Magnetic Resonance Brain Imaging

Inhaltsverzeichnis

1. Frontmatter

2. Chapter 1. Introduction

   Jörg Polzehl, Karsten Tabelow

   Abstract

   Images are common in our lives. They come as simple photographs or as the result of various medical, technical, or scientific experiments and are often very easy to interpret for our visual capabilities as humans. It was a real revolution when Lauterbur and Mansfield invented the use of the magnetic resonance phenomenon to generate images of the human body. It enabled in-vivo images of soft tissues and stipulated a lot of neuroscientific research on structure and function of the human brain. Often statistical models and methods are needed for the understanding of the information that is contained in the images. This has become even more important as neuroimaging evolved from providing images in two dimensions to three-dimensional volumes or time series of volumes or even data in five- or six-dimensional spaces. Then visual inspection becomes difficult if not impossible, and the information has to be aggregated by appropriate methods. In the following chapters, we will demonstrate how such an analysis can be performed for the three MRI imaging modalities that we work with.

3. Chapter 2. Magnetic Resonance Imaging in a Nutshell

   Jörg Polzehl, Karsten Tabelow

   Abstract

   Since its invention in the early seventies by Paul C. Lauterbur (Lauterbur 1973; Mansfield and Grannell 1973) and Peter Mansfield (Mansfield 1977), for which they shared the 2003 Nobel prize in Physiology and Medicine, Magnetic Resonance Imaging (MRI) has evolved into a versatile tool for the in-vivo examination of tissue. MRI is based on the nuclear magnetic resonance phenomenon. Although MRI is based on quantum mechanical properties of the particles at the sub-atom level, the large ensemble of particles in the tissue allows for a semi-classical description that can be relatively easy accessed. We very shortly review the basic ideas of MRI. A number of special MR imaging sequences, i.e., sequences of gradient and RF excitations, have proven to be very important for the neuroscientific research, especially the functional and diffusion weighted MRI and, recently, the Multi-Parameter Mapping. These data and their analysis will be the subject of the main chapters of this book. Here we provide a teaser on the basic acquisition principles.

4. Chapter 3. Medical Imaging Data Formats

   Jörg Polzehl, Karsten Tabelow

   Abstract

   There exists a large variety of data formats used in medical imaging in general and specifically for functional Magnetic Resonance Imaging, diffusion-weighted imaging, Multi-Parameter Mapping, or inversion recovery Magnetic Resonance Imaging. Medical imaging data typically contain the actual data and additionally some metadata. This may be the data dimensionality, the spatial extension of the imaged voxel, but also physical parameters of the image acquisition, or patient data. The way this is stored in the different data formats differs. Here, we discuss DICOM

   Superscript registered trade mark sign

   ), ANALYZE, and NIfTI formats as they are widely used for storing medical imaging data or analysis results that are interchangeable between different analysis software. We demonstrate how these data can be easily accessed from within R. This is amended with a short discussion of the Brain Imaging Data Structure (BIDS) standard.

5. Chapter 4. Functional Magnetic Resonance Imaging

   Jörg Polzehl, Karsten Tabelow

   Abstract

   Functional Magnetic Resonance Imaging (fMRI) maps brain activity by detecting changes in image intensity related to neural activity by the blood oxygenation level–dependent (BOLD) contrast. Functional MRI data essentially consist of time series of 3D images associated with a description of the experimental conditions. The chapter outlines an analysis pipeline for functional Magnetic Resonance Imaging (fMRI) experiments completely based on R packages. Thereby, we focus on a single-subject analysis with the full pipeline of data pre-processing, the General Linear Model, and inference with correction for the multiplicity of the statistical tests. Part of the chapter elaborates on the use of structural adaptive smoothing procedure in fMRI, which we specifically developed. We also include alternative fMRI analysis methods, i.e., other than the mass-univariate approach. The chapter concludes with a section on functional connectivity.

6. Chapter 5. Diffusion-Weighted Imaging

   Jörg Polzehl, Karsten Tabelow

   Abstract

   Diffusion -weighted Magnetic Resonance Imaging (dMRI) has long proven to be a versatile tool for the in-vivo microstructural investigation of the human brain, the spinal cord, or even muscle tissue. In contrast to conventional weighted MRI or functional MRI discussed in the preceding Chap. 4, it is quantitative in the sense that it directly infers on physical quantities with physical units, specifically the diffusion constant. In this chapter, we will first elaborate on the physical background before presenting experimental dMRI data and describe its processing. This includes pre-processing steps, i.e., the removal of artifacts, and the actual modeling of the data to infer on interesting and relevant quantities. We also discuss a structural adaptive smoothing method for dMRI data before concluding the chapter with fiber tracking within the brain white matter and the construction of structural connectivity networks.

7. Chapter 6. Multiparameter Mapping

   Jörg Polzehl, Karsten Tabelow

   Abstract

   Unlike conventional weighted MRI, leading to \(T_1\)-, \(T_2\)-, \(T_2^\star \)-, or proton density (\(P\!D\)) weighted images in arbitrary units, quantitative MRI (qMRI) aims to estimate absolute physical metrics. One example is dMRI considered in Chap. 5. qMRI is of increasing interest in neuroscience and clinical research for its greater specificity and its sensitivity to microstructural properties of brain tissue such as axon, myelin, iron, and water concentration. Furthermore, the measurement of quantitative data allows for comparison across sites, time points, and participants and enables longitudinal studies and multicenter trials. In order to maintain its comparability, quantitative maps obtained from qMRI have to be adjusted for instrumental biases. Then, in combination with biophysical models, qMRI can enable the in vivo characterization of key microscopic brain tissue parameters, which previously could only be achieved with ex vivo histology. Here, we focus on the quantities that are accessible by the multiparameter mapping (MPM) approach. We will also present an adaptive smoothing algorithm for this type of data.

8. Chapter 7. Inversion Recovery Magnetic Resonance Imaging

   Jörg Polzehl, Karsten Tabelow

   Abstract

   In this chapter, we focus on the analysis of data from inversion recovery magnetic resonance imaging (IRMRI). Using a series of acquisitions for different inversion times, it can be used to infer on quantitative parameters like the relaxation time \(T_1\) but also on microstructural tissue properties like its porosity. We will present an estimation method, including structural adaptive smoothing and Rice bias correction to tackle the problems related to the low signal-to-noise ratio, to access these parameter maps.

9. Backmatter