JPS6017568A - Method and device for processing image - Google Patents

Abstract

PURPOSE:To reduce computing speed of image restructure by determining an ending period for calculation while evaluating repeatedly in each calculation the image improved serially by repeated calculation, and obtain a good quality image. CONSTITUTION:A computer 2 performs image reconfiguration projection data collected in a data collecting division 1, and controls the entire equipment. The restructured image is displayed on an image display device 3, and the image and computed results are stored in a storage device 4. Parameters are inputted by an inputting device 5. The computer 2 performs image reconfiguration with serial approximation, during which time the period for finishing calculation is determined while serially evaluating in every repeated calculation the image serially improved in image quality by repeated calculation.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Application of the Invention] The present invention relates to an image processing method and apparatus, and a typical field of application thereof is a computer tomography apparatus and the like. In particular, the present invention relates to a method for obtaining high-quality reconstructed images at high speed by reducing the number of repetitions of successive calculations during image reconstruction in a computerized tomography apparatus that performs image reconstruction using a successive approximation method.

[Background of the invention]

Computed tomography apparatuses, which are a typical field of application of the image processing method and apparatus of the present invention, are generally classified into the following two types based on their image reconstruction algorithms and (.

(1) Successive approximation method (ART method, MEM method, etc.) (2)
Analytical methods (FBP method, FFT method, etc.) Method (1) takes a long calculation time, and in the case of the number of projection data of 180 to 360 currently used in OCT equipment, analytical method (2) Because the accuracy is lower than that of the FBP (Filt
ered Back prc+jeetion: Filter Ni stop back projection) method is mainly used 3, but 17 methods, but when the number of projection data is small, iterative approximation (J'λ method is less rusty). M1=: M〆Leaving (Maxim mountain nE
Entropy Method: It has become clear that the maximum end [1Bee method] is effective.

There are several methods of successive approximation, but Isune compares the projection data created from the distribution of the estimated absorption coefficient with the actually measured projection data and repeats it. Converge the distribution of absorption coefficients to the true distribution,
(G+T, Herma
n, Image l(,eeonstructionf
rom Projects, Academy
c Prcss.

1980).

In image reconstruction using the conventional successive approximation method, the number of repetitions of subtraction to iteratively correct the distribution of the absorption coefficient estimated above is set in advance, so the spatial frequency of the tomographic plane of the human body to be imaged is There have been problems in that the image quality of the reconstructed image varies depending on the characteristics such as tr, and the degree of convergence of the image varies.

That is, when a smooth tomographic plane with few changes in shading is imaged, convergence is quick, while convergence is slow in a tomographic plane with complex changes. Nevertheless, if the 4 calculation is aborted at the preset number of repetitions, the division will be stopped even though the image quality can be improved on rough tomographic planes, whereas on smooth tomographic planes, division will be stopped. This results in wasted division time that does not contribute to improving image quality.

There are several successive approximation methods, but all of them involve repeatedly correcting images by comparing projection data created from the estimated distribution of absorption coefficients with actually measured data (
G, T, Herman, 1mageReeo
nstrjetion from proiecti
ons.

Academic Press, 1980). The amount of calculation required for one correction at this time is, for example, 1 giga operation (image 256 x 56 pixels, number of projection directions 3).
60, it also requires 30 minutes of CPU time using a large computer [Hitachi M-180:]. In conventional methods, such corrections are repeated as many times as expected to improve the distribution of the estimated absorption coefficients, that is, the quality of the estimated image, which has the problem of requiring a huge amount of S4 time. .

[Purpose of the invention]

The present invention has been made to overcome the above-mentioned drawbacks, and in the successive approximation method, it is possible to reduce the number of iterations until the estimated image becomes better, reduce the calculation time for image reconstruction, and obtain a high-quality image. The object of the present invention is to provide an image processing device.

[Summary of the invention]

In order to achieve the above-mentioned object, the present invention is characterized in that the timing for discontinuing the calculation is determined by sequentially evaluating the image whose image quality is successively improved by repeated 2 calculations each time the repeated calculation is performed. There is. In this way, the number of repeated calculations can be changed depending on the characteristics such as the spatial resolution of the tomographic plane to be imaged.

There are two types of evaluation functions:

(1) A method in which the difference between projection data of sequentially output images and actually measured projection data is used for evaluation.

(2) A method in which the difference between the n-th sequential output image and the n+1-th sequential output image is used for evaluation.

Furthermore, when reconstructing a resting tomogram of the heart by successive approximation using only projection data synchronized with a specific phase of heart motion, the initial value is not an image with uniform density, which has been used previously. , which uses an image in which a town is constructed from projection data in all directions using the FBP method. This makes it possible to reduce the number of repetitions until the image quality improves, and further reduce the calculation time.

[Embodiments of the invention]

Hereinafter, the present invention will be specifically explained with reference to Examples.

FIG. 1 is a block diagram showing the configuration of a first embodiment of the present invention. 1 is a data collection unit for collecting projection data, 2 is for reconstructing images from the collected data,
3 is a computer that controls the entire device; 3 is an image display device that displays the reconstructed image; 4 is a memory device that stores the 1 μmj section to 1 μm j unit; and 5 is a computer that inputs various parameters fX, etc. The input value is fX.

The operation of the present invention in the configuration shown in FIGS. 2 to 41 is as follows.
-C will be explained using C1. Yesterday, using Figure 2, we created a picture 1 and an image 1.
↓We will outline the formulation of the J-configuration problem (for details, see Yoshinori Iwatsuki, CT Ski Vsu, Corona Publishing, 1976 2nd grade), Absorption rate based on coordinates (x, y) f (x, y>
Considering the object represented by , we define coordinate systems X and OY that are tilted by the angle θ from the coordinate system xoy. The conversion between this is as follows. Here, if X-rays of intensity Io are irradiated parallel to Y
nθ, )('e+inθ+YCO3θ)dY]...
・・・・・・・・・(2) Then, the projection data is the logarithmic transformation g(X, θ
) as g(X, θ) = / f (XCO3θ−Ysinθ,
X5inθ+Ysinθ)dY・・・・・・・・・・・・
・(Represented by 3. The image reconstruction problem is 0≦:θ〈2π
It is possible to obtain the absorption distribution of the object f (x,) using the projection f - g (X, θ) of .

The successive approximation method, which is one of the methods for solving this problem, will be explained using FIG. In the figure, in this example, block 2 is indicated by a dotted line.
0 is not used. First, the initial image f' of the corrected image 19 (
x+y) gives a uniform distribution image. On the other hand, the projection data of the angle θ is calculated to create the first estimated projection data 10g' (X, θ). This is the formula (
3), go(X,,θ)=f fO(XCO3θ−Y
sinθ, X5inθ→-YCO3θ)dY...
...... (4) is given. This function and the actual projection data collection unit 11
fO(x, y) of the corrected image 19 is corrected by comparing the measured projection data 18g(X, θ) obtained from the above, and f'(x+y) is obtained as a new corrected image 19. By performing such operations sequentially, the modified image 19 can be improved one after another to become sf'(xvy3).

To divide the i-th modified image 19 f''l(x, y), the divine method is proposed first.For example, f
""Xry)-fl(x, y)+[g(X,
θ)-g'(X, θ)]/n fX, θ)...
... (5) cn(x, θn: The projection beam is the subject f (x, y)
There are some that perform successive corrections additively, such as [the length passing through]. Error 1 between measured projection data 18 and estimated projection data 10
2. The relationship between the image correction amount 13, etc. is as shown in FIG.

The present embodiment is characterized in that the timing of discontinuing the calculation of the successive approximation method described above is determined by evaluating the corrected image 19f'(x, y). The flow of the process will be explained using FIG. 4. Input 101 the measured projection data from the projection data collection unit 11 to the Takazu Keishiki 2Fi, and then input the initial image f.
'(X, 3') After setting ffi (102), sequential calculations 103 to 106 begin. Corrected image 19f'(x, y
) is corrected (105, 106) using -01 formula (5) for all scan angle directions, and the corrected image 19 f
Get the circle (x, y). At this time, the modified image 19 f ``'' (x, y) is evaluated (107) using the evaluation formula illustrated below, and f 1-)1 (x + y) is calculated as the final reconstructed image. (108).

Σ, /(g(x, θ) - g'(X, θ))2dX<
K...(6) θ - (g)......(7) (Average density of 'Image' (x+y)) Equation (6) is the total (to) from the modified image 19fl(x,y)
The router projection data g'(X, θ) is the measured projection data g
This is a formula for evaluating whether it is consistent with (X, θ). On the other hand, equation (7) expresses the modified image 19f' (x,
y) and correction +b image I 9 f ”” (X, y
) difference 'CI'L, M-' S, (I(,oot
Mean 5 square ) 4]J in the upper world, 1ㅅ'(', 18-1LIII!l image 1!1 There is a method C to judge the collection of the Song Dynasty.

As described above, it is possible to obtain a high-quality image regardless of the quality of the image and even with unnecessary processing.

Next, a second embodiment of the present invention will be specifically described. FIG. 5 is a block diagram showing the configuration of the second embodiment. As in the case of the first embodiment, 1 is a data collection unit for collecting projection data, 2 is a computer for reconstructing images and editing projection data using the collected data, and 3 is a computer for displaying performance results. An image display device 4 is a storage device for storing images and calculation results, and 5 is an input device for specifying various parameters and areas. Furthermore, in this embodiment, the heart'rIL 6 is connected, and this is for obtaining an electrocardiogram in order to know which phase of the heart movement the projection data from each direction is in. .

Based on the electrocardiogram obtained in step 6, the projection data collected in step 1 is tagged by computer 2 for each class that waits for the same phase.
(Name tag) I'm disappointed.

That is, as shown in FIG. 6, the R wave intervals in the electrocardiogram are divided into n (n is an integer) character, and are divided into tagl, tag2, etc., respectively.
...tag, and the projection data of the same tag section is
This is to combine images by considering them to be of the same phase.

In this embodiment, as in the case of the first embodiment, the formulation using FIG. 2 is used, and the relationships of equations (1) to (3) hold; The feature is that the image is reconstructed using the electrocardiogram data of the figure.

A case in which the successive approximation method, which is one of the methods for solving the problem, is used to reconstruct a resting tomographic image of the heart will be explained with reference to FIG. In the figure, this embodiment has a dotted line block 20 added to the first embodiment. 1. Initial value f' (X,
y) Gives a uniform distribution image by binding. For Kohachi,
For the image that is currently being reconstructed, the electrocardiogram data 200, the projection data with the phase tag (tag i), and the projection data in the direction θ obtained are calculated, and the first estimated projection data 10g' Create (X, θ).

(Equation (4)) Furthermore, from Equation (5) K, fl+j□, -(x
, y) is the same as in the first embodiment, but in this embodiment, as the initial value fO(x, y) of the corrected image 19, all directions without phase division are used. This uses projection data and images reconstructed by the FBP method. This makes it possible to reduce the number of repeated calculations even further.

In other words, the image reconstructed from the projection data in all directions represents the average shape of the moving heart, and is close to the final reconstructed image, so it is possible to obtain a high-quality image even if the number of repeated calculations is small. is obtained.

Although the embodiments have been described in detail above, the image processing method and apparatus of the present invention can be implemented not only in computer tomography apparatuses but also in image measurement and the like that require similar image processing. It can be implemented using either modified software or a node.

〔Effect of the invention〕

According to the present invention, when reconstructing an image by the successive approximation method]7
, high-quality images can be obtained quickly without unnecessary division.

Further, according to the first embodiment, it is possible to reconstruct uniformly high-quality images without performing unnecessary processing, regardless of the characteristics of the tomographic plane that differs depending on the body part to be imaged or the individual.

Further, an example of the effect of the second embodiment is shown in FIG. 7. Figure 7 shows the initial value fO(x,
y) shows how the R, M, and S (Root Mean SquBre) errors change when using a uniform image and an FBP reconstructed image. When using FBP reconstructed images, R,, M, and S suddenly increase.

(This shows that the difference decreases.) When the FBP reconstructed image is used as the initial value in %, the effect of reducing the error value is remarkable when the number of repetitions is small as shown in the figure.

XR, M, S, difference fij: density f ij of pixel (i, j) of fan dome
: Density f of pixel (i, j) of corrected image 9 : Average S of pixel density in area S : Circular area of fan 1

[Brief explanation of drawings]

FIG. 1 is a block diagram of an image display device according to the present invention;
Figure 2 is an explanatory diagram showing the principle of general reconstruction operations;
The figure is a block 70-diagram showing the reconstruction calculation procedure of the successive approximation method according to the present invention, FIG. 4 is a flowchart showing the processing procedure of the present invention, and FIG. 5 is a block diagram of an image display device according to the second embodiment of the present invention. A configuration diagram, FIG. 6 is a waveform diagram showing the relationship between projection data and electrocardiogram synchronization in the second embodiment,
FIG. 7 is a graph showing a specific example of the effect of the second embodiment.

Claims (1)

[Claims] 1. An image processing method for configuring an image using a successive approximation method, which includes the steps of: collecting image data; obtaining an image modification amount from the image data and the image to be constructed; An image processing method in which the hourly rate is to include the steps of modifying the image to be constructed based on the amount of image modification, and sequentially evaluating the modified image to be constructed. 2. Increasing the initial value and filtering. Using an image constructed by a corrected back projection method f: Claim 1 characterized in
Image processing method described in section. 3. means for detecting images showing the state of the subject during operation from a plurality of directions; means for obtaining a synchronization signal matching the operating cycle of the subject; and images detected from the plurality of directions using the synchronization signal; An image processing apparatus characterized by comprising: means for synthesizing the same periodic portions of by a successive approximation method; and means for using an image constructed by a filtered back projection method as an initial value of the successive approximation method.