Reanalysis of Published Histological Data Can Help to Characterize Neuronal Death After Spinal Cord Injury.

Abstract

Neuronal death is a central event in spinal cord injury (SCI) pathophysiology. Despite its importance, we have a fragmentary vision of the process. In our opinion, the research community has accumulated enough information to provide a more detailed, integrated vision of neuronal death after SCI. This work embeds this vision by creating an open repository to store and share data and results from their analysis. We have employed this repository to upload raw images of spinal cord sections from a mouse model of contusive SCI and used this information to compare manual-, threshold-, and neural network-based neuron identifications and to explore neuronal death at the injury penumbra 21 days after injury and the effects of the anti-apoptotic drug ucf-101. Results indicate that, whereas the three identification methods assayed yield coherent estimates of the total number of neurons per section, neural network (NN) outperforms the other two methods. Combining NN identification and image registration has allowed us to characterize neuron distribution among Rexed laminae in the mice T11, revealing spatial patterns in the neuronal death that follows injury and in their survival following ucf-101 treatment.