Access to anterior-segment imaging remains a major bottleneck in global eye care. While slit-lamp biomicroscopy is ubiquitous, it is inherently qualitative and operator dependent, and anterior segment OCT (AS-OCT) remains prohibitively expensive for widespread screening. A new Scientific Reports study describes a potential bridge between these extremes: an ultra–low-cost, AI-powered portable scanning slit-light device capable of delivering quantitative anterior-segment measurements in a handheld format.
Developed by researchers at Tohoku University, the system combines a motorized slit-scanning mechanism with on-device deep learning to extract clinically relevant parameters from short video sequences. The device – costing under $500 – captures a 15-second scanning-slit video, which is then processed using a lightweight U-Net–based model to segment corneal and iris reflections, pupil boundaries, and corneal surfaces.
Unlike conventional slit lamps, the system provides calibrated quantitative outputs. Chief among these is anterior chamber depth (ACD), derived from segmented anatomical landmarks with geometric corrections for slit angle and magnification. In a clinical study involving 170 participants, ACD measurements showed strong agreement with AS-OCT, with mean differences close to zero and most measurements within ±0.3 mm. These results suggest that, at least for screening purposes, the device may approach interchangeability with gold-standard imaging.
Performance for central corneal thickness (CCT) was more limited, with wider variability attributed largely to spatial resolution constraints – approximately 40 µm per pixel – highlighting the current prototype’s limitations for pachymetry.
Beyond biometry, the system captures full-color anterior-segment images, enabling simultaneous qualitative assessment. The device can visualize cataract, narrow angles, corneal opacity, and keratoconus from a single scan, with AI segmentation accurately delineating relevant structures. This multimodal capability – combining geometric measurement with color-based tissue evaluation – distinguishes it from OCT and Scheimpflug systems, which rely on monochromatic imaging.
The workflow is designed for portability and autonomy. The device operates without a chin rest, uses low-intensity illumination for improved patient comfort, and processes data entirely on-device using edge AI hardware, eliminating the need for cloud connectivity. A full acquisition and analysis cycle takes approximately 30–35 seconds per eye.
For ophthalmologists, the implications are significant. By standardizing acquisition and automating analysis, the system reduces operator dependence – a key limitation of both traditional slit lamps and smartphone-based adaptations. At the same time, its low cost and portability position it as a potential tool for community screening, teleophthalmology, and deployment in resource-limited settings, where access to OCT remains scarce.
However, the study authors emphasize that the device is intended as a screening tool rather than a replacement for high-resolution imaging. Nevertheless, by combining affordability, quantitative capability, and AI-driven analysis, this platform represents a meaningful step toward scalable anterior-segment diagnostics – particularly in regions where the burden of preventable blindness remains highest.
