Retinal Nerve Fiber Layer Attenuation Maps Derived From Volumetric OCT Data

Abstract

 

Purpose:

 

To introduce and to evaluate the use of retinal pigment epithelial (RPE) normalized retinal nerve fiber layer (RNFL) attenuation coefficient maps, derived from optical coherence tomography (OCT) data.

 

Methods:

 

Previously, the average OCT-signal of the RNFL was shown to provide complementary data to thickness maps. However, quantitative interpretation of this data is cumbersome, because it depends on the fluctuating strength of the incident beam and on the thickness of the layers (see figure, top row). We therefore derived a model to locally calculate the RNFL attenuation coefficient (μ_att). The model incorporates reflectivity data from the RNFL, uses the RPE as a reference layer and takes into account the thickness of the RNFL to adjust for attenuation of the incoming light beam in the RNFL. By using an automated segmentation algorithm, the resulting μ_att was calculated for every A-line and maps were produced. 10 normal and 8 glaucomatous eyes were imaged on a Spectralis OCT device (Heidelberg Engineering, Germany). Each peri-papillary volumetric scan with a field-of-view of 20° consisted of 193 B-scans (512 A-lines per B-scan, 5 times averaging).In addition to visual inspection of the resulting maps, the data was averaged to produce a single mean RNFL attenuation coefficient per eye.

 

Results:

 

All volumetric data sets were automatically segmented and RNFL attenuation coefficients were used. Examples of the resulting μ_att-maps are shown in the figure (middle row; normal (left) and glaucomatous (right)). The average μ_att was calculated for every eye and their distributions were plotted (see figure, bottom row). The difference between the average μ_att for healthy and glaucomatous eyes was highly significant (Mann-Whitney test, P<0.01).

 

Conclusions:

 

We introduced the attenuation coefficient as a way to describe an optical property of the RNFL. Graphical, spatial maps were produced, showing relatively homogeneous coefficients for normal eyes and clearly pathological patterns in glaucomatous eyes. The difference between normal and glaucomatous eyes was highly significant, suggesting its use as a diagnostical tool. We have shown that, in addition to morphological data, OCT data contains more information that is clinically useful, such as attenuation coefficient maps.  

 

Keywords: image processing • imaging/image analysis: clinical • nerve fiber layer