Magnetic Resonance Imaging

Mary Holleboom

- 1 -

Abstract This paper investigates the development, principles, and applications of magnetic resonance imaging. As a widely used medical technique, this type of imaging utilizes the inherent magnetic properties of protons in human cells. Superconducting magnets are generally enclosed in a large housing unit to scan various parts of the body. Several types of magnetic resonance imaging have been developed to better scan specific organs and tissues. These include functional magnetic resonance imaging, nuclear magnetic resonance, and echo-planar imaging. Although there are no biological hazards yet associated with this technique, extreme caution must be taken when working with such large magnets. They produce strong magnetic fields that can pull both small and large metal objects into the machine. Overall this is an important medical breakthrough that is continually developing to diagnose and prevent serious illnesses.

Keywords Magnetic resonance imaging, Medical applications, Image processing techniques, Protons, Resonance, Gradient

I.  INTRODUCTION

Magnetic resonance imaging, commonly referred to as MRI, is the use of a very strong magnetic field to scan an object and produce an image. It is most frequently used for medical purposes. Unlike radiology methods, MR imaging uses no radiation. Instead, a magnet is housed in a large scanner (Fig. 1), and current flowing through a coil of wire is used to create a magnetic field. This field affects the alignment of the protons in a person’s tissues. An antenna is then used to detect signals sent from the protons, and the signals are converted into an image.

Fig. 1 Magnetic resonance imaging scanner

This technique for imaging internal organs and tissues is quite new. The magnetic resonance phenomenon was first discovered by Felix Bloch and Edward Purcell in 1946. From then until 1970, nuclear magnetic resonance (NMR) was developed for chemical and physical molecular analysis. Paul Lauterbur used a back projection technique to demonstrate MR imaging on small test tube samples in 1973. The basis of current MRI techniques didn’t appear until 1975 when Richard Ernst proposed MR imaging using phase and frequency encoding and the Fourier Transform. Since then, several developments have been made to reach the current status of MRI technology. In 1977 Raymond Damadian demonstrated MRI of the whole body, and Peter Mansfield developed an imaging technique called echo-planar imaging (EPI). This technique was utilized in 1987 to perform real-time movie imaging of a single cardiac cycle. A new development, functional magnetic resonance imaging (fMRI), came about in 1993. This allows the mapping of the function of the various regions of the human brain. Today MRI technology is continually improving and is used in common medical practice.

There are several types of magnetic resonance imaging now in use, including volume imaging, nuclear magnetic resonance, and functional magnetic resonance imaging. They are all based on the same principles and make use of similar equipment. Finally, as with most medical procedures, there are health and safety issues to consider.

II.  imaging principles

A. Spin Physics

Magnetic resonance imaging makes use of a fundamental property called spin. Protons, electrons, and neutrons all posses spin, either + or – ½. Because of the positive and negative factors, spins can pair up and cancel each other. Unpaired, nuclear spins are utilized in NMR. However, NMR can only be performed on isotopes whose natural abundance is high enough for detection.

Within a magnetic field, particles with spin behave in a specific manner. For example, a proton acts like a small magnet with a north and a south pole. When this proton is then placed in an external magnetic field, it aligns itself with the field, as shown in Fig. 2. The particle can also undergo a transition between two energy states when it absorbs a photon. The energy of the photon is related to its frequency. This is called the resonance or the Larmor frequency in NMR and MRI. The signal used for NMR is produced by the difference in energy absorbed and emitted from the spins.






Fig. 2 Proton in an external magnetic field

Spin can also be described at the macroscopic level as a spin packet, or group of spins experiencing the same magnetic field strength. The magnetic field due to the spins is represented by a magnetization vector. T1 and T2 are time constants describing how the equilibrium magnetization is achieved. The z component of the magnetization is referred to as longitudinal magnetization (MZ) (Fig. 3). T1 is the time required to change the longitudinal magnetization by a factor of e and is called the spin lattice relaxation time. It is governed by the equation:

Mz = Mo ( 1 - e-t/T1 ) (1)

where Mo is the equilibrium magnetization and t is the time after displacement.

Fig. 3 Longitudinal magnetization

In the x-y direction, the magnetization is referred to as transverse (MXY). T2 is the time it takes to reduce the transverse magnetization by a factor of e and is called the spin-spin relaxation time. It acts according to the equation:

MXY =MXYo e-t/T2 (2)

where MXYo is the equilibrium transverse magnetization. Both the T1 and T2 processes occur simultaneously, provided T2 is less than or equal to T1.

B. Fundamental Principles

Four basic principles of imaging are the magnetic field gradient, frequency encoding, back projection imaging, and slice selection. First, a magnetic field gradient is a variation in the magnetic field with respect to position (Fig. 4). In MRI, the most useful type of gradient is a one-dimensional linear magnetic field gradient, and is symbolized as Gx, Gy, or Gz. The magnetic field gradient is used to image the positions of the regions of spin.

Fig. 4 Magnetic field gradient

Next, frequency encoding is the basis behind all MR imaging. If a linear magnetic field gradient is applied to the isocenter of a magnet, where (x,y,z) = (0,0,0), the three regions of spin each experience different magnetic fields. This creates an NMR spectrum with multiple signals, whose amplitude is proportional to the number of spins in a plane normal to the gradient. Applying the linear magnetic field gradient to obtain the NMR spectrum is known as frequency encoding.

Back projection imaging then utilizes this spectrum to produce the image. To do so, an object is placed in a magnetic field, and a one-dimensional field gradient is applied at various angle intervals. The NMR spectrum of each gradient is recorded and backprojected through computer space. The image is seen once the background intensity is suppressed.

Finally, to fully utilize this technique, the spins must be imaged in thin slices. Slice selection is the selection of spins in a plane through an object. This is done by applying a one-dimensional, linear magnetic field gradient during the period that the RF pulse is applied. To create the backprojection image, first an apodized sinc pulse-shaped 90o pulse and a slice selection gradient are applied concurrently. Once this gradient is turned off, a frequency encoding gradient is turned on. Fourier transformation is then used to produce the frequency domain spectrum. This process is shown in Fig. 5. Again, backprojection is used to produce the image from the spectrum. Although this is a well-developed technique, Fourier transform imaging methods are usually used in most imaging machines.

Fig. 5 Backprojection imaging sequence timing diagram

C. Fourier Transform Principles

Fourier transform tomographic imaging is the most commonly used MRI method utilized today. It uses another type of magnetic field gradient, called a phase encoding gradient, in addition to the slice selection and frequency encoding gradients. The phase encoding gradient is used to give a specific phase angle to the transverse magnetization vector. It acts in much the same way as the frequency encoding gradient when turned on. In this case each transverse magnetization vector has a unique Larmor frequency. However, the phase encoding gradient is unique in that when it is turned off, each transverse magnetization vector is identical. The phase angle of each vector, measured when the phase encoding gradient is off, is the angle between a reference axis and the magnetization vector.

Fourier tomographic imaging is done using a specific sequence, utilizing radio frequency, magnetic field gradients, and signals. A simplified Fourier transform imaging sequence includes a 90o slice selective pulse, a slice selection gradient pulse, a phase encoding gradient pulse, a frequency encoding gradient pulse, and a signal. The magnitude and duration of the magnetic field gradients are represented by the pulses. A typical imaging sequence would start by turning on the slice selection gradient, while applying the slice selection RF pulse. Once the pulse is done, the slice selection gradient turns off, and the phase encoding gradient is turned on. After that is done, the frequency encoding gradient is turned on. At this point, a signal is recorded. This process is repeated 128 to 256 times, varying the magnitude of the phase encoding gradient each time, in order to obtain sufficient data, free induction decays or signals, for creating an image.

Before actually creating the image, however, the signals must be Fourier transformed. This is done first in the direction in which the spins are located to extract the frequency domain information. Then it is done in the phase encoding direction to obtain information about the spin locations in that direction. The FT data finally becomes an image when the intensities of the data peaks are converted into intensities of pixels. This creates a tomographic image, shown in Fig. 6.

Fig. 6 Tomographic image created from Fourier transformed data

III.  IMAGING HARDWARE

A. Magnets

The magnet is the most expensive part of the imaging hardware. There are three types used for MR imaging. These are superconducting, resistive, and permanent. A superconducting magnet is the strongest of the three. In such an electromagnet, current flows in a circular direction in a coil of wire in order to create a magnetic field. The wire used in the magnet is superconducting wire. This means that it has nearly zero resistance when cooled to a temperature close to absolute zero, achieved by placing it in either liquid helium or liquid nitrogen (Fig. 7). The current in the superconducting wire continues to flow as long as it is kept at the specific temperature. Figure 8 shows an actual superconducting magnet without the housing.

Fig. 7 Internal view of magnetic resonance imaging scanner

Fig. 8 Superconducting magnet without housing

Resistive magnets are also electromagnets; however, they are cooled by the air. This causes a greater resistance to current. This in turn produces a weaker magnetic field. The third type of magnet, permanent, is not an electromagnet, but rather it is made of solid magnetic material. As expected, a permanent magnet creates the weakest magnetic field. Nevertheless, permanent magnets are useful in Open MR scanners due to the fact that they can be arranged in any position. This eliminates the need for the patient to be surrounded by the magnet.

B. Coils

Two important types of coils are gradient coils and RF coils. The gradient coil creates gradients in the equilibrium magnetic field, Bo (Fig. 9). This is possible because they are room temperature coils in a specific configuration. The RF coils, on the other hand, create the B1 magnetic field. This is the same as the magnetic field produced by alternating current through the coil at the Larmor frequency. In addition, RF coils detect transverse magnetization.

Fig. 9 Gradients in an equilibrium magnetic field

An RF coil can be of the transmit and receive, receive only, or transmit only type. The transmit and receive RF coil transmits the B1 field and receives the RF signal from the spins of the imaged object. The transmit only and receive only coils can be used in combination to perform the same tasks. A variety of coils are needed for imaging to compensate for different imaging needs.

An imaging coil consists of both inductive and capacitive elements. This allows it to efficiently store energy. In other words, the inductance and capacitance allow it to resonate. The frequency (ν) at which the imaging coil resonates is determined by the inductance (L) and capacitance (C) as shown in (3).

(3)

There is a multitude of other coils used for specific patients or organs. They include surface coils, bird cage coils, saddle coils, phased-array coils, and litz coils.

IV.  ADVANCED IMAGING TECHNIQUES

Newer imaging techniques are continually being developed. A few of the latest techniques will be discussed here.

A. Volume Imaging

Volume imaging, also called 3-D imaging, obtains magnetic resonance data from a volume instead of a tomographic slice. In other words, a group of slices is used, as seen in Fig. 10. As with previous techniques, an RF pulse and gradient are used; however, the gradient only rotates in the imaged volume. Next, phase encoding gradients in two dimensions are varied between their maximum and minimum values. Finally, the frequency encoding gradient is applied to obtain the signal. Basically this technique is similar to others, except it uses a volume, or thick slices.

Single Slice Contiguous Slices