WO2012039138A1

WO2012039138A1 - Candidate abnormal shadow detection device, detection method, and program, and breast diagnostic imaging system

Info

Publication number: WO2012039138A1
Application number: PCT/JP2011/005326
Authority: WO
Inventors: 俊孝阿賀野; 孝夫桑原; 靖子八尋; 大田　恭義; 玲長谷川
Original assignee: 富士フイルム株式会社
Priority date: 2010-09-23
Filing date: 2011-09-21
Publication date: 2012-03-29

Abstract

[Problem] To alleviate errors in detection when detecting candidates for abnormal shadows in radiographic breast images. [Solution] For two radiographic images for displaying a stereoscopic image using binocular disparity, wherein a breast that is pressed by pressure plates is photographed by respective irradiation with radiation emitted from two foci in two different photographic directions, a detection process which detects candidates for abnormal shadows is carried out on each respective radiographic image in each of the photographic directions. For a first candidate abnormal shadow which is detected in one radiographic image, a corresponding region in the other radiographic image that corresponds to the location of the first candidate abnormal shadow in the one radiographic image is set on the basis of the photography conditions which include pressing thickness of the breast that is pressed by the pressure plates. If no candidate abnormal shadow is present in the corresponding region, the first candidate abnormal shadow is determined to be a false positive.

Description

Abnormal shadow candidate detection apparatus, detection method and program, and breast image diagnosis system

The present invention relates to an abnormal shadow candidate detection apparatus, detection method, and program, and in particular, based on a radiation image of a breast for each imaging direction acquired by imaging breasts from different imaging directions for stereoscopic display. The present invention relates to a method, apparatus, and program for detecting abnormal shadow candidates.

Conventionally, it is known that stereoscopic viewing can be performed using parallax by displaying a plurality of images in combination. Such a stereoscopically viewable image (hereinafter referred to as a stereoscopic image or a stereo image) is generated based on a plurality of images having parallax obtained by photographing the same subject from different directions.

Such generation of stereoscopic images is used not only in the fields of digital cameras and televisions but also in the field of radiographic imaging. That is, the subject is irradiated with radiation from different directions, the radiation transmitted through the subject is detected by the radiation image detector, and a plurality of radiation images having parallax are obtained, and based on these radiation images A stereoscopic image is generated. And by generating a stereoscopic image in this way, a radiographic image with a sense of depth can be observed, and a radiographic image more suitable for diagnosis can be observed.

On the other hand, in the medical imaging field, CAD (computer-aided diagnosis: Computer-aided diagnosis: Computer-aided diagnosis: Computer-aided diagnosis) that automatically detects a candidate for an abnormal shadow that is suspected to be a lesion such as a mass shadow or a microcalcification shadow,
Aided Diagnosis) is being researched.

As an abnormal shadow candidate detection process, an iris filter process suitable for detecting a mass shadow candidate and a morphological calculation process suitable for detecting a microcalcification shadow candidate have been proposed.

However, abnormal shadow candidates are not always reliably detected by the above-described various abnormal shadow candidate detection processes, and detection omission may occur. On the other hand, in the abnormal shadow detection process, there may be a case where a pseudo-shadow that is not an abnormal shadow but has a feature value similar to that of the abnormal shadow is erroneously detected as an abnormal shadow candidate.

In Patent Literature 1, two radiographic images, that is, a lateral image of a breast and a vertical image, are used to detect abnormal shadow candidates, and the positions of abnormal shadow candidates detected from one image are detected. A method and apparatus for improving the detection accuracy of abnormal shadow candidates in a breast radiographic image by mutually setting regions in the other image corresponding to and performing the abnormal shadow candidate detection process on the set region again has been proposed. Yes.

However, the method for detecting an abnormal shadow candidate in Patent Document 1 does not use a stereoscopic image, and a method for improving detection accuracy when using a stereoscopic image has not been proposed.

On the other hand, in Patent Document 2, lesion candidates (abnormal shadow candidates) are detected for each of the left and right images for stereo (stereoscopic) display, and the positions of the detected lesion candidates are associated with each other in the images. There has been proposed a method of reducing false detection by determining that the detection is false detection when the distance between the matched candidates exceeds a predetermined threshold, with close ones as matching candidates.

JP 10-108859 A JP 2004-337200 A

However, Patent Document 2 mainly targets a radiological image of a lung field, and no specific disclosure is made about a method for reducing false detection of a breast.

In view of the above circumstances, the present invention is a device for detecting an abnormal shadow candidate that can more effectively suppress erroneous detection when detecting an abnormal shadow candidate from two radiographic images for stereoscopic vision of the breast. Another object of the present invention is to provide a breast image diagnosis system including a method and a program, and an abnormal shadow candidate detection apparatus.

The abnormal shadow candidate detection apparatus of the present invention is an abnormal shadow candidate detection apparatus for detecting an abnormal shadow candidate from a radiation image of a breast.
Two radiographic images for displaying a stereoscopic image using binocular parallax, and two different radiographs of radiation emitted from two focal positions on a breast compressed by a compression plate Abnormal shadow candidate detection means for performing detection processing for detecting abnormal shadow candidates for each of the radiographic images for each of the imaging directions captured by irradiating each from the direction;
The breast in which the corresponding region in the other radiographic image corresponding to the position where the one image abnormal shadow candidate detected from one radiographic image of the two radiographic images is present is compressed by the compression plate Corresponding region setting means for setting based on the imaging conditions including the compression thickness of the image, and determination means for determining that the one-side image abnormal shadow candidate is false positive when no abnormal shadow candidate is detected in the corresponding region. It is characterized by having.

The imaging conditions include the relationship between the two focal positions, a detector that detects the radiation irradiated and transmitted to the breast, the imaging position of the breast, and the distance between the two focal positions. it can.

It is desirable that the abnormal shadow candidate detection unit performs the detection process with higher detection capability than the detection time of the one-side image abnormal shadow candidate for the corresponding region set by the corresponding region setting unit.

When the abnormal shadow candidate is detected in the corresponding region, the determination unit compares the feature amount between the abnormal shadow candidate and the one image abnormal shadow candidate, and based on the comparison result, the one image It is desirable to determine whether an abnormal shadow candidate is a false positive.

The first method for detecting an abnormal shadow candidate according to the present invention is a method for detecting an abnormal shadow candidate for detecting an abnormal shadow candidate from a radiation image of a breast.
Two radiographic images for displaying a stereoscopic image using binocular parallax, and two different radiographs of radiation emitted from two focal positions on a breast compressed by a compression plate For each of the radiographic images taken for each imaging direction taken by irradiating from each direction, a detection process for detecting a candidate for an abnormal shadow is performed,
The corresponding compression region of the other radiographic image corresponding to the position of the one image abnormal shadow candidate detected from one radiographic image of the two radiographic images in the radiographic image is compressed by the compression plate of the breast. Set based on shooting conditions including
When the abnormal shadow candidate does not exist in the corresponding region, the one-side image abnormal shadow candidate is determined to be false positive.

In the first detection method of an abnormal shadow candidate according to the present invention, after setting the corresponding region, the corresponding region is subjected to a re-detection process with a detection capability higher than the detection capability in the detection process,
The determination may be performed after the re-detection process.

The second abnormal shadow candidate detection method of the present invention is an abnormal shadow candidate detection method for detecting an abnormal shadow candidate from a radiation image of a breast, and is a method for displaying a stereoscopic image using binocular parallax. One of the radiation images in each radiographing direction taken by irradiating the breast pressed by the radiographic image compression plate with radiation emitted from two focal positions from two different radiographing directions. A detection process for detecting an abnormal shadow candidate is performed on the image, and a corresponding region in the other radiographic image corresponding to the position of the one image abnormal shadow candidate detected from the one radiographic image is detected. , Set based on the imaging conditions including the compression thickness of the breast pressed by the compression plate,
For the corresponding region of the other radiographic image, a detection process for detecting a candidate for an abnormal shadow is performed,
When the abnormal shadow candidate does not exist in the corresponding region, the one-side image abnormal shadow candidate is determined to be false positive.

In the second method for detecting an abnormal shadow candidate according to the present invention, the detection processing for the corresponding region of the other radiographic image is performed with a higher detection capability than when detecting the one image abnormal shadow candidate. be able to.

In the first and second detection methods for abnormal shadow candidates according to the present invention, when there is an abnormal shadow candidate in the corresponding region, the feature amounts of the abnormal shadow candidate and the one-side image abnormal shadow candidate are compared. Based on the comparison result, it may be determined whether or not the one-side image abnormal shadow candidate is erroneously detected.

A first program of the present invention is a program for causing a computer to execute an abnormal shadow candidate detection method for detecting an abnormal shadow candidate from a radiation image of a breast,
Two radiographic images for displaying a stereoscopic image using binocular parallax, and two different radiographs of radiation emitted from two focal positions on a breast compressed by a compression plate An abnormal shadow candidate detection process for detecting an abnormal shadow candidate for each of the radiographic images for each of the imaging directions captured by irradiating from each direction,
The breast in which the corresponding region in the other radiographic image corresponding to the position where the one image abnormal shadow candidate detected from one radiographic image of the two radiographic images is present is compressed by the compression plate A corresponding area setting process that is set based on an imaging condition including the compression thickness of the image, and a determination process that determines that the one-side image abnormal shadow candidate is a false positive when there is no abnormal shadow candidate in the corresponding area. It is characterized by executing the detection method including.

A second program of the present invention is a program for causing a computer to execute an abnormal shadow candidate detection method for detecting an abnormal shadow candidate from a radiation image of a breast, and displaying a stereoscopic image using binocular parallax. The two radiographs for performing the above-mentioned radiographing are performed by irradiating the breast pressed by the compression plate with the radiation emitted from the two focal positions from two different imaging directions, respectively. A first abnormal shadow candidate detection process for detecting an abnormal shadow candidate for one of the radiographic images in each direction, the one image abnormal shadow candidate detected from the one radiographic image in the radiographic image The corresponding region in the other radiographic image corresponding to the existing position is set based on the imaging conditions including the compression thickness of the breast compressed by the compression plate. Corresponding region setting processing,
A second abnormal shadow candidate detection process for detecting an abnormal shadow candidate for the corresponding region of the other radiographic image, and when the abnormal shadow candidate does not exist in the corresponding region, the one image abnormal shadow candidate A detection method including a determination process for determining that is a false positive is executed.

In the first and second programs of the present invention, when there is an abnormal shadow candidate in the corresponding region, the feature amounts of the abnormal shadow candidate and the one-side image abnormal shadow candidate are compared, and the result of the comparison It is preferable that the detection method including a determination process for determining whether or not the one-image abnormal shadow candidate is false positive based on the one-way image is executed.

The breast diagnostic imaging system of the present invention irradiates a breast compressed by a compression plate with radiation emitted from two focal positions from two different imaging directions, thereby obtaining a radiographic image for each imaging direction. An image capturing device for capturing, a display device for displaying a stereoscopic image using the two captured radiographic images, and an abnormal shadow candidate detecting device for detecting an abnormal shadow candidate from the two radiographic images of the present invention It is characterized by comprising.

According to the abnormal shadow candidate detection apparatus and method of the present invention, the other radiographic image corresponding to the position of the one image abnormal shadow candidate detected from one radiographic image of the two radiographic images in the radiographic image. Is set based on imaging conditions including the compression thickness of the breast by the compression plate, and when there is no abnormal shadow candidate in the corresponding area, the first abnormal shadow candidate is false positive Since it is determined that there is an error, erroneous detection of an abnormal shadow candidate can be effectively suppressed.

The block diagram which shows schematic structure of 1st Embodiment of the abnormal shadow candidate detection apparatus of this invention. Schematic diagram schematically showing the arrangement of the radiation source, subject and detector during image capture Right-eye image and left-eye image acquired by a mammography apparatus Conceptual diagram of stereoscopic display with two radiographic images The figure which shows the two radiographic images which are the detection targets of an abnormal shadow candidate The flowchart for demonstrating embodiment of the 1st abnormal shadow candidate detection method in the detection apparatus of 1st Embodiment. The flowchart for demonstrating embodiment of the 2nd abnormal shadow candidate detection method in the detection apparatus of 1st Embodiment. The flowchart for demonstrating embodiment of the 3rd abnormal shadow candidate detection method in the detection apparatus of 1st Embodiment. The block diagram which shows schematic structure of 2nd Embodiment of the abnormal shadow candidate detection apparatus of this invention. The flowchart for demonstrating embodiment of the abnormal shadow candidate detection method in the detection apparatus of 2nd Embodiment. 1 is a schematic configuration diagram of an embodiment of a breast image diagnosis system of the present invention. The figure which looked at the arm part of the imaging device of the breast diagnostic imaging system shown in FIG. 11 from the right direction of FIG. The block diagram which shows schematic structure inside the computer of the breast image diagnosis system shown in FIG.

The abnormal shadow candidate detection apparatus and method of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a schematic configuration of an abnormal shadow candidate detection apparatus 50 according to the first embodiment. Note that a program for causing the computer to execute the abnormal shadow candidate detection method is installed and executed, whereby the computer functions as an abnormal shadow candidate detection device. This program is stored in an information storage medium such as a CD-ROM or distributed and installed via a network such as the Internet.

The abnormal shadow imaging candidate detection device 50 according to the present embodiment includes two radiographic images (right eye image SR and left eye image SL) captured for stereoscopic display of the breast, and breast compression at the time of the imaging. Information I relating to the imaging condition including the thickness is input, the abnormal shadow candidate Cn (n = 1, 2,...) Is detected from the radiographic image, and the detected abnormal shadow candidate Cn is output.

The abnormal shadow candidate detection device 50 according to the present embodiment is configured to detect abnormal shadow candidates CRn (n = 1, 2,...), CLn (n = 1, 2,...) For each of the two input radiation images SR and SL. ) And the other radiation corresponding to the existence position of the abnormal shadow candidate CRn detected in one of the radiation images (right-eye image SR in the present embodiment). Corresponding region setting unit 54 that sets the corresponding region An in the image (the image SL for the left eye in the present embodiment) based on the imaging condition including the compression thickness, and whether or not the abnormal shadow candidate CRn is false positive And a determination unit 56.

The information I regarding the image signal and the imaging conditions input to the abnormal shadow candidate detection device 50 will be described. FIG. 2 is a schematic diagram illustrating a configuration of a main part of a mammography apparatus that generates two radiographic images (right-eye image SR and left-eye image SL) for stereoscopic vision.

In FIG. 2, the radiation source 17, which is the main part of the mammography apparatus, the imaging table 14 in which the radiation detector 15 is arranged, and the breast arranged on the imaging table 14 are pressed in the direction of the imaging table 14. A compression plate 18 is shown.

The right-eye image SR and the left-eye image SL are respectively captured by irradiating radiation emitted from each of the two focal positions f1 and f2 shown in FIG. 2 from two different imaging directions. FIG. 3 schematically shows the acquired right-eye image SR and left-eye image SL. The right-eye image SR is photographed by radiation from the focal position f1 in FIG. 2, and the left-eye image SL is photographed by radiation from the focal position f2.

It should be noted that the distance D between the two focal positions f1 and f2 at the time of imaging, the distance SID between the focal positions f1 and f2 and the radiation detection 15, and the distance t0 from the surface 14a of the imaging table 14 to the detection surface 15a of the radiation detection note 15. The compression thickness t of the breast M held between the imaging table 14 and the compression plate 18 is input to the abnormal shadow candidate detection device 50 in association with the radiographic images SR and SL captured as the imaging condition I. The

The abnormal shadow candidate detection means 52 performs detection processing for detecting an abnormal shadow candidate on a medical radiation image, and this detection processing can be performed using various known algorithms. Examples of detection algorithms for detecting abnormal shadow candidates include iris filter processing suitable for detecting mass shadow candidates and morphological calculation processing suitable for detecting microcalcified shadow candidates. These may be used alone or in combination.

In addition, it is preferable that the abnormal shadow candidate detection means 52 can detect an abnormal shadow candidate by changing the detection capability. The detection capability can be improved by lowering the threshold value for determining whether or not it is an abnormal shadow candidate in the detection algorithm.

The corresponding area setting unit 54 compresses the corresponding area An in the left-eye image SL corresponding to the position in the right-eye image SR of the abnormal shadow candidate CRn detected from the right-eye image SR by the compression plate 18. Are set based on the above-described imaging conditions including the compression thickness of the breast M, and when a plurality of abnormal shadow candidates CR1, CR2,... Are detected in the right-eye image SR, each candidate CR1, CR2 is detected. ... Corresponding areas A1, A2,... In the left-eye image SL are set for each.

A corresponding region setting method will be described with reference to FIGS. 4 and 5. FIG. 4 is a diagram schematically showing a stereoscopic display of the breast M by the right-eye image SR and the left-eye image SL, and FIG. 5 is a diagram for the right eye for the stereoscopic display shown in FIG. It is a figure which shows image SR and image SL for left eyes.

As shown in FIG. 4, a corresponding area A1 is set in the left-eye image SL for the abnormal shadow candidate CR1 among the abnormal shadow candidates CRn (CR1 and CR2 in the present embodiment) detected in the right-eye image SR. How to do will be described.

The abnormal shadow candidate CR1 in the right-eye image SR shown in FIGS. 4 and 5 has a length B1 in which a straight line connecting the existing position on the right-eye image SR in FIG. 4 and the right eye overlaps the breast M as the subject. It is predicted that a lesion corresponding to the abnormal shadow candidate CR1 exists on the line segment.

On the other hand, this line segment corresponds to a region A1 having a width ΔX in the X direction on the left-eye image SL in FIGS. 4 and 5, and this region A1 is set as a corresponding region of the abnormal shadow candidate CR1.

The parallax Δd1 on the pressing surface 18a of the compression plate 18 of the breast M and the parallax Δd2 on the surface 14a of the imaging table 14 of the breast M are
Δd1 = D * t2 / t1
Δd2 = D * t0 / (t1 + t)
The existence range of the coordinate XL1 in the x direction in the left eye image of the abnormal shadow candidate CR1 is represented by the parallax Δd1 and Δd2 and the coordinate XR1 in the x direction of the abnormal shadow candidate CR1.
Δd1 + XR1 ≧ XL1 ≧ Δd2 + XR1
It is expressed.

Therefore, in the corresponding area setting unit 54, the coordinates XR1 in the X direction of the abnormal shadow candidate CR1 are substantially obtained, and the distance SID between the focal positions f1 and f2 and the radiation detection 15 included in the information I regarding the imaging conditions, the imaging Parallax Δd1 and Δd2 are obtained from the distance t0 between the surface 14a of the table 14 and the detection surface 15a of the radiation detection note 15 and the compression thickness t of the breast M held between the imaging table 14 and the compression plate 18, A process is performed for obtaining a range that the x-direction coordinate XL1 can take in the left eye image of the abnormal shadow candidate CR1.

The determination unit 56 determines that the first abnormal shadow candidate CRn is false positive (FP) when the abnormal shadow candidate CLn is not detected in the corresponding area An for the first abnormal shadow candidate CRn. . On the other hand, when the abnormal shadow candidate CLn is detected in the corresponding area An, it may be determined to be true positive (TP).

Further, in order to further improve the determination accuracy, the determination unit 56 detects the abnormal shadow candidate CRn in the right eye image and the abnormal shadow candidate CLn in the corresponding region An when the abnormal shadow candidate CLn is detected in the corresponding region An. May be configured to determine whether the abnormal shadow candidate CRn is true or false based on the comparison result.

Examples of the feature amount include shape and / or density, or the type of algorithm that detects a candidate for comparison. For example, when the abnormal shadow candidate CRn is detected by the iris filter process and the other abnormal shadow candidate CLn is detected by the morphological process, the types of algorithms are different, so the abnormal shadow candidates CRn and CLn are different from each other. If they do not match and are determined to be false positives (FP) and both abnormal shadow candidates are detected by iris filter processing, the algorithm type is the same, and abnormal shadow candidates CRn and CLn match. What is necessary is just to determine that it is true positive (TP).

Alternatively, the shapes may be compared as feature amounts. For example, a blood vessel or a mammary gland has a predetermined length, but an image taken from a certain direction may be taken in a circular shape and cannot be distinguished from an abnormal shadow such as a tumor.

For example, in FIG. 5, a circular abnormal shadow candidate CR2 detected in the right-eye image is detected as an elongated abnormal shadow candidate CL2 in the corresponding region A2 of the left-eye image. In such a case, it is possible to determine that the abnormal shadow candidate CR2 is false positive by comparing the shapes as the feature amounts.

A first method for detecting an abnormal shadow candidate in the abnormal shadow candidate detection apparatus 50 of the present embodiment will be described with reference to the flowchart of FIG.

First, the abnormal shadow candidate detection unit 52 performs a detection process for detecting abnormal shadow candidates CRn and CLn for the two radiographic images SR and SL (S10).

Next, the corresponding area setting unit 54 sets the corresponding area An in the other image SL at the position where the abnormal shadow candidate CRn exists in one image SR (S12).

Thereafter, when the abnormal shadow candidate CLn corresponding to the abnormal shadow candidate CRn does not exist (is not detected) in the corresponding area An (S14; No), in the determination unit 56, the abnormal shadow candidate CRn is false positive (FP). (S16) and the process is terminated. On the other hand, when the abnormal shadow candidate CLn corresponding to the abnormal shadow candidate CRn is present (detected) in the corresponding region An (S14; Yes), the abnormal shadow candidate CRn is true positive (TP) in the determination unit 56. (S18), and the process ends.

With the above process, the true / false of the abnormal shadow candidate detected in one image is determined based on whether or not the abnormal shadow candidate is detected in the corresponding region in the other image. It becomes possible to suppress. In addition, since the corresponding region in the other image of the abnormal shadow candidate in one image is set in consideration of the compression thickness of the breast at the time of image capture, the matching accuracy is good and the true of the abnormal shadow candidate is true. The accuracy of false determination can be improved.

A second method for detecting an abnormal shadow candidate in the abnormal shadow candidate detection apparatus 50 of the present embodiment will be described with reference to the flowchart of FIG.

Thereafter, when the abnormal shadow candidate CLn corresponding to the abnormal shadow candidate CRn does not exist (is not detected) in the corresponding area An (S14; No), in the determination unit 56, the abnormal shadow candidate CRn is false positive (FP). (S16) and the process is terminated. On the other hand, when the abnormal shadow candidate CLn corresponding to the abnormal shadow candidate CRn exists (is detected) in the corresponding region An (S14; Yes), the determination unit 56 uses the feature amounts of both abnormal shadow candidates CRn and CLn. If the degree of coincidence of this feature amount is high (S17; Yes), it is determined that the abnormal shadow candidate CRn is true positive (TP) (S18), and the degree of coincidence of this feature is low (S17: No) ), The abnormal shadow candidate CRn is determined to be false positive (FP) (S16), and the process is terminated.

According to this processing method, erroneous detection can be suppressed more than in the first detection method, and the detection accuracy of abnormal shadow candidates can be improved.

A third method for detecting an abnormal shadow candidate in the abnormal shadow candidate detection apparatus 50 of the present embodiment will be described with reference to the flowchart of FIG.

If there is no abnormal shadow candidate CLn corresponding to the abnormal shadow candidate CRn in the corresponding area An (not detected) (S14; No), if it is the first time (S15; Yes), the abnormal shadow candidate detection unit 52 Thus, the corresponding region An is subjected to redetection processing with higher detectability than the detection processing performed previously (S19), and the abnormal shadow candidate CLn corresponding to the abnormal shadow candidate CRn in the corresponding region An also by this re-detection. Is not detected (S15: No), the determination unit 56 determines that the abnormal shadow candidate CRn is false positive (S16), and ends the process.

On the other hand, the processing in the case where the abnormal shadow candidate CLn corresponding to the abnormal shadow candidate CRn exists in the corresponding region An is the same as in the case of the second method described above.

Here, the redetection of the abnormal shadow candidate for the corresponding region An is performed once, but may be performed a plurality of times with gradually increasing detection ability.

According to the present processing method, it is possible to perform the true / false of the abnormal shadow candidate with higher accuracy than the first detection method and the second detection method.

FIG. 9 is a block diagram showing a schematic configuration of an abnormal shadow candidate detection device 50 ′ according to the second embodiment. The detection device 50 ′ includes the

same units

52, 54, and 56 as the detection device 50 according to the first embodiment, but the processing flow is partially different.

A method of detecting an abnormal shadow candidate in the second embodiment will be described with reference to the flowchart of FIG.

First, the abnormal shadow candidate detection unit 52 performs a detection process for detecting an abnormal shadow candidate CRn for one of the two radiographic images SR and SL (here, the right-eye image SR) (S30). .

Next, the corresponding area setting unit 54 sets the corresponding area An in the other image SL at the position where the abnormal shadow candidate CRn exists in one image SR (S32).

The abnormal shadow candidate CLn is detected by the abnormal shadow candidate detection unit 52 for the corresponding area An in the other image SL (S33).

At this time, if the detection ability of the abnormal shadow candidate detection unit 52 is increased as compared with the detection of the abnormal shadow candidate CRn to facilitate detection of the abnormal shadow candidate, the generation of false negative candidates (FN) can be suppressed. .

Thereafter, when the abnormal shadow candidate CLn corresponding to the abnormal shadow candidate CRn does not exist (is not detected) in the corresponding area An (S34; No), the abnormal shadow candidate CRn is false positive (FP) in the determination unit 56. (S36), and the process ends. On the other hand, when the abnormal shadow candidate CLn corresponding to the abnormal shadow candidate CRn exists (is detected) in the corresponding region An (S34; Yes), the determination unit 56 uses the feature amounts of both abnormal shadow candidates CRn and CLn. If the degree of coincidence of this feature amount is high (S37; Yes), it is determined that the abnormal shadow candidate CRn is true positive (TP) (S38), and the degree of coincidence of this feature is low (S37: No), it is determined that the abnormal shadow candidate CRn is a false positive (FP) (S36), and the process ends.

With the above processing, the accuracy of the true / false determination of the abnormal shadow candidate can be improved as in the case of using the abnormal shadow candidate detection apparatus of the first embodiment.

Next, the breast image diagnosis system of the present invention will be described with reference to FIGS.

FIG. 11 is a diagram showing a schematic configuration of the entire breast image diagnostic system of the present embodiment.

As shown in FIG. 11, the breast diagnostic imaging system 1 of the present embodiment includes a breast imaging device 10, a computer 2 connected to the breast imaging device 10, a monitor 3 and an input unit 4 connected to the computer 2. And. The computer 2 constitutes the abnormal shadow candidate detection apparatus 50 of the present invention (see FIG. 12).

As shown in FIG. 11, the mammography apparatus 10 includes a base 11, a rotary shaft 12 that can move in the vertical direction (Z direction) with respect to the base 11, and can be rotated. The arm part 13 connected with the base 11 is provided. FIG. 12 shows the arm 13 viewed from the right direction in FIG.

The arm portion 13 has an alphabet C shape, and a radiation table 16 is attached to one end of the arm portion 13 so as to face the imaging table 14 at the other end. The rotation and vertical movement of the arm unit 13 are controlled by an arm controller 31 incorporated in the base 11.

A radiographic image detector 15 such as a flat panel detector and a detector controller 33 that controls reading of a charge signal from the radiographic image detector 15 are provided inside the imaging table 14. Further, inside the imaging table 14, a charge amplifier that converts the charge signal read from the radiation image detector 15 into a voltage signal, a correlated double sampling circuit that samples the voltage signal output from the charge amplifier, A circuit board provided with an AD conversion unit for converting a voltage signal into a digital signal is also installed.

In addition, the photographing table 14 is configured to be rotatable with respect to the arm unit 13, and even when the arm unit 13 rotates with respect to the base 11, the direction of the photographing table 14 is fixed to the base 11. can do.

The radiation image detector 15 can repeatedly perform recording and reading of a radiation image, and may use a so-called direct type radiation image detector that directly receives radiation and generates charges. Alternatively, a so-called indirect radiation image detector that converts radiation once into visible light and converts the visible light into a charge signal may be used. As a radiation image signal readout method, TFT (thin
Film transistor) A so-called TFT readout method in which a radiation image signal is read out by turning on and off the switch, and a so-called optical readout method in which a radiation image signal is read out by irradiating a reading light. Although it is desirable to use it, the present invention is not limited to this, and other materials may be used.

A radiation source 17 and a radiation source controller 32 are housed in the radiation irradiation unit 16. The radiation source controller 32 controls the timing of irradiating radiation from the radiation source 17 and the radiation generation conditions (tube current, time, tube current time product, etc.) in the radiation source 17.

Further, in the central portion of the arm portion 13, a compression plate 18 disposed above the imaging table 14 to press and compress the breast, a support portion 20 that supports the compression plate 18, and a support portion 20 in the vertical direction ( A moving mechanism 19 for moving in the Z direction) is provided. The position of the compression plate 18 and the compression pressure are controlled by the compression plate controller 34.

The computer 2 includes a central processing unit (CPU) and a storage device such as a semiconductor memory, a hard disk, and an SSD, and the control unit 8a, the radiation image storage unit 8b, and the like shown in FIG. An abnormal shadow candidate detection device 50 is configured.

The controller 8a outputs predetermined control signals to the various controllers 31 to 35 to control the entire system. A specific control method will be described in detail later.

The radiographic image storage unit 8b stores in advance an image signal of a radiographic image for each imaging angle acquired by the radiographic image detector 15.

The abnormal shadow candidate detection device 50 is the abnormal shadow candidate detection device according to the first embodiment of the present invention described with reference to FIG. Instead of the detection device 50, the abnormal shadow candidate detection device 50 'according to the second embodiment of the present invention shown in FIG. 9 may be provided.

The input unit 4 is configured by a pointing device such as a keyboard and a mouse, for example, and receives input of shooting conditions and the like by a photographer, input of shooting start instructions and detection start instructions for abnormal shadow candidates, and the like. .

The monitor 9 displays a stereo image using the two radiographic image signals output from the computer 8. The configuration of the monitor 9 is, for example, a radiographic image based on the two radiographic image signals using two screens. By using a half mirror or polarizing glass, one radiological image is incident on the right eye of the observer, and the other radiological image is incident on the left eye of the observer, thereby displaying a stereo image. A configuration can be employed. Or, for example, two radiographic images may be displayed in a superimposed manner while being shifted by a predetermined amount of parallax, and this may be configured to generate a stereo image by observing with a polarizing glass, or a parallax barrier method and a lenticular method As described above, a stereo image may be generated by displaying two radiation images on a stereoscopically viewable 3D liquid crystal.

Next, the operation of the breast image diagnosis system of this embodiment will be described.

First, the patient's breast M is placed on the imaging table 14, and the breast M is compressed with a predetermined pressure by the compression plate 18.

Next, in the input unit 4, after various shooting conditions are input by the photographer, an instruction to start shooting is input.

When there is an instruction to start imaging at the input unit 4, a stereo image of the breast M is captured. Specifically, first, the control unit 8 a reads a convergence angle θ for capturing a preset stereo image, and outputs information on the read convergence angle θ to the arm controller 31. In this embodiment, θ = ± 2 ° is stored in advance as information on the convergence angle θ at this time. However, the present invention is not limited to this, and an arbitrary convergence angle is set by the photographer in the input unit 4. It can be set.

Then, the arm controller 31 receives the information on the convergence angle θ output from the control unit 8a. The arm controller 31 captures the image of the arm unit 13 based on the information on the convergence angle θ as shown in FIG. A control signal is output so as to rotate + θ ° with respect to a direction perpendicular to the table 14. That is, in the present embodiment, a control signal is output so that the arm unit 13 is rotated + 2 ° with respect to a direction perpendicular to the imaging table 14.

Then, according to the control signal output from the arm controller 31, the arm portion 13 rotates by + 2 °. Subsequently, the control unit 8a outputs a control signal to the radiation source controller 32 and the detector controller 33 so as to perform radiation irradiation and readout of the radiation image signal. In response to this control signal, radiation is emitted from the radiation source 17, a radiation image obtained by photographing the breast from the + 2 ° direction is detected by the radiation image detector 15, and a radiation image signal is read by the detector controller 33. After predetermined signal processing is performed on the radiographic image signal, the radiographic image signal is stored in the radiographic image storage unit 8 b of the computer 2.

Next, as shown in FIG. 2, the arm controller 31 once returns the arm unit to the initial position, and then outputs a control signal so as to rotate by −θ ° with respect to the direction perpendicular to the imaging table 14. That is, in the present embodiment, a control signal is output so that the arm unit 13 is rotated by −2 ° with respect to a direction perpendicular to the imaging table 14.

Then, according to the control signal output from the arm controller 31, the arm portion 13 rotates by -2 °. Subsequently, the control unit 8a outputs a control signal to the radiation source controller 32 and the detector controller 33 so as to perform radiation irradiation and radiation image reading. In response to this control signal, radiation is emitted from the radiation source 17, a radiation image obtained by imaging the breast from the −2 ° direction is detected by the radiation image detector 15, and a radiation image signal is read by the detector controller 33. After predetermined signal processing is performed, it is stored in the radiation image storage unit 8b of the computer 2.

In addition to the imaging conditions previously input from the input unit 4, the distance D between the two focal positions f1 and f2 at the time of imaging, the distance SID between the focal positions f1 and f2 and the radiation detection 15, and the surface 14a of the imaging table 14 To the detection surface 15a of the radiation detection note 15 and information I regarding the compression thickness t of the breast M held between the imaging table 14 and the compression plate 18 are associated with the two radiation image signals. It is stored in the radiation image storage unit 8b of the computer 2.

The two radiographic image signals stored in the radiographic image storage unit 8 b of the computer 8 are read from the radiographic image storage unit 8 b, subjected to predetermined signal processing, and output to the monitor 9. A stereo image of the breast is displayed.

In this breast image diagnosis system 1, when there is an instruction for automatic detection of an abnormal shadow candidate from the input unit 4, the abnormal shadow candidate Cn is detected by the abnormal shadow candidate detection device 50.

The method of detecting an abnormal shadow candidate by the abnormal shadow candidate detection apparatus 50 is as described above.

Note that a predetermined image processing may be performed on a region of interest including the abnormal shadow candidate Cn detected by the abnormal shadow candidate detection device 50, and then a stereo image of the breast may be displayed on the monitor 9. A marker may be displayed so that the abnormal shadow candidate Cn detected in the stereo image can be visually recognized.

Claims

In the abnormal shadow candidate detection apparatus for detecting an abnormal shadow candidate from a radiation image of a breast,
Two radiographic images for displaying a stereoscopic image using binocular parallax, and two different radiographs of radiation emitted from two focal positions on a breast compressed by a compression plate An abnormal shadow candidate detecting means for performing detection processing for detecting an abnormal shadow candidate for each of the radiographic images for each of the imaging directions captured by irradiating from each direction;
The breast in which the corresponding region in the other radiographic image corresponding to the position where the one image abnormal shadow candidate detected from one radiographic image of the two radiographic images is present is compressed by the compression plate Corresponding area setting means for setting based on the imaging conditions including the compression thickness of,
An apparatus for detecting an abnormal shadow candidate, comprising: a determination unit that determines that the one-side image abnormal shadow candidate is a false positive when there is no abnormal shadow candidate in the corresponding region.
The imaging conditions include the relationship between the two focal positions, a detector for detecting the radiation irradiated and transmitted through the breast, the imaging position of the breast, and the distance between the two focal positions. The apparatus for detecting an abnormal shadow candidate according to claim 1.
The abnormal shadow candidate detection means performs the detection process with higher detectability than the one-time image abnormal shadow candidate detection for the corresponding area set by the corresponding area setting means. The apparatus for detecting an abnormal shadow candidate according to claim 1 or 2.
When the determination unit has an abnormal shadow candidate in the corresponding region, the determination unit compares the feature amount of the abnormal shadow candidate and the one image abnormal shadow candidate, and based on the comparison result, the one image abnormal shadow 4. The abnormal shadow candidate detecting apparatus according to claim 1, wherein the candidate determines whether the candidate is a false positive.
In the method for detecting an abnormal shadow candidate for detecting an abnormal shadow candidate from a radiation image of a breast,
Two radiographic images for displaying a stereoscopic image using binocular parallax, and two different radiographs of radiation emitted from two focal positions on a breast compressed by a compression plate For each of the radiographic images taken for each imaging direction taken by irradiating from each direction, a detection process for detecting a candidate for an abnormal shadow is performed,
The corresponding compression region of the other radiographic image corresponding to the position of the one image abnormal shadow candidate detected from one radiographic image of the two radiographic images in the radiographic image is compressed by the compression plate of the breast. Set based on shooting conditions including
A method for detecting an abnormal shadow candidate, characterized in that, when no abnormal shadow candidate exists in the corresponding region, the one-side image abnormal shadow candidate is determined to be false positive.
After setting the corresponding region, for the corresponding region, perform a re-detection process with a detection capability higher than the detection capability in the detection process,
6. The abnormal shadow candidate detection method according to claim 5, wherein the determination is performed after the re-detection process.
In the method for detecting an abnormal shadow candidate for detecting an abnormal shadow candidate from a radiation image of a breast,
Radiation emitted from two focal positions from two different imaging directions to the breast compressed by two radiation image compression plates for displaying stereoscopic images using binocular parallax, respectively. For one radiographic image of the radiographic image captured in each imaging direction by performing a detection process for detecting abnormal shadow candidates,
Imaging conditions including the compression thickness of the breast compressed by the compression plate in the corresponding region in the other radiographic image corresponding to the position of the radiographic image of the one-image abnormal shadow candidate detected from the one radiographic image Set based on
For the corresponding region of the other radiographic image, a detection process for detecting an abnormal shadow candidate is performed,
A method for detecting an abnormal shadow candidate, characterized in that, when no abnormal shadow candidate exists in the corresponding region, the one-side image abnormal shadow candidate is determined to be false positive.
8. The method for detecting an abnormal shadow candidate according to claim 7, wherein the detection process for the corresponding region of the other radiographic image is performed with a higher detection ability than when the one image abnormal shadow candidate is detected.
When there is an abnormal shadow candidate in the corresponding region, the feature amount of the abnormal shadow candidate and the one image abnormal shadow candidate are compared, and the one image abnormal shadow candidate is erroneously detected based on the result of the comparison. The method for detecting an abnormal shadow candidate according to claim 5, wherein it is determined whether or not there is any.
In a program for causing a computer to execute an abnormal shadow candidate detection method for detecting an abnormal shadow candidate from a radiation image of a breast,
Two radiographic images for displaying a stereoscopic image using binocular parallax, and two different radiographs of radiation emitted from two focal positions on a breast compressed by a compression plate An abnormal shadow candidate detection process for detecting an abnormal shadow candidate for each of the radiographic images for each of the imaging directions captured by irradiating from each direction,
The breast in which the corresponding region in the other radiographic image corresponding to the position where the one image abnormal shadow candidate detected from one radiographic image of the two radiographic images is present is compressed by the compression plate A corresponding area setting process that is set based on an imaging condition including the compression thickness of the image, and a determination process that determines that the one-side image abnormal shadow candidate is a false positive when there is no abnormal shadow candidate in the corresponding area. A program characterized by causing a detection method to be executed.
A program for causing a computer to execute an abnormal shadow candidate detection method for detecting an abnormal shadow candidate from a radiation image of a breast,
Two radiographic images for displaying a stereoscopic image using binocular parallax, and two different radiographs of radiation emitted from two focal positions on a breast compressed by a compression plate A first abnormal shadow candidate detection process for detecting an abnormal shadow candidate for one radiographic image of each radiographic image captured by irradiation from each direction;
Imaging conditions including the compression thickness of the breast compressed by the compression plate in the corresponding region in the other radiographic image corresponding to the position of the radiographic image of the one-image abnormal shadow candidate detected from the one radiographic image Corresponding area setting process to set based on
A second abnormal shadow candidate detection process for detecting an abnormal shadow candidate for the corresponding region of the other radiographic image; and if the abnormal shadow candidate does not exist in the corresponding region, the one image abnormal shadow candidate A program for executing a detection method including a determination process for determining that is false positive.
When there is an abnormal shadow candidate in the corresponding region, the feature amount of the abnormal shadow candidate and the one image abnormal shadow candidate is compared, and the one image abnormal shadow candidate is false positive based on the comparison result. The program according to claim 10 or 11, wherein the detection method including a determination process for determining whether or not there is present is executed.
An image capturing device that captures radiation images in each of the capturing directions by irradiating radiation emitted from the two focal positions from two mutually different capturing directions to the breast compressed by the compression plate;
A display device for displaying a stereoscopic image using the two captured radiographic images;
A breast image diagnosis system comprising: the abnormal shadow candidate detecting device for detecting an abnormal shadow candidate from the two radiographic images according to any one of claims 1 to 4.