Segmentation of Magnetic Resonance Brain Imaging based on Graph Theory

Document Type: Original Paper

Authors

1 Department of Electrical Engineering, mashhad branch, Islamic Azad University, mashhad, Iran.

2 Department of Electrical Engineering, Khomeinishahr Branch (Isfahan)

Abstract

Brain images segmentation and extraction of its tissues is an essential requirement. In medical images, especially brain tissues from magnetic resonance imaging (MRI), due to the proximity of the brightness levels of images, the tissues, edges and boundaries between the tissues are ambiguous cannot be well recognized. Accurate segmentation of MRI Brain tissues is a challenge has significant importance in clinical applications and neuroscience research.
In this paper, the graph-based method is proposed to solve the segmentation issue of MRI brain images. In the proposed method, a weighted is assigned to the image. Each edge of the graph is considered as correspondence with an image pixel. The edge weight between two nodes demonstrates the similarity quantity between two pixels of the image. Afterward, a cost function as such points like relative extremes and the turning point, which is equivalent to the derivation of the function is assigned to the graph. Finding the shortest path in a graph is equal to finding the path with the lowest cost sum of all edges in the path. Minimizing this cost function leads to segmenting the image.The advantage of the graph method compared to the other methods is that spatial information is constructed simultaneously with a high rate. Moreover, this method is implemented on the pixels in the space and can partition MRI brain images with low error improving the previous methods. The performed comparisons between the proposed method and active contours driven by local Gaussian distribution fitting energy segmentation method demonstrate that by applying the presented method the accuracy of the MRI brain image segmentation will be improved.
By the proposed approach, the accuracy of segmentation is increased. This method can well distinguish different types of brain tissues from the brain MRI image and has the potential to be employed for clinical applications.

Keywords

Main Subjects



Articles in Press, Accepted Manuscript
Available Online from 25 June 2019
  • Receive Date: 10 November 2018
  • Revise Date: 16 June 2019
  • Accept Date: 25 June 2019