Diabetes, even when managed properly, comes with its share of challenges, such as vision problems and possibly dementia. Yes, you read correctly, dementia. Of course, research is looking into ways to detect both via the eyes.
It has been shown that people with diabetes are more likely to develop dementia. Scientists at the Joslin Diabetes Center in Boston have found that routine imaging can show changes in the retina may be associated with cognitive issues in older persons with type 1 diabetes.
Earlier research revealed a link between proliferative diabetic retinopathy and cognitive impairment in those with type 1 diabetes. Scientists at Joslin wanted to see whether imaging techniques that show changes in the retina would be reflective of changes in cognitive functions. Scientists looked at eye scans from patients who received care at Joslin’s Beetham Eye Institute. One set was based on optical coherence tomography (OCT). The other set was based on OCT angiography, which examines blood vessels in the retina.
This study also had persons who were involved in the Joslin Medalist Study, which looks at the outcomes of those who have had type 1 diabetes for 50 years or more. They did an array of cognitive tests that probed memory function and assessed the time it took to arrange objects, known as psychomotor speed. Researchers found strong associations between performance on memory tasks and changes in the structure of the blood vessel networks in the retina. They also found a links between proliferative diabetic retinopathy and psychomotor speed and that proliferative diabetic retinopathy was linked to memory performance within the larger group of Medalist participants.
Most of the time people are tested for Alzheimer’s when they are showing signs of decline and treatments at that stage aren’t very helpful. In addition, the tests are done via MRI scans, which are difficult and expensive. In comparison, routine eye exams are less expensive and less cumbersome than MRI scans and they detect cognitive changes at an earlier stage.
There are plans to do a larger study to substantiate the potential of eye imaging to show signs of cognitive decline and scientists want to include younger people with type 1 diabetes, as well as look at MRI brain images and study postmortem brain samples donated by Medalists.
As great as the above-mentioned study from Joslin Diabetes Center was, not all diabetes tests provide reliable information. Case in point, artificial intelligence has shown promise in many things. Unfortunately, diagnosing diabetic retinopathy isn’t one of them. Researchers at the University of Washington wanted to learn about the effectiveness several artificial intelligence-based screening algorithms to diagnose diabetic retinopathy. Since there is a shortage of vision care providers, a screening-algorithm would help with diagnosis of diabetic retinopathy, which is one of the most common diabetic eye diseases leading to vision loss, therefore freeing up resources for treatment.
Researchers used algorithm-based technologies on retinal images from almost 24,000 veterans who had diabetic retinopathy screenings from 2006 to 2018 at Veterans Administration healthcare facilities in Seattle and Atlanta. They compared them against the expertise of retina specialists.
What was discovered was that the algorithms didn’t perform as well as the manufactures claimed that they did. Researchers tested the performance of the algorithms and the performance of human screeners and they were compared to the diagnoses ophthalmologists gave when looking at the same images. Three out of the five algorithms studied performed reasonably well when compared to the ophthalmologists’ diagnoses. One did worse and one performed as well as human screeners.
Differences in camera equipment and technique could be the culprit. Scientists feel that this study shows how it important it is for eye care practices who want to use an artificial intelligence screener to test it first and follow guidelines on how to get patient eye images. Also, they found that the algorithms’ performance varied when studying images from patients in the Seattle and Atlanta populations. Researchers think that this indicates that the algorithms need a wider variety of images.
Having diabetes is no picnic. Neither is studying it. While doing research on the disease and its associated conditions doesn’t lead to desired outcomes, it increases knowledge about diabetic conditions. That knowledge ends up informing future research.