Purpose
It has been noted that clinicians have difficulty using structural status to predict functional damage demonstrated in visual fields. Spectral-domain ocular coherence tomography (SD-OCT) has made it possible for researchers to create en face imaging of the retinal nerve fiber layer at different distances from the inner limiting membrane (ILM). With this development, we hypothesize that experienced clinicians can begin to use en face imaging of the RNFL to predict functional loss.
Methods
Data was collected from subjects participating in an ongoing prospective study gathering SD-OCT (Heidelbery Spectralis) scans and comparing perimetric sensitivities on a custom testing station in a range of stimuli. Subjects that had completed 3 visits, 1 month apart and imaged with SD-OCT, were used for this analysis. The area of testing was restricted to 18 of the 54 24-2 locations above and below the nasal portion of the horizontal midline. Four examiners were trained by viewing en face imaging of the RNFL in 3 healthy eyes with the above mentioned visual field (VF) points overlaid on the images. Afterwards the examiners were asked to review the en face image sequences with VF points for 8 subjects with glaucoma and VF loss in the area of testing (see image 1). They were masked to the VF loss and were asked to predict the field results with depth of defect for each point using a grading scale of no loss, mild, moderate, and severe. A Fleiss Kappa analysis was used to compare inter-observer scoring and agreement. The observers’ gradings were also compared to values given by visual field testing.
Results
Comparison between observers showed better than chance agreement (kappa = 0.18 p < 0.0001). The areas graded as having no loss or severe loss demonstrated the best agreement (k=0.32 and 0.15) and agreement was worst in areas of minimal loss (k=0.01). Mean (±SD) depth of visual field defect was -6.2 ± 2.1dB for locations where all examiners predicted a defect (mild - severe), and -1.0±1.6dB for test locations where all examiners predicted no loss (t=9.8 p=<0.0001).
Conclusions
En face RNFL image sequences appear to be a potential new tool for clinicians to translate structure to function in management of glaucoma. Further training for the examiners and refinement of this new potential tool appear to be warranted.