Premium
Evaluation of a global algorithm for wavefront reconstruction for S hack– H artmann wave‐front sensors and thick fundus reflectors
Author(s) -
Liu Tao,
Thibos Larry,
Marin Gildas,
Hernandez Martha
Publication year - 2014
Publication title -
ophthalmic and physiological optics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.147
H-Index - 66
eISSN - 1475-1313
pISSN - 0275-5408
DOI - 10.1111/opo.12097
Subject(s) - wavefront , zernike polynomials , fundus (uterus) , centroid , wavefront sensor , optics , adaptive optics , algorithm , computer science , reflection (computer programming) , mathematics , physics , artificial intelligence , medicine , ophthalmology , programming language
Purpose Conventional aberration analysis by a S hack– H artmann aberrometer is based on the implicit assumption that an injected probe beam reflects from a single fundus layer. In fact, the biological fundus is a thick reflector and therefore conventional analysis may produce errors of unknown magnitude. We developed a novel computational method to investigate this potential failure of conventional analysis. Methods The S hack– H artmann wavefront sensor was simulated by computer software and used to recover by two methods the known wavefront aberrations expected from a population of normally‐aberrated human eyes and bi‐layer fundus reflection. The conventional method determines the centroid of each spot in the SH data image, from which wavefront slopes are computed for least‐squares fitting with derivatives of Z ernike polynomials. The novel ‘global’ method iteratively adjusted the aberration coefficients derived from conventional centroid analysis until the SH image, when treated as a unitary picture, optimally matched the original data image. Results Both methods recovered higher order aberrations accurately and precisely, but only the global algorithm correctly recovered the defocus coefficients associated with each layer of fundus reflection. The global algorithm accurately recovered Z ernike coefficients for mean defocus and bi‐layer separation with maximum error <0.1%. The global algorithm was robust for bi‐layer separation up to 2 dioptres for a typical SH wavefront sensor design. For 100 randomly generated test wavefronts with 0.7 D axial separation, the retrieved mean axial separation was 0.70 D with standard deviations ( S.D. ) of 0.002 D. Conclusions Sufficient information is contained in SH data images to measure the dioptric thickness of dual‐layer fundus reflection. The global algorithm is superior since it successfully recovered the focus value associated with both fundus layers even when their separation was too small to produce clearly separated spots, while the conventional analysis misrepresents the defocus component of the wavefront aberration as the mean defocus for the two reflectors. Our novel global algorithm is a promising method for SH data image analysis in clinical and visual optics research for human and animal eyes.