JPH0324850B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION (Industrial Application Field) The present invention relates to an apparatus for correcting positional/density distortion of a Fourier NMR (nuclear magnetic resonance) image caused by static magnetic field inhomogeneity.

(Prior art) NMR images captured using the Fourier method (including the spin warp method (Fig. 8)) have the following characteristics:
Position/density distortion occurs in the read direction y. If the readout gradient magnetic field is Gy, the original coordinates (x, y ₀ )
The point that falls on the point moves to the point with coordinates (x, y) according to the relationship y = y ₀ + D (x, y ₀ )/Gy, and the concentration at that point also changes to its original value (1-∂/ ∂y′D(x, y′) | _y ′ _=y0 /Gy).

(Problem to be solved by the invention) In order to correct such position/concentration distortion, conventionally, as shown in FIG. I was making corrections, but
In order to correct the image of an arbitrary cross-section, the static magnetic field strength is measured over the entire three-dimensional FOV, the non-uniformity information is memorized, and when a certain cross-section is imaged, the A procedure was required to extract non-uniform information from three-dimensional information. This requires the effort and effort of measuring the static magnetic field strength of the entire FOV before imaging, and a large amount of storage space to store this information.In addition, when correcting images, three-dimensional static magnetic field strength information is required. There was a problem in that complicated and time-consuming calculations were required to extract information on the plane corresponding to the image position from the image position.

The present invention has been made in view of these points, and its purpose is to read out multiple images at the same position and capture images with different gradients, without measuring the three-dimensional static magnetic field strength distribution in advance for correction. Fourier method that can correct position and density distortion based on
An object of the present invention is to provide a device for correcting positional/density distortion of an NMR image.

(Means for Solving the Problems) The first invention to solve these problems is an NMR imaging device using the spin warp method.
In an NMR image position/density distortion correction device, an imaging means for imaging the same cross-section of a subject with different readout gradient magnetic fields and obtaining a plurality of images regarding the same cross-section; calculation means for determining the magnitude of positional deviation of the same point from a plurality of images regarding the same cross-section and determining a non-uniform distribution of static magnetic field strength at each pixel position; In contrast, the present invention is characterized by comprising a correction means for correcting positional/density distortion caused by non-uniform static magnetic field strength using the non-uniform distribution obtained by the calculation means from the same image as the above-mentioned image. The second invention rotates the coordinate axes of the readout gradient magnetic field and the warp gradient magnetic field for the same cross section of the object at the NMR image position/density distortion correction position in the NMR imaging device using the spin warp method. an imaging means that obtains a plurality of images regarding the same cross-section by imaging the same cross-section, and determining the magnitude of positional deviation of the same point from the plurality of images regarding the same cross-section obtained by the imaging means;
calculating means for calculating the non-uniform distribution of static magnetic field strength at each pixel position; and calculating the non-uniform distribution calculated by the calculating means from the same image as the plurality of images obtained by the imaging means and correction means for correcting positional/concentration distortion caused by non-uniform static magnetic field strength.

(Embodiments) Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. First, the principle of the present invention will be explained using an example in which two images are used. Assume that the nonuniform portion of the static magnetic field strength is represented by D (x, y), and the two images (real images) are the readout gradient magnetic field.
When images are taken at Gy ₁ and Gy ₂ (Gy ₁ ≠ Gy ₂ ), the positional distortion of each image is y ₁ = y ₀ + D (x, y ₀ )/Gy ₁ ... Image 1 y ₂ = y ₀ +D(x, y ₀ )/Gy ₂ ... occurs according to Image 2. Therefore, the original coordinates (x, y ₀ )
In image 1, the point that falls is at the coordinates (x, y ₁ ), and
In image 2, move to the point with coordinates (x ₀ , y ₂ ). Also, as the reciprocal of the Jacobian of the transformation at this time, the concentration is multiplied by (1-∂/∂y′D(x,y′)| _y ′ _=y0 /Gy).

Transforming the above equation, D (x, y ₀ ) = (y ₁ − _{y 2} )・
Since Gy ₁・Gy ₂ / (Gy ₂ − _{Gy 1} ), y ₁ = y ₀ + {Gy ₂ / Gy ₂ − Gy ₁ )} × (y ₁ − y ₂ ) (1) y ₂ = y ₀ +{Gy ₁ /Gy ₂ −Gy ₁ )}×(y ₁ −y ₂ ) (2) The positional correspondence relationship is obtained. The value of y ₁ - _{y 2} on the right side of equations (1) and (2) is the amount of positional deviation of the same point in the two images, and therefore, as shown in Figure 2, If the amount of positional deviation can be estimated, the points (x, y ₁ ) and (x, y ₂ ) of each image can be estimated.
can be returned to a state without deviation (coordinates (x, y ₀ )).

Further, at this time, by multiplying the pixel density of each image by the Jacobian of conversion, the density distortion can also be corrected.

An embodiment of the present invention based on such a principle is shown in FIG. In the figure, 1 is an imaging means for obtaining an image from an NMR signal obtained using a pulse sequence based on the spin warp method.
In this method, the same cross section of a subject is imaged with different readout gradient magnetic fields, and a plurality of images regarding the same cross section are obtained. 2 is the imaging means 1
Calculation means 2 calculates the magnitude of positional deviation of the same point from a plurality of images related to the same cross section obtained by It has a matching function and an estimation function. 3 calculates the position/density distortion caused by the non-uniformity of the static magnetic field strength using the non-uniform distribution obtained by the calculation means 3 from the same image as the plurality of images obtained by the imaging means 1. This is a correction means that performs correction. This correction means 3
The correction function will be described later.

In the alignment function, the positions of each pixel in the two images are correlated to obtain the amount of deviation. As shown in Figure 3A, first, the pixel (i,
2×IW+1 pixels are extracted in the readout direction j with j) as the center, and this is used as a template. This template is applied to pixels (i,
Calculate the cross-correlation function between the template and image 2 while shifting the image by ΔW in the ±j direction with j) as the center. If the largest value among the obtained cross-correlation functions is greater than or equal to the predetermined threshold value T, a quadratic curve is fitted to 3 to 5 points including that value as shown in Figure 3B, and 2 The amount of positional shift between the two images at pixel (i, j) of image 1 is obtained from the position of the axis of symmetry of the following curve. Points where the position of the axis of symmetry deviates by more than the fitting point or where the correlation function is less than or equal to the cutoff value T are labeled as points that cannot be aligned.

When the above processing is performed for all pixels,
The amount of positional deviation can be found on a two-dimensional plane.

The estimation function two-dimensionally smoothes the amount of positional deviation determined by the processing of the alignment function as shown in FIG. 4A, and also estimates the amount of deviation at points where alignment is impossible. For this purpose, the pixel point (i, j) as shown in Figure 4 (b) is
The amount of deviation is extracted in a window of M×M pixels centered on . Among these, the positional deviation amounts of the points (shaded areas) where the alignment has been completed are added and averaged. The average value at this time is the smoothed output at the pixel point (i, j) (+ mark point in FIG. 4B). This processing is performed for all pixels, and the result is represented as a two-dimensional smooth curved surface.

In the distortion correction function, the value estimated above y ₁ −
Distortion correction is performed based on y ₂ using equations (1) and (2).

When correcting the positional distortion of image 1, as shown in FIG. 5, the pixel density at a required position in the corrected image is determined by interpolation from the pixel density near the corresponding position in image 1. The density distortion correction is performed by multiplying the pixel density by the separately calculated Jacobian J=1+∂/∂y′D(x,y′)| _y ′ _=y0 /Gy.

In the above processing, according to experiments, IW=
10, ΔW=10, T=0.7, M=51, 320×
It has been confirmed that position and density distortion can be corrected for 320 pixel images.

Note that the present invention is not limited to the embodiments, and may be implemented in the following manner.

It is also possible to use three or more images, in which case the amount of positional deviation can be calculated according to another combination, and the estimation accuracy can be improved by averaging the amounts of positional deviation.

In addition to using the cross-correlation function to align two images, SSDA, which is expected to be faster, can also be used.
(Sequential Similarity Detection
There are many possible methods such as Algorithm), and you can select an appropriate one from among them.

Alignment was performed here using a one-dimensional template, but if you use a two-dimensional template,
By simultaneously performing a certain degree of smoothing and by performing two-dimensional matching, it becomes possible to accurately measure positional deviation even in images with positional distortion in the warp axis direction.

Since the non-uniformity of the magnetic field strength generally changes quite gradually, it is possible to reduce the two images and align them, align them every few points, or calculate the correlation coefficient. do not have.

The estimation function may be performed by fitting a two-dimensional function globally or piecewise.

In order not to change the contribution of the relaxation times T ₁ and T ₂ in each image, it is necessary to change the amount of dephase at the same time as the readout gradient so as not to change the time at which the echo signal appears, or Gy ₂ =
-Gy ₁ can be used for correction. In the case of Gy ₂ =-Gy ₁ , the image is reversed in the reading direction, so it is necessary to correct this before use.

The S/N ratio may be improved by overlapping each image after correction and averaging the images.

When images are captured by rotating the coordinates of the warp axis and the readout axis, and the images are reversely rotated and superimposed, distortion occurs in the same magnitude but in a different direction than before.

Here, if each pixel is aligned in four directions and a point with a high degree of coincidence is found, the magnitude of the non-uniformity of the static magnetic field strength can be determined from the direction and the magnitude of the positional shift.

In this case, the same effect can be obtained by rotating the measurement target instead of rotating the coordinates.

To explain this in detail, since the direction in which distortion occurs is limited to the direction of the readout axis, there is a possibility of deviation in two directions, positive and negative. However, since the readout gradient Gy is the same for both images, the magnitude of the deviation is the same. As shown in FIG. 6, if the rotation angle of the coordinates is θ, the point that originally was at P ₀ will be, for example, point P ₁ in image 1, and will be at point P ₂ in image 2, for example. Therefore, all possible deviations can be considered by performing alignment in the four directions of angles ±α and ±β from _P1 .

Here, it has been confirmed that α=(π+θ)/2 and β=θ/2. Also, non-uniformity D
It is known from the equation given above that the magnitude of the positional shift relative to the distance is D/Gy, so as shown in FIG. 7, _{0 2} = _{0 1} =D/Gy.
Therefore, _{0 1} ² = _{0 2} ² + _{1 2} ² −2・_{0 2}・_1
2
From the formula of cos(π-α), the magnetic field inhomogeneity D(P ₀ ) up to the point P ₀
can be obtained as D(P ₀ )=−Gy・( _{1 2} )/2cosα).

Here, α is the direction in which the corresponding point was found,
Furthermore, _{1 and 2} represent the magnitude of positional deviation, and the method of this method can be applied as is by changing the alignment direction and angle.

(Effects of the Invention) As described above, according to the present invention, since distortion is corrected based on a plurality of real-value images to be corrected, the following effects can be obtained.

(1) There is no need to measure the static magnetic field inhomogeneity distribution before correction.

(2) There is no need to memorize the static magnetic field inhomogeneity distribution for correction.

(3) The effects of (1) and (2) above make it easy to correct arbitrary cross sections.

(4) Since it is possible to always make corrections according to the latest non-uniform distribution, it is also possible to eliminate the effects of changes over time and the proximity or introduction of magnetic materials.

[Brief explanation of the drawing]

Fig. 1 is a diagram showing an embodiment of the present invention, Fig. 2 is a diagram for explaining the principle of the invention, Fig. 3 is a diagram for explaining the alignment function, and Fig. 4 is a diagram for explaining the estimation function. FIG. 5 is a diagram for explaining the distortion correction function. FIGS. 6 and 7 are diagrams for explaining other embodiments of the present invention.
FIG. 8 is a diagram showing a pulse sequence of the spin warp method, and FIG. 9 is a diagram for explaining a conventional position/density distortion correction method. 1... Imaging means, 2... Calculation means, 3... Correction means.

Claims (1)

[Claims] 1. In an NMR image position/density distortion correction device in an NMR imaging device using the spin warp method, the same cross section of a subject is imaged with different readout gradient magnetic fields, and An imaging means for obtaining a plurality of images, and a size of positional deviation of the same point is determined from the plurality of images regarding the same cross section obtained by the imaging means, and a non-uniform distribution of the static magnetic field strength at each pixel position is determined. a calculating means; and a calculating means, for the plurality of images obtained by the imaging means, using the non-uniform distribution obtained by the calculating means from the same image as the image, to calculate the position and position caused by non-uniform static magnetic field strength. An apparatus for correcting position/density distortion of an NMR image, comprising: a correction means for correcting density distortion; 2 In an NMR image position/concentration distortion correction device in an NMR imaging device using the spin warp method, the same cross section of the subject is imaged by rotating the coordinate axes of the readout gradient magnetic field and the warp gradient magnetic field, and multiple images related to the same cross section are An imaging means for viewing one image, and a calculation for determining the magnitude of positional deviation of the same point from a plurality of images regarding the same cross section obtained by the imaging means, and determining the non-uniform distribution of static magnetic field strength at each pixel position. means, for the plurality of images obtained by the imaging means, using the non-uniform distribution obtained by the calculation means from the same image as the image, to calculate the position and concentration caused by non-uniform static magnetic field strength. An apparatus for correcting position/density distortion of an NMR image, comprising: a correction means for correcting distortion;