CN1843295A - Magnetic resonance image scanning automatic positioning method based on human anatomic structure symmetry - Google Patents

Abstract

The invention discloses a method of automatic location for magnetic resonance imaging scanning based on the symmetry of dissect structure of human body. It comprises following steps: (a) getting MRT location image; (b) calculating location information of different position on MRT location image by using the symmetry of dissect structure of human body; (c) outputting location information of MRT location image; and step (b) also comprises steps: (1) dividing target zone on whole image; (2) initiating symmetrical axis of target zone; (3) calculating the point set of said target zone symmetry with said symmetrical axis; (4) defining price function according to said point set, and getting optimum value of price function through optimization; (5) getting the symmetrical axis used as location information in relative with said optimum value. The invention realizes automatic location of MRT scanning, guarantees uniformity of fault scanning image and increases location accuracy and work efficiency.

Description

Based on the symmetric magnetic resonance image scanning automatic positioning method of human anatomic structure

Technical field

The present invention relates to a kind of medical image processing localization method, specifically be based on the symmetric magnetic resonance image scanning automatic positioning method of human anatomic structure.

Background technology

In MRI (Magnetic Resonance Imaging, nuclear magnetic resonance) scanning, need the positioning image of scanning acquisition in advance, in positioning image, set the reference location line, come to determine the position and the direction of MRI tomoscan thus.

In the prior art, generally can be in positioning image generating geometric center with positioning image automatically by system is center, the reference location line parallel with x axle or y direction of principal axis, and the position of reference location line and direction have determined the position and the direction of MRI tomoscan.But in actual applications, the position of clinical requirement reference location line is corresponding with the position of scanning with direction, as with the image of brain transverse section as positioning image, when on this positioning image, setting the reference location line of scanning sagittal plane tomography, the reference location line be centered close to big Schizencephaly center, direction and Schizencephaly is parallel greatly; And with the image of lumbar vertebra coronalplane as positioning image, when on this positioning image, setting the reference location line of scanning sagittal plane tomography, the reference location line be centered close to the spinal cord center, direction is parallel with spinal cord.Therefore, be that center, the reference location line parallel with x or y direction of principal axis can not satisfy clinical demand only with the geometric center of positioning image.

For addressing this problem, the position of manual setting reference location line and the method for direction are provided in the prior art again, to satisfy the needs that different parts carried out the MRI tomoscan.But, adopt the position and the direction of manual setting reference location line to have the following disadvantages:

1,, can't guarantee the concordance of reference location line owing to inevitable personal error;

2, manually-operated positioning accuracy is low;

3, increase doctor's workload, wasted sweep time, reduced doctor's work efficiency.

Summary of the invention

In order to overcome above-mentioned deficiency, the object of the present invention is to provide a kind of doctor of convenience operate, shorten patient total scanning time based on the symmetric magnetic resonance image scanning automatic positioning method of human anatomic structure.

For achieving the above object, the technical solution used in the present invention is:

Comprise following steps: a) obtain the MRI positioning image; B) utilize the symmetry of human anatomic structure, on the MRI positioning image, calculate the locating information of different azimuth automatically; C) locating information of output MRI positioning image.

Above-mentioned steps b) be specially:

1) in entire image, is partitioned into the target area; 2) axis of symmetry of the above-mentioned target area of initialization; 3) calculate in the above-mentioned target area about the symmetric point set of above-mentioned axis of symmetry; 4) based on above-mentioned point set definition cost function, obtain above-mentioned cost function optimal value by optimization process; 5) obtain the axis of symmetry as locating information corresponding with above-mentioned cost function optimal value;

Above-mentioned steps 2) axis of symmetry of the above-mentioned target area of initialization is based on that the main shaft that calculates above-mentioned target area carries out; Above-mentioned steps 4) cost function is defined as the absolute value sum of point of symmetry gray scale difference in the above-mentioned target area.

The present invention has following beneficial effect and advantage:

1. realized the automatic accurate location of MRI scanning.Because position line of the present invention is based on human anatomic structure symmetry feature and generates automatically, but the clinical location requirement of all promising policy does not need to carry out manual setting again;

2. guarantee the concordance of tomoscan image.The present invention can calculate the position line of target area automatically, exactly, guarantee the same doctor in the different time or different doctor at one time in concordance during to same spots localization, thereby guarantee the concordance of tomoscan image;

3. raising positioning accuracy.Personal error when the present invention can reduce manual location aspect clinical practice does not calculate the position line of target area exactly, and is especially good in the positioning image image quality, when the target area feature is difficult to recognize, locating effect is still fine;

4. improved work efficiency: the present invention does not adjust by hand repeatedly owing to the position line that can disposablely determine automatically to satisfy the demand, thereby can reduce patient's sweep time, improves doctor's work efficiency;

5. applied range: the present invention can also expand being generalized to other and possessing the symmetric position of anatomical structure except that the location that can be applicable to brain transverse section, cervical vertebra coronalplane, lumbar vertebra coronalplane.

Description of drawings

Fig. 1 is the inventive method flow chart;

Fig. 2 is an image symmetrical principle key diagram of the present invention;

Fig. 3～5 are design sketch of the present invention ()～(three).

The specific embodiment

As shown in Figure 1, the present invention specifically comprises the steps:

S10: in the MRI entire image of obtaining, be partitioned into the target area, promptly use image segmentation, as (present embodiment adopt big Tianjin method) such as bimodal method, big Tianjin method or iterative methods, the threshold value that target area and background area are cut apart in the computed image uses this threshold value to be partitioned into target area in the image;

S20: calculate the main shaft of target area, as initial axis of symmetry, the axis of symmetry equation can be expressed as: y=ax+b main shaft;

In the present embodiment, main shaft is as the initial value of axis of symmetry, and the part of calculating main shaft can be divided into for two steps, a step be the center of gravity of calculating the target area earlier (x, y), as the point on the main shaft; Second step was an angle [alpha] of calculating main shaft ₀, the angle of calculating main shaft can be according to formula ${\mathrm{\α}}_0=\frac{1}{2}\arctan (-\frac{{2I}_{\mathrm{xy}}}{I_x-I_y})$ Calculate,

Wherein: I _xAnd I _yBe the inertia distance, I _XyBe the product of inertia, their computing formula is $I_x={\&Integral;}_{\mathrm{<figure-callout\; id="2"\; label="A\; y"\; filenames="A20061004612100071.png"\; state="\{\{state\}\}">A</figure-callout>}}{y}^{}{}^2\mathrm{dA}$ $I_y={\&Integral;}_{\mathrm{<figure-callout\; id="2"\; label="A\; x"\; filenames="A20061004612100071.png"\; state="\{\{state\}\}">A</figure-callout>}}{x}^{}{}^2\mathrm{dA}$ With $I_{\mathrm{xy}}={\&Integral;}_A\mathrm{xydA},$ A is the target area;

S30: calculate in the target area about the symmetric point set of axis of symmetry; Look for the equidistant point with axis of symmetry on the vertical direction of axis of symmetry, these points are exactly point of symmetry, and all point of symmetry have just been formed symmetrical point set;

S40: the calculation cost function f (u, v), promptly calculate the absolute value sum of point of symmetry gray scale difference in the target area: $f(u,v)=\underset{i=0}{\overset{n}{\mathrm{\&Sigma;}}}|G(x_{i0},y_{i0},z)-G(x_{\mathrm{<figure-callout\; id="1"\; label="i"\; filenames="A20061004612100071.png"\; state="\{\{state\}\}">i</figure-callout>}1},{\mathrm{<figure-callout\; id="1"\; label="y\; i"\; filenames="A20061004612100071.png"\; state="\{\{state\}\}">y</figure-callout>}}_i,z)|$

Wherein: G (x _I0, y _I0, z), G (x _I1, y _I1, z) being the gray value of point of symmetry respectively, n is a point of symmetry group number in the target area, z is an axis of symmetry;

S50: cost function value whether optimum (it is optimal value that present embodiment adopts minima) among the determining step S40;

S60: when the cost functional value hour, search is finished, and exports the axis of symmetry as locating information corresponding with above-mentioned cost function optimal value;

S70: otherwise revise axis of symmetry, return S30, recomputate cost function search.

Above-mentioned steps S30 is the main shaft of calculating with the major axes orientation and the stasiofax of the target area initial value as axis of symmetry to step S50, uses optimized search algorithm search axis of symmetry position and direction accurately, as the Powell searching algorithm.

Consider that human anatomic structure has symmetric characteristics, this symmetry not only is embodied on the geometry, and on the gray scale of image, also has certain symmetry, as shown in Figure 2, straight line is an axis of symmetry, and circle, triangle, square are respectively about the symmetric three groups of point of symmetry of axis of symmetry, suppose that gray value is respectively 1,2,3, if image is symmetry fully, so the difference of every group of point of symmetry gray value equals zero, and the absolute value sum of the gray scale difference of entire image symmetry point set equals zero.Symmetric two gray values of pixel points might not equate in the real image, but it is close, therefore can in image, move or the axis of symmetry of image rotating, whenever move a position, calculate the summation of entire image about the gray scale difference of symmetric all pixels of axis of symmetry, when this summation reached minima, the axis of symmetry of this position was the target area accurately as the axis of symmetry of locating information.

The present invention calculates the angle [alpha] of the main shaft of target area ₀And center of gravity (x, y) as the axis of symmetry initial value, the expression formula of target area axis of symmetry lateral symmetry point set gray scale difference sum as cost function; Change focus point (x then within the specific limits, y) three parameters of coordinate and axis of symmetry direction, use the minima of optimized search algorithm search cost function, what obtain when search is finished is the axis of symmetry of being asked for, and this axis of symmetry of asking for is the position line of MRI positioning image.

The method is applied to brain transverse section, cervical vertebra coronalplane, lumbar vertebra coronalplane simultaneously and other possess the location at anatomical structure symmetry position, effect such as Fig. 3～shown in Figure 5, be respectively brain transverse section locating effect figure, cervical vertebra coronalplane locating effect figure and lumbar vertebra coronalplane locating effect figure.

Claims (4)

1. one kind based on the symmetric magnetic resonance image scanning automatic positioning method of human anatomic structure, it is characterized in that comprising following steps:

A) obtain the MRI positioning image;

B) utilize the symmetry of human anatomic structure, on the MRI positioning image, calculate the locating information of different azimuth automatically;

C) locating information of output MRI positioning image.

2. described based on the symmetric magnetic resonance image scanning automatic positioning method of human anatomic structure by claim 1, it is characterized in that: its step b) is specially:

1) in entire image, is partitioned into the target area;

2) axis of symmetry of the above-mentioned target area of initialization;

3) calculate in the above-mentioned target area about the symmetric point set of above-mentioned axis of symmetry;

4) based on above-mentioned point set definition cost function, obtain above-mentioned cost function optimal value by optimization process;

5) obtain the axis of symmetry as locating information corresponding with above-mentioned cost function optimal value.

3. described based on the symmetric magnetic resonance image scanning automatic positioning method of human anatomic structure by claim 2, it is characterized in that: described step 2) axis of symmetry of the above-mentioned target area of initialization is based on that the main shaft that calculates above-mentioned target area carries out.

4. described based on the symmetric magnetic resonance image scanning automatic positioning method of human anatomic structure by claim 2, it is characterized in that: described step 4) cost function is defined as the absolute value sum of point of symmetry gray scale difference in the above-mentioned target area.

Images