Abstract
Purpose :
Our goal was to determine whether QuPath open source image analysis software for digital pathology was adaptable for rapid, automated quantification of axons and glia in optic nerves (ONs) from glaucomatous Brown Norway rats.
Methods :
QuPath cell detection parameters and classification features were modified to quantify axon numbers and glial areas in healthy, moderately damaged, and severely damaged ONs. QuPath performance was compared to AxonJ, a Fiji (ImageJ) plugin designed for rodent ON axon analysis. ONs were obtained from Brown Norway rat retired breeders (male and female) that had received intraocular injection of polystyrene microbeads to elevate intraocular pressure (IOP). Five bead concentrations (2.5, 5, 10, 15, and 20%) were tested, 4 rats per group. Opposing and naive eyes served as controls. IOP was measured twice a week. 5 weeks post bead injection, rats were sacrificed and ONs removed 1.5 mm posterior to the globe, fixed in 4% paraformaldehyde, 2% glutaraldehyde in 0.1 M sodium cacodylate buffer, postfixed in osmium tetroxide, stained en bloc with uranyl acetate, and embedded in Epoxy resin. Toluidine blue stained semithin (500 nm) sections were imaged by light microscope (63x oil). Tiled micrographs were stitched together providing high resolution ON cross sections for QuPath and AxonJ analysis.
Results :
Compared to AxonJ, QuPath more accurately quantified axon numbers and glial areas in healthy, moderately and severely damaged optic nerves. QuPath axon counts had fewer false positive and false negative axons. QuPath analysis showed that ON damage increased with IOP. Success rates for IOP elevation increased from 25% to 80% with bead concentration, 2.5% versus 20%, respectively.
Conclusions :
QuPath is an easy to use, effective tool for rapid automated analysis of axons and glia in the Brown Norway rat microbead model of glaucoma. QuPath is superior to AxonJ for ON axon counts and has the added capability to quantify glial areas. Increased bead concentrations improved success rates for IOP elevation.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.