By Allison Proffitt
April 3, 2026 | Research out of Tohoku University in Sendai, Japan, provides a model for easily assessing front of the eye health outside of hospitals. The research team envisions employing the portable AI-powered scanning slit-light device in remote eye-screening camps, elderly-care facilities, pharmacies, or even train stations to make ophthalmic care more accessible, so patients can be assessed any place and any time. The findings were reported in Scientific Reports (DOI: 10.1038/s41598-026-44392-w).
Diseases such as cataracts that affect the front of the eye (also called the anterior segment) are among the leading causes of visual impairment worldwide. Slit-lamp biomicroscopy is the standard for anterior-segment evaluation, but devices are bulky, immobile, and expensive. Anterior-segment optical coherence tomography (AS-OCT) is a quantitative and reproducible diagnostic option, but it is even more cost-prohibitive.
The Tohoku research team designed a handheld system that cost less than $500 to work as an alternative, with reliable results that show strong agreement with AS-OCT scans.
“To our knowledge, this is the first ultra-low-cost, edge-AI slit-lamp quantitatively validated against AS-OCT, establishing a scalable foundation for community-level anterior-segment screening and teleophthalmology,” they write in the paper.
The portable device was designed to remove as much operator dependence and geometric uncertainty as possible. A motorized slit-scanning mirror is controlled by an NVIDIA Jetson Nano developer kit, ensuring precise synchronization with image capture. Onboard AI performs real-time segmentation of corneal and iris reflections, pupil boundaries, and corneal surfaces.
“Instead of relying on single, manually acquired photographs with uncontrolled alignment and unknown scale, the device captures standardized scanning-slit videos under fixed geometry and performs fully automated frame selection, segmentation, and calibration,” the authors write. “A motorized scanning mirror with regulated illumination maintains a fixed optical path, while the acquisition software automatically selects cornea- and iris-centered frames, minimizing operator bias and ensuring consistent alignment.”
By capturing a single scanning-slit video, the system can provide both quantitative measurements and qualitative or AI-assisted evaluation of anterior-segment abnormalities. A key feature of the platform is its lightweight, integrated AI model (LWBNA-unet), which segments important anatomical structures of the eye and supports further screening-oriented disease classification. Because the model is lightweight, accurate analysis can be performed directly on the device itself, without relying on cloud computing. This helps reduce operator dependence while improving portability, privacy, and real-world usability.
The researchers found that device was sufficient for screening-oriented assessment, while also being able to directly visualize clinically important features such as the cornea, iris, lens, ocular surface, pigment variations, and capsular changes—features that are often difficult to appreciate with grayscale OCT alone. The device performed reliably across diverse clinical presentations while maintaining excellent patient comfort, demonstrating its suitability for repeated use and community deployment. The device can also assess angle-closure glaucoma risk, which is of particular concern in Asia, and can lead to sudden, profound vision loss if a full angle-closure event occurs.
Future work will extend this platform beyond quantitative measurements of anterior chamber depth and central corneal thickness toward a fully AI-enabled anterior-segment screening system, the authors say.