Simultaneous decomposition of multiple hyperspectral data sets: signal recovery of unknown fluorophores in the retinal pigment epithelium

Abstract

Abstract:Upon excitation with different wavelengths of light, biological tissues emit distinct but related autofluorescence signals. We used nonnegative matrix factorization (NMF) to simultaneously decompose coregistered hyperspectral emission data from human retinal pigment epithelium/Bruch's membrane specimens illuminated with 436 and 480 nm light. NMF analysis was initialized with Gaussian mixture model fits and constrained to provide identical abundance images for the two excitation wavelengths. Spectra recovered this way were smoother than those obtained separately; fluorophore abundances more clearly localized within tissue compartments. These studies provide evidence that leveraging multiple coregistered hyperspectral emission data sets is preferential for identifying biologically relevant fluorophore information.

Hyperspectral Autofluorescence Imaging of Drusen and Retinal Pigment Epithelium in Donor Eyes With Age-Related Macular Degeneration

“…45 In 9/12 AMD tissues with drusen in this study, S1 at λ ex 436 nm peaked similarly and strongly at 530±10 nm (Figs 3A and 3B, solid green curves). The second abundant constituent of normal RPE, S2, was also represented in AMD tissues, peaking at 570 nm with either excitation, a slightly shorter wavelength than in normal tissues (575 nm).…”

“…45 In brief, NTF is an unsupervised machine-learning technique. NTF for each tissue was initialized with 4 Gaussian-shaped spectra from RPE and one from BrM for each excitation hypercube to search for 5 abundant spectra from each hypercube.…”

“…Localizations were displayed as abundance images in false colors. Because the 3 fundamental novel RPE spectra previously identified in normal tissues 45, 46 were also identified in AMD tissues, we retained the nomenclature and visualization formats from these prior reports, for ease of comparison: S1, green; S2, blue; S3, red; and S0 = BrM, gray. The new spectrum of drusen, SDr, was assigned the color azure.…”

“…44 In AF mode, we have used it to recover the AF spectra of RPE and BrM in normal donor eyes. 45, 46 It is thus potentially one of the most precise methods to analyze retinal disease involving RPE. 47 …”

The Project MACULA Retinal Pigment Epithelium Grading System for Histology and Optical Coherence Tomography in Age-Related Macular Degeneration

“…13 Clinically deployed and experimental technologies show the RPE en face [14][15][16][17][18][19] and in cross-section with other chorioretinal layers. 20,21 A research priority is identifying novel anatomic biomarkers derived from spectral domain optical coherence tomography (SDOCT), 22 including components of the hyperreflective band attributed to RPE and BrM.…”