Great research starts with great data.

Learn More
More >
Patent Analysis of

Nuclear magnetic resonance refocusing pulses for inhomogeneous magnetic fields

Updated Time 12 June 2019

Patent Registration Data

Publication Number

US10001578

Application Number

US14/272325

Application Date

07 May 2014

Publication Date

19 June 2018

Current Assignee

SCHLUMBERGER TECHNOLOGY CORPORATION

Original Assignee (Applicant)

SCHLUMBERGER TECHNOLOGY CORPORATION

International Classification

G01V3/00,G01V3/14,G01V3/32,G01V3/38

Cooperative Classification

G01V3/14,G01V3/32,G01V3/38

Inventor

MANDAL, SOUMYAJIT,KOROLEVA, VAN D. M.,BORNEMAN, TROY,HURLIMANN, MARTIN D.

Patent Images

This patent contains figures and images illustrating the invention and its embodiment.

US10001578 Nuclear magnetic resonance refocusing pulses 1 US10001578 Nuclear magnetic resonance refocusing pulses 2 US10001578 Nuclear magnetic resonance refocusing pulses 3
See all images <>

Abstract

A method for applying a nuclear magnetic resonance (NMR) sequence is described herein. The method includes applying a series of refocusing pulses to a substance within an inhomogeneous static magnetic field. Each refocusing pulse in the series of refocusing pulses includes a central axis, a total pulse duration greater than t180, and at least five segments (e.g., 5, 7, 11, 20, and 21). The phase of each segment is substantially anti-symmetric about the central axis of the refocusing pulse. In a more particular embodiment, the phase of each segment is also symmetric about the central axis of the refocusing pulse and the five segments include a substantially constant amplitude.

Read more

Claims

1. A method for applying a nuclear magnetic resonance (NMR) sequence, the method comprising:applying a series of refocusing pulses to a substance within an inhomogeneous static magnetic field, wherein each refocusing pulse in the series of refocusing pulses comprises: a central axis; a total pulse duration greater than t180; and at least five segments that each comprise a phase, wherein the phase of each segment is substantially anti-symmetric about the central axis of each refocusing pulse.

2. The method of claim 1, wherein the phase of each segment is substantially symmetric about the central axis of each refocusing pulse.

3. The method of claim 2, wherein each segment has a phase of either 0 degrees or 180 degrees.

4. The method of claim 1, wherein the phase of each segment is within 36 degrees of a perfect anti-symmetry.

5. The method of claim 1, wherein each refocusing pulse comprises a substantially constant amplitude.

6. The method of claim 5, wherein the amplitude of each segment is within 10% of an average amplitude of the refocusing pulse.

7. The method of claim 1, wherein the at least five segments comprise segment durations between 0.08 times t180 to 1.1 times t180.

8. The method of claim 1, wherein the total pulse duration is greater than 1.414 times t180.

9. The method of claim 1, further comprising: applying an excitation pulse to the substance, wherein the application of the excitation pulse is performed prior to the application of the series of refocusing pulses.

10. The method of claim 1, where the excitation pulse and the series of refocusing pulses comprise a Carr, Purcell, Meiboom and Gill (CPMG) sequence.

11. The method of claim 1, wherein the substance is a formation and the series of refocusing pulses are applied to the formation using a wellbore tool.

12. A nuclear magnetic resonance (NMR) system comprising: a coil for applying a NMR pulse sequence to a substance; a NMR transmitter coupled to the coil; a processor; anda memory storing instructions executable by the processor to provide the NMR pulse sequence to the NMR transmitter, wherein the NMR pulse sequence comprises a series of refocusing pulses and each refocusing pulse in the series of refocusing pulses comprises: a central axis; a total pulse duration greater than t180; and at least five segments that each comprise a phase, wherein the phase of each segment is substantially anti-symmetric about the central axis of each refocusing pulse.

13. The NMR system of claim 12, wherein the NMR system is part of a wellbore tool for investigating formations.

14. The NMR system of claim 12, wherein the phase of each segment is substantially symmetric about the central axis of each refocusing pulse.

15. The NMR system of claim 14, wherein each segment has a phase of either 0 degrees or 180 degrees.

16. The NMR system of claim 12, wherein each refocusing pulse comprises a substantially constant amplitude.

Read more

Claim Tree

  • 1
    1. A method for applying a nuclear magnetic resonance (NMR) sequence, the method comprising:
    • applying a series of refocusing pulses to a substance within an inhomogeneous static magnetic field, wherein each refocusing pulse in the series of refocusing pulses comprises: a central axis
    • a total pulse duration greater than t180
    • and at least five segments that each comprise a phase, wherein the phase of each segment is substantially anti-symmetric about the central axis of each refocusing pulse.
    • 2. The method of claim 1, wherein
      • the phase of each segment is substantially symmetric about the central axis of each refocusing pulse.
    • 4. The method of claim 1, wherein
      • the phase of each segment is within 36 degrees of a perfect anti-symmetry.
    • 5. The method of claim 1, wherein
      • each refocusing pulse comprises
    • 7. The method of claim 1, wherein
      • the at least five segments comprise
    • 8. The method of claim 1, wherein
      • the total pulse duration is greater than 1.414 times t180.
    • 9. The method of claim 1, further comprising:
      • applying an excitation pulse to the substance, wherein the application of the excitation pulse is performed prior to the application of the series of refocusing pulses.
    • 10. The method of claim 1, where the excitation pulse and the series of refocusing pulses comprise
      • a Carr, Purcell, Meiboom and Gill (CPMG) sequence.
    • 11. The method of claim 1, wherein
      • the substance is a formation and the series of refocusing pulses are applied to the formation using a wellbore tool.
  • 12
    12. A nuclear magnetic resonance (NMR) system comprising:
    • a coil for applying a NMR pulse sequence to a substance
    • a NMR transmitter coupled to the coil
    • a processor
    • anda memory storing instructions executable by the processor to provide the NMR pulse sequence to the NMR transmitter, wherein the NMR pulse sequence comprises a series of refocusing pulses and each refocusing pulse in the series of refocusing pulses comprises: a central axis
    • a total pulse duration greater than t180
    • and at least five segments that each comprise a phase, wherein the phase of each segment is substantially anti-symmetric about the central axis of each refocusing pulse.
    • 13. The NMR system of claim 12, wherein
      • the NMR system is part of a wellbore tool for investigating formations.
    • 14. The NMR system of claim 12, wherein
      • the phase of each segment is substantially symmetric about the central axis of each refocusing pulse.
    • 16. The NMR system of claim 12, wherein
      • each refocusing pulse comprises
See all independent claims <>

Description

TECHNICAL FIELD

This disclosure relates to nuclear magnetic resonance (NMR) measurements, and more particularly to NMR measurements that are performed in inhomogeneous static magnetic fields.

BACKGROUND

Nuclear magnetic resonance (NMR) can be used to determine properties of a substance. Oil and gas field tools use NMR systems to investigate the properties of earth formations, such as the porosity of formations or the composition of fluids within the formations. In one example, the NMR system is lowered into a wellbore that traverses a formation. The NMR system can be lowered into the wellbore as part of a wireline or drilling string. The NMR system is used to investigate the properties of the formation adjacent to the wellbore. The system includes a magnet for applying a static magnetic field to the formation adjacent to the wellbore. The system also includes a coil for applying an oscillating magnetic field to the formation adjacent to the wellbore at a particular frequency. The oscillating field is composed of a sequence of radio frequency (RF) pulses that tip the magnetization of the atomic nuclei within the formation away from an initial magnetization produced by the magnet. The sequence of pulses and the static magnetic field interact with the nuclei in a manner such that a NMR signal composed of “echoes” is generated within at least a portion of the formation. The NMR signal within the formation is detected using the coil and used to determine the properties for the formation.

The static magnetic field produced by the magnet and applied to the formation is inhomogeneous because the field decreases in strength as a function of distance from the magnet. Because of this inhomogeneous field, the NMR signal is produced in a small portion of the formation because the Larmor frequency condition (or resonance condition) is met in a small portion of the field. As is known in the art, the Larmor frequency of the atomic nuclei depends on the strength of the static magnetic field according to the following relationship:

ω0=γ×B0

where B0 is the strength of the static magnetic field, γ is the gyromagnetic ratio of the atomic nuclei of interest, and ω0 is the frequency of the resonant signal that is produced by the atomic nuclei (the Larmor frequency). The small portion of the formation (or substance) that produces the resonant signal is referred to as a “shell” or “slice.” In some cases, the shell has a volume of 10−4 m3 (100 cc). These small shell sizes lead to a degradation in the ratio of signal power to noise power (SNR).

The shell sizes and resulting SNR can be increased by increasing the bandwidth of the RF pulses in the NMR sequence. One approach to expand the bandwidth of RF pulses is to increase the amplitude of the RF pulses in the NMR sequence by a factor, while simultaneously decreasing the duration of the pulses by the same factor. The resulting bandwidth and SNR will then increase by this factor. This solution, however, requires increasing RF field strength and average power consumption, which can be problematic for power constrained applications, such as NMR well logging applications.

Another approach is to modulate the amplitude and/or phase within the RF pulses in a particular manner. For example, the bandwidth of RF pulses can be increased by using composite, shaped, chirped, or adiabatic pulses, as compared to conventional rectangular pulses that are designed to have constant amplitude and phase. Progress has been made in the development of numerically-optimized broadband RF pulses that are robust with respect to variation of the RF field strength and other constraints. These developments have taken advantage of the availability of new algorithms for pulse sequence design based on methods of optimal control theory (OCT), which make it feasible to find new pulse sequences in an efficient manner in a high-dimensional parameter space. However, such work has been largely focused on improving the pulse fidelity, i.e., ability to approximate a given rotation operator over the desired bandwidth, and not on SNR. In addition, such pulses are typically much longer than a conventional pulse. This increase in pulse duration has the undesirable consequence that it entails an increase of the minimum echo spacing and an increase in the power consumption per RF pulse, often resulting in no net increase of SNR per unit time. This fact is of particular concern for mobile and field applications, such as NMR well logging applications.

SUMMARY

Illustrative embodiments of the present disclosure are directed to a method for applying a nuclear magnetic resonance (NMR) sequence. The method includes applying a series of refocusing pulses to a substance within an inhomogeneous static magnetic field. Each refocusing pulse in the series of refocusing pulses includes a central axis, a total pulse duration greater than t180, and at least five segments (e.g., 5, 7, 11, 20, and 21). The phase of each segment is substantially anti-symmetric about the central axis of the refocusing pulse. In a more particular embodiment, the phase of each segment is also symmetric about the central axis of the refocusing pulse and the five segments include a substantially constant amplitude. In some embodiments, the refocusing pulses increase signal to noise ratio (SNR) when the pulses are applied to an inhomogeneous static magnetic field, while also providing for efficient power consumption.

Various embodiments of the present disclosure are also directed to a NMR system that includes: (i) a coil for applying a NMR pulse sequence to a substance, (ii) a NMR transmitter for generating the NMR pulse sequence coupled to the coil, (iii) a processor, and (iv) a memory for storing instructions executable by the processor. The instructions include providing the NMR pulse sequence to the NMR transmitter. The NMR pulse sequence includes a series of refocusing pulses. Each refocusing pulse in the series of refocusing pulses includes a central axis, a total pulse duration greater than t180, and at least at least five segments that each include a phase. The phase of each segment is substantially anti-symmetric about the central axis of each refocusing pulse.

Illustrative embodiments of the present disclosure are further directed to a method for optimizing a NMR pulse sequence for use in an inhomogeneous magnetic field. The method includes performing an optimization for the refocusing pulses to maximize signal to noise ratio for the NMR pulse sequence. The optimization is performed using a number of constraints that include: (i) each refocusing pulse includes a total pulse duration greater than t180, (ii) each refocusing pulse includes at least five segments, and (iii) each segment includes a phase that is substantially anti-symmetric about the central axis of each refocusing pulse.

BRIEF DESCRIPTION OF THE DRAWINGS

Those skilled in the art should more fully appreciate advantages of various embodiments of the disclosure from the following “Description of Illustrative Embodiments,” discussed with reference to the drawings summarized immediately below.

FIG. 1 shows a method of applying an NMR pulse sequence in accordance with one embodiment of the present disclosure;

FIG. 2 shows a NMR pulse sequence in accordance with one embodiment of the present disclosure;

FIG. 3 shows phase alternating refocusing pulses with anti-symmetry in accordance with various embodiments of the present disclosure;

FIGS. 4A and 4B show phase alternating refocusing pulses with symmetry and anti-symmetry in accordance with various embodiments of the present disclosure;

FIGS. 5A-5G show phase alternating refocusing pulses with symmetry and anti-symmetry in accordance with further embodiments of the present disclosure;

FIGS. 6A-6D show signal to noise ratio (SNR) as a function of pulse length for refocusing pulses in accordance with various embodiments of the present disclosure;

FIGS. 7A and 7B show SNR and SNR per unit time for phase alternating refocusing pulses as a function of refocusing pulse length in accordance with various embodiments of the present disclosure;

FIGS. 8A-8C show an experimentally measured spectrum of asymptotic echoes produced by NMR sequences in accordance with various embodiments of the present disclosure;

FIG. 9 shows measured amplitudes of echoes produced by NMR sequences in accordance with various embodiments of the present disclosure;

FIG. 10 shows measured and simulated nutation curves for NMR sequences in accordance with various embodiments of the present disclosure;

FIG. 11 shows a NMR system in accordance with one embodiment of the present disclosure;

FIG. 12 shows a logging-while-drilling (LWD) system in accordance with one embodiment of the present disclosure; and

FIG. 13 shows a LWD NMR logging module in accordance with one embodiment of the present disclosure.

DESCRIPTION OF ILLUSTRATIVE EMBODIMENTS

Illustrative embodiments of the present disclosure are directed to a method for applying a nuclear magnetic resonance (NMR) sequence. The method includes applying a series of refocusing pulses to a substance within an inhomogeneous static magnetic field. Each refocusing pulse in the series of refocusing pulses includes a central axis, a total pulse duration greater than t180, and at least five segments (e.g., 5, 7, 11, and 21). The phase of each segment is substantially anti-symmetric about the central axis of the refocusing pulse. In a more particular embodiment, the phase of each segment is also symmetric about the central axis of the refocusing pulse (e.g., each segment has a phase of either 0 degrees or 180 degrees) and the at least five segments include a substantially constant amplitude. In some embodiments, the refocusing pulses increase signal to noise ratio (SNR) when the pulses are applied to an inhomogeneous static magnetic field, while also providing for efficient NMR transmitter power consumption. Such pulses are referred to herein as “phase alternating refocusing pulses.” Details of various embodiments are discussed below.

FIG. 1 shows a method 100 of applying a NMR pulse sequence in accordance with one embodiment of the present disclosure. The method 100 includes applying a static magnetic field (B0) to an area of interest within a substance 102, such as a reservoir formation containing hydrocarbons. In some embodiments, an inhomogeneous magnetic field is applied to the substance. The term “inhomogeneous” should be considered in the context of the NMR art. Many NMR well logging tools deploy inhomogeneous static magnetic fields due to the limitations and constraints of a wellbore environment. In this context, an inhomogeneous static magnetic field is a static magnetic field that varies in intensity or direction within an area of interest of a substance. In one example, an inhomogeneous static magnetic field within a shell of the substance may vary in intensity by a value approximately equal to or greater than a nominal RF magnetic field (B1) amplitude of an NMR pulse sequence applied to the area of interest. As explained above, the small portion of the substance that produces a resonant signal is referred to as a “shell” or “slice.” The inhomogeneous static field variation within a shell will be proportional to the intensity of B1. For example, with rectangular pulses the variation is approximately 2B1. As B1 increases the shell becomes thicker, so more variation in the static field occurs within the shell.

Once the static magnetic field is applied to the area of interest, the method includes applying a NMR pulse sequence to the area of interest within the substance 104. The NMR pulse sequence includes a series of radio frequency (RF) pulses. FIG. 2 shows one example of a NMR pulse sequence 200. The pulse sequence 200 include an excitation pulse 206 followed by a series of phase alternating refocusing pulses 208. In one specific embodiment, the excitation pulse 206 has a length of t90 and the phase alternating refocusing pulses 208 have lengths greater than t180. The quantity t90 is defined as: t90=π/2(γ×B1), where γ is the gyromagnetic ratio of a nucleus of interest and B1 is the maximum amplitude of an applied circularly-polarized magnetic field in a rotating frame within the area of interest in the substance (e.g., in many cases, B1 is half the amplitude of the linearly-polarized magnetic field produced by a coil). The quantity t180 is defined as: t180=π(γ×B1)=2×t90. The excitation pulse 206 can be a rectangular excitation pulse or an “axis-matching excitation pulse,” as will be further described below. Larger SNR can be obtained by using axis-matching excitation pulses in combination with the phase alternating pulses described herein. Following the excitation pulse 206, each phase alternating refocusing pulse 208 produces a corresponding echo 210. The phase alternating refocusing pulses 208 and corresponding echoes 210 are repeated a number of times (e.g., N=10, N=100, N=1000). In one specific example, the NMR pulse sequence is a Carr, Purcell, Meiboom and Gill (CPMG) sequence.

The NMR pulse sequence 200 applied to the substance generates resonant signals within a portion of the substance (e.g., shell). The resonant signal is composed of the series of echoes 210. In accordance with the method 100 shown in FIG. 1, the resonant signal 210 generated by the NMR pulse sequence 200 is detected. This detected signal can be used to determine NMR properties for the substance (e.g., T1 relaxation time, T2 relaxation time, diffusion, and/or a correlation function of such properties). In turn, the NMR properties can be used to determine physical properties of the substance, such as the chemical composition of the substance and/or the presence of a solid phase, a liquid phase, and/or a gas phase within the substance. Furthermore, in NMR logging applications, this detected signal can be used to study the properties of rocks (e.g., porosity and pore size distribution) and the properties of fluids (e.g., viscosity and diffusion).

Various embodiments of the phase alternating refocusing pulses described herein increase signal to noise ratio (SNR) when the pulses are applied to an inhomogeneous static magnetic field, while also providing for efficient NMR transmitter power consumption. In a homogenous static magnetic field, the portion of the substance that produces a resonant signal (e.g., a shell) will be broad across an area of interest within the substance because the field (B0) is constant. In an inhomogeneous static magnetic field, the field changes across the area of interest and NMR pulse sequences at a particular frequency produce resonant signals in a limited region within the area of interest. These small shell sizes lead to a degradation in the ratio of signal power to noise power (SNR). The phase alternating refocusing pulses described herein provide for expanded “bandwidth” that produces a resonant signal in a larger portion of the area of interest and thus increases SNR. In various embodiments, the phase alternating refocusing pulses produce an increase in SNR without relying on an increase in the amplitude of the pulses in the NMR sequence. In this manner, the phase alternating refocusing pulses produce an increase in SNR, while also providing for efficient power consumption.

FIGS. 3, 4A, 4B, and 5A-5G show phase alternating refocusing pulses in accordance with various embodiments of the present disclosure. The refocusing pulses include a central axis 302 and at least five segments (e.g., 5, 7, 9, 10, 11, 20, and 21) that each comprise a phase (φ). For example, the refocusing pulse in FIG. 3 includes 20 segments, while, in FIGS. 4A and 4B, the refocusing pulses includes 11 segments and 17 segments, respectively. In addition, in various embodiments, the length of each segment is an integer multiple of 0.1 times t180.

The phase (φ) of each segment is substantially anti-symmetric about the central axis of each refocusing pulse. Each segment within the refocusing pulse meets the following condition: φ(t)=−φ(−t) (in modulo 2π), for a pulse centered about t=0. For example, in FIG. 3, segments 304 and 306 are equidistant with respect to the central axis 302 and thus include opposite phases. In this example, the phase of segment 304 is 3 radians and the phase of segment 306 is −3 radians. Perfect anti-symmetry produces the greatest SNR, but improved SNR can also be achieved using pulses with phases that are substantially anti-symmetric. A segment is substantially anti-symmetric about the central axis of the refocusing pulse when the phase of the segment is within 36 degrees of the perfect anti-symmetry. For example, if the phase of a segment is 90 degrees, the phase of a second segment (equidistant with respect to the central axis 302) is substantially anti-symmetric when the phase falls within 36 degrees of 270 degrees, which is the perfect anti-symmetry. Thus, the second segment is substantially anti-symmetric when it has a phase in a range between 234 and 306 degrees.

In some embodiments, the refocusing pulses include phases that are both substantially anti-symmetric and symmetric about the central axis 302 of the refocusing pulse. For example, in FIGS. 4A and 4B, the refocusing pulses include phases that are both anti-symmetric and symmetric about the central axes 302 of the refocusing pulses. To meet this condition, each segment within the refocusing pulses includes a phase of either 0 degrees or 180 degrees (π). For example, in FIG. 4A, segments 308 and 310 are equidistant with respect to the central axis 302 and thus both include phases of 0 degrees. Perfect anti-symmetry and symmetry (segments with either 0 or 180 degrees) produce the greatest SNR, but improved SNR can also be achieved using pulses with phases that are substantially anti-symmetric and symmetric. A segment is substantially anti-symmetric and symmetric about the central axis of the refocusing pulse when the phase of the segment is within 36 degrees of the perfect anti-symmetry and symmetry. For example, if the phase of a segment is 0 degrees, the phase of a second segment (equidistant with respect to the central axis 302) is substantially anti-symmetric and symmetric when the phase falls within 36 degrees of 0 degrees, which is the perfect anti-symmetry and symmetry (e.g., in a range between 324 and 36 degrees).

Illustrative embodiments of the refocusing pulses include total pulse durations greater than t180 (a sum in temporal duration of the segments). In some embodiments, the total pulse durations are greater than 1.414 (the square root of 2) times t180. For example, the refocusing pulse in FIG. 3 has a duration of 2.0 times t180, the refocusing pulse in FIG. 4A has a duration of 2.1 times t180, and the refocusing pulse in FIG. 4B has a duration of 3.1 times t180. Each segment of the refocusing pulses can have a variety of different durations. In some embodiments, each segment has a duration between 0.09 times t180 to 1.1 times t180. In FIG. 4A, the refocusing pulse includes segments with durations of 0.7 times t180, 0.2 times t180, and 0.1 times t180.

FIGS. 5A-5G show examples of seven more phase alternating refocusing pulses with phases that are both anti-symmetric and symmetric about a central axis of each refocusing pulse. These refocusing pulses range in duration between 1.3 times t180 (FIG. 5A) and 5.5 times t180 (FIG. 5G). The phase alternating refocusing pulses also have between 9 segments (FIG. 5A) and 23 segments (FIG. 5G).

In various embodiments, the refocusing pulses have substantially constant amplitude. For example, in one specific embodiment, each segment within the refocusing pulse has the same amplitude. Refocusing pulses with perfectly constant amplitude produce the greatest SNR (e.g. each pulse has an amplitude of 1 Gauss (G)), but improved SNR can also be achieved using pulses with refocusing pulses that include substantially constant amplitudes. A refocusing pulse has substantially constant amplitude when the amplitude of each segment is within 20% of the average amplitude of the pulse (e.g., when the average amplitude is 1.0 G, each segment is within the range of 0.8 G to 1.2 G). In various embodiments, the refocusing pulses have average amplitude between 1 G and 10 G.

Illustrative embodiments of the present disclosure are also directed to determining characteristics of phase alternating refocusing pulses, such as a resulting SNR for the refocusing pulses, when used in combination with various excitation pulses. The description below provides a review of relevant spin dynamics for an ensemble of uncoupled spins ½ placed in inhomogeneous B1 and B0 fields. The exemplary case considered herein includes a generalized CPMG-like sequence that consists of an initial excitation pulse followed by a long train of equally-spaced and identical refocusing pulses. The excitation and refocusing pulses can be simple rectangular pulses, complex composite pulses, or shaped pulses. The exemplary case considers the limit when the echo spacing (tE) is much shorter than the transverse relaxation time (T2) and also short enough to make diffusion effects negligible. In this case, the propagator from echo to echo for any spin can be approximated by an effective rotation R({circumflex over (n)}, θ). The rotation is characterized by an axis {circumflex over (n)}(ω0, ω1) and nutation angle θ(ω0, ω1) that depend on the Larmor frequency (ω0), the strength of the oscillating RF field, (ω1) (which is assumed to be much smaller than ω0), and the echo spacing tE. If the magnetization following the excitation pulse is {right arrow over (M)}(0+), then the magnetization at the nominal center of the k-th echo can be written as:

{right arrow over (M)}k=e−ktE/T2,effR({circumflex over (n)},kθ){{right arrow over (M)}(0+)}  (1)

The effective relaxation time T2,eff depends in general on both T1 and T2. In the limiting case of T1 is equal to T2, then T2,eff is equal to T2. The CPMG component can be identified as the component of the initial magnetization {right arrow over (M)}(0+) that is aligned with the axis {circumflex over (n)}. At the nominal echo time of the k-th echo, tk=ktE, this component is invariant under the multiple application of the rotation R and becomes independent of k except for relaxation. In contrast, the component of {right arrow over (M)}(0+) that is perpendicular to {circumflex over (n)} evolves from echo to echo. This component gives rise to the oscillating Carr-Purcell (CP) part of the signal. For a sample substance in inhomogeneous fields, the CP components from different spins quickly dephase from each other and do not contribute to the detected signal after only a few echoes. The overall signal for the CPMG sequence in inhomogeneous fields shows therefore an initial transient with contributions from both the CPMG and CP components before it enters the asymptotic regime controlled by the CPMG component. In this asymptotic regime, the spectrum of the echo is given by:

Sasy01)=({circumflex over (n)}·{right arrow over (M)}(0+)){circumflex over (n)}  (2)

The quantities on the right-hand side of equation 2 depend on ω0 and ω1. Here {circumflex over (n)}=nx+iny is the component of {circumflex over (n)} that is transverse to B0. In the time domain, the shape of the echo in the asymptotic limit is:

sasy(t)=∫∫01f01)e0t({circumflex over (n)}·{right arrow over (M)}(0+))/{circumflex over (n)},  (3)

where f(ω01) describes the distribution of ω0 and ω1 over the sample.

In a practical implementation of a CPMG measurement, phase cycling of the initial excitation pulse is used. In its simplest implementation, the signals of two sets of CPMG measurements are subtracted from each other. In the second measurement, the phase of the excitation pulse is increased by π. This procedure eliminates various detrimental effects that can interfere with the experiment, including pulse ringing from the refocusing pulses, DC offsets in the acquisition electronics, and/or contributions from undesired coherence pathways that originate from T1 recovery during the train of the refocusing cycles. Standard phase cycling inverts the transverse component of {right arrow over (M)}(0+), while leaving the longitudinal component unaffected. The resulting signal after phase cycling is:

sasy(pc)(t)=∫∫01f01)e0t({circumflex over (n)}·{right arrow over (M)}(0+)){circumflex over (n)}.  (4)

Equations 1 through 4 provided above are general and also apply to CPMG-like sequences where the excitation and refocusing pulses are replaced by more complicated pulses, such as the phase alternating refocusing pulses described herein. For an experimental set-up with a given distribution f(ω0, ω1), the amplitude of the CPMG signal is controlled by the behavior of {right arrow over (M)}(0+) and {circumflex over (n)} over a range of relevant offset frequencies. The excitation pulse determines {right arrow over (M)}(0+), while the refocusing pulse determines {circumflex over (n)}. An ideal or perfect excitation pulse generates {right arrow over (M)}(0+)={circumflex over (x)} over the entire (ω0, ω1) found within the sample. Similarly, an ideal refocusing pulse generates {circumflex over (n)}={circumflex over (x)} at all frequencies. According to equation 4, this is the optimal choice of the CPMG sequence with phase cycling as it optimizes ({circumflex over (n)}·{right arrow over (M)}(0+)). However, inspection of the more general expression in equation 3 shows that, for a given refocusing pulse, the largest asymptotic signal is obtained with an excitation pulse that maximizes ({circumflex over (n)}·{right arrow over (M)}(0+)) rather than ({circumflex over (n)}·{right arrow over (M)}(0+)). The optimal excitation pulse for CPMG applications therefore transforms {circumflex over (z)}→{circumflex over (n)}. In other words, the excitation pulse generates an initial magnetization that is aligned with a refocusing axis produced by a subsequent refocusing cycle. A refocusing cycle is the duration of a refocusing pulse plus the delay between the refocusing pulse and the next pulse in a pulse sequence. In contrast, many conventional systems align the initial magnetization with the transverse plane perpendicular to B0 (i.e., the z-axis), without regard for the refocusing axis produced by the refocusing pulses. Such excitation pulses are referred to herein as “axis-matching excitation pulses” or as “AMEX pulses” and are further described in U.S. Patent Application Publication No. 2013/0162247, published on Jun. 27, 2013, which is hereby incorporated by reference in its entirety. Given that |n|≤1, an ideal axis-matching pulse will generate a CPMG signal that is larger than or equal to a broadband 90°y; pulse that transforms {circumflex over (z)}→{circumflex over (x)}. Axis-matching pulses compensate for the limitations and imperfections of refocusing pulses. Axis-matching pulses are matched to a particular refocusing axis ({circumflex over (n)}) and are, therefore, effective for a given refocusing pulse and echo spacing tE.

Various embodiments of the present disclosure are directed to determining SNR for phase alternating refocusing pulses when used in combination with various excitation pulses. SNR with matched filtering can be determined according to the following relationship:

SNR=-Tacq/2Tacq/2dtsasy(t)2N02.(5)

In equation 5, N02 is the mean squared value of the noise, which is assumed to be additive, white, and completely uncorrelated with the signal, and Tacq is the duration of the acquisition window. For simplicity, the static magnetic field can be characterized by a constant gradient (which results in a uniform distribution of resonant frequencies), while the oscillating RF field is uniform. In this case equation 5 can be rewritten as:

SNR=-dtsasy(t)w(t)2N02=-dω0Sasy(ω0)*W(ω0)2N02.(6)

In equation 6, w(t) is a “window” function that is equal to unity within the acquisition window (|t|<Tacq/2) and is zero otherwise. In addition, the Fourier transform of w(t) is denoted by W(ω0)=(Tacq/2)sinc(ω0Tacq/(2π), and the ‘*’ symbol denotes convolution. The windowing operation smoothens or “blurs” frequency domain structure on scales that are smaller than ˜1/Tacq Parseval's theorem has also been used, which states that the power of a signal is the same in both the time and frequency domains. A similar approach can be used for more general (ω0, ω1) distributions.

The description below calculates SNR for two types of excitation pulses. The first type are perfect 90°y pulses, which transfer {circumflex over (z)}→{circumflex over (x)} and thus generate {right arrow over (M)}(0+)={circumflex over (x)} at all frequencies of interest. The second type are ideal axis-matching pulses, which transfer {circumflex over (z)}→{circumflex over (n)} and thus generate {right arrow over (M)}(0+)={circumflex over (n)} at all frequencies of interest.

For a given refocusing pulse, the ideal axis-matching pulses will produce better results than the perfect 90°y pulses. For practical axis-matching pulses, the performance lies between the perfect 90°y pulses and the ideal axis-matching pulses and this is the reason for considering the two limiting cases. The resulting SNR expressions are given by:

SNR(1)-dω0(nx(ω0)n^(ω0))*W(ω0)2,SNR(2)-dω0n^(ω0)*W(ω0)2.(7)

Note that one of the components of {circumflex over (n)} can be chosen to be positive without loss of generality. Thus, |nx| is used in the convolutions shown in equation 7. These expressions are further simplified for refocusing cycles that are anti-symmetric about the nominal center of the echo (t=0). In such cases, the refocusing axis has no y-component (ny=0). As a result, the asymptotic magnetization lies entirely along the x-axis ({circumflex over (n)}=nx):

SNR(1)-dω0nx2(ω0)*W(ω0)2,SNR(2)-dω0nx(ω0)*W(ω0)2.(8)

Equation 8 shows that the asymptotic SNR is approximately proportional to the integral of |nx|4 for perfect 90-degree excitation pulses, and |nx|2 for axis-matching excitation pulses. However, {circumflex over (n)} denotes a refocusing axis, so it follows that |{circumflex over (n)}|=1. Hence, |nx|≤1 and |nx|2≤|nx| at many or all frequencies. Therefore. the SNR achievable with axis-matching is greater than or equal to that achievable with perfect 90-degree excitation. The difference in performance is particularly marked for short refocusing pulses, which tend to be “imperfect” (with |nx| substantially smaller than 1). In practice, for very short refocusing pulses, axis-matching excitation pulses were found to more than double the asymptotic SNR compared to rectangular excitation pulses.

There is usually a trade-off between the robustness with respect to offset frequency and the overall duration of refocusing pulses. In many cases, longer refocusing pulses will have better performance. However, in many applications, CPMG-like sequences are advantageous because they produce a large number of echoes. For example, it is possible to average echoes to improve the SNR if the transverse relaxation time is much longer than the echo spacing (T2>>tE). In this case, short echo spacings are advantageous as it allows a larger degree of averaging. As a result, the asymptotic SNR per unit time is a more relevant performance metric for CPMG-like sequences. In power units, this quantity is proportional to the SNR per echo multiplied by the rate at which echoes are generated:

SNRtimeSNRechotE(9)

In equation 9, tE is the echo period. Thus, SNRtime ∝1/tE in power units. In general, longer refocusing pulses will result in longer echo periods because echoes are acquired in the gaps between echoes. In addition, the duty cycle (tref/tE) of the NMR transmitter is often limited by heating, pulse droop, or for other reasons. Here tref is the refocusing pulse length. Both effects tend to reduce SNRtime as the refocusing pulse becomes longer. In this description, both SNR and SNRtime are used as figures of merit.

The SNR can be used to optimize the phase alternating refocusing pulses described herein. In certain embodiments, SNR for several different types of refocusing pulses can be determined and the refocusing pulses with the most desirable SNR can be selected for use in an NMR tool. In various embodiments of the optimization described herein, the phase alternating pulses are defined to be piecewise constant over a number of intervals (N) each of length (Δt) which yields an overall pulse length of NΔt. During each interval, the amplitude (A) and the phase (φ) of the oscillating RF field are set to constant values, such that the pulse is described as a list {Aj, φj}, where j runs from 1 to N. The GRAPE algorithm is known in the art and can be used for optimizing such pulses. Further details regarding the GRAPE algorithm are provded in N. Khaneja, et al., Optimal Control of Coupled Spin Dynamics: Design of NMR Pulse Sequences by Gradient Ascent Algorithms, J. Magn. Reson. 172, pp. 296-305 (2005).

A number of performance optimization approaches are possible. The approach pursued in the past was based on finding universal broadband pulses. This result is achieved by maximizing average fidelity over a given bandwidth. Fidelity is a relatively sensitive measure of how well the rotation axis and nutation angle of a pulse are matched to their desired values. Maximizing the fidelity of refocusing pulses amounts to finding pulses that have simultaneously a net rotation axis along {circumflex over (x)} and a net rotation angle of π. An average fidelity of one implies that a perfect π rotation about the x-axis is achieved over the entire bandwidth of interest. In this case, any echo spacing (tE) will yield identical results and the asymptotic echo amplitude will be maximized.

It is not possible to simultaneously obtain refocusing pulses with both the desired rotation axis and the rotation angle over large bandwidths. This results in solutions that form a compromise between these two desirable properties. Nonetheless, the SNR expressions shown in equations 7 and 8 reveal that, for a broadband application, performance is primarily affected by the direction of the rotation axis, not the net rotation angle. Accordingly, various embodiments of the optimization described herein focus on the direction of the axis of the effective rotation, rather than the fidelity. Such an optimization can be run by modifying the GRAPE algorithm to optimize any function of n=nx+iny, as described in greater detail in the APPENDIX below. In particular, the integrals of nx4 or nx2 can be optimized, which maximizes SNR assuming perfect 90y excitation pulses (such as the first type described above) or perfect axis-matching excitation pulses (such as the second type described above), respectively.

Various different constraints can be applied to the optimization. These constraints help reduce the parameter space and thus simplify the optimization. Some or all of the constraints described below can be used during the optimization.

In a first embodiment, the amplitudes of the segments within each phase alternating pulse can be kept constant (at Amax, the maximum allowed value). This constraint results in purely phase-modulated pulses and reduces the number of optimization variables by a factor of 2. This constraint simplifies generation of the pulse by NMR transmitter electronics, particularly in field applications where power-efficient nonlinear RF transmitters are common.

In a second embodiment, the phases of each segment can be kept symmetric with respect to the center of the refocusing pulse (e.g., central axis 302). A constant-amplitude pulse is symmetric when time-dependent phase satisfies the condition φ(t)=φ(−t) for a pulse centered about t=0. Such pulses have performance that is independent of the sign of the resonant offset Δω0. As a result, refocusing cycles that use symmetric pulses have transverse refocusing axes nx that are symmetric about Δω0=0.

In a third embodiment, the phases of each segment can be kept anti-symmetric with respect to the center of the pulse (e.g., central axis 302). A pulse is anti-symmetric when time-dependent phase satisfies the condition φ(t)=−φ(−t) for a pulse centered about t=0. This symmetry ensures that the refocusing axis is confined to the x-z plane and has no component in the y-direction. As a result, the asymptotic magnetization is real (aligned with the x-axis, with no y-component). Imposing the anti-symmetry constraint further reduces the number of optimization variables by a factor of 2.

Pulses with phases that are simultaneously symmetric and anti-symmetric with respect to time are limited to 0 and 180 degrees (π). This is because e+iπ=e−iπ. Such refocusing pulses have refocusing axes that satisfy the following conditions:

nx(Δω0)=+nx(−Δω0)

ny=0

nz(Δω0)=−nz(−Δω0).  (10)

Because the x-component of the refocusing axis is symmetric with offset frequency (nx(Δω0)=nx(−Δω0), where Δω0 ≡ω0−ω0c and ω0c is the nominal Larmor frequency), the asymptotic magnetization Masy has the same symmetry. In this case, the asymptotic time-domain echo masy(t) will be purely real:

masy(t)=-Masy(Δω0)ejΔω0td(Δω0)=20Masy(Δω0)cos(Δω0t)d(Δω0).(11)

The absence of an imaginary component in the time-domain echo simplifies later signal processing. These symmetries also have another practical advantage. The symmetries allow a generalized form of phase cycling, based on phase inversion of the excitation pulse, to be applied to the asymptotic magnetization produced by these pulses. The phase inversion procedure is useful for canceling ringing and static offsets, just like normal phase cycling. However, the phase inversion procedure is generally not useful for refocusing pulses that are not phase alternating. In this case, phase inversion typically cancels out a substantial fraction of the asymptotic magnetization.

The initial magnetization produced by the original and phase-inverted excitation pulse is denoted by {right arrow over (M)}(0+) and {right arrow over (M)}(i)(0+), respectively. The net asymptotic signal obtained after the generalized phase cycling (the signal difference of the sequence run with the initial excitation pulse and that with the phase inverted excitation pulse) can be written as:

2<o ostyle="single">S</o>asy01)=[{right arrow over (M)}(0+{circumflex over (n)}−{right arrow over (M)}(i)(0+{circumflex over (n)}]{circumflex over (n)}.  (12)

There are different ways to impose these symmetry constraints. In some embodiments, a set of random initial phases {φj} can be used, where j runs from 1 to N/2. The anti-symmetry constraint was imposed by optimizing the pulse {φj,−φN/2-j}. The length Δt of each segment within the pulse was assumed to be constant. The symmetry constraint was imposed by adding an appropriate penalty term to the cost function used for optimization. In another embodiment, one can begin with alternating phases {0, π, 0, π, . . . } and optimize the switching times between them (e.g., allow for variable segment lengths Δtj).

The additional term that imposed the symmetric pulse constraint was based on penalizing asymmetry in nx with respect to offset frequency for refocusing pulses of a fixed duration. The resulting cost function was given by:

C=−√{square root over (∫|nx′(Δω0)|2d(Δω0))}+λ√{square root over (∫|nx′(Δω0)−nx′(−Δω0)|2d(Δω0))}.  (13)

In this embodiment, the effective refocusing axis nx′≡nx*W can be used to take into account the effects of a finite acquisition window. The first term in the cost function is proportional to the SNR (in voltage units) of the asymptotic echo, assuming ideal axis-matching excitation pulses (such as the first type described above). In addition, λ is a constant that controls the strength of the second term, which penalizes any asymmetry in nx. An acquisition window length of Tacq=5×t180 and λ=0.5 can be used during optimizations.

Refocusing pulses were generated using optimizations with the following set of constraints:

    • Set 1: The fidelity of these refocusing pulses was optimized, while limiting the normalized maximum RF amplitude to ω1,max=1 No other constraints were placed on the amplitudes or phases of the pulse segments.
    • Set 2: The optimization used the asymptotic SNR for a perfect 90y excitation pulse (as defined in equation 6) as the cost function. Other optimization conditions were unchanged.
    • Set 3: This optimization generates phase alternating pulses with anti-symmetric phases. The optimization used the asymptotic SNR for a perfect 90y excitation pulse as the cost function. Also, the RF amplitude of each segment was fixed at ω1,max=1, thus halving the number of optimization variables. In addition, an anti-symmetric constraint was imposed on the phases of the segments by requiring that φ(i)=−φ(−i), where φ(i) is the phase of the i-th segment and i=0 corresponds to the center of the pulse. This constraint further halved the number of optimization variables.
    • Set 4: This optimization generates phase alternating pulses with both anti-symmetric and symmetric phases. The same constraints were imposed as in the Set 3. In addition, a symmetric constraint was imposed on the phases of the segments by requiring that φ(i)=φ(−i), where φ(i) is the phase of the i-th segment and i=0 corresponds to the center of the pulse. Also, the cost function shown in equation 13 was used to generate the pulses.

The optimization used a uniform distribution of offset frequencies between ±20ω1 and a constant oscillating RF field strength. The length of the free precession interval between adjacent refocusing pulses is represented by tFP. The echo spacing is tE=tFP+tref, where tref is the length of the refocusing pulse. The optimization used a segment length that was multiples of Δt=0.1×t180, unless otherwise specified, and carried out four sets of optimizations, with the pulse length varying between t180 and 10×t180 within each set, as described above.

FIGS. 6A-6D show SNR as a function of pulse length for the various refocusing pulses generated by the four sets of optimizations for uniformly-distributed resonant offsets between ±50ω1, tFP=6×t180, and Tacq=5×t180. All plots have been normalized to the SNR obtained by using a perfect 90y excitation pulse and rectangular 180x refocusing pulses.

FIGS. 6A and 6B were generated using perfect 90y excitation pulses (such as the first type described above). These figures show that, for longer pulses, optimizing the refocusing axis gives significantly better results compared to optimizing for pulse fidelity. Also, the SNR for individual asymptotic echoes initially increases rapidly with pulse length and then levels off. As a consequence, the SNR per unit time shows a maximum between tp=3×t180 and 4×t180. Accordingly, in various embodiments, the total pulse duration of the refocusing pulses is between 3 times t180 to 4 times t180. This maximum shifts towards even shorter pulse lengths if the free precession time tFP is reduced.

FIGS. 6C and 6D were generated using ideal axis-matching excitation pulses (such as the second type described above). The overall behavior of SNR and SNR per unit time versus pulse length is very similar to the first case, but the actual values are considerably larger, as expected. Moreover, these values converge much more slowly with simulation bandwidth than for the first case. Thus, in principle, somewhat more SNR can be obtained by increasing the simulation bandwidth. The SNR per unit time again shows a maximum between tp=3×t180 and 4×t180.

FIGS. 7A and 7B show SNR and SNR per unit time (in power units) of CPMG-like sequences using phase alternating refocusing pulses as a function of refocusing pulse length for tFP=6×t180. Results are shown assuming ideal axis-matching excitation pulses (type 2 described above) for two values of the minimum pulse segment length Δt. The plots have been normalized to the SNR obtained by using a perfect 90y excitation pulse and rectangular 180x refocusing pulses. As expected, the SNR increases as the control bandwidth increases, i.e., Δt decreases. However, there is little improvement for pulses shorter than 2×t80 The performance of such short pulses is mainly limited by the total rotation angle ω1tp and not the control bandwidth.

As explained above, the phase alternating refocusing pulses may be applied with axis-matching excitation pulses. The axis-matching pulses can be optmized for the phase alternating pulses. In one specific example, the axis-matching excitation pulses are optmized using a fixed segment length (e.g., Δt=0.08×t180). The amplitude of each segment is constant (e.g., ω1,max=1) and has the same value as the refocusing pulses. The phase is allowed to take arbitrary values. The performance of the excitation pulse generally improves as the number of segments increases. This behavior occurs because longer pulses have more degrees of freedom that can be optimized, even though it is more difficult to optimize them. For example, going from 110 segments to 180 segments increases the average SNR of solutions by 15%-20%. However, the rate of improvement decreases as the pulse becomes longer than tE. This is because the refocusing axis varies periodically as a function of offset frequency with a period of approximately 1/tE. Thus, in various embodiments, the excitation pulse is at least as long as tE in order to create initial magnetization that matches the periodicity of the refocusing axis. However, long pulses are more difficult to optimize (because they contain more optimization variables), place greater demands on the transmitter electronics, and are unsuitable for measuring small values of T2 because of relaxation during the excitation pulse.

In one specific embodiment, the optimization of axis-matching excitation pulses is performed using MATLAB™'s Optimization Toolbox. In particular, MATLAB's FMINCON function can be used to minimize the cost function in equation 14, which is a weighted sum of the root-mean square (RMS) and average values of the asymptotic signal Sasy(Δω0)=({circumflex over (n)}·{right arrow over (M)}(0+)){circumflex over (n)}:

C=−[√{square root over (∫Sasy2(Δω0)d(Δω0))}+λ∫Sasy(Δω0)d(Δω0)].  (14)

In equation 14, Δω0 is the resonance offset frequency, λ is a real number that controls the strength of the second term in the cost function, and Sasy(Δω0) is purely real (along the x-axis) because ny=0 for phase alternating refocusing pulses. The first term is proportional to the asymptotic SNR in voltage units. The second term is proportional to Sasy(0), the signal amplitude at the nominal center (t=0) of the time-domain echo. This is because Fourier analysis shows that Sasy(0)=∫Sasy(Δω0)d(Δω0). This term allows the optimizer to find better (higher SNR) solutions. λ=0.33 can be used, which results in roughly equal weights being assigned to the two terms in the cost function. Also, an acquisition window length of Tacq=5×t180 can be used in the optimization. Further details regarding axis-matching excitation pulses and optimization are described in U.S. Patent Application Publication No. 2013/0162247, published on Jun. 27, 2013, and S. Mandal et al., Axis-Matching Excitation Pulses for CPMG-like Sequences in Inhomogeneous Fields, Journal of Magnetic Resonance, 237:1-10 (2013).


TABLE 1
Exci-
Re-
tation
focusing
SNR per
Se-
Exci-
pulse
Re-
pulse
SNR
unit time
quence
tation
length
focusing
length
(pre-
(pre-
#
pulse
[t180]
pulse
[t180]
dicted)
dicted)
1
AMEX
14.4
From
2.1
4.65
4.02
FIG. 4A
2
AMEX
14.4
From
2.1
4.78
4.13
FIG. 4A
3
AMEX
14.4
From
3.1
5.07
3.90
FIG. 4B

Table 1 shows predicted SNR and SNR per unit time for NMR pulses seqeuences with a practical axis-matching excitation pulse and phase alternating pulses, as shown in FIGS. 4A and 4B (“AMEX-PA sequences”). The phase alternating pulses include a free precession interval of tFP=6×t180. The axis-matching exitation pulses were generated by constraining the optimization to excitation pulses with 180 segments (total length=14.4×t180) Excitation pulses of this length provide a compromise between pulse length, SNR, and practical concerns for values of tE between 7×t180 and 9×t180. The results in Table 1 have been normalized to a standard CPMG sequence (using rectangular 90 degree and 180 degree pulses). The new sequences provide greater SNR than the standard CPMG sequence. For example, consider phase alternating refocusing pulses of length 2×t180. The second AMEX-PA sequence listed in Table 1 is predicted to produce 4.78 times the SNR (in power units) within a single asymptotic echo than the standard CPMG with rectangular excitation and refocusing pulses. After accounting for the increased echo period relative to the normal CPMG, which has tE=7×t180 the normalized SNR per unit time in this case is 4.13.

The three different AMEX-PA sequences listed in Table 1 were evaluated using an NMR system. FIGS. 8A-8C show an experimentally measured spectrum of the asymptotic echo produced by sequences 1-3, respectively, and the figures compare the experimentally measured spectrum with the theoretically predicted spectrum. The match between theory and experiment is excellent for the three sequences shown in the figure.

FIGS. 8A-8C compare the measured echo amplitude decay of the AMEX-PA sequences in Table 1 with the standard CPMG sequence 90y−(180x)N. In these figures, a free precession interval of tFP=6×t180 was used and phase corrections have been applied to the experimental spectra (both constant and first-order, i.e., varying linearly with offset frequency). Referring to FIG. 9, echo amplitudes were calculated by using the measured asymptotic echo shape as a matched filter. This procedure can be used for processing data when the noise is additive, white, and exhibits Gaussian statistics. In each case, after the first few echoes, the echo amplitudes enter the asymptotic regime where they follow an exponential decay A(0) exp(−ktE/T2) with T2=2.00 sec and k being the echo number. For each sequence, the value of A(0) is marked by a x sign. A(0) is the amplitude extrapolated to t=0 after exponential fitting. The amplitudes were normalized with respect to A(0) of the standard CPMG sequence.

FIG. 9 shows measured amplitudes of the echoes for the three different AMEX-PA sequences during the first 120 ms. The echo amplitudes were extracted using the normalized asymptotic echo shapes as matched filters. The x symbols show the amplitudes extrapolated to t=0, which were obtained by fitting exponential decays to the asymptotic regimes. FIG. 9 also shows that the AMEX-PA sequences have a characteristic transient effect in the first few echoes that is analogous to that observed for the standard CPMG sequence in inhomogeneous magnetic fields. This CP-like component is generated by initial magnetization that is not aligned with the CPMG refocusing axis. FIG. 9 shows that the normalized amplitude of this initial transient is significantly smaller for the AMEX-PA sequences in Table 1. This shows that the sequences place more of the initial magnetization along the CPMG refocusing axis than the standard CPMG sequence.

FIGS. 8A-8C and 9 were generated from a NMR experiment performed using the following NMR system, parameters, and sample. The NMR system was a small-animal imaging probe located within a 1 T superconducting magnet. The sample for the experiment was deionized (DI) water that was placed within a NMR tube with a diameter of 10 mm and length of 23 cm aligned along the z-axis. A set of gradient coils was used to create a constant B0 gradient along the z-axis. The gradient coils produced a uniform distribution of resonant offset frequencies within the sample (a gradient g=1.6 G/cm). This value of gradient minimizes diffusion effects. In addition, to evaluate the shapes of the echoes in detail, a low power level for the refocusing pulses was used, resulting in a nominal 180° rectangular pulse length of t180=500 μs. As a result, the characteristic length of the refocused region was ω1/(γg)=1.50 mm (much shorter than the sample). The acquisition window length Tacq=2.048 ms=4.10×t180 was long enough to ensure that the entire echo shape could be recorded for analysis. Other experimental parameters included tFP=6×t180 and a recycle delay tR=8 sec. A standard eight-part phase cycle was used for the standard CPMG sequence using rectangular excitation and refocusing pulses, while a two-part phase-inversion cycle was used for the AMEX-PA sequences listed in Table 1. Given the quality factor of the coil, the transmit and detection bandwidth was about 1 MHz (three orders of magnitudes larger than ω1/(2π)), and therefore not a limiting factor.

The sensitivity of a given NMR sequence to changes in the static field B1 can be evaluated by plotting a corresponding nutation curve. The nutation curve is defined as the asymptotic echo amplitude as a function of B1 field strength. FIG. 10 shows measured and simulated nutation curves for the third AMEX-PA sequence listed in Table 1 and the standard CPMG sequence (rectangular 90 degree excitation and 180 degree refocusing pulses). FIG. 10 shows that the overall shapes of the nutation curves for the sequences are similar. In particular, the curves have comparable sensitivities to changes in the B1 field. However, for the third AMEX-PA sequence in Table 1, SNR is maximized when B1 is approximately 15% larger than its nominal value, while, for the rectangular sequence, SNR is maximized when B1 is equal to its nominal value. As a result, the SNR produced by the AMEX-PA sequence increases by up to 15% (in power units) as B1 becomes larger than its nominal value, while that of the rectangular sequence decreases. The other AMEX-PA sequences in Table 1 show similar behavior.


TABLE 2
Predicted
Measured
Predicted
Measured
SNR per unit
SNR per unit
Sequence #
SNR
SNR
time
time
1
4.65
4.39
4.02
3.79
2
4.78
4.56
4.13
3.94
3
5.07
4.63
3.90
3.56

Table 2 shows the predicted and measured SNR for the three different AMEX-PA sequences in Table 1. The values shown in the Table 2 have been normalized with respect to the standard CPMG sequence. Each experimental SNR value is given by A(0)2, where A(0) is the amplitude of the corresponding echo decay extrapolated to t=0. The resultant agreement between theory and application is excellent, with an average error of 6.3% between them.

FIG. 11 shows a NMR system 1100 in accordance with one embodiment of the present disclosure. This NMR system 1100 can be used to apply the phase alternating refocusing pulses described above. The NMR system 1100 includes a coil 1102 that is coupled to NMR electronics 1104, 1106, 1108. A sample substance 1101 is located inside and/or outside of the coil 1102. The coil 1102 applies RF pulses and NMR pulse sequences to the substance 1101. The NMR electronics include a transmitter 1104 and a receiver 1106. Each of the transmitter 1104 and the receiver 1106 are coupled to the coil 1102. In some embodiments, however, the NMR system 1100 may include separate transmitter and receiver coils.

The NMR transmitter 1104 includes a NMR transmitter circuit 1110 that is coupled to the coil 1102. The transmitter circuit 1110 generates NMR pulse sequences and provides the NMR pulse sequences to the coil 1102. In some embodiments, the NMR transmitter 1104 uses a “tuned” NMR transmitter circuit 1110. A tuned NMR transmitter is tuned to a particular Larmor frequency using a capacitor that is coupled to the coil. The particular capacitance of the capacitor and the inductance of the coil determine the resonant frequency that is generated by the coil. In other embodiments, the transmitter circuit 1110 is a “non-resonant” transmitter circuit. The circuit is non-resonant because the resonant frequency of the circuit does not need to match the Larmor frequency of interest. Although the non-resonant transmitter circuit and coil 1102 may use capacitors and have some associated capacitance, this capacitance is not specifically selected to match a Larmor frequency of interest. Instead, the transmitter circuit includes a plurality of switches that couple and decouple the coil 1102 with a power source. Operation of the switches generates a particular frequency. Thus, the frequency produced by the transmitter circuit can be modulated directly by a spectrometer. Further details regarding non-resonant transmitters are provided in U.S. Publication No. 2012/0001629 published on Jan. 5, 2012; U.S. application Ser. No. 13/774,457, filed on Feb. 22, 2013, and U.S. patent application Ser. No. 13/963,826, filed on Aug. 9, 2013. These patent applications are incorporated by reference in their entireties.

As shown in FIG. 11, the coil 1102 is also coupled to a NMR receiver 1106 so that NMR resonant signals that are generated within the substance 1101 can be detected, amplified, and analyzed. In one specific embodiment, the receiver 1106 is a broadband NMR receiver, which can receive and process resonant NMR signals over a broad frequency range. The coil 1102 is coupled to the receiver 1106 using a duplexer 1116. The duplexer 1116 decouples the receiver 1106 from the coil 1102 when the coil is operating in a transmitting mode (e.g., transmitting an NMR pulse sequence). In one particular embodiment, the duplexer 1116 includes switches and a switch driver 1118 that opens the switches during a transmitting mode and closes the switches during a receiving mode of operation. In this manner, the duplexer 1116 protects the receiver 1106 during a transmitting mode. A duplexer may not be used when the NMR system 1100 includes separate transmit and receive coils.

The NMR system 1100 also includes a spectrometer 1108 that is used to provide NMR pulse sequences to the NMR transmitter 1104 and to analyze the NMR signal received from the NMR receiver 1106. In various embodiments, the detected NMR signal is output by the NMR receiver 106 in analog form. In such embodiments, the spectrometer 1108 may include a digitizer 1120 (e.g., analog-to-digital converter) for converting the detected NMR signal into digital data. Furthermore, in various embodiments, demodulation of the NMR signal can occur within the spectrometer 1108. In various other embodiments, however, demodulation of the NMR signal can also occur within the NMR receiver 1106. The spectrometer 1108 also includes a post-processor 1122 that is used to interpret the detected digital NMR data and to determine NMR properties from the detected data. This data can be presented to a user using an operator interface with a graphical user interface (GUI). The spectrometer 1108 also includes a pulse sequence generator 1124 that generates NMR pulse sequences based upon parameters selected by an operator at the operator interface. The pulse sequence generator provides the sequences to the NMR transmitter 1104. In one particular embodiment, the spectrometer 1108 is a KEA, which can be obtained from Magritek of Wellington, NZ. The spectrometer 1108 can be controlled from the operator interface using PROSPA™ software, which can also be obtained from Magritek.

Further details of NMR electronics, NMR transmitters, and NMR receivers are described in U.S. Publication No. 2012/0001629, published on Jan. 5, 2012, and U.S. application Ser. No. 13/774,457, filed on Feb. 22, 2013. These patent applications are incorporated by reference in their entireties.

As shown in FIG. 11, the NMR system 1100 includes an electro-magnetic device 1126 for applying a static magnetic field to the substance 1101. In some embodiments, the electro-magnetic device 1126 is a magnet or an array of magnets. The magnets can be formed from a samarium-cobalt (SmCo) magnetic material.

The NMR system 1100 also includes an operator interface 1128 for communicating with the spectrometer 1108. The operator interface 1128 includes a computer system. The computer system may include a computer processor 1130 (e.g., a microprocessor, microcontroller, digital signal processor, or general purpose computer) for executing any of the methods and processes described herein. The computer system may further include a memory 1132 such as a semiconductor memory device (e.g., a RAM, ROM, PROM, EEPROM, or Flash-Programmable RAM), a magnetic memory device (e.g., a diskette or fixed disk), an optical memory device (e.g., a CD-ROM), a PC card (e.g., PCMCIA card), or other memory device. The memory 1132 can be used to store computer instructions (e.g., computer program code) that are interpreted and executed by the processor 1130.

NMR pulse sequences may be implemented as a series of computer instructions (e.g., software or firmware) fixed on a non-transitory tangible medium, such as a computer readable medium (e.g., a memory), or transmittable to the computer system, via a modem or other interface device, such as a communications adapter connected to a network over a tangible medium (e.g., optical or analog communications lines). The series of computer instructions can embody all or part of the NMR pulse sequences and phase alternating refocusing pulses described herein. The processor 1130 may be configured to retrieve the sequences from the memory 132 and provide instructions to the NMR electronics 1104, 1106, 1108 to apply the sequences to the substance 1101. The detected resonant signals may also be communicated from the NMR electronics 1104, 1106, 1108 to the processor 1130 for storage on the memory 1132.

The operator interface 1128 also supports the graphical user interface 1134 (GUI) (e.g., a monitor, a touch screen, a mouse, a keyboard and/or a joystick). The GUI 1134 allows an operator to control and communicate with the NMR electronics 1104, 1106, 1108. In various embodiments, the operator interface 1128 can be used to perform functions selected from the following non-limiting list:

    • Communicate NMR pulse sequences from the operator interface 1128 to the NMR electronics 1104, 106, 108;
    • Communicate instructions to the NMR electronics 1104, 1106, 1108 to initiate and/or terminate NMR measurements;
    • Communicate instructions to change parameters of NMR pulse sequences to the NMR electronics 1104, 1106, 1108 (e.g., pulse amplitude of sequences, pulse lengths, timing between pulses, shape of pulses, and/or frequency of pulses);
    • Communicate detected NMR signal data from the NMR electronics 1104, 1106, 1108 to the operator interface 1128;
    • Perform analysis at the operator interface 1128 of detected NMR signal data to determine NMR properties of substances; and
    • Display various plots of NMR properties to the operator at the operator interface 128.

Illustrative embodiments of the present disclosure are not limited to the NMR system 1100 shown in FIG. 11. Various modifications can be made to the system. For example, in one specific embodiment, the NMR electronics 1104, 1106, 1108 include an additional computer system that supports the NMR electronics. In such an embodiment, the NMR electronics 1104, 1106, 1108 and operator interface 1128 may include their own communication modules, which provide for communication between the NMR electronics and the operator interface. A communications link between the communication modules can be established using, for example, a hard-wired link, an optical link, acoustic link, and/or a wireless link. By using the communication modules, the NMR electronics 1104, 1106, 1108 and the operator interface 1128 can be physically located in two separate locations. For example, in a wellbore application, the NMR electronics 1104, 1106, 1108 can be located downhole, while the operator interface 1128 is located at the surface.

Illustrative embodiments of the present disclosure are further directed to oil and gas field applications, such as wellbore logging tools. In particular, FIG. 12 shows a logging-while-drilling (LWD) system 1200 for investigating, in situ, a substance 1202 within an earth formation 1204 and determining a property of the substance, while a drilling operation is performed. The LWD system 1200 includes a drill string 1208 that is disposed within a wellbore 1206 that traverses the formation 1204. The drill string 1208 includes a drill collar 1210 with a drill bit 1212 disposed at the lower-end of the drill collar. The LWD system 1200 also includes a surface system with a derrick assembly and platform assembly 1214 positioned over the wellbore 1206. The derrick assembly 1214 rotates the drill string 1208 and, as the drill string rotates, the drill bit 1212 drills deeper into the wellbore 1206. An LWD NMR logging module 1216 is disposed within the drill collar 1210 so that the module can log the surrounding earth formation as the drilling operation is performed. The logging module 1216 communicates with surface equipment 1218, which includes an operator interface for communicating with the module. Such an operator interface has already been described with reference to FIG. 11. In various embodiments, the NMR logging module 1216 and operator interface can communicate via any one of a wired-drill pipe connection, an acoustic telemetry connection, optical communication and/or electronic communication.

FIG. 13 shows an LWD NMR logging module 1300 for applying NMR pulse sequences to the formation. The module 1300 includes magnet sections 1304 that generate a static magnetic field within a zone of sensitivity 1306 within the formation 1302. The static magnetic field produced by the magnet section 1304 and applied to the formation 1302 is inhomogeneous because the field decreases in strength as a function of distance from the magnet and into the formation. The module 1300 also includes a drill collar 1308 with an axial slot 1310. A coil 1312 is disposed within the axial slot 1310 and the slot is filled with an insulator, such as ceramic, epoxy, or fiberglass. The coil 1312 is wound around the drill collar 1308 within the axial slot 1310. The axial slot 1310 is sealed using a cover 1314. In some embodiments, the cover 1314 is formed from a non-magnetic material and/or non-conductive material. At one end, the coil 1312 is grounded (e.g., to the drill collar 1308). At the other end, the coil 1312 is coupled to NMR electronics 1316, which include a transmitter and a receiver. The NMR electronics 1316 are coupled to the coil 1312 via, for example, pressure feed-throughs.

The coil 1312 applies an oscillating magnetic field to an area of interest 1320 within the zone of sensitivity 1306 of the formation 1302. In some embodiments, the oscillating magnetic field is axially symmetric to facilitate measurements during rotation of the drill string. The oscillating magnetic field is applied as a NMR pulse sequence, such as a CPMG sequence. The NMR pulse sequence includes phase alternating refocusing pulses as described herein to increase the SNR of resonant signals produced within the inhomogeneous static magnetic field applied by the magnet sections 1304, while also meeting the power constraints of an LWD tool application.

The coil 1312 also detects an NMR signal produced by the oscillating magnetic field within the formation. The NMR signal can be used to determine properties of the formation, such as a porosity of the formation and/or a bulk diffusion coefficient of formation fluids within the formation. Further details of NMR LWD systems are described in U.S. Pat. No. 5,629,623 issued on May 13, 1997 and U.S. Pat. No. 6,392,410, issued on May 21, 2002. Each of these patents is incorporated by reference herein in their entireties. One specific example of a NMR LWD tool is Schlumberger's proVISION™ tool.

The NMR methods and pulses sequences described herein are not limited to any particular device type or system. The NMR methods and pulses sequences described herein can be implemented in surface environments, such as in a laboratory. The NMR methods and pulses sequences described herein can be used in single-sided NMR systems, stray-field NMR systems, or in any other NMR applications where an inhomogenous static magnetic field are used.

The NMR methods and pulses sequences described herein are not limited to any particular wellbore application. The NMR methods and pulses sequences described herein can be used with LWD systems, such as the one shown in FIGS. 12 and 13. Also, the NMR methods and pulses sequences described herein can be applied to wireline systems (e.g., a wireline tool) or measuring-while-drilling systems (e.g., MWD tools). Illustrative embodiments can also be used with any suitable means of conveyance, such as armored cable or coiled tubing.

Although several example embodiments have been described in detail above, those skilled in the art will readily appreciate that many modifications are possible in the example embodiments without materially departing from the scope of this disclosure. Accordingly, all such modifications are intended to be included within the scope of this disclosure.

APPENDIX

The unitary operator that performs a counter-clockwise rotation of angle θ around an axis {circumflex over (n)}, where {circumflex over (n)} is a unit vector, has the form:

U(θ,n^)=e-iθ2n^·σ,(15)

Given U, {circumflex over (n)} up to a sign can be determined. To avoid ambiguity, the constraint nx≥0 is imposed.

Following the GRAPE approach described by Khaneja et al., Optimal Control of Coupled Spin Dynamics: Design of NMR Pulse Sequences by Gradient Ascent Algorithms, J. Magnetic Resonance, 172:296-305 (2005), the gradient of n=nx+iny can be analytically calculated with respect to the controls to the first order and used to maximize any function of n.

The Hamiltonian of the spin system is

H(t)=H0+k=12uk(t)Hk,(16)

where H0 is the natural Hamiltonian, which is −Δω0Iz and Hk represents the control functions, which are Ix and Iy.

Denote the evolution due to free precession as U0. If the controls are discretized in N small time steps and the controls are assumed constant in each time step Δt, then:

U=U0UN . . . Uj . . . U1U0,  (17)

where Uj=exp(−iΔt(Hok=12uk(j)Hk)). If the control uk(j) is perturbed by a small amount δuk(j), the resulting small change in Uj to first order in δuk(j) is

δUj=−iΔtHkUjδuk(j).  (18)

Therefore, to first order, the change in the total unitary operator is

δU=U0UN(-iΔt)HkUjU1U0δuk(j)=U~δuk(j).(19)

U in explicit matrix representation is:

U=e-iθ2n^·σ=Icos(θ2)+isin(θ2)n^·σ=[cos(θ2)+isin(θ2)nzsin(θ2)ny-isin(θ2)nx-sin(θ2)ny-isin(θ2)nxcos(θ2)-isin(θ2)nz].(20)

Hence, nx and ny can be found from U. Taking into account the constraint that nx≥0 provides:

nx=𝔍2(U12)1-2(U11),ny=±2(U12)1-2(U11).(21)

The increments of nx and ny due to a small change in U are:

δnx=12𝔍2(U12)1-2(U11)(2𝔍(U12)δU12(1-2(U11))+2(U11)δU11𝔍2(U12)(1-2(U11))2),δny=±122(U12)1-2(U11)(2(U12)δU12(1-2(U11))+2(U11)δU112(U12)(1-2(U11))2).(22)

From equations (19) and (22), δnx and δny can be found in terms of δuk(j), resulting in:

δnxδuk(j)=12nx(2𝔍(U12)(1-2(U11))𝔍(U~12)+2(U11)𝔍2(U12)(U~11)(1-2(U11))2),δnyδuk(j)=12ny(2(U12)(1-2(U11))𝔍(U~12)+2(U11)2(U12)(U~11)(1-2(U11))2).(23)

The equations in this Appendix are the analytical expressions for the derivatives of nwith respect to the controls, to the first order. Any function of ncan then be numerically optimized using the conjugate gradient method, which converges faster than the steepest descent method.

Read more
PatSnap Solutions

Great research starts with great data.

Use the most comprehensive innovation intelligence platform to maximise ROI on research.

Learn More

Patent Valuation

$

Reveal the value <>

21.0/100 Score

Market Attractiveness

It shows from an IP point of view how many competitors are active and innovations are made in the different technical fields of the company. On a company level, the market attractiveness is often also an indicator of how diversified a company is. Here we look into the commercial relevance of the market.

29.0/100 Score

Market Coverage

It shows the sizes of the market that is covered with the IP and in how many countries the IP guarantees protection. It reflects a market size that is potentially addressable with the invented technology/formulation with a legal protection which also includes a freedom to operate. Here we look into the size of the impacted market.

74.0/100 Score

Technology Quality

It shows the degree of innovation that can be derived from a company’s IP. Here we look into ease of detection, ability to design around and significance of the patented feature to the product/service.

77.0/100 Score

Assignee Score

It takes the R&D behavior of the company itself into account that results in IP. During the invention phase, larger companies are considered to assign a higher R&D budget on a certain technology field, these companies have a better influence on their market, on what is marketable and what might lead to a standard.

16.0/100 Score

Legal Score

It shows the legal strength of IP in terms of its degree of protecting effect. Here we look into claim scope, claim breadth, claim quality, stability and priority.

Citation

Patents Cited in This Cited by
Title Current Assignee Application Date Publication Date
Magnetic resonance transmitter SCHLUMBERGER TECHNOLOGY CORPORATION 09 August 2013 12 February 2015
Amplitude and/or phase modulated NMR pulse sequences BAKER HUGHES INCORPORATED 12 October 2001 17 April 2003
Nuclear magnetic resonance apparatus and method for generating an axisymmetric magnetic field having straight contour lines in the resonance region SCHLUMBERGER TECHNOLOGY CORPORATION 24 May 2001 21 May 2002
Method and apparatus for imaging instruments during interventional MRI using asymmetric spin echo sequences TOSHIBA AMERICA MRI INC. 12 November 1998 18 December 2001
Method, system and software arrangement for measuring magnetic field correlation NEW YORK UNIVERSITY 22 February 2010 14 December 2010
See full citation <>

More like this

Title Current Assignee Application Date Publication Date
System and method of obtaining spatially-encoded NMR parameters from arbitrarily-shaped compartments and linear algebraic modeling THE JOHNS HOPKINS UNIVERSITY 18 May 2015 24 November 2016
Magnetic resonance fingerprinting using a spin-echo pulse sequence with an additional 180 degree RF pulse KONINKLIJKE PHILIPS N.V. 05 November 2015 19 May 2016
Arrays of receive antennas for magnetic resonance measurements BAKER HUGHES INCORPORATED 09 May 2017 30 November 2017
Magnetic resonance imaging system and magnetic resonance imaging method HITACHI MEDICAL CORPORATION 20 July 2005 08 April 2008
Downhole diffusion coefficient measurement NMR SERVICES AUSTRALIA 29 June 2017 04 January 2018
Magnetic resonance fingerprinting with reduced sensitivity to inhomogeneities in the main magnetic field KONINKLIJKE PHILIPS N.V. 13 April 2016 20 October 2016
Multiple depth of investigation nuclear magnetic resonance logging for determining the porosity and PORE type of subterranean formations HALLIBURTON ENERGY SERVICES, INC. 04 March 2016 08 September 2017
T2 inversions with reduced motion artifacts BAKER HUGHES INCORPORATED 30 March 2017 12 October 2017
Low gradient magnetic resonance logging for measurement of light hydrocarbon reservoirs BAKER HUGHES INCORPORATED 12 January 2017 20 July 2017
Active noise suppression for magnetic resonance based magnetic field probes EIDGENÖSSISCHE TECHNISCHE HOCHSCHULE (ETH),UNIVERSITÄT ZÜRICH 11 May 2016 24 November 2016
Combined NMR-resistivity measurement apparatus, systems, and methods HALLIBURTON ENERGY SERVICES, INC. 30 December 2014 07 July 2016
Petrophysically-regularized nuclear magnetic resonance inversion SCHLUMBERGER CANADA LIMITED,SERVICES PETROLIERS SCHLUMBERGER,SCHLUMBERGER TECHNOLOGY B.V.,SCHLUMBERGER TECHNOLOGY CORPORATION 13 January 2016 21 July 2016
External housing for signal to noise improvement HALLIBURTON ENERGY SERVICES, INC. 29 December 2015 06 July 2017
Selective NMR pulse for downhole measurements HALLIBURTON ENERGY SERVICES, INC. 22 December 2015 29 June 2017
Steady state magnetic resonance fingerprinting KONINKLIJKE PHILIPS N.V. 06 February 2017 17 August 2017
Erfassung segmentierter Magnetresonanz-Abbildungs-Herzdaten unter Verwendung einer Echo-Planar-Abbildungs-Impulsfolge GENERAL ELECTRIC CO. 19 February 1999 26 April 2012
Magnetic resonance imaging apparatus and method of obtaining magnetic resonance image by using multiple excitation with delayed spin-echoes SAMSUNG ELECTRONICS CO., LTD. 15 December 2016 24 August 2017
Side-looking NMR logging tool WEATHERFORD TECHNOLOGY HOLDINGS, INC. 02 August 2016 30 March 2017
Nuclear magnetic resonance sensing and fluid sampling device for subterranean characterization HALLIBURTON ENERGY SERVICES, INC. 08 August 2016 15 February 2018
Downhole formation evaluation using nuclear magnetic resonance inversion matrix reduction HALLIBURTON ENERGY SERVICES, INC. 17 April 2017 26 October 2017
See all similar patents <>

More Patents & Intellectual Property

PatSnap Solutions

PatSnap solutions are used by R&D teams, legal and IP professionals, those in business intelligence and strategic planning roles and by research staff at academic institutions globally.

PatSnap Solutions
Search & Analyze
The widest range of IP search tools makes getting the right answers and asking the right questions easier than ever. One click analysis extracts meaningful information on competitors and technology trends from IP data.
Business Intelligence
Gain powerful insights into future technology changes, market shifts and competitor strategies.
Workflow
Manage IP-related processes across multiple teams and departments with integrated collaboration and workflow tools.
Contact Sales
Clsoe
US10001578 Nuclear magnetic resonance refocusing pulses 1 US10001578 Nuclear magnetic resonance refocusing pulses 2 US10001578 Nuclear magnetic resonance refocusing pulses 3