Abstract
Purpose :
The segmentation of macular optical coherence tomography (OCT) scans is a valuable tool for the understanding of various retinal pathologies. There are, however, several artifacts potentially involved in the segmentation process, among them the typically hyperreflective Henle Fiber Layer (HFL), appearing as a part of the outer nuclear layer (ONL) under specific imaging conditions. The appearance of the HFL is dependent on the imaging beam entrance at the pupil and the resultant oblique scanning of the macula. As the HFL can overlap a significant part of the ONL and may thus also alter outer plexiform (OPL) measurements, there is a need to find a marker for this artifact on automated segmentation.
Methods :
Directional OCT (D-OCT) technique was used that involves purposefully altering the entry position of the OCT beam. Imaging was carried out by Spectralis SD-OCT (Heidelberg Engineering, Heidelberg, Germany), using the “Dense” macular protocol. Seven eyes of five healthy subjects were involved in the study (1 male and 4 females, age 21-55 years). Three scans were taken of each eye with a central, temporal and nasal displacement of the imaging beam. The built-in segmentation algorithm of the Spectralis SD-OCT device was used and segmentation maps of the outer plexiform layer (OPL) were exported. The ratios of the superior vs. inferior and temporal vs. nasal ETDRS areas were calculated and compared by paired t-test.
Results :
Imaging artifacts were observed as variable areas of increased thickness, depending on the scanning beam displacement and appearing on the contralateral side relative to the beam (i.e. temporal beam displacement resulting in a nasal artifact). Significant differences were observed with the temporal beam displacement in the comparisons of the inner-temporal vs. inner-nasal and the outer-temporal vs. outer-nasal region ratios (inner-temporal/inner-nasal: 1.01±0.11 vs 0.59±0.19 and outer-temporal/outer-nasal: 0.99±0.11 vs 0.77±0.07, for the middle vs. the temporal beam placement, respectively, p<0.05 for both comparisons).
Conclusions :
Our findings may facilitate the automatic detection of HFL artifacts during the automated segmentation of macular OCT volumes, while, at the same time, they may also help to understand the in vivo morphology of the photoreceptor and Müller cell axons.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.