A Digital Staining Algorithm for Optical Coherence Tomography Images of the Optic Nerve Head

PURPOSE: To digitally stain spectral-domain optical coherence tomography (OCT) images of the optic nerve head (ONH), and highlight either connective or neural tissues. METHODS: OCT volumes of the ONH were acquired from one eye of 10 healthy subjects. We processed all volumes with adaptive compensation to remove shadows and enhance deep tissue visibility. For each ONH, we identified the four most dissimilar pixel-intensity histograms, each of which was assumed to represent a tissue group. These four histograms formed a vector basis on which we ‘projected' each OCT volume in order to generate four digitally stained volumes P1 to P4. Digital staining was also verified using a digital phantom, and compared with k-means clustering for three and four clusters. RESULTS: Digital staining was able to isolate three regions of interest from the proposed phantom. For the ONH, the digitally stained images P1 highlighted mostly connective tissues, as demonstrated through an excellent contrast increase across the anterior lamina cribrosa boundary (3.6 ± 0.6 times). P2 highlighted the nerve fiber layer and the prelamina, P3 the remaining layers of the retina, and P4 the image background. Further, digital staining was able to separate ONH tissue layers that were not well separated by k-means clustering. CONCLUSION: We have described an algorithm that can digitally stain connective and neural tissues in OCT images of the ONH. TRANSLATIONAL RELEVANCE: Because connective and neural tissues are considerably altered in glaucoma, digital staining of the ONH tissues may be of interest in the clinical management of this pathology.