Fast ray-tracing of human eye optics on Graphics Processing Units

Area of Science:

• Ophthalmology
• Computational Optics
• Vision Science

Background:

• Accurate simulation of human optics is crucial for understanding vision and correcting defects.
• Previous analytical models struggle with abnormal ocular structures, limiting their application in vision disorders.
• There is a need for advanced simulation techniques capable of handling complex, arbitrary ocular geometries.

Purpose of the Study:

• To develop a new, efficient technique for simulating retinal image formation.
• To enable accurate modeling of both normal and abnormal ocular structures.
• To facilitate individualized vision correction by integrating patient-specific eye geometry.

Main Methods:

• Developed a computer simulator using ray tracing of millions of rays.
• Employed parallel processing on Graphics Processing Units (GPUs) for efficient computation.
• Incorporated arbitrary ocular structure shapes (e.g., polygon meshes) and patient-specific data from medical imaging.

Main Results:

• Achieved stable retinal image generation within minutes.
• Successfully simulated optical phenomena including depth-of-field, accommodation, chromatic aberrations, and astigmatism.
• Demonstrated application in patient-specific vision correction using reconstructed orbital models.

Conclusions:

• The developed GPU-accelerated ray-tracing technique provides an efficient and accurate method for retinal image simulation.
• This approach overcomes limitations of previous models in handling complex ocular geometries, advancing vision science research.
• The technique holds significant potential for personalized diagnosis and treatment planning in ophthalmology.