Abstract
PURPOSE: To allow exploration of xanthophyll carotenoids in vision and age-related macular degeneration progression using two-wavelength autofluorescence imaging for macular pigment optical density (MPOD), we developed tools for automatically centering and classifying the MPOD distribution pattern. METHODS: A subset of the ALSTAR2 baseline cohort (NCT04112667) and 44 eyes of adults aged 20 to 30 years with healthy maculas were imaged with optical coherence tomography and two-wavelength autofluorescence (MPOD module, Heidelberg Engineering). Images underwent a quality review. Two custom FIJI plugins centered the MPOD distribution by five algorithms (FOVEA, HILLCLIMB, CENTROID, MAX, CONTOUR). Others automatically classified spatial distributions into four patterns from Obana et al: Peak, Ring, Mixed, and Dip. RESULTS: Of 651 qualifying aged eyes and 44 young eyes, the HILLCLIMB and CONTOUR methods best agreed with a manually determined foveal center. Regarding spatial distribution pattern, 445 aged eyes (68.4%) showed peaks, 118 (18.1%) rings, 41 (6.3%) mixed, and 47 (7.2%) dips. In young eyes, 40 (90%) showed peaks, 1 (2.3%) rings, 3 (6.8%) mixed, and none showed dips. Notably, peaks were significantly (P < 0.001) more prominent in men (74.1%) than women (65.0%) and pseudophakic (72.7%) than phakic (62.9%) eyes. CONCLUSIONS: Automatic tools for MPOD centration are reliable and robust. Future studies will use the HILLCLIMB and CONTOUR algorithms. TRANSLATIONAL RELEVANCE: Automated MPOD pattern assignment suggests that the spatial distribution of MPOD varies with gender, lens status, and possibly age. Our analytic software can be applied to large samples for studies of xanthophyll carotenoid impact on vision and age-related macular degeneration progression.