Measuring eye aberrations with Hartmann-Shack wave-front sensors: should the irradiance distribution across the eye pupil be taken into account?

A usual approximation in Hartmann-Shack aberrometry is that the centroid displacements are proportional to the spatial averages of the wave-front slopes at the sampling subapertures. However, these spatial averages are actually weighted by the local irradiance distribution across each microlens. The irradiance across the eye pupil is not uniform in usual reflectometric aberrometers, which is due to several factors including retinal scattering and cone waveguiding directionality. It is shown that neglecting this fact in usual least-squares reconstruction procedures gives rise to a biased estimation of the aberration coefficients. The magnitude of this bias depends on the actual irradiance distribution across the eye pupil, the mode being estimated, the detailed modal composition of the aberrated wave front, and the geometry of the wave-front sampling array. Order-of-magnitude calculations suggest that this bias may well be in the range 5%-10% for relatively smooth irradiance distributions. The systematic nature of this error makes it advisable to check for its presence and, if required, to compensate for it by an adequate choice of the least-squares reconstruction matrix.