Neuron Characterization in Complex Cultures Using a Combined YOLO and U-Net Segmentation Approach.

This work presents a novel approach for the automated characterization of neurons in primary culture from phase-contrast images. Direct characterization of neurons from these images is very challenging due to the complexity of the images. Over time in culture neurons change shape and size, and extend neuronal connections (i.e., neurites) between them. Also, cultured neurons are accompanied by other cell types, mainly glial cells, which may be difficult to distinguish from neurons. In this study we have applied U-Net segmentation to isolate and extract individual neurons, while also addressing the challenges posed by the presence of other cell types and structures in the culture. We then used YOLO object detection to classify and localize neurons accurately. Combining these two models, we have been able to successfully characterize neurons within these complex cultures. Our findings demonstrate the potential of this approach for a more comprehensive analysis of neurons in challenging environments. The present work is part of a larger study aimed to fully analyze neuronal behaviour throughout development.