Gaussian Mixture Model for MRI Image Segmentation to Build a Three-Dimensional Image on Brain Tumor Area

Abstract

A brain tumor is one of the deadly diseases that attack the central and nervous
system. The treatment of brain tumor, need high accuracy and precision. Brain tumor
detection through Magnetic Resonance Imaging (MRI) has two-dimensional output with
three perspectives, namely sagittal, coronal, and axial. These different perspectives need
to be seen one by one to determine the location and size of the tumor. To
solve the problem, this study constructs the three-dimensional visualization perspective of
MRI images. The tumor area in MRI image is segmented as a region of interest (ROI) by
employing the Gaussian Mixture Model (GMM) with Expectation-Maximization as the
optimization technique. These couple segmentation methods have revealed significant gain
as a clear boundary of the tumor area to separate from the healthy part of the brain and
an estimated tumor volume from sagittal, coronal, and axial perspectives. Furthermore,
these findings have been successfully visualized in 3D construction of the tumor position
on the left side of the patient’s head with an estimated volume of 749mm³