Denoise MCG measurements

• US 9,089,274 B2
• Filed: 09/22/2011
• Issued: 07/28/2015
• Est. Priority Date: 01/31/2011
• Status: Active Grant

• Current Assignee: Seiko Epson Corporation (Seiko Group)
• Original Assignee: Seiko Epson Corporation (Seiko Group)
• Inventors: Wu, Chenyu, Xiao, Jing
• Primary Examiner(s): Gupta, Vani
• Application Number:
  US13/239,997
• Publication Number:
  US 20130079622A1
• Time in Patent Office: 1,405 Days
• Field of Search: 600/407, 600/409
• US Class Current: 1/1
• CPC Class Codes: A61B 2562/046 in a matrix array
• CPC Class Codes: A61B 5/0044 for the heart
• CPC Class Codes: A61B 5/05 Detecting, measuring or recording for diagnosis by means of electric currents or magnetic fields; Measuring using microwaves or radio waves impedance plethysmography A61B5/0295; measuring movement of the entire body or parts thereof A61B5/11; detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof A61B5/24
• CPC Class Codes: A61B 5/242 Detecting biomagnetic fields, e.g. magnetic fields produced by bioelectric currents
• CPC Class Codes: A61B 5/243 specially adapted for magnetocardiographic [MCG] signals

Abstract

A magnetocardiogram (MCG) system with reduced noise artifacts is produced by first creating high-resolution image representations of low-resolution measurements obtained with a magnetic field sensor unit. The high-resolution image representations are created by use of a PCA model that has been trained using a library of ideal, no-noise, high-resolution images. The Biot-Sarvart Law is then used to create a 3D model of a current impulse, given the high-resolution image representations. From the 3D current impulse model, ideal sensor unit measurements as they would have been obtained using a theoretical sensor unit observing the 3D current impulse model are synthesized.

Claims

1. A magnetocardiogram (MCG) system comprising:

a sensor unit including an array of M×

M electromagnetic MCG sensors producing a sparse measurement output of M×

M data values, said sparse measurement output constituting a first MCG image, said first MCG image being a two-dimensional (2D) image having M×

M picture elements (e.g. pixels);

a high resolution MCG image synthesizer for receiving said first MCG image and producing a higher resolution image representation of said first MCG image, said higher resolution image representation being a second MCG image of higher pixel density than said first MCG image, said second MCG image being a 2D image having P×

P picture elements where P>

M; and

a de-noising image processing block for receiving said second MCG image and producing a de-noised lower resolution image representation of said second MCG image, said de-noised lower resolution image representation being a third 2D MCG image having M×

M picture elements within a physical sensor area corresponding to the span of said M×

M data values of said first MCG image, and said third MCG image having a reduced noise level as compared to said first MCG image.

2. The MCG system of claim 1, wherein said de-noising image processing block identifies as target picture elements the picture elements of said second MCG image whose image locations correspond to the image locations of the M×

M data values of said first MCG image, and populates said physical sensor area with said target pictures elements.

3. The MCG system of claim 1, wherein the producing of said second MCG image by said high resolution MCG image synthesizer includes:

accessing a linear model defining a model MCG image of higher resolution than said first MCG image, said linear model establishing interpolation patterns between characteristics of the linear model and any data value of said sparse measurement output of M×

M data values; and

producing an intermediate MCG image by projecting said first MCG image onto the subspace of the linear model, and establishing coefficients for said intermediate MCG image in accordance with the linear model and said M×

M data values.

4. The MCG system of claim 3, wherein said high resolution MCG image synthesizer outputs said intermediate image as said second MCG image, and each of said electromagnetic MCG sensors is a superconducting quantum interference device (SQUID).

5. The MCG system of claim 3, wherein said linear model is defined by creating a plurality of synthesized magnetocardiogram images having the same resolution as said second MCG image, said synthesized magnetocardiogram images being based on simulated electrical impulses within a three-dimensional spatial heart volume, as it would be perceived in an expected magnetocardiogram system.

6. The MCG system of claim 5, wherein said synthesized MCG images are based on randomly generated currents within said heart volume, and said linear model is created by using principal component analysis (PCA).

7. The MCG system of claim 3, wherein said interpolation patterns are established by the following steps:

(A) defining the following notation;

N×

N dense Bz magnetic field map to form a vector;

M×

M sparse measurement to form a vector;

K randomly generated single current dipoles Q;

(B) for each randomly generated current Q, computing an N×

N magnetic field map using Biot-Savart equation and stack the resultant image to a vector f₁;

(C) repeating step (B) to obtain K samples and getting a data matrix A=└

f₁, f₂, . . . f_K┘

; and

(D) training a PCA model given input data A, to obtain the eigenmatrix Σ

_f.

8. The MCG system of claim 7, wherein said intermediate MCG image is created by:

given a new dipole and M×

M sparse measurements g_j, finding the corresponding rows in the eigenmatrix, and denoting a resultant submatrix as Σ

_g;

projecting the sparse measurement to the PCA subspace and computing the coefficients as c_g=Σ

_g⁺(g_j−

g_mean), where Σ

_g⁺is the pseudo inverse of Σ

_g; and

using the computed coefficients and original PCA space to reconstruct the dense magnetic field map Bz, as f_j=Σ

_fcg+f_mean.

9. The MCG system of claim 3, wherein said high resolution MCG image synthesizer includes an electric current localizer for determining a position and momentum of an electric current in accord with said intermediate MCG image, said electric current localizer evaluating the electromagnetic output data from each electromagnetic sensor in an x-y orientation (Bxy) assuming single dipole, computing dense Bxy from dense Bz, finding the image maximum in said intermediate MCG image, and using this determined position information as a starting point in an iterative process for identifying a three-dimensional position vector {right arrow over (p)} and momentum vector {right arrow over (J)} for said electric current;

and

said high resolution MCG image synthesizer uses the Biot-Sarvart Law for producing said second MCG image based on the identified three-dimensional position vector {right arrow over (p)} and momentum vector {right arrow over (J)}.

10. The MCG system of claim 9, wherein the identifying of the three-dimensional position vector {right arrow over (p)} and momentum vector {right arrow over (J)} for said electric current further includes:

Given said third MCG image B_z(i,j)(i=1,2, . . . ,N;

j=1,2, . . . ,N), the maximal point of the tangential components B′

_xy(i,j)of B_z(i,j) refers to the 2D position (x_p,y_p) of the electric current, and the tangential components of B_z(i,j) is computed as B_xy(i,j)=√

{square root over ((∂

B_z(i,j)/∂

x)² +(∂

B_z(i,j)/∂

y)² )}{square root over ((∂

B_z(i,j)/∂

x)² +(∂

B_z(i,j)/∂

y)² )}; and

said iterative process for identifying position vector {right arrow over (p)} and momentum vector {right arrow over (J)} for said electric current includes;

(a) defining the Biot-Sarvart Law as {right arrow over (B)}^m={right arrow over (J)}×

{right arrow over (R)}_m={right arrow over (J)} where {right arrow over (B)}^m={right arrow over (B)}({right arrow over (r)}_m),{right arrow over (J)}={right arrow over (J)}({right arrow over (p)}) and

11. The MCG system of claim 1, wherein

said second MCG image spans an imaging area greater than said first MCG image.

12. The MCG system of claim 11, wherein:

said third MCG image spans an imaging area greater than said first MCG image;

the pixel density of the third MCG image'"'"'s physical sensor area is uniformly extended throughout said third MCG image; and

the pixels of said third MCG image that lay beyond the physical sensor area are populated with simulated sensor data determined from said second MCG image.

13. The MCG system of claim 12, wherein said de-noising image processing block identifies as target pixels the picture elements of said second MCG image whose pixel image locations correspond to the pixel image locations of said third MCG image, and populates the pixel image locations of said third MCG with their corresponding target pictures elements.

14. The MCG system of claim 12, wherein the imaging span of said third MCG image is substantially the same as the imaging span of said second MCG image.

15. The MCG system of claim 11, wherein the producing of said higher resolution image representation of said first MCG image by said high resolution MCG image synthesizer includes:

accessing a linear model defining a model MCG image of substantially higher resolution than said first MCG image, said linear model being defined by creating a plurality of synthesized magnetocardiogram images having the same size and resolution as said second MCG image;

said synthesized magnetocardiogram images being based on simulated electrical impulses within a three-dimensional spatial heart volume, as it would be perceived in a hypothetical magnetocardiogram system of similar resolution as said second MCG image;

information from said plurality of synthesized magnetocardiogram images being incorporated into said linear model to establish interpolation patterns between characteristics of the linear model and any data value of said sparse measurement output; and

said producing of said second MCG image including the projecting said first MCG image onto the subspace of the linear model to establishing coefficients for use in the creation of said second MCG image.

16. The MCG system of claim 15, wherein said linear model is defined by:

(A) defining K simulated electrical impulse current dipoles Q;

(B) for each simulated electrical impulse Q, computing an N×

N magnetic field map using Biot-Savart equation and stacking the resultant image to a vector f₁, so as to define a data matrix A=└

f₁,f₂. . . f_K┘

for the K simulated electrical impulses;

(C) defining a mean vector from said data matrix A of vectors; and

(D) defining a covariance matrix by determining the vector components needed for constructing the covariance matrix from said data matrix A, dividing said covariance matrix into L×

L blocks, computing each of said L×

L blocks separately by accessing into an active data memory only the vector components needed the L×

L block being currently processed, storing the computed result for L×

L block being currently processed, and combining the results of all L×

L blocks.

17. A magnetocardiogram (MCG) system comprising:

a sensor unit including M×

M electromagnetic sensors producing a sparse measurement output of M×

M data values, said sparse measurement output constituting a first MCG image;

a high resolution MCG image synthesizer for receiving said first MCG image and producing a higher resolution image representation of said first MCG image, said higher resolution image representation being a second MCG image of higher pixel density than said first MCG image; and

a de-noising image processing block for receiving said second MCG image and producing a de-noised lower resolution image representation of said second MCG image, de-noised lower resolution image representation being a third MCG image having M×

M picture elements within a physical sensor area corresponding to the span of said M×

M data values of said first MCG image and having a reduced noise level as compared to said first MCG image;

wherein the producing of said second MCG image by said high resolution MCG image synthesizer includes;

accessing a linear model defining a model MCG image of higher resolution than said first MCG image, said linear model establishing interpolation patterns between characteristics of the linear model and any data value of said sparse measurement output of M×

M data values; and

producing an intermediate MCG image by projecting said first MCG image onto the subspace of the linear model, and establishing coefficients for said intermediate MCG image in accordance with the linear model and said M×

M data values; and

wherein the producing of said intermediate MCG image includes;

defining the sparse measurement output as a vector g;

defining the linear model as Σ

;

extracting from Σ

the row corresponding to sparse measurement output to form a sub-eigenmatrix Σ

_g;

projecting g onto Σ

_g;

defining the establishment of coefficients as c_g=Σ

_g⁺(g_i−

μ

_g), where Σ

_g⁺ is the pseudo inverse of Σ

_g,μ

_g are extracted coefficients from a mean vector μ

of linear model Σ

; and

defining the intermediate MCG image vector h as h=Σ

·

c_g+μ

.

18. A magnetocardiogram (MCG) system comprising:

a sensor unit including an array of M×

M electromagnetic sensors producing a sparse measurement output of M×

M data values, each of said electromagnetic sensors being a superconducting quantum interference device (SQUID), said sparse measurement output constituting a first MCG image, said first MCG image being a two-dimensional (2D) image having M×

M picture elements (e.g. pixels), one per SQUID device;

a high resolution MCG image synthesizer for receiving said first MCG image and producing a higher resolution image representation of said first MCG image, said higher resolution image representation being a second MCG image of higher pixel density than said first MCG image said second MCG image being a 2D image being a 2D image having a P×

P picture elements where P>

M, said second MCG image being a 2D image spanning an imaging area greater than the image area of said first MCG image and encompassing the image area of said first MCG image.

19. The MCG system of claim 18, wherein the producing of said higher resolution image representation of said first MCG image by said high resolution MCG image synthesizer includes:

accessing a linear model defining a model MCG image of substantially higher resolution than said first MCG image, said linear model being defined by creating a plurality of synthesized magnetocardiogram images having a substantially similar size and resolution as said second MCG image;

said synthesized magnetocardiogram images being based on simulated electrical impulses within a three-dimensional spatial volume as it would be perceived in a hypothetical magnetocardiogram system of similar resolution and image size as said second MCG image;

information from said plurality of synthesized magnetocardiogram images being incorporated into said linear model to establish interpolation patterns between characteristics of the linear model and any data value of said sparse measurement output; and

said producing of said second MCG image including the projecting said first MCG image onto the subspace of the linear model to establishing coefficients for said second MCG image.

20. The MCG system of claim 19, wherein:

said three-dimensional spatial volume is a simulated heart tissue volume;

said synthesized MCG images are based on randomly generated currents within said simulated heart tissue volume; and

said linear model is created by using principal component analysis (PCA).

21. The MCG system of claim 19, wherein said linear model is defined by:

(A) defining K simulated electrical impulse current dipoles Q;

(B) for each simulated electrical impulse Q, computing an N×

N magnetic field map using Biot-Savart equation and stacking the resultant image to a vector f₁, so as to define a data matrix A=└

f₁,f₂, . . . ┘

for the K simulated electrical impulses;

(C) defining a mean vector from said data matrix A of vectors; and

(D) defining a covariance matrix by determining the vector components needed for constructing the covariance matrix from said data matrix A, dividing said covariance matrix into L×

L blocks, computing each of said L×

L blocks separately by accessing into an active data memory only the vector components needed the L×

L block being currently processed, storing the computed result for L×

L block being currently processed, and combining the results of all L×

L blocks.