Abstract
Purpose :
Routine eye exams via slit lamp are critical in screening for conditions like cataracts and corneal injury. The hardware complexity of slit lamps limit their use to ophthalmic clinics by trained professionals, rendering them impractical for resource-constrained settings. We present purely solid-state instrumentation that employs programmable illumination and light steering optics to simulate the motion of a slit, thereby exhibiting functionality similar to a slit lamp with no moving parts. Complete computational control over illumination enables us to generate 3D models and extract meaningful quantitative metrics to aid in disease diagnosis by untrained staff.
Methods :
The prototype comprises a laser projector, an ellipsoidal reflector and a system of cameras to image the eye. The projector and the eye are placed at the foci of the ellipsoid so all the light emitted by the projector converges onto the eye. The ellipsoidal reflector serves two purposes: 1. steering the diverging beam projected into converging slits of light. 2. multiplying the angular spread of the beam projected to achieve a larger sweep angle. Software corrects distortions produced to ensure projection of parallel slits illuminating one cross-section at a time. The images captured by the cameras are then stitched together to generate a 3D model.
Results :
Our prototype exhibits functionality similar to a slit lamp with lower hardware complexity, enabling it to be compact. Based on the eccentricity of the ellipsoid, we are able to multiply the angular spread of the input beam up to a factor of four. The entire data capture process is automated to less than 2 seconds. Complete software control over illumination allows us to precisely determine the orientation of each cross-section. This enables us to reconstruct a complete 3D model and extract quantitative metrics like corneal topography and anterior chamber depth.
Conclusions :
We have proposed a design for examining the anterior segment of the eye in resource-constrained settings. The use of solid state instrumentation instead of mechanically moving parts allows the device to be portable. Computational control over data capture reduces the capture process to under 2 seconds, making it ideal for rapid screening in remote locations. Generation of 3D models and extraction of quantitative metrics makes the captured data interpretable by lesser-skilled staff.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.