Biomedical engineers from Duke University have introduced a new AI-powered imaging system for retinal cells. The development enables obtaining high-precision images 100 times faster compared to traditional methods. The research was published in the journal Science Advances.
The retina is a thin layer of cells at the back of the eye responsible for transmitting visual information to the brain. It is considered a convenient “platform” for studying neurons, including in diseases such as Alzheimer’s or multiple sclerosis. The more accurate the image, the earlier pathological changes can be detected and treatment can begin.
The standard method for observing the retina—AOSLO technology—requires complex and expensive equipment. To obtain more precise data, multiple sensors are typically used, or a single sensor is repositioned after each scan, significantly increasing the procedure time and reducing patient comfort.
Scientists from Duke proposed an alternative—the Deep-Compressed AOSLO (DCAOSLO) system. It utilizes compressed sensing, allowing data to be collected from fewer projections and reconstructing the image using AI algorithms. Instead of traditional point-by-point scanning, the system employs micro-mirrors that direct reflected light onto sensors, simulating the work of 12 devices with only two.
Tests have shown that the new method effectively visualizes photoreceptor and vascular cells in both healthy and diseased patients.
