Design And Implementation Of An Auxiliary Detection System For Lesions Based On U-shaped Network

With the rapid development of computer science and the extension of deep learning applications,deep learning has gained outstanding achievements in computer vision applications.How to combine deep learning with traditional industries has become a hot topic at present.In the field of medical image segmentation,doctors observe and segment the medical images of patients with corresponding medical knowledge,and then devise treatment plans accordingly.However,due to the specificity of medical images,manual segmentation is highly demanding and inefficient for doctors.In order to solve this problem,this paper realizes the application of computer-aided examination technology in lesion segmentation through research,so as to relieve the work pressure of doctors and improve the efficiency of treatment.In this paper,we take the research and system implementation of medical image segmentation algorithm as our research topic,and then conducted the experiments on the brain tumor medical image dataset brats.The specific research content will be developed as follows.We first introduce the background and research significance of lesion segmentation,describes its theoretical basis,and illustrates the advantages and disadvantages of common segmentation algorithms.Then,we proposed an improved segmentation algorithm based on U-shaped network with better performance in numerical and visual terms compared to other classic models.Our Segmentation algorithm achieves higher accuracy,and is sufficient to assist the doctors in completing accurate lesion segmentation.Finally,we carried out the overall design and system analysis work of the lesion assisted testing system.We clarified the system requirements,analyzed the main functional modules of the system,designed the whole software system,and completed the coding with final system testing.