JPH0663038A - Method and device for interpolating data - Google Patents

Abstract

PURPOSE:To execute the improvement so as to obtain the image of the S/N having the extent of a regular axial scan while correcting deviation from the slice position of data caused by a helical scan. CONSTITUTION:The device is provided with a noise signal extracting circuit A 11 and a noise signal extracting circuit B 12 for extracting a noise component and a signal component from data P1, P2, and multipliers, adders, etc., 15, 16. 17, 18 and 19 for executing the addition average of each noise component and also, executing a multiplication processing and an addition processing, etc., so as to execute weighting addition by weight coefficients alpha, 1-alpha corresponding to a distance from a slice position with respect to the signal component.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data interpolation method and apparatus for improving the quality of an image obtained by helical scanning.

[0002]

2. Description of the Related Art X-ray CT irradiates a subject with radiation, detects X-rays that have passed through the subject, performs this over 360 °, and reconstructs an image by performing image reconstruction. It is a device that obtains an image and diagnoses it.

In this X-ray CT, it is necessary to take several tomographic image photographs in order to know the state of a certain organ, but for this purpose, it is necessary to stop breathing several times. There is a helical scan as a method of shortening the imaging time and imaging several slice planes during one breath to improve the throughput.

A helical image obtained by performing a helical scan is obtained by reconstructing an image at an arbitrary slice position z _sp in the moving direction of the object, that is, the z direction, by using projection interpolation.

The scanning method will be described in more detail with reference to FIG. In the coordinate system shown, the x-ray source and detector are x
There is a cradle in the z direction that rotates in the y plane and the subject is on this cradle.

In a normal axial scan (scan for obtaining an image in the xy plane), the cradle is moved so that the slice position z _{sp to} be imaged on the z axis coincides with the rotation plane of the X-ray source and the detector. A 360 ° scan is performed, image reconstruction is performed from the data, and an axial image is obtained. In this case, for the purpose of improving the resolution and the SN ratio, the facing data A and B which are substantially different from each other by 180 ° are averaged. At this time, the random noise included in the data A and B is 1 / √
Reduced to 2. That is, the standard deviation of random noise is 1
/ √2.

On the other hand, the helical scan is a method of continuously scanning while the cradle moves on the z axis. Phase φ in the plane of rotation of the X-ray source during helical scanning
7 shows the relationship among the view j of the detection data, the channel i, and the position z on the z-axis of the rotating surface.

In the figure, (a) is a diagram showing the phase φ in the rotation plane of the X-ray source, and (b) is a diagram showing the position on the z axis of the rotation plane due to the movement of the cradle. (C) The figure shows the view j corresponding to the position on the z-axis of the plane of rotation.
Channel i is shown on the vertical axis.

In this figure, an image at the slice position of z _sp is to be obtained. At this time, for image reconstruction, ± π (or ± 2π, 3π, ...) with φ _sp as the center
, That is, the data from φ _{A to} φ _B , that is, the data from j _A view to j _B view.

However, the data of the j _A view is z _A
Since the j _B view data acquired above is acquired on z _B , if image reconstruction is performed as it is, in the case of an object having a structural change in the z direction, the object stationary on z _sp In contrast to the scanned axial image, the image has many artifacts. Therefore, in general, when it is desired to obtain an image on z _sp by helical scanning, the distance between the positions z ₁ and z ₂ at which each data is acquired and the slice position z _sp for the data from the j _A view to the j _B view. Correct the data by.

There are various methods for this correction, but the correction represented by the formula is generally performed. P (i, j) _sp = αP ₁ + (1-α) P ₂ Here, the directions used to obtain P ₁ , P ₂ : P (i, j) _sp differ by substantially 180 °, The position in the z direction is also different depending on the direction.
° Different, the position in the z direction also changes depending on the direction Projection data of the same channel i: Channel j: View α: Weighting coefficient depending on the distance of P ₁ from z _sp However, different values are taken for i and j.

[0012]

As described above, when the weighted addition is performed according to the equation, the standard deviation S of the random noise based on the value before the weighted addition is within the range of 0≤α≤1.
D (α) is given by the following equation.

SD (α) = √ {α ² + (1-α) ² } = √ {2 (α-0.5) ² +0.5} ≧ 1 / √2 where α = 0.5 Is the standard deviation of the average of 1 / √
Although it is equal to 2, all other values are larger than the average. The curve of the change of the standard deviation of the equation with respect to α is shown in FIG. In the figure, the vertical axis shows the standard deviation with the weight before addition as a reference (α = 0, α = 1 is 1), and the horizontal axis shows α. As shown in the figure,
At 0.5, the standard deviation SD becomes 1 / √2 = 0.707, and all other values are larger than the average.

This is because, in the interpolation by the formula, the calculation of multiplying the projection data P ₁ and P ₂ by the weighting coefficient is performed not only on the image data but also on the random noise included in the image data. Since this is done, statistical erasure of this randomness cannot be expected as much as a normal axial image. As a result, the SN ratio of the image was deteriorated, and it was not possible to obtain an image commensurate with the exposure dose.

The present invention has been made in view of the above points, and an object thereof is to correct the deviation of the data peculiar to the helical scan from the slice position, and to make the image by the helical scan the SN of about the normal axial scan. It is to realize an improved data interpolation method and apparatus so that a ratio image can be obtained. Further, in a broad sense, it is intended to obtain an SN ratio that is about the same as the arithmetic mean of a plurality of data in the interpolation for weighting and interpolating from a plurality of data.

[0016]

According to the present invention for solving the above-mentioned problems, in data interpolation for obtaining interpolation data using a plurality of data containing random noise, a noise component and a signal component are obtained from each of the plurality of data. Is extracted, a weighting factor corresponding to the distance from the interpolation point is determined for each of the plurality of data, and an arithmetic mean is obtained among the plurality of extracted noise components, and In addition, the addition processing is performed on the plurality of noise components and the signal components so as to perform weighted addition accompanied by multiplication of the weighting factors.

Further, in particular, the above is applied to the data interpolation for obtaining the data of the desired slice position from the data obtained by the radiation CT for performing the helical scan.

The noise component is extracted substantially by extracting a high frequency component of the data, and the signal component is extracted by extracting a low frequency component of the data. .

[0019]

Function: A signal component and a noise component are extracted from a plurality of data including random noise, weighted addition is performed on the signal component according to the distance from the interpolation point, and for the noise component, the weighted addition is performed between them. Take the averaging.

[0020]

Embodiments of the present invention will now be described in detail with reference to the drawings. First, an image quality improving method for helical scanning according to an embodiment of the present invention will be described. In the helical scan of the present embodiment, the projection data is subjected to the treatment shown in the following equation.

P (i, j) _sp = (1/2) HPF (P ₁ ) + α · LPF (P ₁ ) + (1/2) HPF (P ₂ ) + (1-α) · LPF (P ₂ ) Here, P ₁ and P ₂ are 180 ° (or 3) as in the conventional example.
Data of two points z ₁ and z ₂ having different positions (60 °).
The contents of the processing by this formula will be described below.

A high-pass filter HPF and a low-pass filter LPF for extracting a noise component composed of high-frequency components and a signal component composed mainly of low-frequency components from the projection data P ₁ are used to extract respective components HPF from P _1.
(P ₁ ) and LPF (P ₁ ) are extracted. In the equation, the first term is, for example, the projection data P _{1 at} z _A in FIG.
The noise component in the data is extracted by HPF.
The term is data in which the internal signal component of the projection data P ₁ is extracted by the LPF.

The third term is the data obtained by extracting the noise component of the projection data P ₂ at z _B by the HPF, and the fourth term is the data obtained by extracting the inner signal component of the projection data P ₂ by the LPF. In the equation, the noise components HPF (P ₁ ) and H
Random noise included in the noise component is reduced to 1 / √2 by averaging PF (P ₂ ). In addition, the signal components LPF (P ₁ ) and LPF (P ₂ ) are respectively z _A
The data of the desired slice position is obtained by weighting and adding according to the distance from.

In general, even if the noise component HPF (P ₁ ) is obtained by a high-pass filter, the high-frequency component of the projection data depending on the subject is included in this, in addition to the random noise. The component is information necessary for improving the resolution. For this reason, P (i, j)
It is used to find _sp without discarding noise components.

As described above, according to the interpolation method of the present embodiment, the noise component is not averaged by weighted addition but averaged, and only the signal component is weighted and added. Therefore, the SN ratio of the data obtained by the helical scan is improved to a level equivalent to the conventional normal axial scan, and an image with a good SN ratio can be obtained without increasing the X-ray dose.

The present invention is not limited to the method of the above embodiment. Although the two-point linear interpolation has been described in the embodiment, it is also effective when one projection data is interpolated using a plurality of projection data of three points or four points or more in different directions of substantially 180 ° or 360 °. Can be used.

Further, although an example in which the HPF and the LPF are used to extract the noise component and the signal component has been described, the processing of the following equation may be performed to replace the HPF. PHPF (P ₁ ) = P ₁ −LPF (P ₁ ) ... This PHPF (P ₁ ) is defined as HPF (P ₁ ) in the equation. However, the computing unit that performs this computation may be considered as an HPF.

Furthermore, the noise component and the signal component may be extracted using a white noise detection filter without using a frequency filter. Next, a block diagram of an interpolator for implementing the above method is shown in FIG. First, prior to the description of the interpolating apparatus of FIG. 1, the data processing portion of the X-ray CT including the above-mentioned interpolating apparatus will be described with reference to the block diagram of FIG. In the figure, reference numeral 1 is a data collection device for collecting data obtained from a subject by a helical scan. This data is stored in the data memory A2. The address of the data memory A2 is configured as shown in FIG. That is, the address is the view number j on the horizontal axis and the channel number i on the vertical axis.

Reference numeral 3 denotes a read address generator for generating a read address for reading data used for interpolation from the data memory A2, and the data read from the data memory A2 is input to the interpolation device 4. The interpolator 4 calculates the equation to obtain the interpolated data.

Reference numeral 5 is a data memory B for storing the interpolated data output from the interpolating device 4, and this data is reconstructed into image data by the image reconstructing device 6 and displayed on the display device 7.

Reference numeral 8 is a controller for controlling the operation and timing of each device. Next, the operation of the apparatus configured as described above will be described. X-ray C that has undergone a helical scan
The data from T is collected by the data collector 1 and stored in the address shown in FIG. 3 in the data memory A2.

Under the control of the controller 8, the read address generator 3 stores the read address in the data memory A.
Data to be used for interpolation is read out by inputting to 2 and input to the interpolating device 4.

The interpolator 4 reads the read data P ₁ , P
To calculate the interpolation data by using the α corresponding to the distance ₂ positions z _1, z ₂ and the slice position z _sp. The interpolated data is once stored in the data memory B5, then reconstructed into image data by the image reconstructing device 6 and displayed on the display device 7. Α in this weighted addition is, for example, linear interpolation according to the distance.

Returning to FIG. 1, details of the interpolator 4 will be described. In the figure, the same parts as those in FIG. 2 are designated by the same reference numerals. In the figure, 11 is input with the projection data P _{1 of the} formula, and noise signal extraction circuits A and 12 for extracting the noise component and the signal component are input with projection data P _{2 of the} formula to extract the noise component and the signal component. It is a noise signal extraction circuit B that does.

Reference numeral 13 indicates the positions of the above-mentioned data P ₁ and P ₂ by the signal from the read address generator 3, and the weight coefficient generators A and 14 for generating the weight coefficient α corresponding to the distance from the interpolation point are The weight coefficient generator B receives the signal from the weight coefficient generator A13 and generates a weight coefficient 1-α.

Reference numeral 15 is a multiplier A for multiplying the signal component of the output of the noise signal extraction circuit A11 by α of the output of the weighting coefficient generator A13, and 16 is a multiplier of the output of the noise signal extraction circuit B12 of the weighting coefficient generator B14. Multiplier B multiplied by 1-α
Is.

Reference numeral 17 denotes a 1/2 multiplier A that multiplies the noise component of the output of the noise signal extraction circuit A11 by 1/2, and 18 denotes 1/2 multiplication that multiplies the noise component of the output of the noise signal extraction circuit B12 by 1/2. It is a vessel B.

Reference numeral 19 is a multiplier A15, a multiplier B16, 1 /
It is an adder that adds the outputs of the 2 multiplier A17 and the 1/2 multiplier B18. Next, the operation of the interpolating device configured as described above will be described. Data P ₁ and P ₂ for calculating interpolation data from the data stored in the data memory A2
Are read by the address from the read address generator 3. The selection of this data is performed by the control of the controller 8.

The data P ₁ is input to the noise signal extraction circuit A11, the signal component and the noise component are extracted, and the product of the signal component with the weighting coefficient α of the output of the weighting coefficient generator 13 is obtained by the multiplier A15. . Noise component is 1/2 multiplier A
Multiply by 1/2 with 17.

The data P ₂ is input to the noise signal extraction circuit B12, the signal component and the noise component are extracted, and the product of the signal component with the weight coefficient 1-α from the weight coefficient generator B14 is obtained by the multiplier B16. , The noise component is 1/2 multiplier B
It is multiplied by 1/2 at 18.

The output of the multiplier A15, the output of the multiplier B16, the output of the ½ multiplier A17, and the output of the ½ multiplier 18 are added by the adder 19 to obtain P (i, j) in the equation. _sp is output.

The circuits of the noise signal extraction circuits A11 and B12 are composed of, for example, the circuit shown in FIG. In the figure, (a) shows the data from the data memory A2 as LPF.
It is a figure which shows the example extracted into the signal component and the noise component by 21 and HPF22.

(B) is a circuit for performing an arithmetic operation of a formula in which the output of the LPF 21 is subtracted from the output P1 of the data memory A2 in the subtractor 23 to obtain a noise component. In this circuit, the HPF 22 may be used instead of the LPF 21 to output noise, and the output of the subtractor 23 may be used as the signal component.

FIG. 6C is a diagram showing a circuit in which the noise component is extracted by using the white noise detection filter 24. As described above, various circuits can be used for the noise signal extraction circuits A11 and B12.

Extracting substantially high frequency components, or
Extracting the low-frequency component substantially means (a),
It includes all of (b) and (c). FIG. 5 is a block diagram of another embodiment. In the figure, reference numeral 31 is a noise signal extraction circuit which receives a large amount of data and extracts noise and signals from each data and outputs the noise and the signal. The extracted signal components are input to the weighting addition circuit 32, weighted coefficients are added according to the distance between each data and the interpolation value to be obtained, and the signals are added and output.

The noise signal extraction circuit 31 is included in the averaging circuit 33.
The noise component of the data from is extracted and input, and the average value of the noise data is output. The weighted signal and the signal averaged by the averaging circuit 33 are added by the adder 34 and output. In this way, noise and signals are extracted from a plurality of data, signal components are weighted and added, noise components are averaged, and weighted and added signals are added to obtain interpolated data. You can

As described above, according to the present embodiment, noise components including random noise are extracted, weighted and averaged without weighting, and only signal components are weighted and added, so that the SN ratio is lowered. In addition, the resolution can be improved and the deterioration of the image can be prevented.

[0048]

As described in detail above, according to the present invention, since noise components are added and averaged and signal components are weighted and added, the S / N ratio of the prior art is corrected while correcting the deviation from the slice position due to helical scanning. It is possible to obtain an image having a better SN ratio than the conventional helical scan, and the practical effect is great. Also, in general data interpolation in which weighted interpolation is performed from a plurality of data, noise can be suppressed to the same extent as the averaging of a plurality of data.

[Brief description of drawings]

FIG. 1 is a block diagram of an interpolation device according to an embodiment of the present invention.

FIG. 2 is a diagram of a circuit for improving the image quality of a helical scan using the interpolation device of the present invention.

FIG. 3 is a configuration diagram of a data memory for sending data to an interpolation device.

FIG. 4 is a diagram showing a configuration of a noise signal extraction circuit,
(A), (b), and (c) are different circuit examples.

FIG. 5 is a diagram of another embodiment of the present invention.

FIG. 6 is a diagram showing a coordinate system of a CT scanner.

FIG. 7 is a diagram showing the relationship between the phase of an X-ray source, the position of a cradle, and detection data in helical scanning.

FIG. 8 is a change curve of standard deviation with respect to a weighting coefficient.

[Explanation of symbols]

 2 data memory 3 read address generator 4 interpolator 11, 12, 31 noise signal extraction circuit 13, 14 weighting coefficient generator 15, 16 multiplier 17, 18 1/2 multiplier 19 adder 32 weighting addition circuit 33 average Circuit 34 adder

Claims (6)

[Claims] 1. A data interpolation method for obtaining interpolation data using a plurality of data containing random noise, the steps of extracting a noise component and a signal component from each of the plurality of data, and each of the plurality of data. The step of determining a weighting factor according to the distance from the interpolation point, and the multiplication of the weighting factor between the plurality of extracted noise components and the addition of the weighting factor among the plurality of signal components. And a step of performing addition processing on the plurality of noise components and signal components so that the weighted addition is accompanied. 2. A data interpolation method for obtaining data at a desired slice position from data obtained by radiation CT for performing a helical scan, wherein a noise component and a signal component are obtained from each of two data points obtained at two slice positions. And a weighting coefficient α and 1-α according to the distance from the slice position of the interpolation point for each of the two points of data, and an arithmetic mean between the two extracted noise components. And between the two signal components, one of the two noise components is such that one is a multiplication by a weighting factor α and the other is a weighted addition involving multiplication by a weighting factor 1-α. And a step of performing addition processing on the two signal components. 3. A noise component is extracted substantially by extracting a high frequency component of data, and a signal component is extracted by
3. The data interpolation method according to claim 1 or 2, wherein the low frequency component of the data is substantially extracted. 4. A data interpolating device for obtaining interpolated data using a plurality of data containing random noise, comprising means (11, 12) for extracting a noise component and a signal component from each of the plurality of data; Means (13, 14) for determining a weighting coefficient according to the distance from the interpolation point for each of the data, and the plurality of noise components and the signals so that an average is obtained between the plurality of extracted noise components. A data interpolating device including means for performing addition processing on components. 5. A data interpolating device for obtaining data at a desired slice position from data obtained by radiation CT performing a helical scan, wherein a noise component and a signal component are obtained from each of two data points obtained at two slice positions. For extracting (11, 1
2), a means (13, 14) for determining weighting factors α and 1-α according to the distance from the slice position of the interpolation point for each of the two data points, and between the two extracted noise components Then, in the above two signal components, one is multiplied by a weighting factor α and the other is a weighted addition involving multiplication by a weighting factor 1-α. Means for performing addition processing on one noise component and two signal components (1
5, 16, 17, 18, 19) and a data interpolating device. 6. A noise component is extracted substantially by extracting a high frequency component of data, and a signal component is extracted by:
6. The data interpolating device according to claim 4, wherein the low frequency component of the data is substantially extracted.