JP2001212118A - Radiation image processing method and apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To automatically obtain an image optimal to diagnosis without complicated operation by automatically determining the optimum image processing conditions regardless of the proportion of a bone part and a soft part contained in an object. SOLUTION: An interest region decision means 20 is so arranged to determine interest regions containing a first interest region corresponding to the structure of an object as main diagnosis subject and a second interest region corresponding to the structure of the object as sub diagnosis subject by analyzing an image data of a radiation image, an image processing condition factor generation means 30 to generate an image processing condition factor based on the image data for each of the determined interest regions, an image processing condition decision means 40 to determine image processing conditions based on a plurality of image processing condition factors generated and an image processing means 60 to perform an image processing of the radiation image using the determined image processing conditions.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a radiation image processing method and a radiation image processing apparatus for processing a radiation image, and more particularly, to a radiation image processing method and a radiation image processing apparatus capable of optimally processing a radiation image.

[0002]

2. Description of the Related Art In recent years, devices capable of directly taking a radiation image as a digital image have been developed. For example, as a device that detects the amount of radiation applied to a subject and obtains a radiation image formed in accordance with the detected amount as an electric signal, a method using a detector using a stimulable phosphor is disclosed in Japanese Unexamined Patent Application Publication No. 2005-110,026. 55-12429 gazette, JP-A-63-189853 gazette,
Many have been disclosed.

In such an apparatus, a stimulable phosphor is applied to a sheet-like substrate, or is fixed by vapor deposition or the like. To absorb.

Thereafter, the stimulable phosphor is excited by light or heat energy to emit the radiation energy accumulated by the stimulable phosphor as a result of the above-mentioned absorption, and the fluorescence is converted to a photoelectric energy. Conversion is performed to obtain an image signal.

On the other hand, an electric charge corresponding to the intensity of the irradiated radiation is generated in the photoconductive layer, the generated electric charge is accumulated in a plurality of two-dimensionally arranged capacitors, and the accumulated electric charge is taken out. Has been proposed.

[0006] Such a radiation image detecting apparatus uses what is called a flat panel detector (FPD). As described in Japanese Patent Application Laid-Open No. 9-90048, this type of FPD can detect fluorescence with a photodiode or a CCD or C-MOS sensor. A similar FPD is described in Japanese Patent Application Laid-Open No. 6-342098.

[0007] In these devices, in order to express a radiographic image in a gradation suitable for diagnosis, an image obtained by the above device is automatically converted so that a doctor can easily see a portion (region of interest) of interest. It is desirable to perform gradation conversion.

[0008] As such, techniques for determining gradation processing conditions based on histogram analysis of an image are described in Japanese Patent Publication No. 6-44796 and Japanese Patent Publication No. 5-69347. Further, a technique in which the enhancement coefficient of the frequency enhancement processing is determined depending on the signal value is described in JP-B-62-62373 and JP-B-62-62376. Further, a technique in which a correction coefficient of the dynamic range compression processing is determined depending on a signal value is described in Japanese Patent Application No. 266318.

[0009]

In general, a radiation image of a human body includes a human body structure as a main diagnostic object and a human body structure as a secondary diagnostic object.

Here, limb bone image: bone part (main) / soft part (secondary), chest image: lung field (main) / mediastinum (secondary).

Further, even if the signal value of the radiation image and the histogram which is the frequency distribution thereof have the same shape, the signal distribution of each human body structure may be different. For example, FIG.
Even if the histogram is as shown in (a), depending on the patient, the case of FIG. 13 (b) and the case of FIG. 13 (c) can be considered.

In such a case, if the same gradation processing is performed, if the bone of the patient A has the proper density, the bone of the patient B will be shifted to the high density region. If the bone of the patient B has the proper density, And the bone of the patient A is shifted to the low concentration region.

A description will now be given with reference to FIG. 14 corresponding to FIG. In the above case, if the same frequency emphasis processing or dynamic range compression processing (the degree of processing differs depending on the signal value) is applied, in the case of gradation processing, FIG.
14 (d) and (e), and in the case of the frequency emphasis processing, as shown in FIGS. 14 (f) and (g), and in the case of the dynamic range compression processing, FIGS. 14 (h) and (i). )become that way.

As a result, the degree of frequency emphasis or dynamic range compression differs between the bone of patient A and the bone of patient B. Frequency emphasis or dynamic range compression is performed between the soft part of patient A and the soft group of patient B. The phenomena such as different degrees occur.

As described above, in the conventional image processing, even in the same region, the finish of the part of the human body structure to be diagnosed varies depending on the patient's body shape, bone thickness, bone density, organ size, and the like. There was a problem that occurred.

On the other hand, if a configuration is prepared in which a plurality of different image processing parameter sets are prepared according to the characteristics of the patient, not only the configuration becomes complicated, but also the problem that the amount of work of the operator increases is newly generated. Become.

SUMMARY OF THE INVENTION The present invention has been made in view of the above-described problems, and automatically determines an optimal image processing condition regardless of the ratio of a bone portion and a soft portion included in a subject. It is an object of the present invention to realize a radiation image processing method and a radiation image processing apparatus capable of automatically obtaining an optimal image for diagnosis without the need.

[0018]

That is, the present invention for solving the above-mentioned problems is as follows.

(1) According to the first aspect of the present invention, by analyzing image data of a radiation image formed corresponding to a radiation dose transmitted through a subject, a first structure corresponding to a subject structure which is a main diagnostic object is analyzed. A plurality of regions of interest including a region of interest and a second region of interest corresponding to a subject structure to be diagnosed are determined, and the image data included in the region of interest is determined for each of the determined regions of interest. Generating an image processing condition factor based on the determined plurality of image processing condition factors, determining an image processing condition based on the generated plurality of image processing condition factors, and performing image processing on the radiation image using the determined image processing condition,
And a radiation image processing method.

According to an eleventh aspect of the present invention, a first region of interest corresponding to a subject structure to be diagnosed is analyzed by analyzing image data of a radiation image formed corresponding to a radiation dose transmitted through the subject. A plurality of regions of interest including a second region of interest corresponding to a subject structure to be diagnosed, and a region of interest determined for each region of interest determined by the region of interest determination unit. Image processing condition factor generating means for generating an image processing condition factor based on the image data included in the region; and determining the image processing condition based on the plurality of image processing condition factors generated by the image processing condition factor generating means. Image processing condition determining means; and image processing means for performing image processing on the radiation image using the image processing conditions determined by the image processing condition determining means. It is a radiation image processing apparatus according to claim.

According to these inventions, a plurality of regions of interest are determined by analyzing image data of a radiation image, an image processing condition factor is generated for each region of interest, and an image processing condition is determined based on the plurality of image processing condition factors. Is determined and image processing is performed.

As a result, since the image processing conditions are determined for each of the plurality of regions of interest, the optimum image processing conditions are automatically determined regardless of the ratio of the bones and the soft parts included in the subject, and the complicated processing is performed. It becomes possible to automatically obtain an optimal image for diagnosis without any operation.

(2) The invention according to claim 2 is that, in the determination of the region of interest, a plurality of small regions are set in an image, and a threshold value is set for each small region based on image data in the small region. Is determined, and by comparing the signal value of the pixel in the small area with the threshold value, it is determined whether or not the pixel is in the target area of interest, and the determined results are integrated. 2. The radiographic image processing method according to claim 1, wherein a region of interest is determined by:

According to a twelfth aspect of the present invention, the region-of-interest determination means sets a plurality of small regions in an image, and sets a threshold value for each small region based on image data in the small region. Threshold value determining means for determining, by comparing the signal value of the pixel in the small area with the threshold value, determining means to determine whether the pixel is in the target region of interest,
12. The radiation image processing apparatus according to claim 11, wherein a region of interest is determined by integrating results determined by the determination unit.

In these inventions, a plurality of small areas are set in an image, a threshold value is determined for each small area based on image data in the small area, and a signal value of a pixel in the small area is determined. Is compared with the threshold value to determine whether or not the pixel is within the target region of interest, and a plurality of regions of interest are determined by integrating the determined results. The image processing condition factors are generated, and the image processing conditions are determined based on the plurality of image processing condition factors to perform the image processing.

As a result, since the image processing conditions are determined by accurately determining a plurality of regions of interest, the optimum image processing conditions are automatically determined irrespective of the ratio between the bones and the soft parts included in the subject. This makes it possible to automatically obtain an optimum image for diagnosis without any complicated operation.

(3) The invention according to claim 3, wherein the determination of the region of interest is performed by scanning an image using a plurality of scanning lines, and for any pixel on each scanning line, a characteristic value due to a signal change between neighboring pixels. Is calculated, using the calculated characteristic value to determine whether or not a pixel on the scanning line is a point on the contour of the target region of interest, and integrating the determined result to determine the interest 2. The radiation image processing method according to claim 1, wherein a contour of the region is determined, and a region surrounded by the contour is set as the region of interest.

According to a thirteenth aspect of the present invention, the region of interest determining means scans the image using a plurality of scanning lines,
A characteristic value calculation unit that calculates a characteristic value due to a signal change between neighboring pixels for an arbitrary pixel on each scanning line, and using the characteristic value calculated by the characteristic value calculation unit, a pixel on the scanning line is used for the purpose. A determination unit that determines whether or not the point is on the contour of the region of interest to be determined.The contour of the region of interest is determined by integrating the results determined by the determination unit, and is surrounded by the contour. 12. The radiation image processing apparatus according to claim 11, wherein a region that has been set is the region of interest.

In these inventions, it is determined whether or not a pixel on the scanning line is a point on the contour of the target region of interest, and the determined result is integrated to determine the contour of the region of interest. Are determined, image processing condition factors are generated for each region of interest, and image processing conditions are determined based on the plurality of image processing condition factors to perform image processing.

As a result, since the determination of the region of interest is accurate, the image processing conditions for each of the plurality of regions of interest are also accurate, and the optimal image processing conditions are automatically obtained, and the optimal image processing conditions are obtained without complicated operations. Images can be obtained automatically.

(4) The invention according to claim 4, wherein the region of interest is determined by setting a plurality of small regions in an image and extracting each small region based on image data in the small region. Is generated as a vector element, a plurality of clusters are generated by statistically classifying a set of the generated feature vectors, and for the plurality of small regions, the feature vector of each small region is The radiation image processing method according to claim 1, wherein the region of interest is determined based on which of the plurality of clusters the region belongs.

According to a fourteenth aspect of the present invention, the region-of-interest determination means sets a plurality of small regions in an image and extracts each small region based on image data in the small region. Feature vector generating means for generating a feature vector having a vector element, and cluster generating means for generating a plurality of clusters by statistically classifying a set of feature vectors generated by the feature vector generating means. 12. The radiation image processing apparatus according to claim 11, wherein, for the plurality of small regions, a region of interest is determined based on which of the plurality of clusters the feature vector of each small region belongs to. is there.

In these inventions, a plurality of regions of interest are determined by classifying feature vectors for each small region, an image processing condition factor is generated for each region of interest, and an image processing condition factor is determined based on the plurality of image processing condition factors. Processing conditions are determined and image processing is performed.

As a result, since the determination of the region of interest is accurate, the image processing conditions for each of the plurality of regions of interest are also accurate, and the optimum image processing conditions are automatically determined, and the optimum image processing conditions are obtained without complicated operations. Images can be obtained automatically.

(5) The invention according to claim 5, wherein the image processing condition factor is generated by extracting a representative signal range for each region of interest based on image data included in the region of interest.
2. The radiation image processing method according to claim 1, wherein an image processing condition factor corresponding to the representative signal range is generated based on a predetermined image processing parameter.

According to a fifteenth aspect of the present invention, the image processing condition factor generating means extracts a representative signal range for each region of interest based on image data included in the region of interest. The radiation image processing apparatus according to claim 11, further comprising means for generating an image processing condition factor corresponding to the representative signal range based on image processing parameters given in advance.

In these inventions, an image processing condition factor is determined from a representative signal range included in a region of interest, and an image processing condition is determined based on the plurality of image processing condition factors to perform image processing. I have.

As a result, since the creation of the image processing condition factors is accurate, the image processing conditions for each of a plurality of regions of interest are also accurate, and the optimum image processing conditions are automatically obtained, and the diagnosis can be performed without complicated operations. An optimal image can be obtained automatically.

(6) The radiation image processing method according to any one of claims 1 to 5, wherein the image processing is gradation processing.

The invention according to claim 16 is the radiation image processing apparatus according to any one of claims 11 to 15, wherein the image processing is gradation processing.

In these inventions, since the image processing is gradation processing, it is possible to automatically obtain an image subjected to gradation processing optimal for diagnosis without complicated operations.

(7) The invention according to claim 7 is characterized in that the image processing is frequency emphasis processing.
A radiation image processing method according to any one of claims 1 to 5.

According to a seventeenth aspect of the present invention, there is provided the radiation image processing apparatus according to any one of the eleventh to fifteenth aspects, wherein the image processing is a frequency enhancement processing.

In these inventions, since the image processing is the frequency emphasis processing, it is possible to automatically obtain an image subjected to the frequency emphasis processing optimal for diagnosis without complicated operation.

(8) The radiation image processing method according to any one of claims 1 to 5, wherein the image processing is a dynamic range compression processing.

The invention according to claim 18 is the radiation image processing apparatus according to any one of claims 11 to 15, wherein the image processing is a dynamic range compression processing.

In these inventions, since the image processing is the dynamic range compression processing, it is possible to automatically obtain an image that has been subjected to the dynamic range compression processing optimal for diagnosis without complicated operations.

(9) The invention according to claim 9 displays the radiographic image or the processed image subjected to the image processing,
9. The display according to claim 1, further comprising displaying information on the region of interest, information on the image processing condition factor, or information on the image processing condition, together with the display of the radiation image or the processed image. Or a radiation image processing method described in

Further, the invention according to claim 19 has an image display means for displaying the radiation image or the processed image subjected to the image processing, and the image display means displays the radiation image or the processed image together with the radiation image or the processed image. Information about the area of interest,
19. The radiographic image processing apparatus according to claim 11, wherein information related to the image processing condition factor or information related to the image processing condition is displayed.

In these inventions, information on the region of interest, information on the image processing condition factor, or information on the image processing condition is displayed together with the display of the image.

As a result, by displaying a region of interest, an image processing condition factor, and an image processing condition that have an important effect on the image processing, it is possible to obtain an image that has been subjected to accurate and optimal image processing, and to provide an image interpretation doctor with the information. It can contribute to providing information useful for diagnosis.

(10) According to a tenth aspect of the present invention, information relating to correction of image processing is input, and the determination of the image processing condition determines the image processing condition corrected according to the input information. The radiation image processing method according to any one of claims 1 to 9, wherein image processing is performed on the radiation image using the corrected image processing conditions.

Further, the invention according to claim 20 has correction information input means for inputting information relating to correction of image processing, wherein the image processing condition determining means determines the image processing condition according to the information input by the correction information input means. 20. The image processing apparatus according to claim 11, wherein a corrected image processing condition is determined, and the image processing unit performs image processing on the radiation image using the corrected image processing condition. It is a radiation image processing apparatus described in the above.

In these inventions, the image processing is performed under the image processing conditions corrected by the input of the operator. As a result, it is possible to obtain an image on which accurate and optimal image processing has been performed, and to contribute to providing a processed image adjusted in accordance with the purpose of diagnosis and the preference of the radiologist.

[0055]

Embodiments of the present invention will be described below in detail with reference to the drawings.

Hereinafter, the configuration of the radiation image processing apparatus will be described in accordance with rough blocks. Note that each unit of the radiation image processing apparatus according to the present embodiment can be configured by hardware, firmware, or software. For this reason, a functional block diagram according to the processing procedure of each means is shown.

[A] Radiation image input means: As shown in FIG. 1, the radiation image input means 10 generates an image having a signal value proportional to the logarithm of the irradiated radiation dose.

As the radiation image input means 10, a device using sensors such as the above-mentioned FPD and CCD, or a known device for reading a stimulable phosphor plate to generate a radiation image can be used. . In each case, a signal value proportional to the logarithm of the radiation dose is obtained, and the larger the dose, the higher the signal value.

Although not shown, in order to reduce the time required for the subsequent processing of each part, the reduced image generating means samples the original radiation image and reduces the number of pixels by sampling the reduced radiation image. May be created, and a region of interest to be described later may be extracted.

[B] Region of Interest Determination Unit: The region of interest determination unit 20 analyzes the radiation image transmitted from the radiation image input unit 10.

As a result, a plurality of regions of interest including the first region of interest corresponding to the structure of the subject to be diagnosed and the second region of interest corresponding to the structure of the subject to be diagnosed are determined.

As an example of the region of interest, for example, when the imaging region is the lower leg bone (refer to the original image in FIG. 3A), a main diagnosis target: a bone part → the region of interest 1 (FIG. 3B) The main diagnostic object: soft part → region of interest 2 (see the hatched portion in FIG. 3C) may determine the region of interest.

The region-of-interest determination means 20 is provided in FIG.
As shown in (a) to (c), the following three aspects are considered.

A local threshold value method: threshold value determining means 32 for setting a plurality of small regions in an image and determining a threshold value for each small region based on image data in the small region; A determination unit 33 that determines whether or not the pixel is in the target region of interest by comparing the signal value of the pixel in the small region with the threshold value (FIG. 2A). reference),
The region of interest is determined by summing the results determined by the determination means 33.

For example, let us consider a case where the imaging part is an elbow joint as shown in FIG. First, as shown in FIGS. 4A to 4C, a subject area (= soft part + bone part) is extracted.
Here, the area where the subject is imaged has a smaller irradiation dose compared to the area where direct radiation is irradiated,
The signal value becomes relatively low. Therefore, an area where the signal value is lower than the surrounding area is extracted as a subject area (FIG. 4).
(C)).

That is, it is executed according to the following procedure. As shown in FIG. 4A, the radiation image is divided into a plurality of small areas. Then, for each small area, an average signal value of the pixel signal values included in the small area is obtained as the first threshold Th1. For each small area, a pixel having a signal value lower than the first threshold value Th1 is detected as a subject area (see FIG. 4B). With respect to the subject area obtained in each of the above-described small areas, an average signal value of all the subject areas is obtained, and is set as a second threshold value Th2. In the entire image, a pixel having a signal value lower than the second threshold value Th2 is detected as a subject area (see FIG. 4C). Then, a method similar to the above is repeated using only the inside of the subject area extracted as described above (FIG. 4).
(D)). By the repetition, a bone region is extracted (FIG. 4E). -The bone region obtained in this manner is defined as a region of interest 1,
A region obtained by excluding the region of interest 1 from the subject region obtained in (c) is defined as a region of interest 2.

In the above description, Th1 and T
An example has been described in which two-stage threshold processing is performed using two types of thresholds h2, but one-stage threshold processing using one type of threshold, or three or more types of thresholds are performed. Multi-level threshold processing using values may be used. Although an example in which the radiation image is divided into a plurality of small regions that do not overlap each other has been described above, small regions that overlap each other may be set. Also,
4 (a) to 4 (c) and FIGS.
In the step (e), the way of dividing the small area may be different from each other. The region of interest determined by performing the local threshold method as described above may be corrected using a generally known boundary tracking method, expansion / contraction processing, or the like. In addition, by inspecting the area and shape of the determined region of interest, the signal distribution, etc., it is determined whether or not the determined region of interest is correct. The value may be changed and the process may be performed again.

Contour extraction method: A characteristic value calculating means 34 for scanning an image using a plurality of scanning lines and calculating a characteristic value of an arbitrary pixel on each scanning line due to a signal change between neighboring pixels.
And determining means 35 for determining whether or not a pixel on the scanning line is a point on a contour of a target region of interest using the characteristic value calculated by the characteristic value calculating means 34 (FIG. 2 (b)), the outline of the region of interest is determined by integrating the results determined by the determination means 35, and the region surrounded by the outline is defined as the region of interest.

For example, let us consider a case where the imaging part is an elbow joint as shown in FIG. As shown in FIG. 5A, the radiographic image or the reduced radiographic image is scanned in the horizontal and vertical directions. In FIG. 5A, a broken line in the horizontal direction and a dashed line in the vertical direction schematically show the result of scanning (scanning line). In the case of a horizontal scanning line, a derivative value S of an arbitrary pixel with respect to the right neighboring pixel is sequentially determined from the left end of the radiation image.
Find 1 If S1 is larger than a positive first threshold value Th1, it is detected as a boundary point BL of the subject area. When S1 is smaller than Th1, it is determined that the pixel is not a boundary point of the subject area, and the pixel is moved to a pixel adjacent to one or more pixels to the right, and the same procedure is performed. The above processing is repeated until BL is found or a predetermined condition is satisfied. Here, the predetermined condition is determined, for example, by the scanning distance, and the scanning distance is 3/3 of the length of the scanning line.
If it exceeds 4, the condition is that scanning is terminated even if BL is not found. When the processing is completed as described above, BL is set at the left end on the scanning line. Next, horizontal scanning is performed in the reverse direction to the above in order from the right end of the image. This time, a derivative value S2 of an arbitrary pixel with respect to a neighboring pixel existing on the left side thereof is obtained. If S2 is smaller than a second threshold value Th2 having a sign opposite to the first threshold value Th1, the subject boundary point BR Detected as・ Repeat until BR is found or a predetermined condition is satisfied in the same manner as the above BL. If BR is not found, BR is set at the right end on the scanning line.・ If BR is on the left side of BL, B
Swap L and BR.・ For vertical scanning lines, follow the same procedure as above.
A boundary point Bt at the upper end of the subject and a boundary point Bb at the lower end are obtained. .BR and BL on adjacent scanning lines are respectively connected,
A region sandwiched between the formed line segments is extracted as a subject region. Similarly, Bt and Bb are connected to each other, and a region sandwiched between the formed line segments is extracted as a subject region (FIG. 5).
(Refer to (b).) The same processing is repeated for the inside of the area obtained by connecting the boundary points obtained by the above processing to obtain the boundary points. In one search performed on one scan line,
The number of boundary points to be detected is two, but finally four boundary points on one scanning line. In the above processing, the detection conditions (threshold and search range) of the two boundary points detected in each search may be different from each other. A region of interest formed by connecting the inner boundary points to the region of interest 1
The region obtained by excluding the region of interest 1 from the region formed by connecting the outer boundary points is defined as the region of interest 2.

The region of interest determined by performing the above-described contour extraction method may be corrected using a contour smoothing process, an expansion / contraction process, or the like. In addition, by inspecting the area and shape of the determined region of interest, the signal distribution, and the like, the determined region of interest is determined to be correct or incorrect, and if determined to be incorrect, the position of the scanning line or the value of the threshold value is determined. May be changed and the process may be performed again.

Hierarchical clustering method: A feature vector generation for setting a plurality of small areas in an image and generating a feature vector having, as a vector element, a feature extracted for each small area based on image data in the small area. Means 36;
A cluster generation unit 37 that generates a plurality of clusters by statistically classifying a set of feature vectors generated by the feature vector generation unit 36 (see FIG. 2C). For the region, the region of interest is determined based on which of the plurality of clusters the feature vector of each small region belongs to.

This hierarchical clustering method is described in Medical Imaging Technology Vol. 15, No. 4, PP.
7,1997 and Medical Imaging Technology Vol.16, No.4, P
P. 455, 1998.

Specifically, the hierarchical clustering method is executed according to the following procedure. As shown in FIG. 6A, a small area Aj of n pixels × n pixels is set in the image. Then, the small area Aj is set over the entire image while being shifted by one pixel, and J small areas Aj (j = 0, 1,..., J-1) are obtained. -The pixel value of the pixel included in the small area Aj is represented as Vj pq.
Here, p = 0, 1,..., N−1 and q = 0, 1,.
(B)). Has a pixel value as a vector element as follows (n
Xn) Create a dimensional pattern vector Pj. Pj = (Vj 00, Vj 10, Vj 20,..., Vj n-1 0, Vj 01,
Vj 11,..., Vj n-1 1,..., Vj n-1 n-1) All pattern vectors Pj (j = 0, 1,..., J-1) are arranged in an (n × N) -dimensional space. Then, it is separated into a plurality of clusters using a statistical method. Note that an annealing method, a K-means method, or the like can be used as a cluster generation method. The bones, soft parts, and other image areas have different textures and belong to different clusters. Therefore, these are separated into regions of interest 1, regions of interest 2, and other regions.

In the above, an example in which the pixel value of the original image is used as a vector element has been described. However, the pixel value obtained by performing image processing such as frequency processing on the original image may be used as the vector element.
The region of interest determined by performing the above-described hierarchical clascling method may be corrected using a generally known boundary tracking method, expansion / contraction processing, or the like. Also, by inspecting the area, shape, signal distribution, etc. of the determined region of interest,
Correctness / incorrectness of the determined region of interest may be determined, and if an error is determined, the process may be performed again by changing the small region size or the cluster generation method.

The above three methods of determining a region of interest have been described above.
May be used in combination.

A method of setting a region of interest corresponding to a desired human body structure using, for example, profile information in the vertical and horizontal directions (see Japanese Patent Application Laid-Open No. 3-218578). A method of setting a region of interest corresponding to a desired human body structure by a method combining threshold processing and labeling (see Japanese Unexamined Patent Publication No. 5-7578) may be used. In addition, as described below, it is preferable to perform an irradiation field recognition process before determining the region of interest, and determine the region of interest in the recognized irradiation field.

In taking a radiographic image, for example, a portion not required for diagnosis is not irradiated with radiation, or a portion not required for diagnosis is irradiated with radiation. To prevent the resolution from deteriorating due to being incident on the required area, a non-transmissive substance such as a lead plate is installed on a part of the subject or the radiation generator to limit the irradiation field of the subject to radiation. Irradiation field aperture is performed. When this irradiation field aperture is performed, it is assumed that level conversion processing and subsequent gradation processing are performed using image data of the irradiation field inside area and the irradiation field outside area. Image processing of a part required for diagnosis cannot be performed properly. Therefore, irradiation field recognition for discriminating the irradiation field inside area and the irradiation field outside area is performed.

As the irradiation field recognition, see, for example,
For example, a differentiation process is performed using image data on a line segment from a predetermined position on an imaging surface toward an end of the imaging surface using a method disclosed in Japanese Patent Application No. 259538. Since the signal level of the differentiated signal obtained by this differentiation processing becomes large at the irradiation field edge portion, the signal level of the differentiation signal is determined to obtain one irradiation field edge candidate point. A plurality of irradiation field edge candidate points are obtained by performing the process of obtaining the irradiation field edge candidate points radially around a predetermined position on the imaging surface. By connecting a plurality of irradiation field edge candidate points obtained in this way with a straight line or a curve, an irradiation field edge portion is obtained.

Further, a method described in Japanese Patent Application Laid-Open No. 5-7579 can be used. According to this method, when the imaging surface is divided into a plurality of small regions, in a small region outside the irradiation field where the irradiation of the radiation is blocked by the irradiation field diaphragm, the radiation dose of the radiation becomes substantially uniform, and the variance of the image data is reduced. Becomes smaller. Also, in a small area inside the irradiation field, the radiation amount is modulated by the subject, so that the variance value is higher than in the outside of the irradiation field. Further, in a small region including the irradiation field edge portion, the portion having the smallest radiation dose and the portion of the radiation dose modulated by the subject are mixed, so that the variance value is the highest.
From this, the small area including the irradiation field edge is determined based on the variance value.

Further, the method disclosed in JP-A-7-181609 can be used. In this method, the image data is rotated and moved about a predetermined rotation center, and the rotation is performed until the boundary of the irradiation field is parallel to the coordinate axes of the rectangular coordinates set on the image by the parallel state detection unit,
When the parallel state is detected, the straight line equation of the boundary before rotation is calculated by the straight line equation calculating means based on the rotation angle and the distance from the rotation center to the boundary line. Then, by determining the area surrounded by multiple boundaries from the linear equation,
The area of the irradiation field can be determined. If the irradiation field edge is a curve, the boundary point extracting means extracts, for example, one boundary point based on the image data, and extracts the next boundary point from a group of boundary candidate points around the boundary point. Similarly, by sequentially extracting the boundary points from the boundary candidate point group around the boundary point, it is possible to determine even if the irradiation field edge portion is a curve.

In the above processing, it is preferable to perform processing using a thinned-out image with a reduced number of pixels in that the processing speed can be improved.

The data of the determined plurality of regions of interest are sent to the image processing condition factor generating means 30 as described above.

[C] Image processing condition factor generating means: In the image processing condition factor generating means 30, for each region of interest determined by the above-described region of interest determining means 20, based on the image data included in the region of interest. Generate an image processing condition factor.

(C-1) Extraction of a representative signal range from a region of interest: A plurality of regions of interest are determined by the above-described processing. Here, a region of interest i (i = 1, 2,...) Are statistically analyzed and the representative signal range (S
iL to SiH).

Here, SiL means a substantially minimum signal value of the region of interest i, and SiH means a substantially maximum signal value of the region of interest i.

Further, the signal value at which the cumulative histogram in the region of interest i becomes a predetermined value may be used. For example, SiL:
20% and SiH: 80%.

Also, Si = SiL = SiH (representative signal value instead of representative signal range) may be satisfied. For example, Si
Is a signal value at which the average signal value of the region of interest i, the median value of the region of interest i, and the cumulative histogram within the region of interest i become predetermined values.

(C-2) Generation of image processing condition factors corresponding to representative signal ranges: The image processing condition factors are determined using the representative signal ranges or signal values obtained above and parameters given in advance. Generate factors.

The following are examples of the parameters given in advance. [Tone processing] For the histogram shown in FIG. 7A, two factors (FIG. 7
(B)) is obtained. That is, the input signal range S1L to S1H
Is defined as S1L 'to S1H', and the output signal range corresponding to the input signal range S2L to S2H is defined as S2L 'to S1L'.
Two factors such as S2H 'are obtained. [Gray scale processing] The output signal value corresponding to the input signal value SiL is SiL ', and the average contrast of the input signal range SiL to SiH is αi. [Gray scale processing] The output signal value corresponding to the input signal value Si. Is Si ′, and the contrast at the input signal value Si is αi
And [Frequency processing] A histogram as shown in FIG.
As shown in FIG. 8B, the input signal ranges S1L to S1
An emphasis coefficient β1 corresponding to 1H and an emphasis coefficient β2 corresponding to the input signal range S2L to S2H are determined. [Frequency processing] The enhancement coefficient corresponding to the input signal value SiL is set to βiL, and the enhancement coefficient corresponding to the input signal value SiH is set to βiH. [Frequency processing] The enhancement coefficient corresponding to the input signal value Si is set to βi.・ [Dynamic range compression processing] Input signal range SiL ~
Let βi be the correction coefficient corresponding to SiH. [Dynamic range compression processing] The correction coefficient corresponding to the input signal value SiL is set to βiL, and the correction coefficient corresponding to the input signal value SiH is set to βiH.

Here, the frequency emphasis processing and the dynamic range compression processing will be specifically described.

In the frequency emphasizing process, for example, in order to control the sharpness by a non-sharp mask process represented by the following equation, the function F is changed to a function disclosed in Japanese Patent Publication No. 62-62373 or Japanese Patent Publication No. 62-62376
It is determined by the method shown in the publication.

Sout = Sorg + F (Sorg-Sus) where Sout is the processed image data, Sorg is the image data before the frequency enhancement process, and Sus is the image data obtained by averaging the image data before the frequency enhancement process. It is sharp data.

In this frequency emphasis processing, for example, F (Sor
g−Sus) is set to β × (Sorg−Sus), and β (emphasis coefficient) is changed almost linearly between the reference values S1 and S2 as shown in FIG. Further, as shown by the solid line in FIG. 10, when low luminance is emphasized, β from the reference value S1 to the value “A” is maximized, and is minimized from the value “B” to the reference value S2. From the value “A” to the value “B”, β changes almost linearly. When emphasizing high brightness, as shown by the broken line, β from the reference value S1 to the value “A” is minimized, and β from the value “B” to the reference value S2 is maximized. Also, the value "A" ~
Up to the value “B”, β changes almost linearly. Although not shown, the value “A” to the value “B” are used when emphasizing the middle luminance.
Is maximized. Thus, in the frequency emphasis processing,
The sharpness of an arbitrary luminance portion can be controlled by the function F.

Here, reference values S1, S2 and values A, B
Is generally obtained based on the analysis result of the signal distribution for each image. Further, the method of the frequency emphasis processing is not limited to the above-described unsharp mask processing, and is disclosed in Japanese Patent Application Laid-Open No. 9-44645.
For example, a technique such as a multi-resolution method disclosed in Japanese Unexamined Patent Publication (Kokai) No. H10-208 may be used.

In the dynamic range compression processing, a function G is determined by the method disclosed in Japanese Patent Publication No. 266318 in order to perform control so that the density is within a legible range by the compression processing represented by the following equation.

Stb = Sorg + G (Sus) where Stb is the processed image data, Sorg is the image data before the dynamic range compression process, and Sus is the image data before the dynamic range compression process. It is sharp data.

Here, as shown in FIG. 11, when the non-sharp data Sus has such a characteristic that G (Sus) increases when the level of the non-sharp data Sus becomes smaller than the level “La”, as shown in FIG. Is considered to have a high groove degree, and FIG.
The image data Sorg shown in FIG. 11A is image data Stb in which the dynamic range on the low density side is compressed as shown in FIG. Further, as shown in FIG. 11D, when the non-sharp data Sus has such a characteristic that the G (Sus) decreases as the level of the non-sharp data Sus becomes smaller than the level “Lb”, as shown in FIG. It is assumed that the concentration is high, and FIG.
The image data Sorg shown in FIG. 1B has a high-density side dynamic range compressed as shown in FIG. Here, the levels “La” and “Lb” are generally obtained based on the analysis result of the signal distribution for each image.

In the above processing, the parameters given in advance may be determined on the basis of imaging information such as an imaging part, a designated body position, imaging conditions, and an imaging method. When these pieces of information are input in the radiation image input unit 10 as management information accompanying the image,
This management information can be used.

[D] Determination of image processing conditions: The image processing condition determining means 40 determines image processing conditions based on a plurality of image processing condition factors generated by the image processing condition factor generating means 30. Further, it is also possible to correct the plurality of generated image processing condition factors or final image processing conditions based on the information input from the correction information input means 50.

[0100] More specifically, as described below, the final image processing conditions are determined by integrating the above image processing condition factors.

[Gradation Processing] A gradation conversion curve is created by connecting the partial gradation conversion curves (line segments) obtained above. -Create a polygonal line connecting the line segments and perform smoothing processing by averaging or the like.・ Take discrete representative points on the line segment and connect the representative points smoothly.

For example, interpolation processing such as spline interpolation,
Approximation to multidimensional curved wool, approximation to sigmoid function, approximation to function modeling film gradation characteristics
No. 42964, JP-A-11-88688). The gradation conversion curve is a continuous function over the entire signal range of the input signal value, and its differential function is preferably continuous. It is preferable that the sign of the differential function of the gradation conversion curve is constant over the entire signal range of the input signal value. The tone conversion curve in the signal range (near the minimum signal value / near the maximum signal value) that is not included in the cotton obtained above may be determined by extrapolation, or the predetermined minimum output signal value and maximum It may be determined based on the output signal value.

In the case of the processing condition factors shown in FIG. 7, a polygonal line connecting two line segments is created. Then, the output for the minimum signal value is fixed to a predetermined signal value, the section from the minimum signal value to S1L is connected by a straight line, and the output for the maximum signal value is fixed to the predetermined signal value, and the S1L to maximum signal value is fixed. The sections of the value are connected by a straight line (FIG. 7B). After that, a smoothing process is performed (FIG. 7C).

[Frequency Processing] The partial enhancement function (line segment) β obtained above is connected to create an enhancement function. -The enhancement function is preferably a continuous function over the entire signal range of the input signal value. -The sign of the emphasis function is preferably constant over the entire signal range of the input signal value. The enhancement function of the signal range (around the minimum signal value / near the maximum signal value) not included in the line segment obtained above may be determined by extrapolation, or may be determined by a predetermined minimum enhancement coefficient and maximum enhancement coefficient. It may be determined based on this.

In the case of the enhancement coefficient shown in FIG. 8, the following monotone increasing function is determined based on two line segments. That is, in the section from S1L to (S1H + S2L) / 2, β1, (S1H
+ S2L) / 2 In the section from 2 to S2H, after creating a step function of β2, in order to prevent false contour,
Attach a predetermined slope. Further, the minimum signal value ~ S1L and S2H ~ maximum signal value are extrapolated.
Note that a curve may be obtained by performing a smoothing process.

[Dynamic Range Compression Processing] A partial correction function (line segment) obtained above is connected to create a correction function. The connection method is the same as in the above-described frequency processing.

[E] Image Processing: The image processing conditions determined by the image processing condition determining means 40 are
Sent to 0. The image processing means 60 performs image processing using the radiation image as the original image sent from the radiation image input means 10 and the image processing conditions to obtain a final output image. In this image processing means 60, gradation processing,
Frequency emphasis processing, dynamic range compression processing, and a combination thereof are performed.

[F] Other Embodiments: In the image display means 70 for displaying a radiographic image or a processed image subjected to image processing, the radiographic image or the processed image is displayed together with information on the region of interest or the image. It is possible to display information on processing condition factors or information on the image processing conditions.

As a result of such display, a region of interest, an image processing condition factor, and an image processing condition that have an important effect on image processing are displayed, so that an image subjected to accurate and optimal image processing can be obtained. This can contribute to providing information useful for diagnosis to the radiologist.

Further, by combining the image display means 70 and the correction information input means 50 as described above, the operator can interactively correct the image processing conditions while observing the processed image. Specifically, CR
It has image display means such as a T monitor or a liquid crystal monitor, and correction information input means using a pointing device such as a touch screen, a track pad, or a mouse, and displays a region of interest, an image processing condition factor, or final image processing conditions. When the operator corrects the image, the processed image based on the corrected image processing condition is displayed on the image display means.

[0111]

As described above, by analyzing the image data of the radiation image formed corresponding to the amount of radiation transmitted through the subject, the first region of interest corresponding to the subject structure, which is the main diagnostic object, is determined. Determining a plurality of regions of interest including a second region of interest corresponding to a subject structure to be diagnosed, and performing image processing based on image data included in the region of interest for each of the determined regions of interest. A specific processing example in which a condition factor is generated, an image processing condition is determined based on the generated plurality of image processing condition factors, and image processing is performed on the radiation image using the determined image processing condition. Is shown in FIG.

Here, when the histograms have the same shape as in FIG. 12A, the signal distribution of each human body structure is different, and the case of FIG. 12B and the case of FIG. Suppose.

In such a case, an image processing condition factor is generated from a plurality of regions of interest, and an image processing condition is determined based on the image processing condition factor.
2 (d) and (e), and in the case of frequency emphasis processing, as shown in FIGS. 12 (f) and (g), and in the case of dynamic range compression processing, FIGS. 12 (h) and (i). )become that way.

Here, FIGS. 12 (e), (g), (h)
Indicate the image processing condition based on the image processing condition factor of the patient B, and the characteristic of the dotted line is the conventional image processing condition.

Here, as an example of the gradation processing, the minimum output signal value of the bone region and the maximum output signal value of the soft region are respectively set to predetermined values, and the average contrast of the bone region and the average contrast of the soft region are set. Are shown as examples of gradation conversion curves, each of which has a predetermined value.

As an example of the frequency emphasis processing, the emphasis coefficient corresponding to the bone region and the emphasis coefficient corresponding to the soft region are each set to a predetermined value, and a monotonically increasing line is connected between them by a straight line having a predetermined inclination. Here is an example of an enhancement function.

As an example of the dynamic range compression processing, an example of a monotonically decreasing correction function in which the correction amount corresponding to the bone region is set to 0, and the correction coefficient corresponding to the soft region is a predetermined positive value.

As described above, it is possible to perform different processing for each of the patients, and it is possible to process the bones and the soft parts to appropriate gradation, density, and dynamic range for each patient. That is, it is possible to automatically obtain the optimal image processing conditions regardless of the ratio of the bone part and the soft part included in the subject, and to automatically obtain the optimal image for diagnosis without complicated operation.

[0119]

As described above in detail, according to the present invention, the optimum image processing conditions are automatically determined regardless of the ratio of the bones and the soft parts included in the subject, and no complicated operation is required. It has become possible to automatically obtain the optimal image for diagnosis.

[Brief description of the drawings]

FIG. 1 is a functional block diagram illustrating a configuration of a radiation image processing apparatus according to an embodiment of the present invention.

FIG. 2 is a functional block diagram illustrating a configuration of a main part of the radiation image processing apparatus according to the exemplary embodiment of the present invention.

FIG. 3 is an explanatory diagram showing how a region of interest is determined in the embodiment of the present invention.

FIG. 4 is an explanatory diagram showing how a region of interest is determined in the embodiment of the present invention.

FIG. 5 is an explanatory diagram showing how a region of interest is determined in the embodiment of the present invention.

FIG. 6 is an explanatory diagram showing how a region of interest is determined in the embodiment of the present invention.

FIG. 7 is an explanatory diagram showing image processing condition factors and image processing conditions in the embodiment of the present invention.

FIG. 8 is an explanatory diagram showing image processing condition factors and image processing conditions in the embodiment of the present invention.

FIG. 9 is an explanatory diagram showing an enhancement coefficient and image data according to the embodiment of the present invention.

FIG. 10 is an explanatory diagram showing an enhancement coefficient and image data according to the embodiment of the present invention.

FIG. 11 is an explanatory diagram showing dynamic range compression in the embodiment of the present invention.

FIG. 12 is an explanatory diagram showing a specific example of an image processing condition in the embodiment of the present invention.

FIG. 13 is a characteristic diagram illustrating an example of a histogram.

FIG. 14 is a characteristic diagram illustrating an example of a histogram.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 Radiation image input means 20 Region of interest determination means 30 Image processing condition factor generation means 40 Image processing condition determination means 50 Correction information input means 60 Image processing means 70 Image display means

Claims (20)

[Claims] 1. A first region of interest corresponding to a subject structure, which is a main diagnosis target, by analyzing image data of a radiation image formed corresponding to a radiation dose transmitted through a subject, and a secondary diagnosis target And determining a plurality of regions of interest including a second region of interest corresponding to the subject structure, and generating an image processing condition factor for each of the determined regions of interest based on the image data included in the region of interest. An image processing method comprising: determining image processing conditions based on the plurality of generated image processing condition factors; and performing image processing on the radiation image using the determined image processing conditions. 2. The method according to claim 1, wherein the determination of the region of interest includes setting a plurality of small regions in the image, determining a threshold value for each small region based on image data in the small region, By comparing the signal value of the pixel with the threshold value, it is determined whether or not the pixel is within the target region of interest, and the region of interest is determined by integrating the determined results. The radiation image processing method according to claim 1, wherein: 3. The determination of the region of interest includes: scanning an image using a plurality of scanning lines; calculating a characteristic value due to a signal change between neighboring pixels for an arbitrary pixel on each scanning line; Using the value, determine whether the pixel on the scanning line is a point on the contour of the target region of interest, determine the contour of the region of interest by integrating the determined results, The radiation image processing method according to claim 1, wherein an area surrounded by an outline is set as the region of interest. 4. A method for determining a region of interest, comprising: setting a plurality of small regions in an image; and generating a feature vector having, as a vector element, a feature extracted for each small region based on image data in the small region. A plurality of clusters are generated by statistically classifying the generated set of feature vectors; and for the plurality of small regions, a feature vector of each small region belongs to any of the plurality of clusters. The radiation image processing method according to claim 1, wherein a region of interest is determined based on the distance. 5. The method of claim 1, further comprising: extracting a representative signal range for each region of interest based on image data included in the region of interest; and generating a signal range based on image processing parameters given in advance. The radiation image processing method according to claim 1, wherein an image processing condition factor corresponding to the representative signal range is generated. 6. The radiation image processing method according to claim 1, wherein said image processing is gradation processing. 7. The image processing is frequency enhancement processing.
The radiation image processing method according to claim 1, wherein: 8. The radiation image processing method according to claim 1, wherein the image processing is a dynamic range compression processing. 9. Displaying the radiation image or the processed image on which the image processing has been performed, and displaying the radiation image or the processed image together with the information on the region of interest or the information on the image processing condition factor, or the image 9. The radiation image processing method according to claim 1, wherein information on a processing condition is displayed. 10. Information about correction of image processing is input, wherein the determination of the image processing condition determines an image processing condition corrected according to the input information, and using the corrected image processing condition. 10. The radiation image processing method according to claim 1, wherein image processing is performed on the radiation image. 11. A first region of interest corresponding to a subject structure, which is a main diagnosis target, by analyzing image data of a radiation image formed corresponding to a radiation dose transmitted through a subject, and a secondary diagnosis target Region of interest determining means for determining a plurality of regions of interest including a second region of interest corresponding to a subject structure, and image data included in the region of interest for each region of interest determined by the region of interest determination means Image processing condition factor generating means for generating an image processing condition factor based on the image processing condition factor determining means for determining an image processing condition based on a plurality of image processing condition factors generated by the image processing condition factor generating means; Image processing means for performing image processing on the radiographic image using the image processing conditions determined by the image processing condition determining means. Ray image processing apparatus. 12. A threshold determining means for setting a plurality of small areas in an image, and for each small area, determining a threshold based on image data in the small area. Determining means for comparing the signal value of the pixel in the small area with the threshold value to determine whether the pixel is in the target area of interest. The radiation image processing apparatus according to claim 11, wherein a region of interest is determined by integrating the results obtained. 13. A characteristic value calculating unit configured to scan an image using a plurality of scanning lines and calculate a characteristic value of an arbitrary pixel on each scanning line due to a signal change between neighboring pixels, Using the characteristic value calculated by the characteristic value calculation means,
A determination unit that determines whether a pixel on the scanning line is a point on the contour of the target region of interest, and determining the contour of the region of interest by integrating the results determined by the determination unit. The radiation image processing apparatus according to claim 11, wherein the determined area is defined as a region surrounded by the contour line. 14. The region-of-interest determining means sets a plurality of small regions in an image and generates a feature vector having, as a vector element, a feature extracted for each small region based on image data in the small region. Feature vector generation means, and cluster generation means for generating a plurality of clusters by statistically classifying a set of feature vectors generated by the feature vector generation means, and for the plurality of small areas, The radiation image processing apparatus according to claim 11, wherein a region of interest is determined based on which of the plurality of clusters the feature vector of each small region belongs to. 15. The image processing condition factor generating means includes a representative signal range extracting means for extracting a representative signal range for each region of interest based on image data included in the region of interest. The radiation image processing apparatus according to claim 11, wherein an image processing condition factor corresponding to the representative signal range is generated based on the image processing parameters. 16. The radiation image processing apparatus according to claim 11, wherein the image processing is a gradation processing. 17. The radiation image processing apparatus according to claim 11, wherein the image processing is frequency enhancement processing. 18. The radiation image processing apparatus according to claim 11, wherein the image processing is a dynamic range compression processing. 19. An image display means for displaying the radiation image or the processed image subjected to the image processing, wherein the image display means displays information on the region of interest or the image together with the radiation image or the processed image. 19. The radiation image processing apparatus according to claim 11, wherein information related to a processing condition factor or information related to the image processing condition is displayed. 20. A correction information input unit for inputting information relating to correction of image processing, wherein the image processing condition determination unit determines an image processing condition corrected according to the information input by the correction information input unit. 20. The radiographic image processing apparatus according to claim 11, wherein the image processing unit performs image processing on the radiographic image using the corrected image processing conditions.