Method for calculating signal-to-noise ratio in parallel acquisition image reconstruction

• US 7,719,275 B2
• Filed: 05/15/2008
• Issued: 05/18/2010
• Est. Priority Date: 05/17/2007
• Status: Active Grant

• Current Assignee: Siemens Healthineers AG (Siemens AG)
• Original Assignee: Siemens AG
• Inventors: Zhang, Bi Da, Fu, Cai Xia, Wang, Jian Min, Zhang, Qiang
• Primary Examiner(s): Shrivastav; Brij B
• Assistant Examiner(s): Vargas; Dixomara
• Application Number:
  US12/152,571
• Publication Number:
  US 20080285833A1
• Time in Patent Office: 733 Days
• Field of Search: 324300-322, 600410-435
• US Class Current: 324/309
• CPC Class Codes: G01R 33/5611 Parallel magnetic resonance imaging, e.g. sensitivity encoding [SENSE], simultaneous acquisition of spatial harmonics [SMASH], unaliasing by Fourier encoding of the overlaps using the temporal dimension [UNFOLD], k-t-broad-use linear acquisition speed-up technique [k-t-BLAST], k-t-SENSE structural details of arrays of sub-coils G01R33/3415

Abstract

The invention discloses a method for calculating the signal-to-noise ratio (SNR) in parallel acquisition image reconstruction, comprising: determining a reconstruction expression for a linear operation of the image reconstruction; determining a weighted coefficient according to the reconstruction expression; calculating the SNR according to the weighted coefficient and the raw data. The SNR not only is relevant to the geometric shape and position of the coils, but also is influenced by the reconstruction method and the sampling mode. The SNR is calculated based on contribution of the raw data at positions in the reading direction from all the phase-coding lines in all acquisition channels. It reflects more precisely the loss of the SNR in the parallel acquisition image reconstruction, especially the changes in the SNR caused by the number of the reference lines combined during the reconstruction.

Claims

1. A method for calculating a signal-to-noise ratio of a parallel-acquired image of an object in a parallel acquisition image reconstruction, comprising:

acquiring the parallel-acquired image by a MRI device;

determining a reconstruction expression for a linear operation of the image reconstruction, ρ

=Mr, wherein ρ

is a reconstructed image of the parallel-acquired image, M is the reconstruction expression, and r is raw data of the parallel-acquired image by an image processing unit;

determining a weighted coefficient of the image reconstruction according to the reconstruction expression by the image processing unit; and

calculating the signal-to-noise ratio of the parallel-acquired image according to the weighted coefficient and the raw data by the image processing unit,

wherein the raw data comprises a uniform under-sampling data in K-space, or

wherein the raw data comprises a mixed sampling data by combining the uniform under-sampling data and low frequency full-sampling data in the K-space.

2. The method as claimed in claim 1, further comprising evaluating a loss of the signal-to-noise ratio in the parallel acquisition image reconstruction according to a ratio of the signal-to-noise ratio of the parallel-acquired image to a signal-to-noise ratio of a standard-acquired image.

3. The method as claimed in claim 1, wherein the reconstruction expression is a reconstruction matrix.

4. The method as claimed in claim 1, wherein the image reconstruction is calculated on the K-space.

5. The method as claimed in claim 4, wherein the reconstruction expression is determined by a formula:

M=C·

F·

G, wherein;

G represents a K-space data-filling matrix which fills in uniform under-sampled K-space positions using a fitting coefficient while keeping a low frequency full-sampled data,

C represents a channel combining operation matrix which is equivalent to squared and combined SOS carried out by using complex conjugation of sensitivity of a coil of the MRI device, and

F represents a Fourier transformation matrix.

6. The method as claimed in claim 1, wherein the image reconstruction is calculated on image space.

7. The method as claimed in claim 6, wherein the reconstruction expression is an inversion of a sampling matrix:

M=inv(E′

·

E)·

E′

.

8. The method as claimed in claim 7, wherein the sampling matrix is determined by a formula:

E=U·

F<sup>−

1</sup>·

P, wherein;

U represents an under-sampling matrix,

F represents a Fourier transformation matrix, and

P represents a sensitivity distribution matrix of the coil of the MRI device.

9. The method as claimed in claim 8, wherein the under-sampling matrix comprises a uniform under-sampling matrix.

10. The method as claimed in claim 8, wherein the under-sampling matrix comprises a mixed under-sampling matrix by combining uniform under-sampling and low frequency full-sampling in K-space.

11. The method as claimed in claim 1, wherein the raw data comprises data at a same position in a reading direction from different acquisition channels and different phase-coding lines for sampling in a rectangular coordinate system.

12. The method as claimed in claim 11, wherein the reading direction is horizontal and a phase-coding direction is vertical.

13. The method as claimed in claim 12,

wherein a weighted coefficient for a pixel at a position (jRO,iPE) with a horizontal position set as jRO and a vertical position set as iPE is determined by a formula:

w_jRO^iPE=(w₁, w₂, . . . , w<sub>nCh×

nPE</sub>), and

wherein the pixel is determined by a formula;

Im a_jRO^iPE=w_jRO^iPE·

r_jRO,

wherein;

nCh represents a number of acquisition channels,

nPE represents a number of phase-coding lines,

w_jRO^iPE represents a row vector of the weighted coefficient of the pixel at the position (jRO,iPE),

(w₁, w₂, . . . , w<sub>nCh×

nPE</sub>) represents iPEth row in the reconstruction expression M_jRO,

r_jRO represents a column vector of raw data at a same position of jRO in the reading direction in K-space from different channels and different phase-coding positions, and

Im a_jRO^iPE represents the pixel at the position (jRO, iPE) in the reconstructed image.

14. The method as claimed in claim 13, wherein the signal-to-noise ratio of the parallel-acquired image is calculated by a formula:

$S<br><br><br><br>N<br><br><br><br>R_{\mathrm{jRO}}^{\mathrm{iPE}}=\frac{<br><br>w_{\mathrm{jRO}}^{\mathrm{iPE}}·<br><br>r_{\mathrm{jRO}}<br><br>}{\sqrt{w_{\mathrm{jRO}}^{\mathrm{iPE}}·<br><br>\mathrm{\psi <br><br>}·<br><br>{\left(w_{\mathrm{jRO}}^{\mathrm{iPE}}\right)}^{\mathrm{\prime <br><br>}}}<br><br>\sqrt{\mathrm{mPE}}},$

wherein;

SNR_jRO^iPE represents the signal-to-noise ratio of the pixel at the position (jRO,iPE) of the parallel-acquired image,

mPE represents a number of actually acquired phase-coding lines, and

ψ

represents a noise correlation matrix.

15. The method as claimed in claim 1, wherein the raw data comprises K-space data from all acquisition channels for sampling in a non-rectangular coordinate system.