MRI Techniques for Orthopedic Imaging

Chapter 1: MRI Techniques for Orthopedic Imaging

1.4 - Spin Echo Techniques

1.4.1 - Conventional Spin Echo

A typical pulse sequence timing diagram for a spin echo sequence is shown in Figure 1.15.


Pulse sequence timing diagram

In this sequence, a 90? excitation RF pulse is first played out, followed by a 180? refocusing pulse at a time TE/2 later. For 2D imaging, the 90? excitation pulse is slice-selective, and creates precessing components of magnetization in the transverse plane throughout the entire slice. Immediately following the excitation pulse, all the protons in the slice precess at the same frequency, and have the same phase. These protons begin gradually to precess at resonant frequencies determined by their particular magnetic environments, and the total signal decays in intensity as these components fall farther out of step with each other over time*. The envelope describing this decay is characterized by the time constant, T2* (See Section 1.3.2), resulting from a combination of T2 and T2' effects.


In a spin echo sequence, a 180? pulse is used to "refocus" the static T2' effects. This means that the magnitude of the echo formed at TE is determined solely by the intrinsic T2 of the tissue, and is not affected by main field imperfections, or susceptibility gradients in the tissue. This is accomplished in the following way: After the 90? excitation pulse, the transverse magnetization begins to dephase according to T2* as described above, with the slower components falling behind the faster components. The 180? refocusing pulse instantaneously reverses the order of the magnetic components, such that the faster components are suddenly repositioned behind the slower as shown in Figure 1.16.

Figure 1.16 (A) The transverse magnetization is dephasing due to differing resonance frequencies of the protons contributing to the MR signal. (B) A 180 degree RF pulse is used to flip the spins around the x-axis, and back into the transverse plane. This reverses the order of the magnetization components, such that the faster components are now behind the slower ones. (C) The faster magnetization components catch up with the slower components, creating what is known as a "Hahn echo".

Because the static effects discussed above result in a constant increment or decrement to the precession frequency of the protons, the slower components will still be slower than the faster components, and will begin to lose their lead on the faster components. Assuming that the individual protons are not able to move between regions of constant field strength during the time TE†,the faster magnetic components "catch up with" the slower components exactly at time TE such that a "Hahn echo" is formed (i.e. the transverse magnetization attains a second maximum). In spin echo imaging, this echo is collected as the MR signal and is written into k-space as a single line of information. In this way, the T2' effects are virtually eliminated from the image . T2 effects are not static throughout the excitation-to-acquisition time, and are therefore not refocused by the 180? pulse. The longer the chosen TE, the more T2 decay that occurs, and the smaller the MR signal at TE. TEs of 50ms-80ms with long TRs are used routinely in orthopedic imaging to provide T2-weighted images. T2-weighted images are often used with fat suppression to identify areas of bone marrow edema and abnormal fluid collections or fluid in tendon or fibrocartilage tears.

For 2D imaging sequences, the 90? pulse is used to select the desired 2D plane, or "slice" for imaging. This is accomplished by applying a gradient in the slice direction while the RF pulse is played out. The slice-select gradient maps the spins at each location in the slice direction to a particular frequency. The RF pulse is shaped such that it has a particular bandwidth and profile that corresponds to the range of frequencies in the desired slice (Figure 1.17).

Figure 1.17
Figure showing a typical RF pulse profile. The slice thickness is controlled by the range of frequencies contained in the RF excitation pulse in combination with the strength of the slice-selective gradient, Gz.

In spin echo imaging, the excitation-to-acquisition step described above is repeated until all the lines of k-space are acquired. A delay time must be included after each echo is acquired to allow recovery of the magnetization along the +z axis via T1 processes before the next excitation pulse. The time between excitations is called TR, and one echo is acquired per slice per TR. Each echo is encoded with partial information about the spatial distribution of protons in the y-direction. This is achieved by applying a "phase-encoding" gradient that causes the transverse magnetization to slightly dephase. The amount of dephasing that results for the applied gradient amplitude and duration reveals a clue to the spatial distribution of protons along the y-direction. We refer to this process of encoding the echo with a specific amount of dephasing as encoding with a particular ky value. In each subsequent TR, the amount of dephasing is increased step-wise by increasing the phase encode gradient amplitude, until enough complementary information about the y-distribution of protons is acquired to create an image, i.e. until all required values of ky are acquired. The number of values for ky that is required is equal to the desired matrix size of the image in the y-direction.

During the acquisition of each echo, a readout gradient is applied to encode the MR signal with spatial information about the distribution of protons along the x-direction. The echo is sampled at discrete points separated by a constant time interval ("sampling" time). The rate at which the signal is sampled is determined by the chosen RBW (See Section 1.2.4). Mathematically, the collection of information about the x and y distributions of protons is equivalent. Each phase-encode step encodes the echo with a single piece of information about the y distribution of the protons, i.e. a specific value of ky is encoded. Similarly, each single point collected during the acquisition window yields an equivalent piece of information about the x distribution, i.e. each data point is encoded with a different value of kx. Each single data point is therefore encoded with a pair of specific values of kx and ky. It is necessary that we obtain data for all combinations of (kx, ky) within the specified range in order to reconstruct the image. By convention, we hold ky constant for each echo, and collect all the kx data for that particular ky.

The degree of recovery of the longitudinal magnetization between each excitation is determined by how long a TR is chosen and by the T1 of the tissues. The shorter the TR relative to T1, the less magnetization will be available on the longitudinal axis at the time of the next excitation pulse. Hence, short-TR pulse sequences (TR = 500 – 800 ms) with short TE (TE = 10 - 20 ms) provide T1-weighted images. Figure 1.18 shows an example of a T1-weighted SE knee image. Selecting a long TR and a short TE yields an image that is proton-density-weighted (PD-weighted).


Figure 1.18

In orthopedic imaging, T1-weighted images are used mainly for anatomic detail because of their high SNR, and soft tissue characterization. PD-weighted spin echo images can be obtained with no additional scan time when a T2-weighted image is obtained using the Double-Echo technique discussed below.

The relationship of signal intensity to TE and TR in a spin echo sequence is expressed as:



(1.3)

where is the number of protons. For , the 3rd term is essentially unity (i.e. no T1-weighting), for , the 2nd term is close to unity and we have little T2-weighting.

 

For 2D imaging of a single slice, there exists some time between TE and TR where the MR scanner is neither playing out gradient or RF pulses, nor collecting an MR signal. This time can be used to excite and acquire signal from other slices if more than one slice is desired. For a spin echo sequence, the maximum number of slices which can be excited in a single acquisition is determined by the chosen TR and TE: NUMBER OF SLICES = TR/( TE + T0 ) (1.4)

In practice, the overhead time, T0 varies according to the options selected. It is larger, for example, if Fat Sat is chosen. The reason is that execution of the additional RF pulses required for Fat Sat takes additional time. In addition to the time required to play out RF and gradient pulses, the specific absorption rate (SAR) of RF power must be considered. The SAR limit prevents us from heating the patient's tissue through excessive deposition of RF power. For every pulse sequence and patient weight chosen, the system computes the maximum number of slices per TR that can be permitted. If more slices are prescribed than will fit into the TR period, the Signa scanner automatically puts the slices into concatenated multiple acquisition passes.


* This decaying MR signal is called a FID. In conventional spin echo sequences, the acquisition window is not open during the FID, but in gradient echo sequences, the FID is collected and used to reconstruct an image.

† In fact, diffusion of water molecules through these regions of differing field strength causes a small net loss of magnetization that is not recovered by the 180 deg refocussing pulse. A discussion of the effects of diffusion in these pulse sequences is beyond the scope of this chapter but is covered in detail in Woessner DE. Effects of diffusion in NMR spin-echo experiments. J. Chem. Phys. 34(6):2057-2061, 1961.

†† Because the T2' effects are really only zero at the center of the echo and the acquisition window has a finite length during which rephasing and dephasing are occuring, there is still some dephasing from inhomogeneities in spin echo and FSE images, but these are greatly reduced from GRE.