MRI METHODS AND APPARATUS FOR FLEXIBLE VISUALIZATION OF ANY SUBSET OF AN ENLARGED TEMPORAL WINDOW

• US 20150276909A1
• Filed: 05/09/2014
• Published: 10/01/2015
• Est. Priority Date: 03/25/2014
• Status: Active Grant

• Current Assignee: BETH Israel Deaconess Medical Center (Caregroup Incorporated)
• Original Assignee: BETH Israel Deaconess Medical Center (Caregroup Incorporated)
• Inventors: NEZAFAT, Reza, MANNING, Warren J., KAWAJI, Keigo
• Granted Patent: US 9,702,956 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: A61B 5/024 Detecting, measuring or recording pulse rate or heart rate A61B5/0205, A61B5/021 take precedence
• CPC Class Codes: A61B 5/055 involving electronic [EMR] or nuclear [NMR] magnetic resonance, e.g. magnetic resonance imaging
• CPC Class Codes: G01R 33/4826 in three dimensions
• CPC Class Codes: G01R 33/56308 Characterization of motion or flow; Dynamic imaging
• CPC Class Codes: G01R 33/5635 Angiography, e.g. contrast-enhanced angiography [CE-MRA] or time-of-flight angiography [TOF-MRA]
• CPC Class Codes: G01R 33/56509 due to motion, displacement or flow, e.g. gradient moment nulling G01R33/567 takes precedence
• CPC Class Codes: G01R 33/5673 Gating or triggering based on a physiological signal other than an MR signal, e.g. ECG gating or motion monitoring using optical systems for monitoring the motion of a fiducial marker

Abstract

A method and apparatus for MRI retrospective reconstruction determines a period of minimal coronary artery motion within the acquired temporal window with a rotation angle optimized for a K-space four-dimensional (4D) volume by a segmented 4D radial stack-of-stars (SOS) acquisition during a temporal acquisition window (taw). Radial stacks of the image of an object are acquired by performing the radial (SOS) acquisition to determine a plurality of k_z plane samples of the object, in which each k_z plane is repeatedly determined during a sub-window of the temporal acquisition window. Consecutive volumes of k_z centric slices of the image at each temporal sub-window of the temporal acquisition window are generated and summed to fill a 3D k-space volume. The radial (SOS) acquisition of the image is determined utilizing a customized rotation angle θ providing a uniform distribution of k-space spokes of the k_z centric slices for each sub-window of the (taw).

Claims

1. A method for controlling a magnetic resonance (MR) imaging system to provide retrospective reconstruction of an object image of a four-dimensional (4D) volume by an optimized rotation angle, the method comprising:

acquiring a temporal acquisition window based on an associated motion period of the object image,

acquiring a plurality of image slices of the object, in which each k_z plane is repeatedly determined during a sub-window of the temporal acquisition window;

in response to a request, selecting a temporal acquisition sub-window for retrospective processing of the image slices to determine a period of minimal objection motion within the acquired temporal window.

2. The method according to claim 1, wherein the rotation angle is optimized for a segmented 4D radial stack-of-stars acquisition during the temporal acquisition window by:

acquiring radial stacks of the image of the object by performing a radial stack-of-stars acquisition to determine a plurality of k_z plane samples of the object, in which each k_z plane is repeatedly determined during a sub-window of the temporal acquisition window;

generating a plurality of consecutive volumes of k_z centric slices of the image at each temporal sub-window of the temporal acquisition window that are summed to fill a 3D k-space volume; and

wherein the radial stack-of-stars (SOS) acquisition of the image is determined utilizing a customized rotation angle θ

to provide a uniform distribution of k-space spokes of the k_z centric slices of the image for each sub-window of the temporal acquisition window.

3. The method according to claim 2, wherein the k_z plane samples of the k_z centric slices of the image are acquired by performing Cartesian sampling of the object in a Z plane and radial sampling of the object in an X, Y plane.

4. The method according to claim 2, comprising performing radial sampling of the object by two planes and performing Cartesian sampling of the object by a third plane.

5. The method according to claim 2, wherein each sub-window of the temporal acquisition window is determined utilizing Golden Angle (GA) interleaving to determine the customized rotation angle θ

for radial spoke rotation.

6. The method according to claim 2, wherein the temporal acquisition window is enlarged according to a quantity of multiple heart beats per k_z slice.

7. The method according to claim 2, wherein an optimal k-space distribution over any temporal acquisition sub-window is determined as a function of:

a rotation angle θ

between consecutive radial spokes, a number of projection lines per heart beat (nTFE), a number of heart beats per k_z slice (nHB), and a number of projections in each heart beat used for reconstruction.

8. The method according to claim 2, wherein the temporal acquisition window is a sum of the temporal sub-windows.

9. The method according to claim 7, wherein the optimal k-space distribution comprises:

10. The method according to claim 2, wherein the reconstruction is a subset of the temporal acquisition window in which nRCW≦

nTFE in which nTFE comprises a number of projection lines per heartbeat, and nRCW is a number of projections in each heartbeat in reconstruction window.

11. The method according to claim 9, wherein Δ

Θ

<sub>θ

, nTFE</sub>^{nRHB, nRCW} is invariant to a reconstruction window position within the temporal acquisition window.

12. The method according to claim 3, wherein in the SOS acquisition is configured for enforcing uniform spacing and minimizing a clustering of spokes of a radial slice in accordance with following:

C(nTR,nHB)=arg min_θ

Σ

_{nRCW=N min}^{N max}∥

Δ

Θ

<sub>θ

, nTR</sub>^{nHB, nRCW}−

Δ

Θ

_{Ll, nTR}^{nHB, nRCW}∥

₂²+λ

·

max (|Θ

<sub>θ

, nTR</sub>^{nHB, nRCW}−

Θ

_{Ll, nTR}^{nHB, nRCW}|)

wherein Θ

<sub>θ

, nTR</sub>^{nHB, nRCW} comprises a vector corresponding to angles of a number of heart beats per k_z slice x number of projections in each heartbeat used for reconstruction (reconstruction window RCW) in which nHB·

nRCW radial spokes acquired within the reconstruction window over multiple heartbeats when a θ

rotation is used between subsequent spokes, Θ

_{Ll, nTR}^{nHB, nRCW} is a same vector of radial angles for a linear case with a 180°

/(nHB·

nRCW) rotation between subsequent spokes, and Δ

Θ

is a first derivative of Θ

, wherein a first left term enforces uniform spacing, while a second right term minimizes clustering.

13. The method according to claim 11, wherein a GA/n angle yielding a minimal objective function comprises a minimal objective function yielding a uniform distribution of radial spokes.

14. A method for controlling a magnetic resonance (MR) imaging system to provide interactive visualization of 4D MRI imaging for flexible gridding image reconstruction by a Graphics Processing Unit (GPU) for lag-free performance, the method comprising:

forwarding selected k-space points and coordinates from a selected slice of an image of an object and temporal window to a GPU reconstruction and visualization engine executed by the GPU comprising circuitry configured by machine executable code that is executed by the GPU for operation;

performing by the GPU rapid gridding on a per-point basis of the selected slice of the image of an object by indexing raw k-space data into k_x, k_y, k_z and t coordinates;

processing by the GPU a 2D image according to input parameters including the k_x, k_y, k_z and t coordinates mounted onto a GPU memory;

performing by the GPU a gridding operation on a per-point basis and accumulating gridded data from each GPU kernel onto a final Cartesian grid;

performing by the GPU an image reconstruction sequence, and mounting the reconstructed image back to a CPU memory for display by a display device.

15. The method according to claim 14, where performing the image reconstruction sequence includes performing (1) zero-padding, (2) inverse Fast Fourier Transform (FFT), (3) cropping to a correct field of view (FOV), and (4) roll-off correction.

16. The method according to claim 14, wherein performing the gridding operation by the GPU includes parallelizing the gridding operation on the per-point basis of the k_x, k_y, k_z and t coordinates.

17. The method according to claim 16, further comprising a graphic user interface (GUI) including machine executable code loaded for execution by a processor comprised of circuitry and output by a display, and wherein the method includes in response to forwarding by the GUI all selected k-space points and coordinates from a selected image slice and temporal window to the GPU for execution to reconstruct the image from the selected image slice and displaying the reconstructed image on the display.

18. The method according to claim 14, wherein indexing the raw k-spaced data includes generating a look-up table for referencing of k-spaced data from any desired temporal window;

employing a coil compression method to load a 3D k-spaced volume from channels of an MRI apparatus onto a standard workstation having a processor with circuitry configured for running a visualization platform;

compressing an indexed data to generate a hybrid k-space domain; and

mounting a compressed data onto a CPU memory for visualization output by the display device.

19. The method according to claim 17, wherein in response to GUI selection of the image slice and the temporal window, the GPU reconstruction and visualization engine performs the gridding operation of selected non-Cartesian k-space data, and outputs the reconstructed image for immediate display by the display device, wherein the display device comprises a GUI panel display.

20. The method according to claim 14, wherein the k-spaced points are acquired by performing a radial stack-of-stars sampling acquisition and was integrated with the GPU processing for rapid non-Cartesian gridding and real time visualization.

21. A method for controlling a magnetic resonance (MR) imaging system to address an inaccurate trigger delay derived from Scout Cine Steady State Free Precession (SSFP) in a whole-heart coronary MRI with an enlarged acquisition window, the method comprising:

setting an enlarged acquisition window in accordance with a subject'"'"'s observed heart rate;

determining a subject-specific trigger delay from an initial breath-held 2D cine at a beginning of an imaging protocol;

adjusting the subject-specific trigger delay to begin a predetermined time prior to a determined quiescent period from a scout cine scan;

performing a radial stack-of-stars scan of a whole heart coronary sequence;

performing k-space sampling over the enlarged acquisition window;

reconstructing offline images of the subject for display by a display device in which a temporal reconstruction window is optimized in response to user selected adjustment of the offline images of the subject being displayed by the display device.

22. The method according to claim 21, wherein the enlarged acquisition window was set to 36 times repetition (TR) based on a subject'"'"'s heart rate.

23. The method according to claim 21, wherein the enlarged acquisition window was set to 48 times repetition (TR) based on a subject'"'"'s heart rate.

24. The method according to claim 21, where the predetermined time prior to the determined quiescent period comprises 50 ms.

25. The method according to claim 21, wherein the predetermined time prior to a determined quiescent period ranges from 10-100 ms.

26. The method according to claim 21, wherein the temporal reconstruction window for a right coronary artery (RCA) and a left anterior descending artery (LAD) is initially set to a 120 ms window when performing optimizing of the temporal reconstruction window.

27. The method according to claim 21, wherein the k-space sampling performed over the enlarged acquisition window spans a greater time than a quiescent period derived from a scout sine scan.

28. An apparatus for providing flexible visualization of any subset of an enlarged temporal window of Magnetic Resonance (MR) imaging, the apparatus comprising:

an image generating unit including a Graphics Processing Unit (GPU);

a user interface that detects input from an operator the of the apparatus;

an image outputting unit including a display screen that displays an image of a temporal acquisition window of MR generated by the image generating unit; and

a control unit configured to control at least the image generating unit, image outputting unit and the user interface,

wherein in response to selection or input via the user interface, the control unit controls the image generating unit to generate reconstruction windows with a stack-of-stars acquisition and interleaving of radial stacks of MR images utilizing a customized rotation angle θ

to provide a uniform distribution of k-space spokes of k_z centric slices of a scanned image for each sub-window of a temporal acquisition window, and

the control unit is configured to process retrospectively a determination of a period of minimal coronary artery motion within an acquired temporal window in response to a request via the user interface.

29. The apparatus according to claim 28, wherein further comprising a GPU reconstruction and visualization engine executed by the GPU comprising circuitry configured by machine executable code that is executed by the GPU to perform rapid gridding on a per-point basis of a selected slice of the image of an object by indexing raw k-space data into k_x, k_y, k_z and t coordinates.

30. The apparatus according to claim 28, wherein a reconstructed image corresponds to cardiac data.

31. The apparatus according to claim 30, further comprising a magnetic resonance image (MRI) system for obtaining image data, which includes:

a main magnet that generates a static magnetic field for arranging directions of magnetic dipole movements of atomic nuclei;

a gradient coil that generates a gradient magnetic field;

an RF coil under control of the control unit;

a first signal generating unit in communication with the control unit generate a first signal to generate the magnetic field in the gradient coil and RF coil; and

a signal collecting unit.