Code for: Multi-modal screening for synergistic neuroprotection of mild extremely preterm brain injury: Cell counting code repository

Preterm brain injury affects both white and grey matter, including altered
cortical development and gyrification, with associated neurodevelopmental
sequelae such as cerebral palsy and learning deficits. The preterm brain
also displays regionally heterogeneous responses to both injury and
treatment, supporting the need for drug combinations to provide global
neuroprotection. We developed an extremely preterm-equivalent organotypic
whole hemisphere (OWH) slice culture injury model using the gyrencephalic
ferret brain to probe treatment mechanisms of promising therapeutic agents
and their combination. Regional and global responses to injury and
treatment were assessed by cell death quantification, machine
learning-augmented morphological microglia assessments, and digital
transcriptomics. Using two promising therapeutic agents, azithromycin (Az)
and erythropoietin (Epo), we show minimal neuroprotection by either
therapy alone, but evidence of synergistic neuroprotection by Az*Epo both
globally and regionally. This effect of Az*Epo involved emergent
augmentation of transcriptomic responses to injury related to neurogenesis
and neuroplasticity and downregulation of transcripts involved in cytokine
production, inflammation, and cell death. This study supports the use of
the ferret OWH slice culture model to provide a powerful high-throughput
platform to examine combinations of therapeutics for extremely preterm
brain injury. This repository supports our associated publication by
Jin et al. in Bioengineering & Translational Medicine by
supplying code used to process histology images. This codebase serves
to identify nuclei in a DAPI-stained histology slide, measure
geometry and intensity-based properties, and then develop and apply a
random forest model to classify nuceli as pyknotic or non-pyknotic based
on their measured properties.