Submit Manuscript  

Article Details


An Automotive Approach for Brain Tumor Segmentation Based on Gaussian Distribution and Level Set Method

[ Vol. 10 , Issue. 4 ]

Author(s):

R. Karuppathal and V. Palanisamy   Pages 290 - 296 ( 7 )

Abstract:


In recent years a great work of the research in the field of medical imaging was focused on the brain tumor segmentation. In this paper, a novel region based active contour model for Magnetic Resonance Images (MRIs) brain tumor segmentation based on a level set origination is proposed and implemented. The image intensities are explained based on the Gaussian distributions through diverse means and variances. The attained local mean and variances are defined as variables and the Moore Gaussian distributions are described by the level set function. The energy minimization is attained by the curve evolution of level set and the approximation of the local intensity means and variances, it is an iterative process. To handle the intensity inhomogeneities and noise, the means and variances are measured as spatially varying functions. The tumor segmentation is an important early phase to solve the segmentation problem effectively. Hybrid Median Filter (HMF) is used to preserve the edges. Also, Sussman boundary condition is explored to accurately extract the tumor portions rather than the unwanted segmentations. The number of iterations used in this novel framework is quite lesser than the existing approach. Hence, the time taken for proposed tumor segmentation technique is also lesser than the existing method. The proposed system Gaussian Distribution with Level Set Method (GD-LSM) results accurate segmentation, sensitivity and specificity outcomes.

Keywords:

Active Contour, Brain Tumor Segmentation, Curve Evolution, Gaussian Distributions, Hybrid Median Filter (HMF), Level set, and Magnetic Resonance Images (MRIs).

Affiliation:

PSNA College of engineering & Technology, Dindigul, Tamilnadu, India and Principal, Info Institute of Engineering, Coimbatore, Tamilnadu, India.

Graphical Abstract:



Read Full-Text article