Development of Level Set in Image Segmentation with the Portable Extensible Toolkit for Scientific Computation

作者：Lomthong Phusanisa^*; Huabsomboon Pallop; Tamagawa Masaaki

来源：Journal of Medical Imaging and Health Informatics, 2016, 6(6): 1519-1525.

DOI：10.1166/jmihi.2016.1842

• Image Segmentation
• Level Set
• Re-Initialization
• PETSc

The level set method is one class of the segmentation algorithms in medical imaging and computer science. The Aim of medical image segmentation is to separate a given image into the essential segments expressed various problem including tumor segmentation, shape analysis and diagnosis some diseases. To implement the standard level set method, re-initialization is needed occasionally and it makes quite time consuming during detecting boundary curves. Fast medical image segmentation is essential for medical technologist to diagnose and understand some diseases better. So it is an extensive problem to reduce the computational time for re-initialization process. Message Passing Interface (MPI) approach is represented as a fast computing technique. This paper presents the Portable Extensible Toolkit for Scientific Computation (PETSc) for developing a large scale level set in image segmentation. PETSc is a parallel algorithm based on MPI for solving nonlinear systems. By comparing with traditional algorithm, experimental results show that the parallel algorithm is effective in terms of time reduction with the same segmentation accuracy.

• 出版日期2016-10