Optimizing the characterization and quantification of retinal ganglion cell somas in healthy and injured retinas using cellpose.

Abstract

Quantification of retinal ganglion cell (RGC) soma number and characterization of somal features are commonly used output metrics for investigation of optic neuropathies. Many investigators still perform these quantifications by hand, which is time consuming and prone to bias. Cellpose is an open-source Python package that can perform cellular segmentation and holds promise for automating somal analyses. Here, we designed a custom script incorporating the Cellpose package and custom-trained Cellpose models that are capable of automatic characterization and quantification of RGC somas. Our script, using our models, is capable of automatically counting RGC somas, along with characterizing RGC somal size. Further, we show that Cellpose can quantify RGCs using multiple cell-type specific markers and has the potential to quantify RGCs across the entire retina. Together, our custom Cellpose models and script which generates and analyzes Cellpose outputs provides a powerful tool for all-in-one RGC somal analysis.