Magnetic Resonance Imaging (MRI)

In recent years, magnetic resonance imaging (MRI) has become primarily a medical imaging technique to visualize the structure and function of the body or brain. It is also important for clinical diagnosis, medical treatment and further residential care. It was developed by Dr. Paul Lauterber in 1972 [4]. The major principle technique behind MRI is the development of nuclear magnetic resonance (NMR). In the past, magnetic resonance was used only for studying the chemical structure of substances. Until the 1970s NMR could be used to produce images of the body by Lauterbur’s and Mansfield’s great work.

However, as the word nuclear was associated in the public mind with ionizing radiation exposure. It is generally now referred to simply as MRI.

There are three major components of an magnetic resonance imaging scanner : A static magnetic field, an RF transmitter and receiver, and three orthogonal, controllable magnetic gradients. Fig. 1.4 is a 3.0 T MR scanner. The image quality is directly proportional to the magnetic field strength. Higher magnetic fields increase signal-to-noise ratio, permitting higher resolution or faster scanning. However, The higher the magnetic field strength is,

1.1 Background 7

Figure 1.4: A MR scanner. Photo acquires from the web site of Research Center for Integrative Neuroimaging and Neuroinformatics (RCINN) National Yang-Ming University.

(Graphic source : http://www.ym.edu.tw/rcinn/introduction.htm)

the better quality image we can acquire. A field strength of 1.0 - 1.5 T is a good compro-mise between cost and performance for general medical use. However, for certain specialist uses higher field strengths are desirable, with some hospitals now using 3.0 T scanners. Fig 1.2 is an MR image scanned by a 3.0 T MR scanner in National Yang-Ming University.

With the rapid growth in MR imaging, it is a widely used technology in medical diagnosis, pathological study, and medical treatment. From head to foot, from cancer to cardiovas-cular vessel disease, from diagnosing to afterward following observations, it has already become an important imaging technique indispensable to modern medical centers.

There are many advantages of MR technology. First of all, it is noninvasive when detecting signals inside the body, so it is more safety for people under operations and diagnosis. Unlike computed tomography (CT), it uses non-ionizing radiation. Instead, it uses a powerful magnetic field to align the nuclear magnetization of hydrogen atoms in water molecules in the body. Which means MR imaging has no harm to patients who take the MR scan. Another advantage of MR imaging is that it provides much greater contrast between each different soft tissues of the body than CT does. It is especially useful in neurological (brain), musculoskeletal, cardiovascular, and oncological imaging and gives a great assistance in diagnosis of tumor or brain discrepancy in bipolar disorder (BD) or major depressive disorder (MDD). Another advantage of MR imaging is that it has no

side-effects which calls supplemental harm to patients. However, a disadvantage of MRI scanner is that the instrument is quite expensive. A new 1.5 tesla scanner approximately costs one million US dollars and two million US dollars for a new 3.0 tesla scanners. Constructing a MRI suite can cost a hundred thousand US dollars.

In human brain diagnosis and its researches, more and more studies are using CT, 2-D MR image or even higher resolution MR images. Because of the great contrast between different soft tissues of the brain, we can more easily distinguish gray matter, white matter and cerebrospinal fluid from brain. Due to the improvement of image resolution and devel-opment of the image processing tools and computers could handle numerous and complex operations, a number of unbiased whole brain morphometric analysis methods were pro-posed to characterize brain discrepancy.

1.2 Morphometrics

Morphometrics is a method to analyze the variation and the change in size or shape of organisms or brain. With the MR imaging technique, morphometric analysis are now com-monly performed on in-vivo studies and particularly useful in analyzing the fossil record.

It gives a quantitative element to describe the discrepancy of objects and allows more rigor-ous comparisons. In morphometric analysis, we can describe complex shapes or variation in size, and use the numerical comparison between different objects. Furthermore, sta-tistical analysis can highlight areas where change is concentrated and quantify the level of significance. The morphometric analysis of brain images originally requires manually defining a number of regions of interest (ROI). It means that the method is based on an defined region of interests and analysis each object in this pre-defined ROI to perform the statistical differences on the volumes in each object [5].

However, the method has many potential drawbacks and limitations, including the de-mand of subject is always high, subjectivity, lacking of reproducibility. Quite a few

lim-1.2 Morphometrics 9

itations are that it is impossible to know which area in brain might be atrophy or enlarge by diseases or a surgical trauma, we could not know the relations between diseases and the area which has been analyzed by the method in advance. Although several regions are known related to the disease, the measurement may include other surrounding regions blur the results and reduces statistical power. Therefore, since a priori knowledge of regions of interest for the disease is quite important, and according to sufficient previous studies, we can make up to the deficiency of priori knowledge.

More and more complex automatically/semiautomatically morphometric methods in-clude the techniques of spatial normalization and tissue segmentation are proposed to an-alyze shape transformation or brain structure discrepancy. These methods can be divided into two categories: The first category uses the deformation fields computed by spatial normalization to compare the differences, which are to detect the differences in shape of the brain. The other category uses the normalized images to make comparisons, that is to detect the differences in brain tissue under an identical space.

The first category of morphometric method includes methods that measure the spatial transformation, which is analyzed by the deformation field deformed from templates of brain to each individual subject in the study. Several methods have been proposed, such as deformation-based morphometry (DBM) [6, 7] and tensor-based morphometry (TBM) [8, 9]. These approaches are the most direct way of measuring brain shape. But the method is based on the perfect registration between the template and the subject. Otherwise, a bad registration with small errors may reduce the accuracy of the method.

The other category of morphometric method includes the well-known method: voxel-based morphometry (VBM) [10, 11]. It uses a spatial transformation to normalize images into an identical space. Due to the use of the normalized images, the overall shape dif-ferences between subjects can be removed. It means that each subjects registered to the template will be in the same template space and in the same shape, and we can make com-parisons of brain tissue in normalized images. The method have been commonly used in several studies within the past decade.