Abstract
Purpose: :
The confocal scanning laser ophthalmoscope (cSLO), like any imaging instrument, gives a representation of the retina based on the image formation process of the instrument; however there is no direct way of verifying how faithful these images are to the actual retinal morphology. The retinal health assessment made either by automated algorithms or by clinicians is therefore dependent on the accuracy of these images. A virtual computational platform is presented that simulates the image formation process of the cSLO, thus producing simulated cSLO images from known numerical 3-dimensional (3-D) structures. This platform allows the detailed study of the limitations of the cSLO and the artifacts it introduces under different optical configurations, and enables a quantifiable assessment of the various algorithms used on cSLO images to extract morphological parameters of the retina and assess retinal health.
Methods: :
The virtual platform is designed in a modular fashion so that different aspects of the image formation process within the cSLO can be included. The basic module simulates a cSLO image from first principles, using Fourier optics techniques and it enables scanning through a 3-D numerical structure defined by the user. Pupil size and geometry, ocular aberrations and wavelength are all variable user inputs that are taken into account. A small library of objects (3-D structures) is available for simulation. The second module currently implemented allows the user to define the size of the confocal pinhole to be used in the simulation, which affects image resolution, contrast and noise.
Results: :
The simulation platform outputs simulated slices of the 3-D structure at different axial depths, and generates axial cross-sections at different lateral positions of the structure. The axial and lateral resolution limit is quantified from these images for different aberration maps, pupil geometries and sizes, and confocal pinhole sizes.
Conclusions: :
A virtual platform has been presented that simulates the image formation process within the cSLO, and simulated images for a number of different optical configurations are presented. This platform serves as a powerful tool in understanding the limitations of the cSLO and the artifacts it generates, and provides a methods for verifying algorithms that are used on real cSLO images obtained in clinical practice.
Keywords: imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) • imaging/image analysis: non-clinical • retina