JPH0481981A - Processor for digital radiation image signal - Google Patents

Abstract

PURPOSE:To improve the picture quality of a reproducing image by changing an interpolating operation expression in accordance with an image data level before interpolation, in an interpolating operation executed in order to change a picture element. CONSTITUTION:A CPU 15 performs an interpolating operation to radiation image information (image data) stored in an image memory 14, and simultaneously, performs various image processings being suitable for a diagnostic purpose, and the image data to which the image processing is performed is stored in the image memory 14 again. In this case, at the time of executing the interpolating operation in order to change the number of picture elements, an interpolating operation expression is changed in accordance with a picture element data level before the interpolating operation, therefore, the processing for emphasizing or weakening a spatial frequency component can be executed in accordance with the picture element data level (density or luminance). In such a way, an easily visible reproducing image can be realized.

Description

[Detailed Description of the Invention] <Industrial Application Field> The present invention relates to a digital radiation image signal processing device, and more specifically, the present invention relates to a digital radiation image signal processing device that can improve the image quality of a reproduced image by performing an interpolation operation to change the number of pixels. The present invention relates to a device that does this.

<Conventional technology> Radiographic images such as X-ray images are often used for disease diagnosis. irradiate,
This generated visible light, which was then irradiated onto a film using silver salt to develop the film, similar to ordinary photography.
So-called radiographs have been widely used.

However, in recent years, methods have been devised to directly extract images from the phosphor layer without using a film coated with silver salt.

In this method, the radiation transmitted through the subject is absorbed by a phosphor, and then this phosphor is excited with light or thermal energy, so that the radiation energy accumulated by the phosphor is converted into fluorescence. There is a method of emitting fluorescence and photoelectrically converting the fluorescence to obtain an image signal.

Specifically, for example, U.S. Pat. has been done. This method uses a radiation image conversion panel with a stimulable phosphor layer formed on a support.The stimulable phosphor layer of this conversion panel is exposed to radiation that has passed through the object, and A latent image is formed by accumulating radiation energy corresponding to the degree of penetration, and then
By scanning this photostimulation layer with photostimulation excitation light, the accumulated radiation energy is emitted and converted into light,
This optical signal is photoelectrically converted to obtain a radiation image signal.

The radiographic image signal obtained in this way may be used as is, or after being subjected to image processing, it may be transferred to a silver halide film, C
It is often visualized by RT or the like, or digitized for image processing by a computer. In addition, digitized radiation image signals are stored in semiconductor storage devices and magnetic storage devices.

The image may be stored in an image storage device such as an optical disk storage device or a magneto-optical storage device, and then taken out from these image storage devices as needed and output to a silver halide film, CRT, etc. for visualization.

In addition, the silver halide film that records radiographic images is coated with a laser beam.
There is also a method of irradiating light from a light source such as a fluorescent lamp to obtain transmitted light through a silver halide film, photoelectrically converting the transmitted light to obtain a radiation image signal, and further digitizing it.

As described above, the configuration of the apparatus for obtaining digital radiographic image signals from a silver halide film on which a radiographic image has been recorded is to scan a light beam one-dimensionally on the silver halide film, and at the same time scan the silver halide film in the scanning direction. A silver halide film is transported in a direction perpendicular to the light source, and the transmitted light is detected by a photodetector installed on the opposite side of the light source. There is a device in which an aperture that guides transmitted light to a photodetector is moved parallel to the rotation axis of the drum at the same time as the drum is rotated.

<Problems to be Solved by the Invention> By the way, when reproducing the digital radiation image signal obtained as described above, it is necessary to adjust the density of the region of interest (the region including the image part necessary for medical diagnosis) in the reproduced image. In order to achieve a uniform finish and output the shadows of human body structures and lesions more clearly, image processing such as gradation processing and spatial frequency processing is performed before outputting to a CRT etc. for visualization and diagnosis. I am trying to provide it.

The gradation processing is a process that determines the correspondence between the image signal level and the output density (or brightness) so that a desired gradation can be obtained. Methods for determining gradation processing conditions (see Japanese Patent Laid-Open No. 63-31641, etc.) have been proposed, and the present applicant has also previously determined the desired image area from the positional information and signal values of the image signal. However, a method has been proposed in which tone processing conditions are determined based on the characteristics of image data within the area (Japanese Patent Application No. 1-306462).

Furthermore, examples of spatial frequency processing include unsharp mask processing (see Japanese Patent Publication No. 62-62373, etc.), which is processing for emphasizing or attenuating specific frequency components of an image.

On the other hand, when making a diagnosis based on radiographic images, there are cases where it is necessary to enlarge the image to a larger size or reduce it to a smaller size for reproduction. It is common practice to increase or reduce the number of pixels by interpolation, which is an operation to discretize the image again at a small or large sampling pitch (see Japanese Patent Application Laid-Open No. 175575/1983).

However, although various interpolation calculation formulas are used in conventional interpolation calculations for radiation image signals, the interpolation calculation formula is uniform within an image (rResto
ring 5line interpolation
ofCT ImagesJ IEEE TRAN
SACTION ON MEDICALIMAG[NG
,VOL,MI-2,NO,3,SEPTEMBER1
No. 983, etc.), gradation processing, spatial frequency processing, etc., do not have the effect of making the reproduced image an easy-to-see image suitable for diagnosis.

The present invention was made in view of the above-mentioned circumstances, and focuses on the fact that it is possible to emphasize or attenuate spatial frequency characteristics through interpolation calculations. It is an object of the present invention to provide a processing device capable of processing signals into easy-to-see image signals suitable for diagnosis.

<Means for Solving the Problems> Therefore, in the present invention, as shown in FIG. 1, a digital radiation image signal processing device that changes the number of pixels by interpolating a radiation image signal consisting of digital data for each pixel. interpolation calculation means for changing the number of pixels by interpolating pixel data of a digital radiation image; and interpolation calculation formula changing means for changing the interpolation calculation formula in the interpolation calculation means according to the pixel data level before interpolation calculation. The structure was made to include and.

<Operation> According to the digital radiation image processing device having such a configuration,
When performing interpolation to change the number of pixels, the interpolation formula is changed depending on the pixel data level before the interpolation, so spatial frequency components can be emphasized depending on the pixel data level (density or brightness). For example, in medical radiation images, it is possible to perform attenuation processing, such as emphasizing frequency components for areas at the image data level that contain a lot of diagnostic information, or emphasizing frequency components for areas at the image data level where noise is noticeable. It becomes possible to reduce noise by attenuating frequency components, and it is possible to realize a reproduced image that is easier to see while changing pixels through interpolation calculations.

<Example> The present invention will be explained in detail below.

FIG. 2, which shows one embodiment, shows a radiation image information recording and reading device including a digital radiation image signal processing device according to the present invention, which is applied to chest radiography of a human body for medical use. .

Here, the radiation source l is controlled by the radiation control device 2 and emits radiation (
(Generally, X-rays) are irradiated. The recording/reading device 3 is equipped with a conversion panel 4 on a surface facing the radiation source 1 with the subject M in between, and the conversion panel 4 converts energy according to the radiation transmittance distribution of the subject M with respect to the radiation dose irradiated from the radiation source 1. is accumulated in the photostimulable layer, and a latent image of the subject M is formed there.

The conversion panel 4 is provided with a stimulable layer on a support by vapor deposition of a stimulable phosphor or coating of a stimulable phosphor paint, and the stimulable layer blocks out adverse effects and damage caused by the environment. It is shielded or covered with a protective member in order to do so.

As the stimulable phosphor material, for example, JP-A-61-7
Materials such as those disclosed in Japanese Patent Laid-open No. 2091 or Japanese Patent Application Laid-open No. 75200/1986 are used.

A light beam generator (gas laser, solid-state laser, semiconductor laser, etc.) 5 generates a light beam whose emission intensity is controlled, and the light beam reaches the scanner 6 via various optical systems, where it is After being deflected, the light path is further deflected by a reflecting mirror 7 and guided to the conversion panel 4 as stimulated excitation scanning light.

The condenser 8 has a condensing end, which is an optical fiber, located close to the conversion panel 4 scanned with the stimulated excitation light, and has a condensing end that is proportional to the latent image energy from the conversion panel 4 scanned with the light beam. Receives stimulated luminescence of luminous intensity. 9 is a light condenser 8
It is a filter that allows only light in the stimulated emission wavelength range to pass from the light introduced from the filter 9, and the light that has passed through the filter 9 is
The light enters the photomultiplier 10 and is photoelectrically converted into a current signal corresponding to the incident light.

The output current from Photomaru IO is the current/voltage converter 1.
After being converted into a voltage signal in step 1 and amplified in amplifier 12,
The A/D converter 13 converts it into digital data (digital radiation image signal). This digital data is then sequentially stored in the image memory 14.

15 is a CPU that performs interpolation calculations on the radiation image information (image data) stored in the image memory 14, and at the same time performs various image processing suitable for diagnostic purposes (for example, gradation processing, frequency processing, movement, rotation, rotation, etc.). , statistical processing, etc.), and the image data subjected to image processing is stored again in the image memory 14. Reference numeral 16 is an interface for transmitting the radiation image signal in the image memo 1/4 to the printer 17.

18 is a reading gain adjustment circuit, and this reading gain adjustment circuit 18 adjusts the light beam intensity of the light beam generating section 5;
The gain of the photomultiplier IO is adjusted by adjusting the power supply voltage of the photomultiplier high-voltage power supply 19, the gain adjustment of the current/voltage converter 11 and the amplifier 12, and the input dynamic range of the A/D converter 13 are adjusted. The reading gain of the signal is adjusted comprehensively.

The image memory 14. CPU15. interface 1
6 is more specifically constructed as shown in FIG.

That is, the digital radiographic image signal from the A/D converter 13 is subjected to gradation processing in the gradation processing section 22, and then is temporarily stored in the line memory 21 constituting the image memory 14.
With reference to the interpolation characteristic table LUTα24 constituting the interpolation calculation formula changing means that stores the relationship between the image data level and the interpolation calculation formula (coefficients that determine the interpolation calculation formula) used for the interpolation calculation, the interpolation work 25 is changed for calculation. An interpolation calculation is performed by an interpolation calculation unit 26 as an interpolation calculation means controlled by a control logic 23 as a work area, and the line memory 2
1 and 2 are stored in the output line memory 27 constituting the image memo January 4.

In FIG. 3, the interpolation process is performed after the gradation process, but this order may be reversed.

In addition, 28 is a ROM attached to the CPU 15,
29 is a RAM also attached to the CPU 15.

The digital radiation image signal stored in the output line memory 27 and subjected to gradation processing and interpolation calculation is outputted to the printer 17 via the interface 16 and hard-copied. For example, the fourth
It is configured as shown in the figure.

Furthermore, what is connected via the interface 16 is the CR
The storage device may be a monitor such as T, or a storage device (filing system) such as a semiconductor storage device.

In the printer 17 shown in FIG.
The digital radiation image signal read out via the interface 16 is first subjected to various signal correction processes in the signal correction circuit 31 via the buffer memory 30, and then sent to the D
/A converter 32 converts the signal into an analog signal. Then, in order to modulate the laser beam according to this analog signal, the output of the D/A converter 32 is input to a modulator drive circuit 33, and this modulator drive circuit 33 adjusts the output level of the D/A converter 32. A corresponding drive voltage is output to the optical modulator 34.

The optical modulator 34 modulates the laser light emitted from the laser light source 35 according to the image signal level based on the drive voltage, and the modulated laser light is a deflection mirror (polygon mirror) rotated by a motor (not shown). The light is reflected by 36 polygonal reflecting surfaces and distributed in the main scanning direction. Note that a galvanometer mirror may be used as the deflection mirror.

The reflected light from the deflection mirror 36 passes through an fθ lens 37 and is adjusted to a constant scanning speed, and the scanning light is received by a recording medium (photosensitive material) 38 that is conveyed in the sub-scanning direction, whereby recording is performed. A two-dimensional radiographic image is recorded on the medium 38, and then the recording medium 38 is developed to obtain a hard copy of the digital radiographic image.

Next, details of the interpolation calculation according to the present invention performed based on the configuration shown in FIG. 3 will be explained.

First, before explaining an example in which a well-known Peruvian spline interpolation formula is used as an interpolation formula, the Peruvian spline interpolation formula will be briefly explained ( r
The Be1l-Spline, a digit
lfiltering / 1nterpolatio
n algorithm J Enric.

Dolazza, SP [E proceeding, M
I-IF, 1988).

That is, the frequency spectrum of the original discrete signal 1o(x) is 1. (ω), the original signal 1o(x
) is determined by the point spread function (the paint-sprea
d function) as G(ω
), a continuous interpolated signal obtained by interpolation 1. (x)
The frequency spectrum of 1. (ω), then 1, (ω
)=D・ID(ω)・G(ω), which is a discrete signal I obtained by interpolation.

Frequency spectrum of (X)1. (ω) is the discrete signal 1. after interpolation. If the sampling pitch of (x) is d, then the point spread function (the point
By replacing the frequency spectrum G(ω) of the t-spread function with the normalized frequency spectrum F(ω, β) in the point spread function of the Peruvian spline, the following calculation formula for Peruvian spline interpolation is obtained.

1, (ω) = Here, the point spread function F(X, β) of the Peruvian spline
is the well-known Peruvian number B (x), the spline function 5 (
It is a function expressed by the following formula using X) and a positive real number β.

F (X, β) = (3/β)S(x)+ (1/2) (1-3/β) (B (x+D/2)
10B (x-D/2)) Here, when β, which is a parameter of the point spread function in the Peruvian spline, is changed, the frequency spectrum F(ω,
β) changes, and as a result, when it is smaller than β months, a correction is applied along with interpolation that emphasizes the frequency characteristics in the spatial frequency region from 0 to 1/D, and when β is larger than 1, it attenuates the same frequency characteristics. Correction will be applied along with interpolation.

By the way, in medical radiation images, for example, it may be necessary to attenuate the spatial frequency characteristics in low-density areas where noise is easily noticeable, while it may be necessary to emphasize the spatial frequency characteristics in medium-density areas that contain a lot of diagnostic information. If the parameter β of the point spread function in the Peruvian spline interpolation is changed in accordance with such a requirement, it is possible to perform spatial frequency processing suitable for diagnosis while performing supplementary law. For example, in chest radiography, the low-density area is the abdomen where radiation is difficult to pass through and noise is likely to occur.Especially when the reproduced image is expanded to a larger size by increasing the number of pixels, Due to the nature of vision, noise in such areas becomes more noticeable. Furthermore, the medium-density area is the lung field, which is the region of interest, and the high-density area is the area through which radiation passes without passing through the subject.

Therefore, as the interpolation characteristic table LUTα in which the relationship between the image data level and the interpolation calculation formula is stored in advance, the relationship between the level of the digital radiation image signal before interpolation (image data level) and the above β is determined, for example, as shown in FIG. - Set as shown in Figure 9, determine the Peruvian spline interpolation formula (frequency spectrum of point spread function in Peruvian spline) by referring to β corresponding to the signal level at that time, and apply this interpolation formula to Interpolation processing may be performed based on this. For example, if β smaller than l is searched from the table in a region of interest that is a medium density region, it is possible to emphasize the spatial frequency characteristics in such a region, improve sharpness, and improve diagnostic ability. In addition, if a β larger than 1 is searched from the table in areas where noise is likely to occur in low-density areas, the spatial frequency characteristics will be attenuated in low-density areas, noise will be reduced, and graininess will be improved. can do.

The maximum value of β is 0.8 to 3.0, and the minimum value is 0.3.
It is preferable to set it to ~2.0, but what value within that range is optimal varies depending on the region of the image and the purpose of diagnosis, and is not uniquely determined.

In addition, in the above embodiment, the parameter β of the point spread function in Peruvian spline interpolation is changed based on the signal level before interpolation, but linear interpolation, spline interpolation, etc. The interpolation dimension (point spread function; the paint-s
It is also possible to switch and use interpolation calculations that have different dimensions (number of pread functions) and different directions of changes in spatial frequency characteristics due to interpolation.

For example, linear interpolation can be used to ensure sharpness at medium and high density signal levels corresponding to the region of interest, while spline interpolation can be used to improve graininess in low density areas where noise is noticeable. While enlarging pixels through calculation, it is possible to reduce noise without impairing the sharpness of the region of interest in the reproduced image, making the reproduced image easier to see and improving diagnostic performance.

In other words, if interpolation or linear interpolation based on a polygonal line function connecting two adjacent points is performed on all images, the spatial frequency characteristics will change only slightly, so sharpness will be maintained high, but graininess will be reduced. becomes worse depending on the enlargement ratio of the image. In addition, since interpolation is performed based on a function that connects quadratic curves determined from three neighboring points, if spline interpolation, which has the effect of attenuating frequency characteristics, is performed on all images, the graininess is excellent, but the sharpness is However, if such characteristics are selectively used depending on the requirements determined by the image signal level, it is possible to achieve both graininess and sharpness at a high level.

It should be noted that, as mentioned above, when using different types of interpolation calculation formulas by switching according to the data level of the image, it is clear that the above-mentioned linear interpolation and spline interpolation are not limited to the above. Volume interpolation and spline interpolation may be used separately.

Next, a more specific example in which the interpolation calculation formula is changed based on the signal level before interpolation as described above (including a change in the coefficient that determines the interpolation calculation formula like β) will be described as a comparative example. This will be explained by showing its effects in comparison with the following.

Incidentally, spline interpolation, linear interpolation, and cubic convolution interpolation used in each interpolation operation shown below are all known interpolation functions ("Restoring
5pline Interpolation of C
T ImagesJIEII:E TRANSACTI
ON ON MEDICAL IMAGING, VOL
, MI-2゜NO,3,5EPTεMBER1983,
rcubic Convolution for Dig
ital Image Processing J I
EEE TRANSACTIONON ACUST[
CS, AND 5IGNAL PROCESSING
, VOL, ASSP-29, No. 61981, etc.).

First, for a normal human chest frontal image signal (original signal) of 2048 pixels x 464 pixels obtained by the apparatus shown in FIG. The average gamma (the slope of the output density characteristic curve where the horizontal axis is the image signal value and the vertical axis is the output density) is approximately 2.
．． 0. Appropriate gradation modulation was performed so that the lowest density below the diaphragm was approximately 0.3, and an image signal O was obtained.

Then, the image signal O after the gradation processing is subjected to interpolation processing to enlarge the image size to 4096 pixels x 928 pixels to obtain an image signal, as shown in Fig. 4. A hard copy CA was obtained by printing on a 14 inch x 1 inch silver halide film using a standard printer 17.

Here, the interpolation calculation when obtaining the image signal A was performed as follows.

■) Signal value θ ~ 0.2x S max → Spline interpolation ■) Signal value 0.2 X S max'x - S max
-$Linear interpolation Here, Smax is the maximum value in the image signal, and since the interpolation calculation formula is changed depending on the signal level as described above, the hard copy CA is subjected to the interpolation calculation processing according to the present invention. This is a hard copy of the image signal.

Furthermore, spline interpolation processing is applied to the image signal 0 over the entire image, and the result is the same < 4096 pixels x 9 pixels.
An image signal P of 28 pixels was obtained, and this image signal P was reproduced on a silver halide film in the same manner as above to obtain a hard copy CP.

Furthermore, the image signal O is subjected to linear interpolation processing over all images to obtain an image signal Q of <4096 pixels x 4928 pixels, and this image signal Q is reproduced on a silver halide film in the same manner as above. and obtained a hard copy CQ.

The hard copy CP and hard copy CQ do not change the interpolation function depending on the signal level, so they are comparative examples to be compared with the embodiments of the present invention.

Table 1 Table 1 shows the results of a visual evaluation comparison of the hard copies CA, CP, and CQ obtained by performing different treatments as described above.

Here, the graininess is evaluated by visually comparing the graininess and ranking 1 to 1 in order of superior graininess. 2. This is a ranking of 3. Similarly, as a result of visual comparison, the sharpness was ranked as 1, 2, and 3 in descending order of sharpness, and those that were evaluated to be equivalent in terms of graininess and sharpness were The ranking is tied. As for the overall evaluation, as a result of the radiologist's observation, the hard copy that was judged to be the easiest to diagnose was marked with an ○ mark.

As is clear from Table 1 above, the hard copy CP that performs spline interpolation over the entire image is excellent in terms of graininess, but is the poorest in sharpness, and the hard copy that performs linear interpolation over the image CQ
is superior in sharpness but has the poorest graininess, and is either hardcopy CA, which uses spline interpolation or linear interpolation depending on the signal level, or hardcopy CP, which is inferior to hardcopy CP in graininess but has the lowest sharpness. Excellent hard copy CP that performs spline interpolation or linear interpolation for all image signals.

The results showed that it has better diagnostic performance than CQ.

In other words, for low-density areas where noise is noticeable but is not an area of interest, spline interpolation, which has the effect of attenuating frequency characteristics, is applied to improve graininess (reduce noise), while for other densities In the area, sharpness is ensured by linear interpolation, which emphasizes frequency characteristics more than spline interpolation, and the effect of frequency characteristic correction of both spline interpolation and linear interpolation is exhibited, resulting in excellent diagnostic performance. We were able to obtain a reproduced image.

In addition, apart from the above examples and comparative examples, for the image signal O after gradation processing, ■) signal value O ~ 0.2XSmax → spline interpolation ■)
Signal value 0.2 It was then reproduced on silver halide film to obtain a hard copy CB.

Furthermore, the image signal 0 is subjected to cubic convolution interpolation processing over all image signals to obtain 40
An image signal R of 96 pixels x 928 pixels was obtained, and a hard copy CR was similarly obtained based on this image signal R.

The hard copy CB (example) obtained by changing the interpolation function according to the signal level as described above, and the hard copy obtained only by cubic convolution interpolation, which is a comparative example in which no processing according to the present invention was performed. Table 2 shows the evaluation comparison results with hard copy CP using only CR and spline interpolation.

Table 2 Compared to the case where only cubic convolution interpolation, which has the characteristics of
It can be seen that reproducing an image processed by switching between cubic convolution interpolation and spline interpolation depending on the signal level is better in terms of diagnosability as it achieves both graininess and sharpness.

Further, as an embodiment of the interpolation process of the present invention, Peruvian spline interpolation is performed for the image signal 0 by changing the parameter β of the point spread function as shown in FIG. 10 according to the image signal level. 4096 pixels
An image signal C of 928 pixels was obtained, and a hard copy CC was similarly obtained based on this image signal C.

As a comparative example of the hard copy CC as such an embodiment, first, fix β to 2.0 and perform Peruvian spline interpolation to obtain an image signal S of 4096 pixels x 4928 pixels. A hard copy C8 was similarly obtained based on S. In addition, in other comparisons shown in Table 2, as a comparison example emphasizing the frequency characteristics, the above β was fixed to 1.0 and Peruvian spline interpolation was applied, and the same < 40
An image signal T of 96 pixels x 928 pixels was obtained, and a hard copy CT was similarly obtained based on this image signal T.

Here, hard copy CC is a case in which Peruvian spline interpolation is performed by changing the β according to the image level,
Table 3 below shows the results of a comparative evaluation of hard copies C8 and CT, which are the cases where β is fixed and Peruvian spline interpolation is performed.

Table 3 In the visual evaluation of hard copies so far, the comparison was made in the case where the interpolation function was changed according to the image level as an example, but as is clear from Table 3 above, the same interpolation function was used. By appropriately changing the coefficient (β in Peruvian spline interpolation) that determines the interpolation function according to the image signal level, even if the coefficient is constant, an image with better diagnostic properties can be obtained than when the coefficient is constant. It is clear that something can be done.

Furthermore, by increasing or decreasing the β that determines the frequency processing characteristics in Peruvian spline interpolation, it is possible to obtain an image with excellent graininess or sharpness, or it is difficult to adequately satisfy both. This shows that it is effective to change β according to the image signal level, and in this case, the low-density area where you want to reduce noise is more important than the medium-high density area that includes the region of interest. By setting a large β, while ensuring the sharpness of the region of interest,
Noise can be reduced in low density areas.

Next, based on the image signal of 4096 pixels x 928 pixels obtained by applying Peruvian spline interpolation to the pattern of change with respect to the image signal level of β as shown in FIG. On the other hand,
A hard copy CU as a comparative example was obtained based on an image signal U of 4096 pixels x 4928 pixels obtained by performing Peruvian spline interpolation with β = 0.7 over the entire image,
The results in Table 4 were obtained by comparing the three cases together with hard copy CT when β was fixed at 1.0.

Table 4 Here too, we obtained the result that images with excellent diagnosticity can be obtained by changing β according to the image signal level, but in Figure 10, which corresponds to hard copy CC, and hard copy CD. Comparing Tables 3 and 4 with reference to the corresponding Fig. 11, it is found that compared to the case where β is set to 1 and the frequency characteristics are not changed, it is better to leave β at 1 in the middle and high concentration ranges and keep the frequency characteristics unchanged. By increasing β in It can be seen that the sharpness can be improved while maintaining the image quality.

In other words, no matter how large β is set on the low concentration side as a change pattern of β with respect to the signal level, if the range of change of β is greater than 1, the frequency characteristics will be weakened. The effect is large and the graininess is further improved, and when the change range of β is below the month range, the effect of emphasizing the frequency characteristic becomes large and the sharpness is further improved.

Table 5 Furthermore, as shown in FIG. 12, by changing the β according to the image signal level and performing Peruvian spline interpolation,
An image signal E of 6 pixels x 928 pixels was obtained, and a hard copy CE as an example was obtained based on this image signal E, and 4096 pixels were obtained by performing Peruvian spline interpolation with β = 0.5 over the entire image. Hard copy CV as a comparative example based on the image signal V of pixel X4928 pixels
The results shown in Table 5 were obtained by comparing the three cases together with the hard copy CT when β was fixed at 1.0.

Here again, the results showed that by changing β according to the image signal level, it was possible to obtain images with better diagnostic performance than when β was fixed.

Furthermore, as shown in FIG. 13, Peruvian spline interpolation is performed by changing the β according to the image signal level.
An image signal E of 6 pixels x 928 pixels is obtained, and a hard copy CF as an example is obtained based on this image signal F, while a hard copy CU is obtained when β=0.7 is fixed over the entire image, The results in Table 6 were obtained by comparing the three cases together with hard copy CT when β was fixed at 1.0.

As shown in the table, the results show that by changing β according to the image signal level, it is possible to obtain images with excellent diagnostic performance that have both graininess and sharpness compared to when β is fixed. Ta.

As is clear from Tables 1 to 6 above, hard copies CA, CB, and CC are examples of the present invention.

Compared to a hard copy (comparative example) obtained using a -like interpolation calculation formula over the entire image signal range, CD, CE, and CF showed improvement in graininess without sacrificing sharpness, and improved diagnostic performance. is excellent. Because the interpolation function (interpolation calculation formula) is changed according to the signal value, noise in low-density areas that do not contribute to diagnosis is maintained while maintaining the sharpness of areas important for diagnosis (for example, lung fields in the chest). This is because it has become possible to eliminate the noise or to improve the sharpness of the signal region corresponding to the diagnostically important region without increasing the noise in the low-density region.

Furthermore, as in the above embodiment, if a Peruvian spline is used in the interpolation formula to change the value of β according to the signal value, the only change to the formula is to replace the coefficients.
This is preferable because there is no need to prepare and switch between multiple interpolation calculation units.

In this example, a system for obtaining a digital radiographic image using a stimulable phosphor was used, but a system for obtaining a digital radiographic image signal by photoelectrically converting the transmitted light of a silver halide film on which a radiographic image is recorded may also be used. This does not limit the configuration for obtaining digital radiation image signals.

Further, after the interpolation calculation according to the present invention, spatial frequency processing may be further performed, and furthermore, the interpolation calculation according to the present invention may be performed based on the image signal that has been subjected to spatial frequency processing in advance. Also good.

Further, the digital radiation image signal subjected to the interpolation calculation according to the present invention may be immediately hard-copied by the printer 17 as described above, but it may be reproduced on a CRT or temporarily stored in a filing system. read it out and make a hard copy when necessary, or copy it to a CR
It may be displayed on T.

<Effects of the Invention> As explained above, according to the present invention, in the interpolation calculation performed without pixel change, the interpolation calculation formula is changed according to the image data level before interpolation, so that frequency emphasis or It is now possible to use attenuation differently depending on the requirements based on the signal level, making it possible to reduce noise while ensuring sharpness, and improve diagnostic performance based on reproduced images in medical applications. effective.

[Brief explanation of drawings]

Fig. 1 is a block diagram showing the configuration of the present invention, Fig. 2 is a system block diagram showing an embodiment of the invention, and Fig. 3 is a block diagram showing the configuration of the present invention.
Figure 4 is a detailed system block diagram of the part that performs interpolation calculations in the illustrated system. Figure 4 is a system block diagram showing the configuration of the printer illustrated in Figure 2. Figure 5 is a system block diagram showing the configuration of the printer shown in Figure 2. Diagram showing changes in frequency spectrum when
Figure 9 is a diagram showing various patterns when changing the β according to the image data level (signal value), and Figures 10 to 13 are diagrams showing hard copies of images obtained by interpolation calculations. FIG. 4 is a diagram showing a change pattern of the β used when 14... Image memory 15... CPU 16
...Interface 17...Printer 21
．． 27... Line memory 22... Gradation processing section
23... Control logic 24... Interpolation characteristic table 25... Interpolation work 26... Interpolation calculation section

Claims (1)

[Scope of Claims] A digital radiation image signal processing device that changes the number of pixels by performing an interpolation calculation on a radiation image signal consisting of digital data for each pixel, the processing device changing the number of pixels by performing an interpolation calculation on the pixel data of the digital radiation image. A digital radiation image characterized by comprising: an interpolation calculation means for changing the pixel data level before the interpolation calculation; and an interpolation calculation formula changing means for changing the interpolation calculation formula in the interpolation calculation means according to the pixel data level before the interpolation calculation. Signal processing device.