Heuristically improved Bayesian segmentation of brain MR images

Abstract

One of the major tasks or even the most prevalent task in medical image
processing is image segmentation. Among them, brain MR images suffer
from some difficulties such as intensity inhomogeneity of tissues, partial
volume effect, noise and some other imaging artifacts and so their segmentation is more challenging. Therefore, brain MRI segmentation
based on just gray values is prone to error. Hence involving problem specific heuristics and expert knowledge in designing segmentation algorithms seems to be useful. A two-phase segmentation algorithm based
on Bayesian method is proposed in this paper. The Bayesian part uses the
gray value in segmenting images and the segmented image is used as the
input to the second phase to improve the misclassified pixels especially in
borders between tissues. Similarity index is used to compare our algorithm
with the well known method of Ashburner which has been implemented in
Statistical Parametric Mapping (SPM) package. Brainweb as a simulated
brain MRI dataset is used in evaluating the proposed algorithm. Results
show that our algorithm performs well in comparison with the one implemented in SPM. It can be concluded that incorporating expert knowledge and problem specific heuristics improve segmentation result.
The major advantage of proposed method is that one can update the
knowledge base and incorporate new information into segmentation
process by adding new rules.

Keywords: Magnetic Resonance Imaging (MRI); Segmentation; Bayesian
classifier; Heuristics