Image segmentation and research on virus propagation method based on Unet algorithm

Abstract

To investigate the propagation performance of the improved Unet network technology in the recognition and segmentation of hemorrhagic regions in brain CT images. Methods A total of 476 brain CT images of patients with spontaneous intracerebral hemorrhage were retrospectively included. The improved Unet network was used to identify and segment the hemorrhagic areas of the patients' brain CT images. Clinicians manually marked the image data of the hemorrhagic areas. 430 pieces of data from 106 patients were selected to enter the training set, and 46 pieces of data from 11 patients were entered into the test set. After the experimental data set was enhanced by data, it underwent network training and model testing to determine the virus cell spreading performance, and segmented the results. Comparison with Unet network, FCN-8s and Unet++ network. Results In the segmentation of the hemorrhagic region of brain CT images by the improved Unet network, the three evaluation indexes of similarity coefficient, forward prediction coefficient and sensitivity coefficient reached 0.8738, 0.901 1 and 0.864 8 respectively, which were improved respectively compared with FCN-8s network. 8.80%, 7.14% and 8.96%, which are 4.56%, 4.44% and 4.15% higher than the Unet network respectively.