Premium
A medical image segmentation method based on hybrid active contour model with global and local features
Author(s) -
Li Yuanmu,
Wang Zhanqing
Publication year - 2020
Publication title -
concurrency and computation: practice and experience
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.309
H-Index - 67
eISSN - 1532-0634
pISSN - 1532-0626
DOI - 10.1002/cpe.5763
Subject(s) - initialization , active contour model , level set (data structures) , segmentation , computer science , image segmentation , curvature , level set method , term (time) , artificial intelligence , computation , energy functional , computer vision , image (mathematics) , algorithm , mathematics , geometry , mathematical analysis , physics , quantum mechanics , programming language
Summary In this article, we proposed an improved region‐based active contour model based on curve evolution theory and variational level set method. Our method can be used to segment image with intensity inhomogeneity, such as medical computed tomography images. Our model contains a local intensity fitting term that makes the evolution curve stop at boundaries of the object and a global expansion term that makes the evolution curve have the chance to get to every location in the image. Therefore, our model has a good performance to solve the problem of flexible initialization, which exposed in region‐scalable fitting energy model. For the curvature term that occurs during the calculation, we calculated it with a more efficiency and accuracy method. Compared with other models, our model shows good segmentation result and less computation expense. Finally, we will present some experimental results, especially the result of contrast experiment.