Fluorescence image region segmentation method and system based on deep learning

The invention discloses a fluorescence image region segmentation method and system based on deep learning, and aims to determine the position of a blood vessel region image by using a deep learning model and improve the blood vessel detection speed and accuracy. Specifically, in an improved dense U-Net deep neural network model, a label of a single image and time sequence information contained in front and back frames of an image adjacent to the image can be extracted so as to adapt to the characteristic that fluorescence development in the task has flowability, and the requirement for labeling work is small. A dense connection structure is added in the model, so that multi-scale features can be utilized more fully to adapt to the task of blood vessel segmentation in which coarse branches and fine branches exist at the same time, and the accuracy of image segmentation can be remarkably improved. And the number of layers of the improved dense U-Net deep neural network model is reduced from four layers to three