Purchase this article with an account.
Daniel E. Maidana, Shoji Notomi, Tiannan Zhou, Takashi Ueta, Pavlina Tsoka, Hidetaka Matsumoto, Keiko Kataoka, Haijiang Lin, Joan W Miller, Demetrios G. Vavvas; High-throughput Retinal Layer Thickness Quantitation in Retinal Degenerative Diseases.. Invest. Ophthalmol. Vis. Sci. 2016;57(12):5948. doi: https://doi.org/.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Retinal nuclear layer thickness is a widely used criterion to evaluate retinal degeneration onset and progression in animal models. Quantitation of outer nuclear layer (ONL) and inner nuclear layer (INL) thickness in retina sections is usually performed manually, ideally by two observers. This process is time consuming, prone to inaccurate measurements, and not entirely reproducible. In this context, we describe an automated quantitation approach to address these difficulties.
We designed an ImageJ macro to quantitate ONL and INL area and thickness in retinal cross-sections of animal retinal degeneration models. The script was coded using IJ1 programming language. To validate this tool, we randomly selected 75 digital retinal images from the Angiogenesis Laboratory fluorescence microscopy database. Two observers, inexperienced and experienced, quantitated retinal layer thickness twice, using 6 calipers per retinal layer. Measurements between observers and macro were analyzed with Bland-Altman plots and mean difference.
The ImageJ macro was initially calibrated using a sham digital image containing an object of known pixel dimensions and bi-dimensional spatial orientation. When processing retinal digital images, thickness was comprehensively measured using 150 calipers per segmented layer. After image processing, the macro reported values for area and mean thickness per layer and corresponding standard error of the mean (SE). The mean ± SE for ONL thickness was 65.48 ± 8.48 µm, 62.01 ± 6.57 µm, and 60.48 7.15 µm, for the inexperienced, experienced, and ImageJ macro, respectively. Mean ± SE for INL thickness was 38.89 ± 6.84 µm, 35.47 ± 5.57 µm, and 38.13 ± 7.32 µm, for the inexperienced, experienced, and ImageJ macro, respectively. Between observers, the mean difference was 3.47 ± 0.48 µm for the ONL and 3.42 ± 0.46 µm for the INL. Comparing observers’ measurement with macro, mean difference was 3.26 ± 0.87 for the ONL and 0.95 ± 0.78 for the INL.
The customized and validated ImageJ macro achieved a fitted segmentation and comprehensively assessed retinal layers thickness, in agreement with observer values. We believe that this standardized high-throughput measurement tool could be advantageous to compare results across different research groups, as it will be freely distributed as open-source.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.
This PDF is available to Subscribers Only