Custom extraction of macular ganglion cell-inner plexiform layer thickness more precisely co-localizes structural measurements with visual fields test grids

www.nature.com/scientificreports OPEN Custom extraction of macular ganglion cell‑inner plexiform layer thickness more precisely co‑localizes structural measurements with visual fields test grids Janelle Tong1,2, David Alonso‑Caneiro3, Nayuta Yoshioka1,2, Michael Kalloniatis1,2 & Barbara Zangerl1,2* We aimed to evaluate methods of extracting optical coherence tomography (OCT)-derived macular ganglion cell-inner plexiform layer (GCIPL) thickness measurements over retinal locations corresponding to standard visual field (VF) test grids. A custom algorithm was developed to automatically extract GCIPL thickness measurements from locations corresponding to Humphrey Field Analyser 10-2 and 30-2 test grids over Goldmann II, III and V stimulus sizes from a healthy cohort of 478 participants. Differences between GCIPL thickness measurements based on VF test grids (VF-based paradigms) and the 8 × 8 grid, as per instrument review software, were analyzed, as were impacts of fovea to optic disc tilt and areas over which GCIPL thickness measurements were extracted. Significant differences between the VF-based paradigms and the 8 × 8 grid were observed at up to 55% of locations across the macula, with the greatest deviations at the fovea (median 25.5 μm, 95% CI 25.24–25.72 μm, P < .0001). While significant correlations with fovea to optic disc tilt were noted at up to 33% of locations distributed 6°–8° from the foveal center, there were no marked differences in GCIPL thickness measurements between VF-based paradigms using different stimulus sizes. As such, standard high-density OCT measurement paradigms do not adequately reflect GCIPL measurements at retinal locations tested with standard VF patterns, with the central macular region contributing most to the observed differences and with further correction required for fovea to optic disc tilt. Spatial direction of GCIPL thickness measurements will improve future comparisons of structure and function, thereby improving methods designed to detect pathology affecting the inner retina. Glaucoma is a progressive optic neuropathy exhibiting selective loss of retinal ganglion cells (GCs), and is a major contributor to significant visual morbidity in developed nations1, 2. A hallmark characteristic of glaucoma is concordance between structural and functional damage; that is, defects of the optic nerve head, retinal nerve fiber layer (RNFL) and inner retina mirror associated deficits in visual field (VF) sensitivity3, 4. In contrast, structural evidence of glaucoma in the absence of notable functional deficits or vice v ersa5, termed structure–function discordance, introduces a level of diagnostic a mbiguity5. As such, there is increased interest in understanding the structure–function relationship in detail, with the intention of facilitating earlier detection of glaucoma. The ability of optical coherence tomography (OCT) to acquire in vivo, quantitative RNFL and inner retinal thickness measurements has contributed to its ongoing use in investigations of the structure–function relationship5. The macula is a particular region of interest due to its high density of retinal G Cs6, and given the association between macular inner retinal thickness measurements and central VF loss7, 8, early detection of 1 Centre for Eye Health, University of New South Wales, Sydney, NSW 2052, Australia. 2School of Optometry and Vision Science, University of New South Wales, Sydney, NSW, Australia. 3Contact Lens and Visual Optics Laboratory, Queensland University of Technology, Brisbane, QLD, Australia. *email: Scientific Reports | (2020) 10:18527 | https://doi.org/10.1038/s41598-020-75599-0 1 Vol.:(0123456789) www.nature.com/scientificreports/ Demographic variable Mean ± SD Range Age 47.83 ± 16.02 20.13 to 84.91 Spherical equivalent refractive error (D) − 0.62 ± 1.86 − 6.00 to + 3.75 Fovea to optic disc tilt (°) 6.64 ± 3.33 − 5.60 to 16.40 Demographic variable OD (%) OS (%) Eye included 253 (52.9) 225 (47.1) Demographic variable Male (%) Female (%) Gender 204 (42.7) 274 (57.3) Table 1.  Demographic information of the study cohort. SD standard deviation, D diopters, ° degrees, OD right eye, OS left eye, % percentage. glaucomatous loss at the macula is paramount. As such, several studies have capitalized on the ability of commercial OCT software to obtain macular inner retinal thickness measurements to describe the macular structure–function relationship9–12. While using measurement paradigms from commercially available OCT software has the advantage of direct applicability to clinical settings, areas over which measurements are averaged are semi-arbitrary and do not directly coincide with retinal locations stimulated by test targets as per standard VF patterns. In particular, averaged foveal inner retinal measurements commonly include the GC-poor foveal pit, and as a result do not adequately reflect the GC-rich locations directly adjacent to the foveal pit6, 13 that constitute the primary retinal loci contributing to high foveal VF sensitivities14, 15. As region-averaged OCT-derived measurements do not directly reflect the cellular tissues contributing to visual function, the phenomenon of structure–function discordance may be in part due to discrepancies in locations over which measurements are derived. Indeed, several studies have highlighted the necessity of comparing structural and functional measures from corresponding retinal l ocations7, 16–18, and we recently partially accounted for this issue via manual extraction of foveal ganglion cell layer (GCL) thicknesses19, 20, with these models demonstrating excellent structure–function concordance (coefficients of determination 0.94–0.98)19. Additionally, given that different VF stimulus sizes are often tested in investigations of the structure–function relationship15, 20–23, the area over which structural measurements are derived may also be pertinent in limiting discordance. Practically speaking, it is important to consider whether high-density measurement paradigms available on commercial OCT software are sufficient surrogates for structural measurements matching retinal locations stimulated by VF test targets, as are utilized in numerous studies10, 24, 25, thereby enabling direct translation of research findings to clinical applications. Alternatively, if systematic differences between these methods exist, additional efforts are necessary to ensure structure and function are precisely matched. Given the discrepancies in macular OCT measurements used in structure–function investigations, we hypothesized that comparisons between OCT measurements directly corresponding to retinal locations stimulated by VF test patterns and a high-density measurement grid available on OCT review software will yield statistically and clinically significant differences. We tested this hypothesis by developing a customized algorithm enabling automated extraction of inner retinal thickness measur (...truncated)