AxoNet 2.0: A Deep Learning-Based Tool for Morphometric Analysis of Retinal Ganglion Cell Axons.

Purpose: Assessment of glaucomatous damage in animal models is facilitated by rapid and accurate quantification of retinal ganglion cell (RGC) axonal loss and morphologic change. However, manual assessment is extremely time- and labor-intensive. Here, we developed AxoNet 2.0, an automated deep learning (DL) tool that (i) counts normal-appearing RGC axons and (ii) quantifies their morphometry from light micrographs. Methods: A DL algorithm was trained to segment the axoplasm and myelin sheath of normal-appearing axons using manually-annotated rat optic nerve (ON) cross-sectional micrographs. Performance was quantified by various metrics (e.g., soft-Dice coefficient between predicted and ground-truth segmentations). We also quantified axon counts, axon density, and axon size distributions between hypertensive and control eyes and compared to literature reports. Results: AxoNet 2.0 performed very well when compared to manual annotations of rat ON (R2 = 0.92 for automated vs. manual counts, soft-Dice coefficient = 0.81 ± 0.02, mean absolute percentage error in axonal morphometric outcomes < 15%). AxoNet 2.0 also showed promise for generalization, performing well on other animal models (R2 = 0.97 between automated versus manual counts for mice and 0.98 for non-human primates). As expected, the algorithm detected decreased in axon density in hypertensive rat eyes (P ≪ 0.001) with preferential loss of large axons (P < 0.001). Conclusions: AxoNet 2.0 provides a fast and nonsubjective tool to quantify both RGC axon counts and morphological features, thus assisting with assessing axonal damage in animal models of glaucomatous optic neuropathy. Translational Relevance: This deep learning approach will increase rigor of basic science studies designed to investigate RGC axon protection and regeneration.

Evaluation of Spatially Targeted Scleral Stiffening on Neuroprotection in a Rat Model of Glaucoma.

Purpose: Scleral stiffening may protect against glaucomatous retinal ganglion cell (RGC) loss or dysfunction associated with ocular hypertension. Here, we assess the potential neuroprotective effects of two treatments designed to stiffen either the entire posterior sclera or only the sclera adjacent to the peripapillary sclera in an experimental model of glaucoma. Methods: Rat sclerae were stiffened in vivo using either genipin (crosslinking the entire posterior sclera) or a regionally selective photosensitizer, methylene blue (stiffening only the juxtaperipapillary region surrounding the optic nerve). Ocular hypertension was induced using magnetic microbeads delivered to the anterior chamber. Morphological and functional outcomes, including optic nerve axon count and appearance, retinal thickness measured by optical coherence tomography, optomotor response, and electroretinography traces, were assessed. Results: Both local (juxtaperipapillary) and global (whole posterior) scleral stiffening treatments were successful at increasing scleral stiffness, but neither provided demonstrable neuroprotection in hypertensive eyes as assessed by RGC axon counts and appearance, optomotor response, or electroretinography. There was a weak indication that scleral crosslinking protected against retinal thinning as assessed by optical coherence tomography. Conclusions: Scleral stiffening was not demonstrated to be neuroprotective in ocular hypertensive rats. We hypothesize that the absence of benefit may in part be due to RGC loss associated with the scleral stiffening agents themselves (mild in the case of genipin, and moderate in the case of methylene blue), negating any potential benefit of scleral stiffening. Translational Relevance: The development of scleral stiffening as a neuroprotective treatment will require the identification of better tolerated stiffening protocols and further preclinical testing.

Assessment of Visual and Retinal Function Following In Vivo Genipin-Induced Scleral Crosslinking.

Purpose: Genipin has been proposed as a possible neuroprotective therapy in myopia and glaucoma. Here, we aim to determine the effects of prolonged genipin-induced scleral stiffening on visual function. Methods: Eyes from Brown Norway rats were treated in vivo with either a single 15 mM genipin retrobulbar injection or sham retrobulbar injection and were compared to naïve eyes. Intraocular pressure, optomotor response, and electroretinograms were repeatedly measured over 4 weeks following retrobulbar injections to determine visual and retinal function. At 4 weeks, we quantified retinal ganglion cell axon counts. Finally, molecular changes in gene and protein expression were analyzed via real-time polymerase chain reaction (RT-PCR) and proteomics. Results: Retrobulbar injection of genipin did not affect intraocular pressure (IOP) or retinal function, nor have a sustained impact on visual function. Although genipin-treated eyes had a small decrease in retinal ganglion cell axon counts compared to contralateral sham-treated eyes (-8,558 ± 18,646; mean ± SD), this was not statistically significant (P = 0.206, n = 9). Last, we did not observe any changes in gene or protein expression due to genipin treatment. Conclusions: Posterior scleral stiffening with a single retrobulbar injection of 15 mM genipin causes no sustained deficits in visual or retinal function or at the molecular level in the retina and sclera. Retinal ganglion cell axon morphology appeared normal. Translational Significance: These results support future in vivo studies to determine the efficacy of genipin-induced posterior scleral stiffening to help treat ocular diseases, like myopia and glaucoma.

AxoNet: A deep learning-based tool to count retinal ganglion cell axons.

In this work, we develop a robust, extensible tool to automatically and accurately count retinal ganglion cell axons in optic nerve (ON) tissue images from various animal models of glaucoma. We adapted deep learning to regress pixelwise axon count density estimates, which were then integrated over the image area to determine axon counts. The tool, termed AxoNet, was trained and evaluated using a dataset containing images of ON regions randomly selected from whole cross sections of both control and damaged rat ONs and manually annotated for axon count and location. This rat-trained network was then applied to a separate dataset of non-human primate (NHP) ON images. AxoNet was compared to two existing automated axon counting tools, AxonMaster and AxonJ, using both datasets. AxoNet outperformed the existing tools on both the rat and NHP ON datasets as judged by mean absolute error, R2 values when regressing automated vs. manual counts, and Bland-Altman analysis. AxoNet does not rely on hand-crafted image features for axon recognition and is robust to variations in the extent of ON tissue damage, image quality, and species of mammal. Therefore, AxoNet is not species-specific and can be extended to quantify additional ON characteristics in glaucoma and potentially other neurodegenerative diseases.