A rapid and accurate method to quantify neurite outgrowth from cell and tissue cultures: Two image analytic approaches using adaptive thresholds or machine learning

Abstract

BackgroundAssessments of axonal outgrowth and dendritic development are essential readouts in many in vitro models in the field of neuroscience. Available analysis software is based on the assessment of fixed immunolabelled tissue samples, making it impossible to follow the dynamic development of neurite outgrowth. Thus, automated algorithms that efficiently analyse brightfield images, such as those obtained during time-lapse microscopy, are needed. New methodWe developed and validated algorithms to quantitatively assess neurite outgrowth from living and unstained spinal cord slice cultures (SCSCs) and dorsal root ganglion cultures (DRGCs) based on an adaptive thresholding approach called NeuriteSegmantation. We used a machine learning approach to evaluate dendritic development from dissociate neuron cultures. ResultsNeuriteSegmentation successfully recognized axons in brightfield images of SCSCs and DRGCs. The temporal pattern of axonal growth was successfully assessed. In dissociate neuron cultures the total number of cells and their outgrowth of dendrites were successfully assessed using machine learning. Comparison with existing methodsThe methods were positively correlated and were more time-saving than manual counts, having performing times varying from 0.5−2 min. In addition, NeuriteSegmentation was compared to NeuriteJ®, that uses global thresholding, being more reliable in recognizing axons in areas of intense background. ConclusionThe developed image analysis methods were more time-saving and user-independent than established approaches. Moreover, by using adaptive thresholding, we could assess images with large variations in background intensity. These tools may prove valuable in the quantitative analysis of axonal and dendritic outgrowth from numerous in vitro models used in neuroscience.