Biomedical Engineering Reference
In-Depth Information
promising. Deformable models proposed in [6, 12, 13] depend on evolving an ini-
tial co-dimensional object (curve in 2D or surface in 3D) under certain constraints
until it covers the structure of interest. Although this approach is very flexible in its
initialization, it requires tuning of a large number of parameters. Baillard et al. [7]
presented an approach that does not require that any parameters be tuned. The
proposed algorithm controls the evolution of the surface by taking into considera-
tion probabilistic region information. The designed level set model is based on the
estimated density functions using the stochastic expectation maximization. But
this method can only work for bimodal images, and this may be too restrictive for
many applications. Farag et al. [9] generalized the previously discussed approach
to be applied to multimodal images. Xu et al. [14] introduced a method for recon-
struction of the cerebral cortex from MR imaging based on fuzzy segmentation
and deformable models. This chapter will utilize the advanced level set techniques
used for brain segmentation and analysis to develop an integrated system that will
be used in the analysis and quantification of brain developmental disorders, and
especially dyslexia, which is the main focus of this chapter. Unlike the previously
published approaches, we will focus on investigating the relationship between the
microscopic findings that have been investigated through the minicolumnar struc-
tures of autistic and dyslexic brains [3, 15] and the Magnetic Resonance Images
employing level sets. We will present this subject matter in the following order:
The neuroanatomy background:
This will handle the definition of the mini-
columns, and how they relate to the developmental brain disorders of inter-
est. The hypotheses made to quantify these diseases either using Magnetic
Resonance Imaging or Diffusion Tensor Imaging will be included.
Problem statement, dataset description:
Thiswill introduce a description of the
data, the data collection procedure, and the MRI protocol, as well as the
basic pipeline for the approach used to analyze the data.
Volumetric MRI analysis of dyslexic subjects using a level set framework:
This section will introduce the details of the approach used in the analysis
and the results on both synthetic and real MRI data.
Concluding remarks:
This section will present the conclusion of the work and
propose possibilities for future work.
2.
THE NEUROANATOMY BACKGROUND
Brain function research suggests that the minicolumns, not individual neu-
ronal cells, are the basic operational units in the brain. For instance, Mountcastle
[17] reported that the effective unit of operation in the human brain is not the sin-
gle neuron and its axon, but groups of cells with similar functional properties and
