Single Cell Morphology and Dynamic Signal Analysis Based on U-Net and DeepCell

Single-cell morphology and dynamic signal analysis are essential to reveal cell biological mechanisms and discover new phenotypes. However, there are challenges in cell segmentation and tracking, morphological feature extraction, and quantification. This paper proposes a single-cell morphology and dynamic signal analysis algorithm based on U-Net and DeepCell. First, according to the network structure and loss function of U-Net, a single cell morphology segmentation model is established, and morphological information such as cell contour, location, size, and shape is obtained. Second, to eliminate the signal interference between cells, we propose a single-cell signal-tracking model based on DeepCell. The accurate identification and quantitative analysis of cell signals are obtained by analyzing the morphological similarity and motion continuity of cells. Then, considering the morphological and signal characteristics of single cells, we construct a single-cell morphology and dynamic signal analysis model that comprehensively considers morphological changes and signal dynamics, making it more accurate in describing cells' function, state, and biological behavior. Finally, we built a single-cell morphology and dynamic signal analysis system based on U-Net and DeepCell algorithms. Simulation and experiments prove the effectiveness of the proposed algorithm.