Research On Vascular Network Developmental Pattern Based On Deep Learning

The developmental pattern of vascular network can reflect the normal growth process of blood vessels and provide an important reference for studying the growth mechanism of life and tumor.The development of vascular network is mainly controlled by a series of biomechanical and biochemical functions,such as blood flow rate and metabolic demands,and has complex characteristics of multiple time scales and spatial pattern changes.Due to the lack of image technology to evaluate the morphological changes of vascular network,the early research on the development mode of vascular network mainly focuses on molecular and cellular factors.With the continuous development of deep learning,more and more researchers apply it to medical images,especially to the combination of vascular network images,and get good research results.By training the deep learning model,the model can accurately identify the vascular network at different development stages,which is helpful to analyze the development pattern of the vascular network.The main research work and achievements completed in this paper are as follows:(1)Data enhancement for raw data.Aiming at the problem of low diversity of original data of chicken embryo vascular network,this paper proposes a data set enhancement method for deep learning to evaluate the development level of vascular network.This method can effectively increase the diversity of vascular network images.(2)Identification of vascular network at different development stages based on convolutional neural network.In this paper,convolutional neural network is applied to the identification of vascular network at different development stages.This experiment mainly consists of three parts: model construction,training model and test model.(3)Perform visualization operations on the test data.On the basis of the identification of vascular network at different stages,this paper further studies and performs visual operation on the development pattern of vascular network.Through the visualization operation of the test set,it is convenient to explain the prediction results of the convolutional neural network model in the later stage of this paper.Through the above work,the experiment has obtained good results.Through visualization technology,the characteristics and growth trends of vascular networks at different developmental stages are better demonstrated.