Construction of an image recognition algorithm based on neurite outgrowth of human motor neurons and its application in toxicological evaluation

Objective To develop an MATLAB algorithm for the automated measurement of motor neuron (MN) neurites and apply this algorithm to evaluate the organophosphate flame retardant tris (2-chloroethyl) phosphate (TCEP) on the growth of MN neurites. Methods Human embryonic stem cells were induced to gradually differentiate into MN. After the neurites were labelled with βⅢ-tubulin for fluorescence image processing, an image process algorithm was developed to automatically analyze the neurite and changes in neurite network area after pollutant treatment. During MN differentiation, the cells were treated with different concentrations of TCEP (0, 25, 50 and 100 μmol/L). Results MNs were successfully induced with the expression of choline acetyltransferase. The developed image recognition algorithm could analyze images in batches, and calculate the pixel area occupied by neural network, and reserve weak neurites by optimizing neurite retention threshold to improve the accuracy of measurement. The quantification data from image process algorithm showed that TCEP significantly decreased the percentage of neurite network area since the dose started from 50 μmol/L (P < 0.05). Conclusion An image process algorithm is successfully developed for automated measurement of neurites based on the human MN neurite model. Moreover, this algorithm can be applied to the toxicity assessment of TCEP.