Deep learning based multifunctional synaptic analysis of electron microscope images

Synaptic analysis with regard to its number and fine morphology are the basis for the understanding of neurological functions and diseases. Manually analyzing synaptic structures in images obtained from electron microscope (EM) often encountered with limitations such as low efficiency and subjective bias. In this study, we developed a multifunctional synaptic analysis system based on several advanced deep learning (DL) models. The system achieved synapse counting in low-magnification EM images and synaptic ultrastructure analysis in high-magnification EM images.