Semantic Segmentation Model of Fluorescent Neuronal Cells in Mouse Brain Slices Under Few Samples.

Image of mouse neuronal cells is an important tool for studying mice. Examining mouse neuronal cells is an essential step in both pharmacological and toxicological tests. The morphology of individual neurons and the state of connections between multiple neurons are essential measures of the physiological state of the mouse. Staining neuronal cells samples and observation through microscopy is the mainstay of this field. However, this step is tedious, monotonous, and requires a high level of practical experience from the researcher. In recent years, the analysis of cell morphology using computer vision has proven to be an efficient and accurate solution. This paper presents a deep neural network-based semantic segmentation model; it does not use the popular attention mechanism but breaks down the process into two steps to achieve satisfactory performance while maintaining the number of parameters at a low level.