EPI calibration method to minimize ghosting in reconstructed images

• US 6,700,374 B1
• Filed: 03/29/2000
• Issued: 03/02/2004
• Est. Priority Date: 03/29/2000
• Status: Expired due to Fees

• Current Assignee: Marconi Medical Systems, Inc. (Koninklijke Philips N.V.)
• Original Assignee: Koninklijke Philips Electronics N.V. (Koninklijke Philips N.V.)
• Inventors: Dannels, Wayne R., Wu, Dee H.
• Primary Examiner(s): Lefkowitz, Edward
• Assistant Examiner(s): Fetzner, Tiffany A.
• Application Number:
  US09/537,139
• Time in Patent Office: 1,434 Days
• Field of Search: 324/309, 324/307, 324/312, 324/306, 324/318, 324/300, 324/303, 382/128
• US Class Current: 324/312
• CPC Class Codes: G01R 33/56554 caused by acquiring plural, differently encoded echo signals after one RF excitation, e.g. correction for readout gradients of alternating polarity in EPI

Abstract

A method of magnetic resonance imaging is provided. It includes supporting a subject in an examination region of an MRI scanner (A), and applying an EPI pulse sequence with the MRI scanner (A) to induce a detectable magnetic resonance signal from a selected region of the subject. The magnetic resonance signal are received and demodulated to generate raw data. Applied to the raw data are a pair of ghost reducing correction factors (θ,Δ). The pair of corrections factors (θ,Δ) included a phase correction (θ) and a read delay (Δ). The phase correction (θ) compensates for phase errors in the raw data, and the read delay (Δ) effectively shifts a data acquisition window (120) under which the raw data was collected to thereby align the raw data in k-space. The correction factors affect how data is loaded into k-space to generate k-space data, and the k-space data is subjected to a reconstruction algorithm to generate image data. Thereafter, values for the pair of correction factors (θ,Δ) are derived from the image data.

Claims

1. A method of magnetic resonance imaging comprising:

(a) applying an EPI pulse sequence with an MRI scanner to induce a detectable magnetic resonance signal from a selected region of a subject positioned in the MRI scanner;

(b) receiving and demodulating the magnetic resonance signal to generate raw data;

(c) applying a pair of ghost reducing correction factors to the raw data, said pair of corrections factors including a phase correction factor and a data alignment correction factor, wherein the phase correction factor compensates for phase errors in the raw data and the data alignment correction factor effectively shifts a data acquisition window under which the raw data was collected to thereby align the raw data with respect to itself in k-space;

(d) loading the correction factor affected data into k-space to generate a k-space data matrix;

(e) subjecting the k-space data matrix to an algorithm to reconstruct actual image data therefrom; and

,

(f) calculating values for the pair of correction factors using the reconstructed actual image data from step (e).

2. The method according to claim 1, wherein steps (a) through (f) make up an iterative calibration procedure, such that with each ensuing repetition, the calculated values from step (f) are used for the pair of correction factors applied in the step (c) that follows.

3. A method of magnetic resonance imaging comprising:

(a) applying an EPI pulse sequence with an MRI scanner to induce a detectable magnetic resonance signal from a selected region of a subject positioned in the MRI scanner;

(b) receiving and demodulating the magnetic resonance signal to generate raw data;

(c) applying a pair of ghost reducing correction factors to the raw data, said pair of corrections factors including a phase correction and an alignment correction, wherein the phase correction compensates for phase errors in the raw data and the alignment correction effectively shifts a data acquisition window under which the raw data is collected to thereby adjust an alignment of the raw data in k-space;

(d) loading the correction factor affected data into k-space to generate a k-space data matrix;

(e) subjecting the k-space data matrix to a reconstruction algorithm to generate image data; and

,

(f) calculating values for the pair of correction factors from the image data, including;

evaluating the image data to obtain a measurement of an amount of ghosting present in the image data; and

,

employing the measurement of the amount of ghosting as a cost function in an optimization algorithm which generates the values for the pair of correction factors.

4. The method according to claim 3, wherein steps (c) through (f) make up an iterative calibration procedure, such that with each ensuing repetition, the derived values from step (f) are used for the pair of correction factors applied in the step (c) that follows.

5. The method according to claim 4, wherein the iterative calibration procedure is repeated until a change in successive calculated values for the pair of correction factors is within a determined tolerance.

6. The method according to claim 4, wherein the iterative calibration procedure is repeated until a change in the cost function is within a determined tolerance.

7. The method according to claim 3, wherein said image data represents an array of pixels having complex intensity values with corresponding magnitudes, and the step of evaluating the image data comprises summing the magnitudes.

8. The method according to claim 3, wherein the optimization algorithm is a nonlinear mathematical optimization.

9. The method according to claim 3, wherein the optimization algorithm is a Nelder-Mead optimization.

10. The method according to claim 3, wherein the cost function is monitored as an indication of changed experimental conditions warranting re-calibration.

11. A method of magnetic resonance imaging comprising:

(a) supporting a subject in an examination region of an MRI scanner;

(b) conducting an iterative calibration procedure, said iterative calibration procedure comprising;

(i) applying a calibration EPI pulse sequence with the MRI scanner to induce a detectable magnetic resonance signal from a selected region of the subject, said calibration EPI pulse sequence having phase encoding gradient pulses of a first amplitude;

(ii) receiving and demodulating the magnetic resonance signal to generate raw data;

(iii) applying a pair of ghost reducing correction factors to the raw data, said pair of corrections factors including a phase correction and an alignment correction, wherein the phase correction compensates for phase errors in the raw data and the alignment correction effectively shifts a data acquisition window under which the raw data was collected to thereby align the raw data with respect to itself in k-space;

(iv) loading the correction factor affected data into k-space to generate a k-space data matrix;

(v) subjecting the k-space data matrix to a reconstruction algorithm to generate actual image data; and

,

(vi) calculating values for the pair of correction factors from the actual image data such that with each ensuing repetition of the calibration procedure the calculated values are used for the pair of correction factors applied in the step (iii) that follows; and

,

(c) upon completion of the iterative calibration procedure, designating the last calculated values for use as the correction factors in a subsequent imaging experiment directed to the selected region of the subject.

12. The method according to claim 11, wherein the imaging experiment comprises:

applying an image acquisition EPI pulse sequence with the MRI scanner to induce a detectable magnetic resonance signal from the selected region of the subject said image acquisition EPI pulse sequence having phase encoding gradient pulses of a second amplitude;

receiving and demodulating the magnetic resonance signal to generate raw data;

applying the pair of ghost reducing correction factors to the raw data using the designated values;

loading the correction factor affected data into k-space to generate a k-space data matrix;

subjecting the k-space data matrix to a reconstruction algorithm to generate image data; and

,

generating from the image data a ghost reduced human-viewable image representation of the selected region of the subject.

13. The method according to claim 12, wherein the first amplitude is different than the second amplitude.

14. The method according to claim 13, wherein the first amplitude is less than the second amplitude thereby increasing a field of view during the calibration procedure relative to a field of view during the imaging experiment.

15. The method according to claim 11, wherein the calibration procedure is conducted a series of times such that designated values for pairs of correction factors are experimentally obtained for a number of corresponding selected regions of the subject, said number of corresponding selected regions of the subject being taken from a series of selected regions.

16. The method according to claim 15, wherein the number of corresponding selected regions of the subject includes all the selected regions in the series thereof.

17. A method of magnetic resonance imaging comprising:

(a) supporting a subject in an examination region of an MRI scanner;

(b) conducting an iterative calibration procedure, said iterative calibration procedure comprising;

(i) applying a calibration EPI pulse sequence with the MRI scanner to induce a detectable magnetic resonance signal from a selected region of the subject, said calibration EPI pulse sequence having phase encoding gradient pulses of a first amplitude;

(ii) receiving and demodulating the magnetic resonance signal to generate raw data;

(iii) applying a pair of ghost reducing correction factors to the raw data, said pair of corrections factors including a phase correction factor and a data alignment correction factor, wherein the phase correction factor compensates for phase errors in the raw data and the data alignment correction factor effectively shifts a data acquisition window under which the raw data was collected to thereby align the raw data with respect to itself in k-space;

(iv) loading the correction factor affected data into k-space to generate a k-space data matrix;

(v) subjecting the k-space data matrix to a reconstruction algorithm to generate actual image data; and

,

(vi) calculating values for the pair of correction factors from the actual image data such that with each ensuing repetition of the calibration procedure the calculated values are used for the pair of correction factors applied in the step (iii) that follows;

(c) upon completion of the iterative calibration procedure, designating the last calculated values for use as the correction factors in a subsequent imaging experiment directed to the selected region of the subject; and

,

(d) conducting the calibration procedure a series of times such that designated values for pairs of correction factors are experimentally obtained for a number of corresponding selected regions of the subject, said number of corresponding selected regions of the subject being taken from a series of selected regions, wherein the number of corresponding selected regions of the subject includes less than all the selected regions in the series thereof.

18. The method according to claim 17, wherein for those selected regions in the series thereof that do not experimentally obtain designated values for their pairs of correction factors, designated values for their correction factors are obtained via an interpolation from the experimentally obtained designated values.

19. The method according to claim 18, wherein the interpolation is carried out along a smoothly varying curve fit to the experimentally obtained designated values.

20. An MRI scanner comprising:

a main magnet that generates a substantially uniform temporally constant main magnetic field through an examination region wherein an object being imaged is positioned;

a magnetic gradient generator that produces magnetic gradients in the main magnetic field across the examination region;

a transmission system which includes an RF transmitter that drives an RF coil which is proximate to the examination region;

a sequence control which manipulates the magnetic gradient generator and the transmission system to produce an EPI pulse sequence, said EPI pulse sequence inducing a detectable magnetic resonance signal from the object;

a reception system which includes a receiver that receives and demodulates the magnetic resonance signal to obtain raw data;

a k-space storage device into which the raw data is loaded as a k-space data matrix;

a reconstruction processor that subjects the k-space data matrix to a reconstruction algorithm to generate image data which is loaded into an image data storage device;

an output device that produces human-viewable image representations of the object from the image data; and

,

an automatic ghost reducing means which automatically generates from the image data correction factors that are applied to the raw data for reducing ghosting in the image representations, wherein the correction factors include a phase correction factor and a data alignment correction factor, said phase correction factor compensating for phase errors in the raw data and said data alignment correction factor effectively shifting a data acquisition window under which the raw data is collected to thereby adjust an alignment of the raw data.

21. The MRI scanner according to claim 20, wherein the image data represents an array of pixels having complex intensity values with corresponding magnitudes, and the automatic ghost reducing means includes a summing processor that sums the magnitudes to thereby evaluate an amount of ghosting.

22. The MRI scanner according to claim 21, wherein the automatic ghost reducing means further includes an optimization engine that employs the sum of magnitudes from the summing processor as a cost function in a non-linear optimization to generate the correction factors.

23. The MRI scanner according to claim 22, wherein the automatic ghost reducing means further includes a correction factor storage device from which the generated correction factors are taken for application to the raw data.

24. A magnetic resonance imaging apparatus comprising:

means for applying an EPI pulse sequence with an MRI scanner to induce a detectable magnetic resonance signal from a selected region of a subject positioned in the MRI scanner;

means for receiving and demodulating the magnetic resonance signal to generate raw data;

means for applying a pair of ghost reducing correction factors to the raw data, said pair of corrections factors including a phase correction and an alignment correction, wherein the phase correction compensates for phase errors in the raw data and the alignment correction effectively shifts a data acquisition window under which the raw data was collected to thereby align the raw data with respect to itself in k-space;

means for loading the correction factor affected data into k-space to generate a k-space data matrix;

means for subjecting the k-space data matrix to a reconstruction algorithm to generate actual image data; and

,

means for calculating values for the pair of correction factors using the actual image data.