Signal amplitudes obtained from retinal scanning depend on numerous factors. Working with polarized light to interrogate the retina, large parts of which are birefringent, is even more prone to artifacts. This article demonstrates the necessity of using normalization when working with retinal birefringence scanning signals in polarization-sensitive ophthalmic instruments. After discussing the pros and cons of employing a normalization signal obtained by means of added optoelectronic hardware, the study shifts over and focuses on a numerical normalization method based on merely the s- and p-polarization components without additional optical or electronic hardware. This minimizes the adverse effects of optical asymmetries, the presence of certain instrumental noise, device-to-device variability, pupil diameter, retinal reflectivity, subject-to-subject variations, the position of the eye in the exit pupil of the device, and even signal degradation by cataracts. Results were experimentally and numerically tested on human data from 15 test subjects and clearly demonstrated the signal standardization achieved by numerical normalization. This is expected to lead to substantial improvement in algorithms and decision-making software, especially in ophthalmic screening instruments for pediatric applications, without added hardware cost. The proposed normalization method is also applicable to other polarization-sensitive optical instruments.
Read full abstract