be valuable if it proves accurate, repeatable, and connected to established biomarkers.
Dr. Dumitrascu also raises an important ethical point. Telling people they may have presymptomatic Alzheimer’s risk could create anxiety, especially while treatment and prevention options continue to evolve. In the transcript, she frames the most important application as supporting preventive clinical trials and monitoring response to therapies, rather than simply giving people distressing information without a clear next step.
That perspective is important. Earlier information only helps if it is paired with responsible counseling, medical context, and meaningful pathways for action.
What eye care professionals should say carefully
For optometrists, ophthalmologists, and other eye care professionals, this research reinforces the broader connection between eye health and systemic health. The retina can reveal important information about ocular and vascular health, and ongoing research is exploring whether it may also help researchers understand neurodegenerative disease.
Still, the language around this topic needs care. Patients may see headlines about eye scans and Alzheimer’s disease and assume that a routine eye exam can already detect Alzheimer’s. That is not what the Mayo Clinic Press transcript establishes.
A better message is more precise: researchers are studying whether retinal imaging and AI can identify patterns associated with Alzheimer’s disease and brain health. The approach remains investigational and requires further validation before it can guide routine clinical decision-making.
That balanced framing protects patients from overinterpretation while still recognizing the promise of the research.
A research frontier, not a replacement for diagnosis
The future described in the Mayo Clinic Press article is hopeful, but it remains under development. Retinal imaging could eventually become part of a broader toolset that includes cognitive testing, medical history, blood-based biomarkers, brain imaging, specialist evaluation, and longitudinal monitoring.
The most useful version of this technology is not one where AI replaces clinicians. It is one where AI helps clinicians and researchers analyze complex images quickly and consistently, while trained medical professionals interpret those results in context.
That context matters because Alzheimer’s disease is complex. Many patients have mixed pathology, including both neurodegeneration and vascular disease. The transcript notes that differentiating Alzheimer’s disease from mixed dementias and cerebrovascular conditions is one of the ongoing challenges in the field.
This is another reason retinal imaging research should be viewed as one piece of a larger clinical and scientific picture.
Looking through the eye toward brain health
The idea that the eye can offer clues about the brain is not new, but retinal imaging and AI are giving researchers new ways to explore it. The Mayo Clinic Press episode shows how eye care, neurology, computer science, and population health can come together around a shared goal: detecting risk earlier and supporting better research.
For now, the responsible takeaway is careful optimism. Eye scans may one day help researchers and clinicians identify Alzheimer’s-related risk earlier, especially when paired with well-validated AI and appropriate follow-up. But the field still needs larger studies, diverse data, explainable models, and comparison against established biomarkers.
The eye may become an important window into brain health. The work now is making sure that window is clear, accurate, equitable, and clinically useful.
