JPH01291388A - Method and device for discriminating streak from two-dimensional distribution of physical quantity - Google Patents

Abstract

PURPOSE:To discriminate linear and streak-like shapes at a high speed by extracting the physical quantity having a two-dimensional expanse from a data storage and generating projection data corresponding to each direction, and comparing it with data of the vicinity or a wide range. CONSTITUTION:A data storage (DS) 1 stores the physical quantity (input data) having the expanse of a three-dimensional space, a projection data generating device (PRJ) extracts object two-dimensional data, and after various processings corresponding thereto have been performed, projection data corresponding to each direction theta is derived by an operation by a Fourier-transform or a direct operation. Also, a shape discriminating device DET discriminates and decides whether linear and streak-like shapes exist or not from the projection data in each theta direction and length of each projection by data of a data storage device DS 2. Subsequently, discriminating and deciding information (or the correction quantity, the manipulated variable, etc., corresponding thereto) is outputted to a data storage DS 3 (or an arithmetic unit, an operating device, etc.). In such a way, a streak-like false image can be discrminated distinctly and at a high speed.

Description

Detailed Description of the Invention (Field of Industrial Application) The present invention provides a method for rapidly identifying (including detection, hereinafter simply referred to as identification) linear and streak-like shapes from a two-dimensional distribution of physical quantities, and a method thereof. An apparatus for using the method.

(Prior Art) In the tomographic images of X-ray tomography devices (hereinafter referred to as X-ray CT), high-speed identification of straight and streak-like shapes in the image is an important issue for the purpose of reducing streak-like artifacts. Therefore, such a method is desired. Similarly, medical image information, meteorological image information, resource satellite or meteorological satellite information,
Photographs, TV images, two-dimensional heat distribution and stress distribution of objects, etc. 2
In the image display of various physical quantities with a dimensional spread, there is also a need for high-speed identification of linear and streak-like shapes.

Hereinafter, in order to avoid complicating the explanation, in this specification, among various physical quantities, the explanation will be limited to image information obtained by X-ray CT.

In X-ray CT, as is well known, an X-ray source irradiates a subject from an X-ray source in each direction, the transmitted X-rays are detected by a detector, the data is reconstructed into an image, and the image is displayed.

By the way, projection data (
Also called projection data or projection data. Hereinafter, this is simply referred to as projection data. ) uses projection data measured by offset detection due to the requirement for higher resolution and lower image artifacts for the same geometrical conditions of the measurement system. Here, 1/4-1/4 is typical of the offset detection measurement method.
This section describes sample data measurement and collection using the offset method. For example, in the third-generation Xl1lCT, this sample data collection method rotates an X-ray source and a multi-channel detector that face each other with the object in between, and collects sample data in multiple directions. In this case, the X-rays emitted from the X-ray source through the center of rotation into the central channel of the multichannel detector are detected so that they are incident on a point shifted from the center of the channel by 1/4 of the channel spacing, that is, the sample spacing. The device is positioned to collect data.

In the multidirectional projection data obtained in this way,
In the case of parallel beams, those whose projection directions are opposite are
Since the path of the line beam is shifted by 1/2 of the detector channel spacing, that is, the sample spacing, when data with such a relationship is combined, the detector channel spacing, that is, the sample spacing can be finely divided by 1/2. The image will be reconstructed based on data equivalent to that obtained, and an image with high spatial resolution and few artifacts can be obtained.

It is well known that in XICT, streak-like artifacts occur due to various causes and are a major hindrance to image diagnosis. In order to identify and reduce such streak-like artifacts, high-speed generation of offset-detected projection data is indispensable, and the present inventor has developed a high-speed projection data calculation based on the Fourier transform method, which is the basic technology, and its device. has already been granted patent application No. 62-33544.
The proposal is made by No. 9. This projection data generation method using the Fourier transform method can be applied to parallel beams, reduces the amount of calculations by mainly using the so-called fast Fourier transform, enables high-speed calculations, and provides accurate and high-precision calculation results. It has certain advantages. The device configuration in this case can be a simple and economical configuration mainly consisting of a general-purpose Fourier transform device.

Here, projection and offset detection will be explained.

In Figure 2, xy is a rectangular coordinate, and XY is a xyX coordinate θ
The coordinate system is rotated by (counterclockwise direction θ>0). Spatial distribution μ(×.

For (2), the following line integral a(X, θ) on the straight line x-X of the XY coordinates is called projection data.

...(1) In FIG. 2, O is the origin of the xy, XY coordinate system.

Now, when measuring projection data or measuring needle lines at discrete points on the XY coordinates, on the X coordinates; ・d(i=O9±1.±2
．． ...; d is the sample interval) The set of projection data at the point is (a(X+, θ)=a(i-d, θ)=8
+:i=0. ±1. ±2. ...) becomes. That is, the measurement point (calculation point) of the projection data in this case corresponds to an integer point whose unit is d of the X coordinate.

On the other hand, on the X coordinate i-d +Xo (i = O, ±1
゜±2. ...; Xo is a constant), the set of projection data is (a(Xl', θ) = a(t-d +Xo, θ〉
=fll'; 1-0. ±1. ±2. ...) becomes. That is, the measurement points (calculation points) of the projection data in this case are points separated by Xo from each integer point whose unit is d of the X coordinate. In other words, the latter is
This is a data group in which an offset of o is detected.

(Problems to be Solved by the Invention) By the way, in X-ray CT, partial volume effect (a phenomenon in which a local change in the shape of the object to be measured or a local change in the distribution of the X-ray absorbing substance causes a directional abnormality in projection data) ,
It is well known that streak-like false images occur due to the presence of highly X-ray absorbing substances such as metals, X-ray hardening (beam hardening), movement of the subject such as a living body, and various other causes. There is a need to quickly identify linear and streak-like shapes in order to remove or reduce the amount of damage, and therefore there has been a demand for quickly identifying these shapes.

The present invention was developed in view of the above-mentioned problems, and its purpose is to quickly identify a straight line, a streak shape, or a shape formed by combining straight lines (such as a fan shape) from various physical quantities having a two-dimensional spread. (algorithm) and also provide a device for identifying streaks from physical quantities using the method.

(Means for Solving the Problems) The present invention for solving the above-mentioned problems is applicable to cases where the distribution of two-dimensional physical quantities has characteristics such as a straight line, a streak shape, or a shape resulting from a combination of straight lines (such as a fan shape). , identify (detect) it
In a method for identifying streaks from physical quantities,
All or a part of the two-dimensional physical quantities are extracted, projection data in each direction or any direction is calculated from these data, and the shape is identified by comparing the projection data. That is.

The apparatus for implementing this method also includes a projection data generating means that extracts target two-dimensional data from physical quantities in a multidimensional space, performs calculations on the data, and obtains projection data of the extracted physical quantities in each direction. , shape identification means for identifying a linear shape, a streak shape, a shape obtained as a combination of straight lines, etc. in physical quantities having a two-dimensional spread by a shape identification calculation performed based on the projection data, and at least one It is characterized by comprising a data storage means.

As a means for obtaining the projection data, all or a part of the physical quantities are extracted, the value of the data of the unextracted part is set to O or a constant, etc., and a closed region is extracted from this (continuous region conversion).
Level shift for each area.

Furthermore, after performing all or part of the processing such as adding O data in Fourier transform and converting the data array, Fourier transform is performed to obtain components on orthogonal coordinates in the frequency space, and this is converted to data on polar coordinates. By performing offset calculation and inverse Fourier transform, projection data can be obtained extremely quickly.

Furthermore, instead of performing calculations using Fourier transform, projection data may be obtained by direct calculations.

(Operation) Physical quantities with a two-dimensional spread are extracted from the data storage device, projection data corresponding to each direction is generated using the Fourier transform method or direct calculation method, and comparison is made with nearby or wide-range data to determine the linear shape. Identify the shape of.

(Example) Hereinafter, an example of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram illustrating an apparatus for carrying out the method of one embodiment of the present invention. In the figure, DS1 is a data storage device that stores physical quantities (input data) that span a three-dimensional space.

=15- PRJ extracts the target two-dimensional data from the physical quantities stored in the data storage device DS1, performs calculations according to the model (1) in which various processes are applied to the data, and calculates the values of each direction θ. This is a projection data generation device that obtains projection data in each X direction corresponding to θ. The generated projection data is stored in the data storage device DS2.

Possible methods for calculating projection data include a projection calculation method using the Fourier transform method, which is a typical analytical method, and a direct projection calculation method.

Δ, Projection calculation by Fourier transform method FIG. 8 shows an example of the configuration of the projection data generation device PRJ that performs projection calculation by the Fourier transform method.

In the figure, ΔRNG is the target two-dimensional space x from the data storage device DSI storing physical quantities in the three-dimensional space
Extract all or - part of the physical quantity μ(x, y) of y, set the value of the data of the unextracted part to 0 or a constant, etc.
Extract this as it is or use it to extract (continuous regions), level shift each region, add O value data for Fourier transform, and transform the data array, and store the processed data in the data storage device DS1A. This is a data extraction array processing device that stores data in The data to be extracted may be data of the entire area or data of a partial area in the two-dimensional space.

A specific example of extracted partial area data is shown in FIG. In the figure, (a) is a three-dimensional image of the head, (b)
(a) is a cross-sectional image taken by plane A from the three-dimensional image in figure A, and (C) is a partial image of soft tissue (area 1.2.3) extracted from the image of cross-section A. The data extraction array processing device ARNG extracts the data in areas 1, 2, and 3 of this diagram (C), and generates data by adding the values of other parts as O. This can be used as is or from this. Extraction of closed regions (continuous regions).

Performs all or part of the processing such as level shift for each area. The data obtained as a result of the processing up to this point is shown in Figure 3 (
This is actual data corresponding to area A of a>. In order to further increase the number of samples and data in Fourier transform, a value 0 is added to the outside of region A to generate data corresponding to region [B] in Figure 3 (1), and a data arrangement convenient for Fourier transform is created. Arrange in. FFT is a two-dimensional Fourier transform device that performs two-dimensional Fourier transform as shown in equation (2) on the two-dimensional image data shown in FIG. 3(b).

Here, μ(X, V) is the value of the physical quantity at the point (x, y). The actual Fourier transform is calculated at high speed as a discrete two-dimensional Fourier transform (fast Fourier transform) for discrete points. The two-dimensional Fourier transform can also be calculated by repeating the one-dimensional Fourier transform.

The two-dimensional Fourier transform device FFT is connected to the data storage device DS1.
Fourier transform is performed on the data stored in A as input data, and the final result of the two-dimensional Fourier transform is stored again in the data storage device DS1A.

The result of the Fourier transform is mapped onto the two-dimensional frequency plane ξη of FIG. 4(a). When the two-dimensional Fourier transform is performed as a repetition of the one-dimensional Fourier transform, the Fourier transform result of the projection data is expressed by the ξ-axis and θ in FIG. 4(a). It is mapped as a set of points on the ω axis that form an angle of . That is, the Fourier transform result of the projection data is mapped as a set of points on the polar coordinate ωθ. RPC is a point on polar coordinates (ω, θ
) is an orthogonal coordinate-polar coordinate data conversion device that calculates data on orthogonal coordinates ξη by calculation. The rectangular coordinate-polar coordinate data conversion device RPC is Fourier transformed by a two-dimensional Fourier transform device FFT and stored in the data storage device DS1.
Process the data stored in A. For example, the following calculation formula can be cited as an arithmetic expression when points on orthogonal coordinates are obtained discretely and points on polar coordinates are also obtained discretely.

...(3-1) [1(ω, θT+) → G+ (ξ3.ηo) F(
ω, θn)−Fr (ω,, θn)j −Fl
(ω□, θη) ξ；ωTlICO8θn η−ω1rlSin θ
n... (3-3> Here, F (ωyo, θn), Fr (ω3.θη), and F+ (ωyo, On) are the points P (ω
□, frequency component (complex number) at θn>.

These are its real component and its imaginary component. Also, 2O- G (ξ3.ηi), G7 (ξ3.ηo) and G+ (
ξ3. ηo) is the orthogonal coordinate ξη of each of the four points P(ω7.On)
The frequency component (complex number), its real component, and its imaginary component are n, l
is an integer, and L is chosen depending on the accuracy of the approximate calculation. (For example, -3) The calculation result data is stored in the data storage device @DSIB.
is stored in

OFP is an offset calculation processing device that receives data stored in the data storage device DSIB and performs offset calculation to obtain projection data corresponding to the offset detection shown in FIG. The following calculation is performed on the data F(ω, θ) on the polar coordinate ωθ of the frequency plane, and the processing result F(
ω, θ)' are stored in the data storage device DSIC.

F(ω, θ)′ −Fr(ω, θ)′+j−F1 (ω, θ)′=F (
ω, θ)−eJaIXo・(4) F “(ω, θ)′ = Fr(ω, θ)−cos(ωXo) −Fl((7J, θ)・sin
>ω≧O... (4-1) F+ (ω, θ)t = Fr(ω, θ)-sin Dz)Xo)+"1
(ω, θ) ・cos (ωXo)ω≧O...
(4-2) Fr (-ω, θ)'-F, (ω, θ)'ω〉0
... (4-3) F+ <-ω, θ)'=-F+(ω, θ)'ω>
○ ... (1-4) F(ω, θ), F(ω, θ)' are complex numbers, Fr (ω
, θ), Fr(ω, θ)' is the real part.

Fl (ω, θ) and F+(ω, θ)' indicate the imaginary part.

Xo is the offset amount (the unit is length, for example +++m) in offset detection. For discrete data with sample interval d (unit: mm, for example), γ=ωmXo−(4
πm/2LX')X.

−(4πm/(Nd))Xo ・= (4
-5) However, m = 0.1.2. -, N/2 Here, N is the total number of discrete frequency data (Figure 4 (a)
(total number of discrete data on the ω-axis). Offset I
X o is taken as (+) in the positive direction of the X axis. 1/4-1
/4 offset detection, if xO=-d/4, then from (4-5> equation, r=-πm/N (m = 0.1
, -, N/2) γ--π/2 (When III = N/2) Equation 4) gives a different phase change for each frequency component ω with respect to ω, θ). The off-hit arithmetic processing unit OFP often performs various filter processes for changing frequency characteristics before or after the calculations of equations (4) and (4-1) to (4-4).

IFFT is the straight line ω on the polar coordinate ωθ plane in Figure 4(a).
A Fourier inverse transform device that performs inverse Fourier transform on the frequency component F(ω, θ)' on (θ-θ) and obtains the projection data b (X, θ) on the real space XY in Fig. 4(b). , performs the following calculation on data input from the data storage device DS1C.

b (X, θ)... (5) Here, () means the real part of the Fourier transform result of F(ω, θ)', and C is a proportionality constant. The calculation results of the inverse Fourier transform device IFFT are stored in the data storage device DS2. The projection calculation using the Fourier transform method described above is performed using the discrete Fourier transform (D F T ; D
1screte FourierT ransror
As m), high-speed calculation is performed using a so-called fast Fourier inverse transform method. The 0-cho is a control device that performs unified control of the operation of each device of the projection data generation device PRJ, data transfer control between each storage device and each processing device, data input/output control with other external devices, etc. be. In FIG. 8, thick lines connecting each device indicate data paths, and thin lines indicate control signal paths.

B. Projection data calculation by direct calculation This method is a method of directly calculating the projection according to equation (1). The amount of calculation is not small, and high-speed calculations depend on equipment configuration using high-speed elements and parallelization of equipment, but it is possible to use not only parallel beams but also fan-shaped beams (fan beams).
However, it has the flexibility of calculations that can be applied accurately. Physical quantity μ(
When X, V) are discrete data, the formula for generating projection data is... Assume that μ(x, V) is located at an integer point on the xy coordinates (vixels), and consider a unit square with this integer point as the center.In this unit square, μ(
X, V) are assumed to have constant values. l) l shadow a(
X, θ) is calculated on the straight line XX on the orthogonal coordinates XY that forms an angle θ with the X axis, and for all unit squares that intersect with this line, μ(yl.

yl)=μ(X-cosθ-yl-s:nθ, x・s
inθ+Yl-CO3θ) = u (X, θ, Y1> and the straight line length dY+ can be calculated according to equation (1)'. The projection data calculation method by direct calculation is not limited to the above method, There are also more complex and highly accurate methods (
In that case, it often involves an increase in the amount of calculations and processing time). Projection data calculation using the Fourier transform method is basically parallel beam projection data calculation as described above, but it can also be applied to fan beam projection data calculation. In this case, parallel beam projection data in each direction is determined by the Fourier transform method, and then projection data of the fan-shaped beam in each direction is determined from this parallel beam data group by linear transformation or other transformation method.

DET is a shape identification device that uses shape identification calculations from projection data of physical quantities to identify linear shapes, streak shapes, shapes obtained as a combination of straight lines, etc. in two-dimensional physical quantities. Including equipment. The same applies hereafter). The input to the shape identification device DET is the projection data stored in the data storage device O82, and the result of shape identification is stored in the data storage device DS3. If a control operation such as correction or manipulation is required according to the result of shape identification, the amount of correction, operation amount, etc. as the result of calculation by the calculation device may be output (in this case, the result of shape identification is (It may not be stored in the data storage device DS3)
. An example of the discrimination calculation for straight lines and linear shapes is as follows. Projection data a(X, θ) in the θ direction and distance X,
For the length of the projection L (X, θ), find the next amount in X-X K.

...(6) +1(XK, fl) is the neighborhood nickel El of X=XK
It is the average value of (X, θ), N is the number of neighboring data, and δ is a so-called threshold. Although n, N, and δ are usually constants, they can also be dynamically changed depending on conditions. n=M+1. If N=2M+1, +1(X
K, θ) is the average value of 2M+1, which is a total of 9 M values before and after XK. L (XK, θ) is the physical quantity μ(x, y
) is the projection data obtained by setting all the values to a constant value of 1, and L (XK, θ) is the projection data a(XK
, θ). Therefore, 1) (XK.

θ) is the projection data per unit length obtained by dividing the projection data in the θ direction
(XK, θ) is the difference between the absolute value of the difference between the projection data per unit length on X = X K and the average value of the projection data per unit length on the straight line in the vicinity thereof.

The following conditions are determined for D<X, , θ).

D (X, , θ)≧0, that is, 11) (XK, θ) - lower (XK, θ) 1≧δ... (6
-1) D (XK, θ)〈0 i.e. + 11 (X, θ) - stone (XK, θ) 1〈δ...
(6-2) When the condition of Equation (6-1) is satisfied, a straight line exists at X = X K and there is a linear shape change), Equation (6
When the condition -2) is satisfied, it is assumed that there is no straight line in X-XK (there is no linear shape change).

Next, the operation of the above embodiment will be described with reference to FIG. 6 for an example in which projection data is calculated at high speed using the Fourier transform method.

The data extraction array processing device ARNG is the data storage device DS.
The physical quantity μ(x, y) of the target two-dimensional space×y is extracted from the physical quantities of the multidimensional space stored in 1. Furthermore, data of the entire area or data of a partial area within this two-dimensional space may be extracted. In the case of extracting partial data, for example, only data of regions having the same type of composition and therefore the values of physical quantities are quite close (less changes) are to be extracted. To give a specific example, as shown in Fig. 9, from the three-dimensional image (a) of the X-ray CT head to the two-dimensional image (b) regarding the cross-sectional view △, (c)
This is a case where soft tissue (areas 1, 2.3) is extracted as a partial image. The value of the data in the part that is not extracted is set to 0. The following processing is performed on the extracted data prior to the projection calculation.

(i) Extract continuous area data, excluding isolated data and data in minute areas. As a result, the extracted data becomes data in a continuous region or a closed region. At this time, the number of areas is one or more.

(ii) Calculate the average value of the physical quantity for each region, and subtract this average value from the original value of each data for each region. Alternatively, for each closed region, the average value of data in the surrounding area or a value close to it is subtracted from the original value (the surrounding area of each closed area is set to O or a value close to it). That is, the level is shifted for each closed area. The data in the portion that is not extracted remains O.

Projection generation based on subregion extraction, (i) and (it) processing provides highly sensitive and highly efficient streak identification.

This is extremely important from the standpoint of eliminating and reducing side effects such as ringing. The data subjected to the above processing (data corresponding to area A in FIG. 3(a)) is stored in the data storage device DS1Δ (step 1).

The data extraction array processing device ARNG is the data storage device DS.
To the data read from 1△, a value 0 is added outside the area △ to increase the number of sample data, etc., to generate data corresponding to area B in Figure 3 (b), and data suitable for Fourier transformation. Make it an array.

The two-dimensional Fourier transform device FFT converts this data into the formula (2
) and then stored in the data storage device DS1A (step 2). As a result, the data on the two-dimensional real space x y is the two-dimensional frequency space ξη
mapped onto the top.

The rectangular coordinate-polar coordinate data conversion device RPC converts the data in the data storage device DSIA into the formula (3-1>
, performs orthogonal coordinate-polar coordinate transformation corresponding to (3-2) (
Step 3). This obtains each ω component in the θ direction.

In the case of offset projection, the offset calculation processing unit
The calculation of equation (4) is performed using P, and the discrete Fourier inverse transform corresponding to equation (5) is applied to this using the Fourier inverse transform device IFFT.
(Step 4).

As a result, projection data on real space in the θ direction is obtained.

Steps 3 and 4 are performed in all directions of θ, and once completed (step 5), move on to the next step. All projection data in all θ directions are calculated by steps 1 to 4.
Obtain a(X, θ) (a(X, θ) mi b(X, θ)). This projection data is stored in the data storage device DS2.

The data extraction array processing device ARNG sets all the data values of the extracted part to 1 for the data area corresponding to area A in FIG.
Let the value of the unextracted part be O, and add the value O to the outside of the area to generate data corresponding to area B in Figure 3(b), making it a data array suitable for Fourier transform. . The two-dimensional Fourier transform device @FF King performs a two-dimensional discrete Fourier transform corresponding to equation (2) on this data and stores it in the data storage device OS'1 A (step 6).
.

Next, the orthogonal coordinate-polar coordinate data conversion device RPC converts the data in the data storage device 1fO81Δ into equation (3-1), (
3-2> is performed (step 7). This obtains each ω component in the θ direction.

In the case of offset projection, the off-hit arithmetic processing unit OF
P performs an operation equivalent to equation (4), and applies discrete Fourier inverse transform corresponding to equation (5) to the Fourier inverse transform device IF.
Performed by FT. As a result, the projection a on the real space in the θ direction is
Obtain the length L (X, θ) of (X, θ) (Step 8)
. Steps 7 and 8 are performed in all directions (step 9), and the length L(X, θ) of all projection data a(X, θ) in all θ directions is obtained by steps 6 to 8. This data is stored in the data storage device DS2.

The shape identification device DET performs the following processing using the data in the data storage device DS2. Projection data a(X,
θ) and the length of each projection L (X, θ), equation (6-
1> and (6-2>), the presence or absence of a straight line (linear shape) is discriminated (step 10).
direction and complete the entire movement (step 11)
. The identification determination information (or the corresponding correction amount, operation amount, etc.) is stored in the data storage device DS3 (
or an arithmetic device, an operating device, etc.).

Note that the present invention is not limited to the above-described embodiments, and various modifications as shown below are possible, for example.

(a) Other Examples (i) Equations (6), (6-1>, and (6-2) are equations for detecting and determining accurate and faithful straight lines (linear shapes); In many cases, as in the case of
) etc., if the shape of the region is appropriately selected (for example, circular), L (X, θ) in the vicinity of X=XK can be detected.
→L (XK, θ>=C can be considered. However, C
is a constant, so equation (6) in this case is L (X, θ)
The quantity p(XK, θ) that ignores the new l1l(XK).

Even if it is treated as θ), it is sufficiently useful for detection and identification.

That is, ...(6)' In this case, the length L(X, θ) of each projection a(X, θ)
This eliminates the need for calculations, allowing faster calculations. Processing in this case is performed according to FIG. This operational flowchart is the same as that of the embodiment except that steps 6 to 9 are omitted, so the explanation thereof will be omitted.

(Note 1) The soft tissue at the base of the brain measured by X-ray CT is a collection of human tissues that have almost the same X-ray absorption coefficient, and the identification of linear artifacts in this soft tissue is based on a flat pattern. This is an example.

(It) Although the processing time is long, an example in which straight and streak shapes are accurately identified from physical quantities is shown in FIG. This example is an example in which the apparatus shown in FIGS. 1 and 8 is used to generate projection data for each closed region and identify linear and streak-like shapes. This operation is the same as the operation in FIG. 6 except for step 0.1.12, so the same explanation will be omitted.

In step O, all or a portion of the physical quantities of the target two-dimensional plane are extracted from the physical quantities of the three-dimensional space. The value of the data in the part that is not extracted is set to O or a constant. In step 1, one closed region (continuous region) is obtained from the extracted two-dimensional physical quantities, and processing such as level shift is performed on this closed region. Steps 1 to 11 are performed for each closed area, and when steps 1 to 11 are completed for each closed area, a check is made at step 12, and the process is terminated.

(b) Modification example ■ Algorithm deformation Shapes such as straight lines, streaks, etc., shapes by combining straight lines (
(6), (6)' for fan-shaped etc.

Although (6-1) and (6-2> are very effective, various modifications can be made to the determination algorithm.

(i) In equations (6) and (6)', 11 (XK, θ) is calculated not only on the average value of a data group in a narrow range, but also on a constant, the average value of a neighboring data group in a wide area, or a neighboring data group. It may be a value obtained by applying Furthermore, data in the vicinity may be obtained from multiple directions (multiple views). Also, n and N
and δ are appropriate constants that can be dynamically selected during calculation depending on the conditions.

(ii) The following algorithm may be used to detect a shape that changes stepwise.

(iii) D (XK, θ) may not be the average value of the neighboring data, but may be something like the average value of the maximum value and the minimum value of the neighboring data.

(1v) In the shape identification equation, the threshold value may be changed depending on the sign of the comparison result.

That is, if p(XK, θ)≧1)(XK, θ), then D(XK
, θ) = I) (XK, θ) - lower (x4. θ) - δ1p (XK,
If θ)<D(XK, θ), then D (XK, θ) = -1) (XK, θ) tenths (XK, θ)...(6)
' δ1. δ2 is each threshold value (δ1>O4 δ2>O).

(V) To extract a target physical quantity in a two-dimensional space from a physical quantity in a multidimensional space and further extract a specific part of the two-dimensional space, specify the location and space of the part to be extracted; the physical quantity of the part to be extracted; It goes without saying that this can be done partly by specifying the value of the data, or by a combination thereof.

Specific examples of the data values of the physical quantities of the part to be extracted include a set of data within a certain value range (a set of data between the lower limit value and upper limit value of one set, a set of data between the lower limit value and the upper limit value of multiple sets). (e.g., a set of data within the range of ).

■Variation of operations and processing In the operations shown in Figures 6 and 7, the data extraction array processing device △RNG extracts all or part of the target two-dimensional physical quantities from the physical quantities in the multidimensional space in step 1. The value of the data in the unextracted portion is set to O or a constant, and this is used as it is or from this. In step 2, the data extraction array processing device ARNG further performed all or part of processing such as addition of zero-value data for Fourier transformation and data array conversion.

By the way, the data extraction sequence processing device ARNG performs step 1.
It is also possible to perform all of the above-mentioned processing. That is, in step 1, all or part of the target physical quantities in the two-dimensional space are extracted from the physical quantities in the multidimensional space, and the value of the data in the unextracted part is set to O or a constant, and this is used as is or closed from this. Area (continuous area)
, level shift for each area, addition of O-value data for Fourier transform, and data array conversion, all or part of the processing is performed. (In step 2, addition of O-value data for Fourier transformation, data array conversion, etc. are not performed.) Also in the operation shown in Figure 10, addition of O-value data for Fourier transformation in step 2, data arrangement, etc. It goes without saying that processing such as conversion can be added after the processing in step 1 and processed in step 1.

■Device modification (i > Division, integration, and addition of data storage devices (a) A configuration in which one data storage device is divided into multiple parts. Also, the speed of the data storage device.

A hierarchical structure based on economic efficiency, etc.

(b) A device configured by integrating two or more data storage devices.

(c) Buffer memory, cache memory, etc. are added as appropriate.

(ii) Division, integration, and parallelization of equipment (a) A configuration in which each device is divided functionally, for example.

(b) Fourier transform device of the projection data generation device PRJ;
Cartesian coordinate-polar coordinate data conversion device.

A parallel configuration of devices such as Fourier inverse transform equipment.

(c) A device configured by integrating multiple devices.

(d) A unified control device consisting of a microprocessor, microprogram memory, decoder, etc.

e) The projection data generation device PRJ (Fig. 1) is divided into a data extraction processing device PKIJP and a bronze medal data generation device PRJ. That is, when using the Fourier transform method,
In FIG. 8, the data extraction array processing device ARNG is replaced by the data extraction array processing device PKUP and the data extraction array processing device △
Divided into RANG. Therefore, in this case, the projection data generation device PRJ is the data extraction array processing device ARAN.
G, two-dimensional Fourier transform device FFT, rectangular coordinate-polar coordinate data conversion device RPC, offset calculation processing device OFP, Fourier inverse transform device IFFT, and control device C
Tl- and data storage device DS1△, ~, DSIC,D
It consists of 82 mag.

In this case, the data extraction processing device PKUP extracts the physical quantities in the two-dimensional space from the physical quantities in the multidimensional space, sets the value of the data in the unextracted part to 0 or a constant, and uses this as it is or further converts it into a closed region (continuous region). ) extraction, claim 6
Performs all or part of processing such as level shifting for each area.

The data extraction array processing device @ARANG performs addition of O value data for Fourier transformation, data array conversion, etc.

In addition, devices such as a two-dimensional Fourier transform device FFT perform the processing described above with the configuration described above.

(iii) Addition of devices (a) Addition of large capacity storage devices such as disks (b) Addition of external storage devices such as magnetic tape device MT, floppy disk device FD, etc. (c) Addition of dedicated operation computers, CRTs, etc. (2) Addition of a graph display device, etc. (e) Addition of an image display device with a graphic information reading and information input device (man-machine interface device with an image display device).While displaying three-dimensional data on the image display device, 62 An attachment of a device consisting of all or part of a device that can display a dimensional data plane on the same screen or another display device, select and specify an extraction plane, and specify a further part to be extracted while displaying this extracted two-dimensional plane. =43- addition. Or an image display device that displays an image, a parameter reading display device for an extraction area (or a parameter reading device)
, addition of a device consisting of an extraction region parameter input device, etc.

(iv) Projection data generation device PRJ When realizing the projection data generation device PRJ as a high-speed projection calculation device using the Fourier transform method, a high-speed projection data calculation device having the configuration shown in FIG. 8 and its modifications may be used. For example, a two-dimensional Fourier transform device FFT and a Fourier inverse transform device IFFT are configured with a one-dimensional Fourier transform device DFT and used in a time-division manner.

(V) Separation and integration of buses, control lines, and signal lines.

(vi) One that uses a general-purpose information processing device CPU, a dedicated array processor AP, etc. for data processing.

(vii) The same arithmetic processing is realized using different configurations (for example, information processing unit@cPU, array processor AP, and other devices and software).

(c) Application Example An X-ray CT apparatus in which the present apparatus and a corrected projection data generation apparatus are added to an X-ray CT is an application example of the apparatus according to the present method as a CT capable of reducing various streak-like artifacts.

(Effects of the Invention) As described in detail above, according to the present invention, the following effects can be obtained.

■Using a simple identification algorithm, shapes such as straight lines, streaks, and composite shapes of straight lines (fan shapes, etc.) can be identified extremely effectively.

■Shape identification can be performed with high sensitivity and accuracy.

■Extremely high-speed processing is possible by using fast Fourier transform, simple discrimination calculations, etc.

■It is possible to realize an economical device with a simple configuration using a general-purpose arithmetic unit centered on a Fourier transform device.

■Many useful uses are possible. X using this device
In X-ray CT, the partial volume effect is considered to be an essential problem in X-ray measurement.

It clearly identifies streak-like false images that occur due to the presence of highly X-ray absorbing substances such as metals, X-ray hardening (beam hardening), movement of the subject such as a living body, and other causes. This makes it possible to realize X-ray CT and various image processing devices that can effectively reduce the amount of radiation, and is expected to have great clinical effects.

Furthermore, the reduction of false images in X-ray CT and various imaging devices using this method can be achieved without sacrificing resolution at all.

[Brief explanation of the drawing]

Fig. 1 is a block diagram of a device according to an embodiment of the present invention, Fig. 2 is an explanatory diagram of projection data and offset detection, Fig. 3 is a diagram of real data and data obtained by Fourier transform, and Fig. 4 is a frequency space diagram. Fig. 5 is an explanatory diagram of the relationship between orthogonal coordinates and polar coordinate components and projection in real space, Fig. 5 is an explanatory diagram of projection in direct calculation, Fig. 6 is a flowchart of the operation of the embodiment of Fig. 1, Fig. 7 is a flowchart of the operation when calculation of the projection length is not required, FIG. 8 is a block diagram of an embodiment of the projection data generation device when performing projection 4 calculation using the Fourier transform method, and FIG. 9 is a diagram of the data extraction. FIG. 10, a diagram showing a specific example, is a flowchart of operations when generating projection data and identifying linear shapes for each closed region. DSl, DSIA, D81B, DSIC,, D82, D
S3...Data storage device PRJ...Projection data generation device DET...Shape identification device ARNG...Data extraction array processing device FFT...2
Dimensional Fourier transform device RPC... Cartesian coordinate-polar coordinate data conversion device OFP・
...Offset arithmetic processing device 1, FFT...Fourier inverse transform device Patent applicant: Yokogawa Medical Systems Co., Ltd. Kadoiro 1 Figure → 耶＠IOFlow---Data flow cylinder 27 ＾hahe＾＾ CL < co u○. 1.1 0゛tto葎

Claims (21)

[Claims] (1) When the two-dimensional distribution of physical quantities has characteristics such as a straight line, a streak shape, or a shape resulting from a combination of straight lines (fan shape, etc.),
In the method of identifying or detecting the shape from the physical quantities (hereinafter simply referred to as identification), all or a part of the physical quantities of the two-dimensional distribution are extracted, and from these data, projection in each direction or an arbitrary direction is performed. A method for identifying streaks from a two-dimensional distribution of physical quantities, characterized in that data is obtained by calculation, and the shape is identified by comparing the projection data. (2) The method for identifying streaks from a two-dimensional distribution of physical quantities according to claim 1, characterized in that at least one portion in which the value of the physical quantity changes relatively small is used as a part of the physical quantity of the two-dimensional distribution. . (3) A claim characterized in that at least one physical quantity in a closed region, which is a continuous region in which changes in the value of the physical quantity are relatively small, or in an open region, which is a discontinuous region, is used as part of the physical quantities in the two-dimensional distribution. A method for identifying streaks from a two-dimensional distribution of physical quantities according to item 1. (4) The two-dimensional physical quantity according to claim 1, 2 or 3, characterized in that when extracting a part of the physical quantity of the two-dimensional distribution, the data value of the part that is not extracted is set to 0, a constant, or empty. How to identify streaks from a distribution. (5) The method for identifying streaks from a two-dimensional distribution of physical quantities according to claim 1, wherein the physical quantity used for calculating the projection data is the extracted original physical quantity itself. (6) The physical quantities used for calculating the projection data are the average value of the physical quantities of each region from each original physical quantity for each region for all the extracted physical quantities, or the value obtained by performing calculations on them, or the value obtained by calculating them. The value of the physical quantities in the peripheral area, their average value, the value obtained by performing calculations on them, or the average value of all extracted physical quantities, or the value obtained by subtracting a constant, etc., that is, the physical quantity that is level-shifted for each region. A method for identifying streaks from a two-dimensional distribution of physical quantities according to claim 1. (7) The physical quantities used to calculate the projection data are either the original physical quantities of the extracted subareas themselves, physical quantities that have been level-shifted for each area of the extracted subareas, or the parts that are not extracted. 2 of the physical quantities according to claim 1, characterized in that the value of is 0 or a constant, etc.
How to identify streaks from dimensional distributions. (8) A method for identifying streaks from a two-dimensional distribution of physical quantities according to claim 1, characterized in that the projection data is generated by a Fourier transform method. (9) The method for identifying streaks from a two-dimensional distribution of physical quantities according to claim 1, wherein the projection data is generated by a direct projection calculation method. (10) The method for identifying streaks from a two-dimensional distribution of physical quantities according to claim 1, characterized in that the projection data is generated as a group of parallel projection data parallel to each other. (11) Streak from the physical quantity according to claim 1, characterized in that the projection data is generated by direct projection calculation as a group of fan-shaped projection data that spreads out in a fan shape, or by calculation from a group of parallel beam projection data in a plurality of directions. How to identify. (12) A method for identifying streaks from a two-dimensional distribution of physical quantities according to claim 1, characterized in that the projection data is generated by offsetting the position. (13) Shape identification is performed by comparing projection data at any position in any direction with the average value of a group of nearby projection data or a group of projection data over a wide range, or a value obtained by performing calculations on these, or a constant. Claim 1 characterized in that:
A method for identifying streaks from a two-dimensional distribution of physical quantities described. (14) Shape identification is performed by comparing projection data at an arbitrary position in a certain arbitrary direction with the average value of a group of neighboring projected data and a group of neighboring projected data in different directions, or a value obtained by performing a calculation on these, or a constant, etc. 2. A method for identifying streaks from a two-dimensional distribution of physical quantities according to claim 1. (15) A method for identifying streaks from a two-dimensional distribution of physical quantities according to claim 13 or 14, wherein a threshold value is set as a criterion for determination in the shape identification performed by comparing projection data. (16) Projection data generation means for extracting all or part of the target physical quantities in the two-dimensional space from the physical quantities in the multidimensional space, and calculating the extracted data to obtain projection data regarding the extracted physical quantities in each direction; shape identification means for identifying a linear shape, a streak shape, a shape obtained as a combination of straight lines, etc. in physical quantities having a two-dimensional spread by a shape identification calculation performed based on the projection data; and at least one 1. A device for identifying streaks from a two-dimensional distribution of physical quantities, characterized in that the device comprises: data storage means. (17) The projection data generation means extracts all or a part of the target physical quantities in the two-dimensional space from the physical quantities in the multidimensional space, sets the value of the data of the unextracted part to 0 or a constant, and uses this as it is or From this, data extraction array processing means performs all or part of processing such as extraction of closed regions (continuous regions), level shifting for each region, addition of zero value data for Fourier transform, and data array conversion. , a two-dimensional Fourier transform means that performs a two-dimensional Fourier transform on two-dimensional data, a rectangular coordinate-polar coordinate data conversion means that calculates data of a point on polar coordinates from data on a rectangular coordinate, and offset detection. offset calculation processing means for performing an offset calculation to obtain projection data;
It is characterized by being comprised in whole or in part of an inverse Fourier transform means for performing inverse Fourier transform on frequency components on a straight line on a polar coordinate plane to obtain projection data in real space, and at least one data storage means. An apparatus for identifying streaks from a two-dimensional distribution of physical quantities according to claim 16. (18) The projection data generation means extracts all or a part of the target physical quantities in the two-dimensional space from the physical quantities in the multidimensional space, sets the value of the data of the unextracted part to 0 or a constant, and uses this as it is or It is characterized by comprising a data extraction array processing means that performs all or part of processing such as extraction of a closed region (continuous region) and level shift for each region, and a calculation means that directly calculates projection data. An apparatus for identifying streaks from a two-dimensional distribution of physical quantities according to claim 16. (19) Shape identification means is performed by comparing projection data at an arbitrary position in a certain arbitrary direction with an average value of a group of nearby projection data or a group of projection data over a wide range, or a value obtained by performing calculations on them, or a constant. 19. The apparatus for identifying streaks from a two-dimensional distribution of physical quantities according to claim 16, 17 or 18. (20) The shape identification means compares the projection data at an arbitrary position in a certain arbitrary direction with the average value of the neighboring projection data group and the neighboring projection data group in different directions, or a value obtained by performing an operation on them, or a constant, etc. 19. The apparatus for identifying streaks from a two-dimensional distribution of physical quantities according to claim 16, 17 or 18. (21) Claim 16, 17, or 18, characterized in that the shape identification means is a means for setting a threshold value and using it as a criterion for determination in the shape identification performed by comparing projection data.
A device that identifies streaks from a two-dimensional distribution of the described physical quantities.