Influence of neutral density filters on RNFL thickness as assessed with OCT

Glaucoma causes thinning of the retinal nerve fiber layer due to loss of
retinal ganglion cells. Currently, retinal nerve fiber layer (RNFL) thickness
measurements are used as a measurement of glaucomatous damage, both for
detecting glaucoma (screening) and for progression detection (monitoring), once
glaucoma has been diagnosed. Optical coherence tomography (OCT) has become the
most important instrument for the assessment of the RNFL thickness. In the
beginning, OCT was cheered because of its presumed objectivity (test result
independent of patient performance) and reproducibility. However, much like
other imaging techniques, OCT is subject to artifacts and variability.

Media opacities like cataract are a common cause of error in the measurement of
RNFL thickness, causing an underestimation of layer thickness. Cataract is
often present in patients with glaucoma, especially in the eyes of older
patients. It can act as a confounding factor and lead to an incorrect diagnosis
of glaucoma or give inaccurate results on glaucoma progression. Because the
RNFL thickness measurement with OCT is important for early detection of
glaucoma and detecting glaucoma progression, it is of interest to know what the
exact influence of media opacities is on RNFL thickness measurements and how
the analysis of OCT images can be improved in order to overcome the influence
of media opacities.

In this study, we assess the effect of media opacities by employing neutral
density (ND) filters (*sunglasses*) in OCT imaging in healthy subjects. These
ND filters cause the light to be attenuated and therefore degrades the signal
to noise ratio (SNR) in the same way as changed media opacity in cataract. The
degrading effect of ND filters on OCT layer thickness measurements has been
shown before (Darma 2015). We aim (1) to reproduce their findings and (2) to
analyse the data in a different way, in order to avoid this image quality bias.
For the reproduction (aim #1), we use the proprietary software of the OCT
machine and freely available segmentation software, being the Iowa Reference
Algorithm (Garvin et al. 2009), which was also used by Darma (2015), and the
DOCTRAP software (Chiu et al 2010). For the alternative analysis (aim #2), we
follow the theoretical framework as recently developed in our department
(Jansonius 2016). Two different OCT machines will be used: the machine
(*Topcon*) used by Darma (2015) and a more recent machine (*Canon*), the
current state of the art.

To determine the influence of signal-to-noise ratio (SNR) and layer thickness
assessment software on the RNFL thickness as assessed with OCT.

Cross-sectional study

Healthy subjects will have one visit to the Laboratory of Experimental
Ophthalmology to perform OCT measurement. Subjects will undergo screening to
confirm their eye health, which will comprise a questionnaire (see Appendix
F1), visual acuity (letter chart) testing, a screening visual field test, and
an intraocular pressure (IOP) measurement. Screening will take around 15
minutes. The eye will not be touched during the screening.
For the OCT measurement, the pupil is dilated with a mydriatic drug, 1%
tropicamide. After 20 minutes of waiting the OCT measurements will be taken,
with and without a ND filter in front of the eye. An OCT measurement takes
approximately six minutes (this includes three images per condition of which
the median value of each analysis will be used for the final comparison and
includes also time to change filters, saving data, explanation, and rest: an
actual OCT scan is like taking a picture: seconds rather than minutes); with
and without ND filters (optical density 0, 0.3, 0.6, and 0.9) and two different
devices (Topcon and Canon) it will thus take 48 minutes.

If abnormal screening results are obtained for healthy subjects, they will be
referred to their GP. Detection of signs of an eye condition may cause
psychological stress. However, an early diagnosis will allow treatments to be
initiated earlier and therefore more preservation of visual functioning. For
the recruitment of healthy subjects, poster adverts (see Appendix E3) will be
placed in and around the UMCG. Healthy subjects will spend 1.5 hour (15 min
screening, 20 min waiting (pupil dilation), and 48 min OCT scanning) in our
lab to complete the required measurements.