WO2018030519A1 - Breast region detecting system, breast region detecting method, and program - Google Patents

Breast region detecting system, breast region detecting method, and program Download PDF

Info

Publication number
WO2018030519A1
WO2018030519A1 PCT/JP2017/029125 JP2017029125W WO2018030519A1 WO 2018030519 A1 WO2018030519 A1 WO 2018030519A1 JP 2017029125 W JP2017029125 W JP 2017029125W WO 2018030519 A1 WO2018030519 A1 WO 2018030519A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
breast region
breast
enhanced
enhanced image
Prior art date
Application number
PCT/JP2017/029125
Other languages
French (fr)
Japanese (ja)
Inventor
山本将勝
Original Assignee
コニカミノルタメディカルソリューションズ株式会社
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by コニカミノルタメディカルソリューションズ株式会社 filed Critical コニカミノルタメディカルソリューションズ株式会社
Priority to JP2018533566A priority Critical patent/JP7010225B2/en
Publication of WO2018030519A1 publication Critical patent/WO2018030519A1/en

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B6/00—Apparatus for radiation diagnosis, e.g. combined with radiation therapy equipment
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T1/00—General purpose image data processing

Definitions

  • the present invention relates to a breast region detection system, a breast region detection method, and a program for detecting a breast region from a digital mammography image.
  • an interpreting physician sometimes watches a high-luminance screen for interpretation of a digital mammography image. Further, an interpreting physician may perform image processing such as contrast change or black-and-white reversal on an image displayed on the screen in order to make the lesion in the breast region easy to see.
  • the density of the background region other than the breast region is increased (for example, white) by such image processing, it is very dazzling and increases the eye burden on the interpretation doctor. For this reason, it is desired to make the density of the background region lower (for example, black) than the breast region.
  • the maximum density value excluding the distribution corresponding to the background region is obtained, and the digital mammography image is binarized using the maximum density value as a threshold value.
  • a region having a density lower than the threshold value can be detected as a breast region.
  • the said threshold value cannot be set appropriately and a breast area
  • region may not be detected more correctly.
  • the threshold may be set manually by the system administrator. In this case, the threshold value cannot be set appropriately due to human error, and the breast region may not be detected more accurately. Further, for example, when the mammography apparatus is upgraded, the optimum value of the threshold value may change. In this case, it is necessary to reset the threshold value each time, and the chance of human error increases.
  • an object of the present invention is to provide a breast region detection system, a breast region detection method, and a program capable of more accurately detecting a breast region in solving the above-described problems.
  • a breast region detection system includes: An edge-enhanced image generation unit that generates an edge-enhanced image of a breast region based on a digital mammography image; A reference image generation unit that generates a reference image by performing a process of filling a breast region with a color different from a background region on the edge-enhanced image; A breast region detection unit for detecting a breast region based on the reference image; It is characterized by providing.
  • a program includes a computer, An edge-enhanced image generation unit that generates an edge-enhanced image of a breast region based on a digital mammography image; A reference image generation unit that generates a reference image by performing a process of filling a breast region with a color different from a background region on the edge-enhanced image; A breast region detection unit for detecting a breast region based on the reference image; It is a program that functions as:
  • a breast region detection method generates an edge-enhanced image of a breast region based on a digital mammography image
  • a reference image is generated by performing processing for filling the breast region with a color different from the background region on the edge-enhanced image, Detecting a breast region based on the reference image; It is characterized by including.
  • the present invention it is possible to provide a breast region detection system, a breast region detection method, and a program that can detect a breast region more accurately.
  • FIG. 1 It is a block diagram which shows the basic composition of the breast area
  • FIG. 10 is a diagram showing a state in which a background region is displayed in black based on the breast region detected by the breast region detection system of FIG. 1 in the digital mammography image after black and white reversal shown in FIG. 9.
  • an edge enhanced image generation unit that generates an edge enhanced image of a breast region based on a digital mammography image;
  • a reference image generation unit that generates a reference image by performing a process of filling a breast region with a color different from a background region on the edge-enhanced image;
  • a breast region detection unit for detecting a breast region based on the reference image;
  • a breast region detection system is provided.
  • the breast region detection unit changes a threshold value of the density value with respect to the digital mammography image, and acquires a binarized image closest to the reference image as a breast region candidate image.
  • region detection system as described in a 1st aspect which detects a breast area
  • the breast region detection unit performs a labeling process on the breast region candidate image and detects a region having the largest area as a breast region. Provide a detection system.
  • the edge enhanced image generation unit includes: A skinline-enhanced image generation unit that generates a skinline-enhanced image that emphasizes the skinline of the breast region; A structure-enhanced image generating unit that generates a structure-enhanced image in which the structure in the breast region is emphasized; An enhanced image combining unit that generates the edge enhanced image by combining the skinline enhanced image and the structure enhanced image; A breast region detection system according to any one of the first to third aspects is provided.
  • the skin line enhanced image generation unit generates the skin line enhanced image by performing a Sobel filter process on the digital mammography image.
  • An area detection system is provided.
  • the structure emphasized image generation unit generates the structure emphasized image by performing a Laplacian filter process on the digital mammography image.
  • a breast region detection system is provided.
  • the reference image generation unit generates the reference image by performing a closing process on the edge-enhanced image, according to any one of the first to sixth aspects.
  • a breast region detection system is provided.
  • a computer An edge-enhanced image generation unit that generates an edge-enhanced image of a breast region based on a digital mammography image;
  • a reference image generation unit that generates a reference image by performing a process of filling a breast region with a color different from a background region on the edge-enhanced image;
  • a breast region detection unit for detecting a breast region based on the reference image;
  • the breast region detection unit changes a threshold value of the density value with respect to the digital mammography image, and acquires a binarized image closest to the reference image as a breast region candidate image. Then, the program according to the eighth aspect is provided for detecting a breast region based on the breast region candidate image.
  • the breast region detection unit performs a labeling process on the breast region candidate image and detects a region having the largest area as a breast region. provide.
  • the edge enhanced image generation unit includes: A skinline-enhanced image generation unit that generates a skinline-enhanced image that emphasizes the skinline of the breast region; A structure-enhanced image generating unit that generates a structure-enhanced image in which the structure in the breast region is emphasized; An enhanced image combining unit that generates the edge enhanced image by combining the skinline enhanced image and the structure enhanced image; A program according to any one of the eighth to tenth aspects is provided.
  • the skinline enhanced image generation unit generates the skinline enhanced image by performing a Sobel filter process on the digital mammography image. I will provide a.
  • the structure emphasized image generation unit generates the structure emphasized image by performing a Laplacian filter process on the digital mammography image. Provide a program.
  • the reference image generation unit generates the reference image by performing a closing process on the edge enhanced image. Provide a program.
  • an edge-enhanced image of a breast region is generated based on a digital mammography image
  • a reference image is generated by performing processing for filling the breast region with a color different from the background region on the edge-enhanced image, Detecting a breast region based on the reference image;
  • a breast region detection method is provided.
  • detecting a breast region based on the reference image comprises By changing a threshold value of density value for the digital mammography image, a binary image closest to the reference image is obtained as a breast region candidate image, Detecting a breast region based on the breast region candidate image; The breast area
  • FIG. 1 is a block diagram showing a basic configuration of a breast region detection system according to an embodiment of the present invention.
  • the breast region detection system 1 is a system that detects a breast region based on a digital mammography image captured by a mammography apparatus or the like.
  • the breast region detection system 1 includes a preprocessing unit 2, an edge enhanced image generation unit 3, a reference image generation unit 4, and a breast region detection unit 5.
  • the pre-processing unit 2 is configured to perform a reduction process for reducing the number of pixels on a digital mammography image that is a target for detecting a breast region. By this reduction processing, fine noise in the digital mammography image is removed. Moreover, the speed of subsequent image processing can be increased by this reduction processing.
  • the edge-enhanced image generation unit 3 is configured to generate an edge-enhanced image of the breast region based on the digital mammography image.
  • the edge enhanced image generation unit 3 includes a skin line enhanced image generation unit 31, a structure enhanced image generation unit 32, and an enhanced image composition unit 33.
  • the skinline enhanced image generation unit 31 is configured to generate a skinline enhanced image in which the skinline SL of the breast region E1 is enhanced.
  • the skin line SL is a portion that becomes a boundary between the breast region E1 and the background region E2.
  • the skinline enhanced image generation unit 31 is configured to generate a skinline enhanced image (FIG. 2) by performing a Sobel filter process that is a first-order differential filter on a digital mammography image.
  • the “Sobel filter” is a filter that calculates a spatial first derivative and detects a contour.
  • the skin line SL can be emphasized by applying a Sobel filter to the digital mammography image in, for example, four directions, up, down, left, and right.
  • the skinline enhanced image generation unit 31 may be configured to perform binarization processing on the skinline enhanced image (FIG. 2). As a result, as shown in FIG. 3, it is possible to generate a skinline enhanced image in which not only the skinline SL but also structures in the breast region E1 are enhanced.
  • the structure emphasized image generation unit 32 is configured to generate a structure emphasized image in which the structure in the breast region E1 is emphasized.
  • the structure-enhanced image generation unit 32 is configured to generate a structure-enhanced image by performing Laplacian filter processing that is a second-order differential filter on the digital mammography image.
  • the “Laplacian filter” is a filter that detects a contour by calculating a spatial second derivative, and is a filter that extracts a portion where the amount of change in density difference is extremely large.
  • the structure emphasized image generation unit 32 binarizes the structure emphasized image. It may be configured to perform processing. Thereby, the structure emphasis image in which the structure in the breast region E1 is more emphasized can be generated.
  • the emphasized image combining unit 33 is configured to generate an edge emphasized image by combining the skinline emphasized image (FIG. 3) and the structure emphasized image (FIG. 4).
  • the skinline emphasized image FIG. 3
  • the structure emphasized image FIG. 4
  • the reference image generation unit 4 performs a process of filling (filling) the breast region E1 with a color (for example, white) different from the background region E2 on the edge-enhanced image (FIG. 5).
  • a reference image is generated.
  • the reference image generation unit 4 is configured to generate a reference image (FIG. 6) by performing a closing process on the edge enhanced image (FIG. 5).
  • “expansion processing” is a type of morphology processing, and refers to processing in which “expansion” is performed N times and then “shrinkage” is performed N times. According to the closing process, it is possible to obtain effects such as filling a figure or combining cut parts.
  • the breast region detection unit 5 is configured to detect the breast region E1 based on the reference image (FIG. 6).
  • the breast region detection unit 5 changes the threshold value of the density value with respect to the digital mammography image, and acquires the binarized image closest to the reference image (FIG. 6) as the breast region candidate image. It is configured.
  • the breast region detection unit 5 sequentially increases the threshold value of the density value, sequentially compares the binarized image obtained thereby and the reference image, and searches for a binarized image with the smallest number of non-overlapping pixels. Then, the searched binary image is acquired as a breast region candidate image.
  • the threshold value is a value serving as a reference for whether or not to display a specific color (for example, white) in the binarized image. For example, a portion where the density value is less than or equal to the threshold is displayed in black, while a portion where the density value is greater than the threshold is displayed in white. That is, when the threshold value is small, the binarized image is displayed in black, and as the threshold value is increased, the breast region is gradually displayed in white.
  • the breast region detection unit 5 is configured to detect a breast region based on the acquired breast region candidate image.
  • the breast region candidate image has a plurality of breast region candidates E11 and E12 (portions displayed in white) as shown in FIG. 7 due to noise, identification information, and the like in the mammography image. There is.
  • the breast region detection unit 5 is configured to perform a labeling process on the breast region candidate image (FIG. 7) and detect the region E11 having the largest area as a breast region.
  • the “labeling process” refers to a process of distinguishing areas in a binarized image by assigning the same number to pixels in which a portion of a color (for example, white) different from the background area is continuous. That is, the breast region detection unit 5 is configured to detect a region having the largest area as a breast region among regions assigned the same number by the labeling process.
  • the digital mammography image is continuous with no density difference between the background region and the breast region as shown in FIG. 13, and the edge enhancement is performed even if the breast region is unclear. Since the image generation unit 3 is provided, an edge-enhanced image of the breast region can be obtained. In addition, since the reference image generation unit 4 is provided, it is possible to obtain a reference image in which the breast region is filled with a color different from the background region with respect to the edge enhanced image. With this reference image, the breast region can be detected more accurately.
  • the breast region detection system since the breast region is detected without using the density value histogram, it is possible to suppress the influence of imaging conditions and the like, and to further improve the breast region. It can be detected accurately.
  • the breast region detection system since the breast region is detected based on a reference image (FIG. 6) that can be automatically created, an artificial detection error is suppressed. be able to.
  • the breast region detection system according to the embodiment of the present invention, a labeling process is performed on the breast region candidate image, and the region E11 having the largest area is detected as a breast region.
  • the breast region can be detected more accurately even if the background region includes a region caused by noise, identification information, or an extra structure.
  • FIG. 8 is a diagram showing an example of a digital mammography image.
  • FIG. 9 is a diagram illustrating a state in which black and white inversion has been performed on the digital mammography image illustrated in FIG. 8.
  • FIG. 10 is a diagram illustrating a state in which the background region is displayed in black based on the breast region detected by the breast region detection system in the digital mammography image after black and white reversal illustrated in FIG. 9. As shown in FIG. 10, according to the breast region detection system, it is possible to obtain an image in which the inside of the breast region is easy to see with little burden on the eyes of the interpretation doctor.
  • the preprocessing unit 2, the edge-enhanced image generation unit 3, the reference image generation unit 4, and the breast region detection unit 5 can be realized by an MPU, a memory, or the like, for example. Further, the processing procedures of the units 2 to 5 can be realized by software or hardware (dedicated circuit) recorded in a storage medium such as a ROM, for example.
  • FIG. 11 is a flowchart showing an example of a breast region detection method according to the embodiment of the present invention.
  • the preprocessing unit 2 performs a reduction process for reducing the number of pixels on a digital mammography image that is a target for detecting a breast region (step S1).
  • a reduction process for reducing the number of pixels on a digital mammography image that is a target for detecting a breast region.
  • fine noise in the digital mammography image is removed.
  • the speed of subsequent image processing can be increased by this reduction processing.
  • the original size of the digital mammography image is, for example, about 2000 to 10,000 pixels.
  • the size of the digital mammography image after the reduction process is, for example, about 256 to 512 pixels.
  • the edge-enhanced image generation unit 3 generates an edge-enhanced image (FIG. 3) of the breast region E1 based on the digital mammography image after the reduction process (Steps S2 to S4).
  • the skin line-enhanced image generation unit 31 performs a Sobel filter process that is a first-order differential filter on the digital mammography image after the reduction process, thereby enhancing the skin line SL of the breast region E1.
  • a line-enhanced image (FIG. 2) is generated (step S2).
  • the skinline enhanced image generation unit 31 performs binarization processing on the skinline enhanced image (FIG. 2), and not only the skinline SL but also the structure in the breast region E1 is enhanced.
  • a skinline enhanced image (FIG. 3) is generated.
  • the structure-enhanced image generating unit 32 emphasizes structures (mammary gland, fat, etc.) in the breast region E1 by performing Laplacian filter processing that is a second-order differential filter on the digital mammography image after reduction processing.
  • a structure emphasized image (FIG. 4) is generated (step S3).
  • the enhanced image composition unit 33 synthesizes the skinline enhanced image (FIG. 3) and the structure enhanced image (FIG. 4) to generate an edge enhanced image (FIG. 5) (step S4).
  • the reference image generation unit 4 generates a reference image (FIG. 6) by performing a closing process on the edge enhanced image (FIG. 5) (step S5).
  • the breast region detection unit 5 detects a breast region based on the reference image (FIG. 6) (steps S6 to S8).
  • the breast region detection unit 5 changes the threshold value of the density value for the digital mammography image, and searches for a threshold value that becomes a binary image closest to the reference image (FIG. 6) (step S6).
  • the breast region detection unit 5 sequentially increases the threshold value of the density value, sequentially compares the binarized image obtained thereby and the reference image, and searches for the threshold value that minimizes the number of non-overlapping pixels.
  • a flowchart is shown in FIG.
  • the breast region detection unit 5 acquires a binarized image based on the searched threshold value as a breast region candidate image (FIG. 7) (step S7).
  • the breast region detection unit 5 performs a labeling process on the acquired breast region candidate image (FIG. 7), and detects the region E11 having the largest area as a breast region (step S8).
  • the breast region detection method since the breast region is detected without using the histogram of density values, the influence of the imaging condition or the like is suppressed, and the breast region E1 is more accurately detected. Can be detected.
  • the edge-enhanced image (FIG. 5) is generated by synthesizing the skinline-enhanced image (FIG. 3) and the structure-enhanced image (FIG. 4).
  • the present invention is not limited to this.
  • the skinline enhanced image (FIG. 3) when the structure of the breast region E1 is sufficiently displayed to the extent that the breast region E1 is filled with a specific color in the subsequent closing process, the skin is displayed.
  • the line enhanced image may be an edge enhanced image.
  • the structure-emphasized image (FIG. 4) when the structure of the breast region E1 is sufficiently displayed to the extent that the breast region E1 is filled with a specific color in the subsequent process, the structure is displayed.
  • the object enhanced image may be an edge enhanced image.
  • an edge-enhanced image (FIG. 5) is generated by combining two or more types of emphasized images, a situation occurs in which an appropriate edge-enhanced image cannot be obtained due to the influence of shooting conditions and the like. This can be suppressed more reliably.
  • the edge-enhanced image (FIG. 5) is generated by performing the Sobel filter process and the Laplacian filter process, but the present invention is not limited to this.
  • the edge-enhanced image may be an image in which the structure of the breast region E1 is sufficiently displayed so that the breast region E1 is filled with a specific color in the subsequent process. That is, the edge-enhanced image (FIG. 5) may be generated by performing one or more other types of image processing.
  • another first-order differential filter process for example, a Prewitt filter process or the like
  • another secondary differential filter process may be performed instead of the Laplacian filter process.
  • the breast region E1 is completely filled with a specific color by the closing process.
  • the breast region E1 is specified by the closing process. It is also assumed that it is not completely filled with the color. In this case, the breast region E1 may be completely filled with a specific color by another process or by using a closing process and another process in combination.
  • the binarized image closest to the reference image is the breast region candidate image, but the present invention is not limited to this.
  • a predetermined standard for example, the contents of error check described later
  • the absence of a hole (a part not completely filled with a specific color) as shown in FIG. It may be a breast region candidate image. That is, an error check may be performed after step S5 in FIG. 11 to detect a breast region directly from the reference image if a predetermined reference is satisfied.
  • FIG. 15 shows a setting screen for error check processing.
  • the “predetermined reference” is, for example, the presence or absence of a hole in the breast region (see FIG. 12), the size of the breast region, the deviation of the vertical position of the breast region, and the complexity of the breast region.
  • the error check determines that an error has occurred, for example, an error is displayed on the screen, or the original digital mammography image is displayed on the screen. As a result, it is possible to inform the interpretation doctor that the accuracy of detection of the breast region is not sufficient.
  • the functions of the units 2 to 5 of the breast region detection system 1 may be realized by software.
  • the software may be provided by downloading or the like.
  • the software may be provided by being recorded on a computer-readable storage medium such as a CD-ROM.
  • region detection system 1 which concerns on this embodiment is the following programs. That is, the program includes an edge-enhanced image generation unit that generates an edge-enhanced image of a breast region based on a digital mammography image, and a process of filling the edge-enhanced image with a color different from a background region.
  • the present invention is useful for, for example, a breast region detection system, a breast region detection method, and a program that detect a breast region from a digital mammography image.

Abstract

This breast region detecting system is provided with: an edge emphasized image generating unit which generates an edge emphasized image of a breast region on the basis of a digital mammography image; a reference image generating unit which generates a reference image by subjecting the edge emphasized image to a process to fill the breast region with a color different from a background region; and a breast region detecting unit which detects the breast region on the basis of the reference image.

Description

乳房領域検出システム、乳房領域検出方法、及びプログラムBreast region detection system, breast region detection method, and program
　本発明は、デジタルマンモグラフィ画像から乳房領域を検出する乳房領域検出システム、乳房領域検出方法、及びプログラムに関する。 The present invention relates to a breast region detection system, a breast region detection method, and a program for detecting a breast region from a digital mammography image.
　従来、読影医（放射線診断医）は、デジタルマンモグラフィ画像の読影のために、高輝度の画面を注視することがある。また、読影医は、乳房領域内の病変を見易くするため、画面に表示する画像に対してコントラストの変更や白黒反転などの画像処理を行うことがある。 Conventionally, an interpreting physician (radiologist) sometimes watches a high-luminance screen for interpretation of a digital mammography image. Further, an interpreting physician may perform image processing such as contrast change or black-and-white reversal on an image displayed on the screen in order to make the lesion in the breast region easy to see.
　しかしながら、このような画像処理によって乳房領域以外の背景領域の濃度を高く（例えば、白く）した場合には、非常に眩しく、読影医の目の負担が大きくなる。このため、乳房領域よりも背景領域の濃度を低く（例えば、黒く）することが望まれている。 However, when the density of the background region other than the breast region is increased (for example, white) by such image processing, it is very dazzling and increases the eye burden on the interpretation doctor. For this reason, it is desired to make the density of the background region lower (for example, black) than the breast region.
　背景領域の濃度を低くする画像処理を行うためには、乳房領域をより正確に検出する必要がある。乳房領域を検出する技術として、例えば、以下のような判別分析法を利用した技術が知られている（例えば、特開２０１１－１２５３６３号公報参照）。 In order to perform image processing to reduce the density of the background area, it is necessary to detect the breast area more accurately. As a technique for detecting a breast region, for example, a technique using the following discriminant analysis method is known (see, for example, JP 2011-125363 A).
　まず、デジタルマンモグラフィ画像（マンモグラム）を構成する画素の濃度値のヒストグラムを作成する。その後、当該ヒストグラムに基づいて、背景領域（濃度値が高い部分）に対応する分布を除く最大濃度値を求め、当該最大濃度値を閾値としてデジタルマンモグラフィ画像を二値化する。これにより、閾値よりも低濃度の領域を乳房領域として検出することができる。 First, create a histogram of the density values of the pixels that make up the digital mammography image (mammogram). Thereafter, based on the histogram, the maximum density value excluding the distribution corresponding to the background region (the portion having a high density value) is obtained, and the digital mammography image is binarized using the maximum density value as a threshold value. Thereby, a region having a density lower than the threshold value can be detected as a breast region.
特開２０１１－１２５３６３号公報JP 2011-125363 A
　しかしながら、前記技術では、乳房の大きさ、撮影装置、撮影条件等によっては、図１３に示すように、背景領域と乳房領域との濃度差がなく連続していることがある。このため、前記閾値を適切に設定できず、乳房領域をより正確に検出できない場合がある。 However, in the above technique, depending on the size of the breast, the imaging device, the imaging conditions, etc., as shown in FIG. For this reason, the said threshold value cannot be set appropriately and a breast area | region may not be detected more correctly.
　また、前記閾値の設定は、システムの管理者が手動で行うことがある。この場合、人為的なミスにより前記閾値を適切に設定できず、乳房領域をより正確に検出できない場合がある。また、例えば、マンモグラフィ装置などをバーションアップした際には、前記閾値の最適値が変わることがある。この場合、その都度、閾値を設定し直す必要があり、人為的なミスが発生する機会が増えることになる。 Also, the threshold may be set manually by the system administrator. In this case, the threshold value cannot be set appropriately due to human error, and the breast region may not be detected more accurately. Further, for example, when the mammography apparatus is upgraded, the optimum value of the threshold value may change. In this case, it is necessary to reset the threshold value each time, and the chance of human error increases.
　従って、本発明の目的は、前記課題を解決することにあって、乳房領域をより正確に検出することができる乳房領域検出システム、乳房領域検出方法、及びプログラムを提供することにある。 Therefore, an object of the present invention is to provide a breast region detection system, a breast region detection method, and a program capable of more accurately detecting a breast region in solving the above-described problems.
　前記目的を達成するために、本発明の一態様に係る乳房領域検出システムは、
　デジタルマンモグラフィ画像に基づいて乳房領域のエッジ強調画像を生成するエッジ強調画像生成部と、
　前記エッジ強調画像に対して乳房領域を背景領域とは異なる色で埋める処理を行うことにより基準画像を生成する基準画像生成部と、
　前記基準画像に基づいて乳房領域を検出する乳房領域検出部と、
　を備えることを特徴とする。 In order to achieve the above object, a breast region detection system according to an aspect of the present invention includes:
An edge-enhanced image generation unit that generates an edge-enhanced image of a breast region based on a digital mammography image;
A reference image generation unit that generates a reference image by performing a process of filling a breast region with a color different from a background region on the edge-enhanced image;
A breast region detection unit for detecting a breast region based on the reference image;
It is characterized by providing.
　本発明の一態様に係るプログラムは、コンピュータを、
　デジタルマンモグラフィ画像に基づいて乳房領域のエッジ強調画像を生成するエッジ強調画像生成部と、
　前記エッジ強調画像に対して乳房領域を背景領域とは異なる色で埋める処理を行うことにより基準画像を生成する基準画像生成部と、
　前記基準画像に基づいて乳房領域を検出する乳房領域検出部、
　として機能させるプログラムであることを特徴とする。 A program according to one embodiment of the present invention includes a computer,
An edge-enhanced image generation unit that generates an edge-enhanced image of a breast region based on a digital mammography image;
A reference image generation unit that generates a reference image by performing a process of filling a breast region with a color different from a background region on the edge-enhanced image;
A breast region detection unit for detecting a breast region based on the reference image;
It is a program that functions as:
　本発明の一態様に係る乳房領域検出方法は、デジタルマンモグラフィ画像に基づいて乳房領域のエッジ強調画像を生成し、
　前記エッジ強調画像に対して乳房領域を背景領域とは異なる色で埋める処理を行うことにより基準画像を生成し、
　前記基準画像に基づいて乳房領域を検出する、
　ことを含むことを特徴とする。 A breast region detection method according to an aspect of the present invention generates an edge-enhanced image of a breast region based on a digital mammography image,
A reference image is generated by performing processing for filling the breast region with a color different from the background region on the edge-enhanced image,
Detecting a breast region based on the reference image;
It is characterized by including.
　本発明によれば、乳房領域をより正確に検出することができる乳房領域検出システム、乳房領域検出方法、及びプログラムを提供することができる。 According to the present invention, it is possible to provide a breast region detection system, a breast region detection method, and a program that can detect a breast region more accurately.
本発明の実施形態に係る乳房領域検出システムの基本構成を示すブロック図である。It is a block diagram which shows the basic composition of the breast area | region detection system which concerns on embodiment of this invention. スキンライン強調画像の一例を示す図である。It is a figure which shows an example of a skinline emphasis image. 図２のスキンライン強調画像に二値化処理を行って得られたスキンライン強調画像の一例を示す図である。It is a figure which shows an example of the skinline emphasized image obtained by performing the binarization process on the skinline emphasized image of FIG. 構造物強調画像の一例を示す図である。It is a figure which shows an example of a structure emphasis image. エッジ強調画像の一例を示す図である。It is a figure which shows an example of an edge emphasis image. 基準画像の一例を示す図である。It is a figure which shows an example of a reference | standard image. 乳房領域候補画像の一例を示す図である。It is a figure which shows an example of a breast area | region candidate image. デジタルマンモグラフィ画像の一例を示す図である。It is a figure which shows an example of a digital mammography image. 図８に示すデジタルマンモグラフィ画像に対して白黒反転を行った状態を示す図である。It is a figure which shows the state which performed the black and white inversion with respect to the digital mammography image shown in FIG. 図９に示す白黒反転後のデジタルマンモグラフィ画像において、図１の乳房領域検出システムによって検出した乳房領域に基づいて背景領域を黒色で表示した状態を示す図である。FIG. 10 is a diagram showing a state in which a background region is displayed in black based on the breast region detected by the breast region detection system of FIG. 1 in the digital mammography image after black and white reversal shown in FIG. 9. 本発明の実施形態に係る乳房領域検出方法の一例を示すフローチャートである。It is a flowchart which shows an example of the breast area | region detection method which concerns on embodiment of this invention. 乳房領域が特定の色で完全に埋められていない状態を示す図である。It is a figure which shows the state in which the breast area | region is not completely filled with the specific color. デジタルマンモグラフィ画像において背景領域と乳房領域との濃度差がなく連続している状態を示す図である。It is a figure which shows the state which has no density difference of a background area | region and a breast area | region in a digital mammography image, and is continuous. 乳房領域候補画像を求めるために、基準画像に最も近くなる閾値を探索する一例を示すフローチャートである。It is a flowchart which shows an example which searches the threshold value nearest to a reference | standard image, in order to obtain | require a breast area | region candidate image. 検出した乳房領域が所定の基準を満たしているか否かを確認するエラーチェック処理の設定画面の一例を示す図である。It is a figure which shows an example of the setting screen of the error check process which confirms whether the detected breast area | region satisfy | fills the predetermined reference | standard.
　本発明の第１態様によれば、デジタルマンモグラフィ画像に基づいて乳房領域のエッジ強調画像を生成するエッジ強調画像生成部と、
　前記エッジ強調画像に対して乳房領域を背景領域とは異なる色で埋める処理を行うことにより基準画像を生成する基準画像生成部と、
　前記基準画像に基づいて乳房領域を検出する乳房領域検出部と、
　を備える、乳房領域検出システムを提供する。 According to the first aspect of the present invention, an edge enhanced image generation unit that generates an edge enhanced image of a breast region based on a digital mammography image;
A reference image generation unit that generates a reference image by performing a process of filling a breast region with a color different from a background region on the edge-enhanced image;
A breast region detection unit for detecting a breast region based on the reference image;
A breast region detection system is provided.
　本発明の第２態様によれば、前記乳房領域検出部は、前記デジタルマンモグラフィ画像に対して濃度値の閾値を変更して、前記基準画像に最も近い二値化画像を乳房領域候補画像として取得し、当該乳房領域候補画像に基づいて乳房領域を検出する、第１態様に記載の乳房領域検出システムを提供する。 According to the second aspect of the present invention, the breast region detection unit changes a threshold value of the density value with respect to the digital mammography image, and acquires a binarized image closest to the reference image as a breast region candidate image. And the breast area | region detection system as described in a 1st aspect which detects a breast area | region based on the said breast area | region candidate image is provided.
　本発明の第３態様によれば、前記乳房領域検出部は、前記乳房領域候補画像に対してラベリング処理を行い、最も面積の大きい領域を乳房領域として検出する、第２態様に記載の乳房領域検出システムを提供する。 According to a third aspect of the present invention, the breast region detection unit performs a labeling process on the breast region candidate image and detects a region having the largest area as a breast region. Provide a detection system.
　本発明の第４態様によれば、前記エッジ強調画像生成部は、
　前記乳房領域のスキンラインを強調したスキンライン強調画像を生成するスキンライン強調画像生成部と、
　前記乳房領域内の構造物を強調した構造物強調画像を生成する構造物強調画像生成部と、
　前記スキンライン強調画像と前記構造物強調画像とを合成することにより前記エッジ強調画像を生成する強調画像合成部と、
　を備える、第１～３態様のいずれか１つに記載の乳房領域検出システムを提供する。 According to the fourth aspect of the present invention, the edge enhanced image generation unit includes:
A skinline-enhanced image generation unit that generates a skinline-enhanced image that emphasizes the skinline of the breast region;
A structure-enhanced image generating unit that generates a structure-enhanced image in which the structure in the breast region is emphasized;
An enhanced image combining unit that generates the edge enhanced image by combining the skinline enhanced image and the structure enhanced image;
A breast region detection system according to any one of the first to third aspects is provided.
　本発明の第５態様によれば、前記スキンライン強調画像生成部は、前記デジタルマンモグラフィ画像に対してソーベルフィルタ処理を行うことにより前記スキンライン強調画像を生成する、第４態様に記載の乳房領域検出システムを提供する。 According to a fifth aspect of the present invention, in the breast according to the fourth aspect, the skin line enhanced image generation unit generates the skin line enhanced image by performing a Sobel filter process on the digital mammography image. An area detection system is provided.
　本発明の第６態様によれば、前記構造物強調画像生成部は、前記デジタルマンモグラフィ画像に対してラプラシアンフィルタ処理を行うことにより前記構造物強調画像を生成する、第４又は５態様に記載の乳房領域検出システムを提供する。 According to a sixth aspect of the present invention, in the fourth or fifth aspect, the structure emphasized image generation unit generates the structure emphasized image by performing a Laplacian filter process on the digital mammography image. A breast region detection system is provided.
　本発明の第７態様によれば、前記基準画像生成部は、前記エッジ強調画像に対してクロージング処理を行うことにより前記基準画像を生成する、第１～６態様のいずれか１つに記載の乳房領域検出システムを提供する。 According to a seventh aspect of the present invention, the reference image generation unit generates the reference image by performing a closing process on the edge-enhanced image, according to any one of the first to sixth aspects. A breast region detection system is provided.
　本発明の第８態様によれば、コンピュータを、
　デジタルマンモグラフィ画像に基づいて乳房領域のエッジ強調画像を生成するエッジ強調画像生成部と、
　前記エッジ強調画像に対して乳房領域を背景領域とは異なる色で埋める処理を行うことにより基準画像を生成する基準画像生成部と、
　前記基準画像に基づいて乳房領域を検出する乳房領域検出部、
　として機能させるためのプログラムを提供する。 According to an eighth aspect of the present invention, a computer
An edge-enhanced image generation unit that generates an edge-enhanced image of a breast region based on a digital mammography image;
A reference image generation unit that generates a reference image by performing a process of filling a breast region with a color different from a background region on the edge-enhanced image;
A breast region detection unit for detecting a breast region based on the reference image;
Provide a program to function as
　本発明の第９態様によれば、前記乳房領域検出部は、前記デジタルマンモグラフィ画像に対して濃度値の閾値を変更して、前記基準画像に最も近い二値化画像を乳房領域候補画像として取得し、当該乳房領域候補画像に基づいて乳房領域を検出する、第８態様に記載のプログラムを提供する。 According to the ninth aspect of the present invention, the breast region detection unit changes a threshold value of the density value with respect to the digital mammography image, and acquires a binarized image closest to the reference image as a breast region candidate image. Then, the program according to the eighth aspect is provided for detecting a breast region based on the breast region candidate image.
　本発明の第１０態様によれば、前記乳房領域検出部は、前記乳房領域候補画像に対してラベリング処理を行い、最も面積の大きい領域を乳房領域として検出する、第９態様に記載のプログラムを提供する。 According to a tenth aspect of the present invention, there is provided the program according to the ninth aspect, wherein the breast region detection unit performs a labeling process on the breast region candidate image and detects a region having the largest area as a breast region. provide.
　本発明の第１１態様によれば、前記エッジ強調画像生成部は、
　前記乳房領域のスキンラインを強調したスキンライン強調画像を生成するスキンライン強調画像生成部と、
　前記乳房領域内の構造物を強調した構造物強調画像を生成する構造物強調画像生成部と、
　前記スキンライン強調画像と前記構造物強調画像とを合成することにより前記エッジ強調画像を生成する強調画像合成部と、
　を備える、第８～１０態様のいずれか１つに記載のプログラムを提供する。 According to an eleventh aspect of the present invention, the edge enhanced image generation unit includes:
A skinline-enhanced image generation unit that generates a skinline-enhanced image that emphasizes the skinline of the breast region;
A structure-enhanced image generating unit that generates a structure-enhanced image in which the structure in the breast region is emphasized;
An enhanced image combining unit that generates the edge enhanced image by combining the skinline enhanced image and the structure enhanced image;
A program according to any one of the eighth to tenth aspects is provided.
　本発明の第１２態様によれば、前記スキンライン強調画像生成部は、前記デジタルマンモグラフィ画像に対してソーベルフィルタ処理を行うことにより前記スキンライン強調画像を生成する、第１１態様に記載のプログラムを提供する。 According to a twelfth aspect of the present invention, in the program according to the eleventh aspect, the skinline enhanced image generation unit generates the skinline enhanced image by performing a Sobel filter process on the digital mammography image. I will provide a.
　本発明の第１３態様によれば、前記構造物強調画像生成部は、前記デジタルマンモグラフィ画像に対してラプラシアンフィルタ処理を行うことにより前記構造物強調画像を生成する、第１１又は１２態様に記載のプログラムを提供する。 According to a thirteenth aspect of the present invention, in the eleventh or twelfth aspect, the structure emphasized image generation unit generates the structure emphasized image by performing a Laplacian filter process on the digital mammography image. Provide a program.
　本発明の第１４態様によれば、前記基準画像生成部は、前記エッジ強調画像に対してクロージング処理を行うことにより前記基準画像を生成する、第８～１３態様のいずれか１つに記載のプログラムを提供する。 According to a fourteenth aspect of the present invention, in any one of the eighth to thirteenth aspects, the reference image generation unit generates the reference image by performing a closing process on the edge enhanced image. Provide a program.
　本発明の第１５態様によれば、デジタルマンモグラフィ画像に基づいて乳房領域のエッジ強調画像を生成し、
　前記エッジ強調画像に対して乳房領域を背景領域とは異なる色で埋める処理を行うことにより基準画像を生成し、
　前記基準画像に基づいて乳房領域を検出する、
　ことを含む、乳房領域検出方法を提供する。 According to the fifteenth aspect of the present invention, an edge-enhanced image of a breast region is generated based on a digital mammography image,
A reference image is generated by performing processing for filling the breast region with a color different from the background region on the edge-enhanced image,
Detecting a breast region based on the reference image;
A breast region detection method is provided.
　本発明の第１６態様によれば、前記基準画像に基づいて乳房領域を検出することは、
　前記デジタルマンモグラフィ画像に対して濃度値の閾値を変更して、前記基準画像に最も近い二値化画像を乳房領域候補画像として取得し、
　前記乳房領域候補画像に基づいて乳房領域を検出する、
　ことを含む、第１５態様に記載の乳房領域検出方法を提供する。 According to a sixteenth aspect of the present invention, detecting a breast region based on the reference image comprises
By changing a threshold value of density value for the digital mammography image, a binary image closest to the reference image is obtained as a breast region candidate image,
Detecting a breast region based on the breast region candidate image;
The breast area | region detection method as described in a 15th aspect is provided.
　以下、本発明の実施形態について、図面を参照しながら説明する。なお、この実施形態によって、本発明が限定されるものではない。 Hereinafter, embodiments of the present invention will be described with reference to the drawings. In addition, this invention is not limited by this embodiment.
　《実施形態》
　本発明の実施形態に係る乳房領域検出システムについて説明する。図１は、本発明の実施形態に係る乳房領域検出システムの基本構成を示すブロック図である。 <Embodiment>
A breast region detection system according to an embodiment of the present invention will be described. FIG. 1 is a block diagram showing a basic configuration of a breast region detection system according to an embodiment of the present invention.
　本実施形態に係る乳房領域検出システム１は、マンモグラフィ装置などで撮影されたデジタルマンモグラフィ画像に基づいて乳房領域を検出するシステムである。 The breast region detection system 1 according to the present embodiment is a system that detects a breast region based on a digital mammography image captured by a mammography apparatus or the like.
　図１に示すように、乳房領域検出システム１は、前処理部２と、エッジ強調画像生成部３と、基準画像生成部４と、乳房領域検出部５とを備えている。 As shown in FIG. 1, the breast region detection system 1 includes a preprocessing unit 2, an edge enhanced image generation unit 3, a reference image generation unit 4, and a breast region detection unit 5.
　前処理部２は、乳房領域を検出する対象となるデジタルマンモグラフィ画像に対して、画素数を減少させる縮小処理を行うように構成されている。この縮小処理により、デジタルマンモグラフィ画像中の微細なノイズが除去される。また、この縮小処理により、後続の画像処理の速度を速くすることができる。 The pre-processing unit 2 is configured to perform a reduction process for reducing the number of pixels on a digital mammography image that is a target for detecting a breast region. By this reduction processing, fine noise in the digital mammography image is removed. Moreover, the speed of subsequent image processing can be increased by this reduction processing.
　エッジ強調画像生成部３は、デジタルマンモグラフィ画像に基づいて乳房領域のエッジ強調画像を生成するように構成されている。本実施形態において、エッジ強調画像生成部３は、スキンライン強調画像生成部３１と、構造物強調画像生成部３２と、強調画像合成部３３とを備えている。 The edge-enhanced image generation unit 3 is configured to generate an edge-enhanced image of the breast region based on the digital mammography image. In the present embodiment, the edge enhanced image generation unit 3 includes a skin line enhanced image generation unit 31, a structure enhanced image generation unit 32, and an enhanced image composition unit 33.
　スキンライン強調画像生成部３１は、図２に示すように、乳房領域Ｅ１のスキンラインＳＬを強調したスキンライン強調画像を生成するように構成されている。スキンラインＳＬは、乳房領域Ｅ１と背景領域Ｅ２との境界となる部分である。本実施形態において、スキンライン強調画像生成部３１は、デジタルマンモグラフィ画像に対して１次微分フィルタであるソーベル（Sobel）フィルタ処理を行うことによりスキンライン強調画像（図２）を生成するように構成されている。「ソーベルフィルタ」は、空間１次微分を計算し、輪郭を検出するフィルタである。デジタルマンモグラフィ画像に対してソーベルフィルタを、例えば上下左右の４方向に適用することで、スキンラインＳＬを強調することができる。 As shown in FIG. 2, the skinline enhanced image generation unit 31 is configured to generate a skinline enhanced image in which the skinline SL of the breast region E1 is enhanced. The skin line SL is a portion that becomes a boundary between the breast region E1 and the background region E2. In the present embodiment, the skinline enhanced image generation unit 31 is configured to generate a skinline enhanced image (FIG. 2) by performing a Sobel filter process that is a first-order differential filter on a digital mammography image. Has been. The “Sobel filter” is a filter that calculates a spatial first derivative and detects a contour. The skin line SL can be emphasized by applying a Sobel filter to the digital mammography image in, for example, four directions, up, down, left, and right.
　また、図２に示すように、スキンライン強調画像生成部３１が生成したスキンライン強調画像では、低濃度ではあるものの、乳房領域Ｅ１内の構造物（乳腺、脂肪等）のエッジも強調される。このため、本実施形態において、スキンライン強調画像生成部３１は、スキンライン強調画像（図２）に対して二値化処理を行うように構成されてもよい。これにより、図３に示すように、スキンラインＳＬだけではなく、乳房領域Ｅ１内の構造物も強調されたスキンライン強調画像を生成することができる。 Further, as shown in FIG. 2, in the skinline enhanced image generated by the skinline enhanced image generation unit 31, the edge of the structure (mammary gland, fat, etc.) in the breast region E1 is also enhanced, although the density is low. . For this reason, in the present embodiment, the skinline enhanced image generation unit 31 may be configured to perform binarization processing on the skinline enhanced image (FIG. 2). As a result, as shown in FIG. 3, it is possible to generate a skinline enhanced image in which not only the skinline SL but also structures in the breast region E1 are enhanced.
　構造物強調画像生成部３２は、図４に示すように、乳房領域Ｅ１内の構造物を強調した構造物強調画像を生成するように構成されている。本実施形態において、構造物強調画像生成部３２は、デジタルマンモグラフィ画像に対して２次微分フィルタであるラプラシアン（Laplacian）フィルタ処理を行うことにより構造物強調画像を生成するように構成されている。「ラプラシアンフィルタ」は、空間２次微分を計算して、輪郭を検出するフィルタであり、濃度の差分の変化量が極端に大きくなっている部分を抽出するフィルタである。 As shown in FIG. 4, the structure emphasized image generation unit 32 is configured to generate a structure emphasized image in which the structure in the breast region E1 is emphasized. In the present embodiment, the structure-enhanced image generation unit 32 is configured to generate a structure-enhanced image by performing Laplacian filter processing that is a second-order differential filter on the digital mammography image. The “Laplacian filter” is a filter that detects a contour by calculating a spatial second derivative, and is a filter that extracts a portion where the amount of change in density difference is extremely large.
　なお、ラプラシアンフィルタ処理により生成された構造物強調画像において、乳房領域Ｅ１内の構造物のエッジが低濃度である場合、構造物強調画像生成部３２は、構造物強調画像に対して二値化処理を行うように構成されてもよい。これにより、乳房領域Ｅ１内の構造物がより強調された構造物強調画像を生成することができる。 Note that, in the structure emphasized image generated by the Laplacian filter processing, when the edge of the structure in the breast region E1 has a low density, the structure emphasized image generation unit 32 binarizes the structure emphasized image. It may be configured to perform processing. Thereby, the structure emphasis image in which the structure in the breast region E1 is more emphasized can be generated.
　強調画像合成部３３は、図５に示すように、スキンライン強調画像（図３）と構造物強調画像（図４）とを合成することによりエッジ強調画像を生成するように構成されている。なお、上述のスキンライン強調画像のみを使って、その胸壁側を乳房領域として確定することは通常は出来ない場合が多い。その理由は、乳房の大きさ、撮影装置、撮影条件等によっては、図１３に示すように、背景領域と乳房領域との濃度差がなく連続していることがあり、このために乳房領域を正確に検出できない場合があるからである。 As shown in FIG. 5, the emphasized image combining unit 33 is configured to generate an edge emphasized image by combining the skinline emphasized image (FIG. 3) and the structure emphasized image (FIG. 4). In many cases, it is not usually possible to determine the breast wall side as a breast region by using only the above-described skinline enhanced image. The reason is that depending on the size of the breast, imaging device, imaging conditions, etc., as shown in FIG. 13, the background area and the breast area may be continuous with no difference in density. This is because it may not be detected accurately.
　基準画像生成部４は、図６に示すように、エッジ強調画像（図５）に対して乳房領域Ｅ１を背景領域Ｅ２とは異なる色（例えば、白色）で埋める（塗りつぶす）処理を行うことにより基準画像を生成するように構成されている。本実施形態において、基準画像生成部４は、エッジ強調画像（図５）に対してクロージング処理を行うことにより基準画像（図６）を生成するように構成されている。 As shown in FIG. 6, the reference image generation unit 4 performs a process of filling (filling) the breast region E1 with a color (for example, white) different from the background region E2 on the edge-enhanced image (FIG. 5). A reference image is generated. In the present embodiment, the reference image generation unit 4 is configured to generate a reference image (FIG. 6) by performing a closing process on the edge enhanced image (FIG. 5).
　なお、二値化画像内の図形を１画素分膨らませる処理を「膨張」といい、逆に１画素分縮める処理を「収縮」という。このような処理を一般にモフォロジー処理という。「クロージング処理」は、モフォロジー処理の一種であり、「膨張」をＮ回行った後、「収縮」をＮ回行う処理をいう。クロージング処理によれば、図形を穴埋めしたり、切断部分を結合したりするなどの効果を得ることができる。 Note that the process of expanding the figure in the binarized image by one pixel is called “expansion”, and the process of reducing the figure by one pixel is called “contraction”. Such processing is generally called morphology processing. “Closing processing” is a type of morphology processing, and refers to processing in which “expansion” is performed N times and then “shrinkage” is performed N times. According to the closing process, it is possible to obtain effects such as filling a figure or combining cut parts.
　乳房領域検出部５は、基準画像（図６）に基づいて乳房領域Ｅ１を検出するように構成されている。本実施形態において、乳房領域検出部５は、デジタルマンモグラフィ画像に対して濃度値の閾値を変更して、基準画像（図６）に最も近い二値化画像を乳房領域候補画像として取得するように構成されている。例えば、乳房領域検出部５は、濃度値の閾値を順次増加させ、それにより得られる二値化画像と基準画像とを順次比較して、重複しない画素数が最も少なくなる二値化画像を探索し、当該探索した二値化画像を乳房領域候補画像として取得する。 The breast region detection unit 5 is configured to detect the breast region E1 based on the reference image (FIG. 6). In the present embodiment, the breast region detection unit 5 changes the threshold value of the density value with respect to the digital mammography image, and acquires the binarized image closest to the reference image (FIG. 6) as the breast region candidate image. It is configured. For example, the breast region detection unit 5 sequentially increases the threshold value of the density value, sequentially compares the binarized image obtained thereby and the reference image, and searches for a binarized image with the smallest number of non-overlapping pixels. Then, the searched binary image is acquired as a breast region candidate image.
　なお、閾値は、二値化画像中において特定の色（例えば、白色）で表示するか否かの基準となる値である。例えば、濃度値が閾値以下である部分は、黒色で表示される一方、濃度値が閾値より大きい部分は、白色で表示される。すなわち、閾値が小さいとき、二値化画像はほとんど黒色で表示され、閾値を大きくしていくと、乳房領域が徐々に白色で表示されることになる。 Note that the threshold value is a value serving as a reference for whether or not to display a specific color (for example, white) in the binarized image. For example, a portion where the density value is less than or equal to the threshold is displayed in black, while a portion where the density value is greater than the threshold is displayed in white. That is, when the threshold value is small, the binarized image is displayed in black, and as the threshold value is increased, the breast region is gradually displayed in white.
　また、乳房領域検出部５は、当該取得した乳房領域候補画像に基づいて乳房領域を検出するように構成されている。なお、乳房領域候補画像には、マンモグラフィ画像中のノイズや識別情報等に起因して、図７に示すように、複数の乳房領域候補Ｅ１１，Ｅ１２（白色で表示される部分）が存在することがある。 Further, the breast region detection unit 5 is configured to detect a breast region based on the acquired breast region candidate image. It should be noted that the breast region candidate image has a plurality of breast region candidates E11 and E12 (portions displayed in white) as shown in FIG. 7 due to noise, identification information, and the like in the mammography image. There is.
　このため、本実施形態において、乳房領域検出部５は、乳房領域候補画像（図７）に対してラベリング処理を行い、最も面積の大きい領域Ｅ１１を乳房領域として検出するように構成されている。「ラベリング処理」とは、二値化画像において、背景領域とは異なる色（例えば、白色）の部分が連続した画素に同じ番号を割り振ることで領域を区別する処理をいう。すなわち、乳房領域検出部５は、ラベリング処理により同じ番号を割り振った領域のうち、最も面積の大きい領域を乳房領域として検出するように構成されている。 Therefore, in the present embodiment, the breast region detection unit 5 is configured to perform a labeling process on the breast region candidate image (FIG. 7) and detect the region E11 having the largest area as a breast region. The “labeling process” refers to a process of distinguishing areas in a binarized image by assigning the same number to pixels in which a portion of a color (for example, white) different from the background area is continuous. That is, the breast region detection unit 5 is configured to detect a region having the largest area as a breast region among regions assigned the same number by the labeling process.
　本発明の実施形態に係る乳房領域検出システムによれば、デジタルマンモグラフィ画像が図１３に示すように背景領域と乳房領域との濃度差がなく連続し、乳房領域が不鮮明であっても、エッジ強調画像生成部３を備えているので、乳房領域のエッジ強調画像を得ることができる。また、基準画像生成部４を備えているので、エッジ強調画像に対して乳房領域を背景領域とは異なる色で埋めた基準画像を得ることができる。この基準画像により、乳房領域をより正確に検出することができる。 According to the breast region detection system according to the embodiment of the present invention, the digital mammography image is continuous with no density difference between the background region and the breast region as shown in FIG. 13, and the edge enhancement is performed even if the breast region is unclear. Since the image generation unit 3 is provided, an edge-enhanced image of the breast region can be obtained. In addition, since the reference image generation unit 4 is provided, it is possible to obtain a reference image in which the breast region is filled with a color different from the background region with respect to the edge enhanced image. With this reference image, the breast region can be detected more accurately.
　また、本発明の実施形態に係る乳房領域検出システムによれば、濃度値のヒストグラムを使用せずに乳房領域を検出するようにしているので、撮影条件等による影響を抑えて、乳房領域をより正確に検出することができる。 Further, according to the breast region detection system according to the embodiment of the present invention, since the breast region is detected without using the density value histogram, it is possible to suppress the influence of imaging conditions and the like, and to further improve the breast region. It can be detected accurately.
　また、本発明の実施形態に係る乳房領域検出システムによれば、自動的に作成可能な基準画像（図６）に基づいて乳房領域を検出するようにしているので、人為的な検出ミスを抑えることができる。 In addition, according to the breast region detection system according to the embodiment of the present invention, since the breast region is detected based on a reference image (FIG. 6) that can be automatically created, an artificial detection error is suppressed. be able to.
　また、本発明の実施形態に係る乳房領域検出システムによれば、乳房領域候補画像に対してラベリング処理を行い、最も面積の大きい領域Ｅ１１を乳房領域として検出するようにしている。これにより、背景領域にノイズや識別情報や余計な構造物に起因する領域があっても、乳房領域をより正確に検出することができる。 Further, according to the breast region detection system according to the embodiment of the present invention, a labeling process is performed on the breast region candidate image, and the region E11 having the largest area is detected as a breast region. As a result, the breast region can be detected more accurately even if the background region includes a region caused by noise, identification information, or an extra structure.
　なお、図８は、デジタルマンモグラフィ画像の一例を示す図である。図９は、図８に示すデジタルマンモグラフィ画像に対して白黒反転を行った状態を示す図である。図１０は、図９に示す白黒反転後のデジタルマンモグラフィ画像において、本乳房領域検出システムによって検出した乳房領域に基づいて背景領域を黒色で表示した状態を示す図である。図１０に示すように、本乳房領域検出システムによれば、読影医の目の負担が少なく、乳房領域内が見易い画像を得ることができる。 FIG. 8 is a diagram showing an example of a digital mammography image. FIG. 9 is a diagram illustrating a state in which black and white inversion has been performed on the digital mammography image illustrated in FIG. 8. FIG. 10 is a diagram illustrating a state in which the background region is displayed in black based on the breast region detected by the breast region detection system in the digital mammography image after black and white reversal illustrated in FIG. 9. As shown in FIG. 10, according to the breast region detection system, it is possible to obtain an image in which the inside of the breast region is easy to see with little burden on the eyes of the interpretation doctor.
　なお、前処理部２、エッジ強調画像生成部３、基準画像生成部４、及び乳房領域検出部５は、例えば、ＭＰＵやメモリ等により実現可能である。また、各部２～５の処理手順は、例えば、ＲＯＭ等の記憶媒体に記録されているソフトウェアやハードウェア（専用回路）により実現可能である。 Note that the preprocessing unit 2, the edge-enhanced image generation unit 3, the reference image generation unit 4, and the breast region detection unit 5 can be realized by an MPU, a memory, or the like, for example. Further, the processing procedures of the units 2 to 5 can be realized by software or hardware (dedicated circuit) recorded in a storage medium such as a ROM, for example.
　次に、本発明の実施形態に係る乳房領域検出方法について説明する。図１１は、本発明の実施形態に係る乳房領域検出方法の一例を示すフローチャートである。 Next, a breast region detection method according to an embodiment of the present invention will be described. FIG. 11 is a flowchart showing an example of a breast region detection method according to the embodiment of the present invention.
　まず、前処理部２が、乳房領域を検出する対象となるデジタルマンモグラフィ画像に対して、画素数を減少させる縮小処理を行う（ステップＳ１）。この縮小処理により、デジタルマンモグラフィ画像中の微細なノイズが除去される。また、この縮小処理により、後続の画像処理の速度を速くすることができる。デジタルマンモグラフィ画像の元のサイズは、例えば、約２０００～１００００画素である。デジタルマンモグラフィ画像の縮小処理後のサイズは、例えば、約２５６～５１２画素である。 First, the preprocessing unit 2 performs a reduction process for reducing the number of pixels on a digital mammography image that is a target for detecting a breast region (step S1). By this reduction processing, fine noise in the digital mammography image is removed. Moreover, the speed of subsequent image processing can be increased by this reduction processing. The original size of the digital mammography image is, for example, about 2000 to 10,000 pixels. The size of the digital mammography image after the reduction process is, for example, about 256 to 512 pixels.
　次いで、エッジ強調画像生成部３が、縮小処理後のデジタルマンモグラフィ画像に基づいて乳房領域Ｅ１のエッジ強調画像（図３）を生成する（ステップＳ２～Ｓ４）。 Next, the edge-enhanced image generation unit 3 generates an edge-enhanced image (FIG. 3) of the breast region E1 based on the digital mammography image after the reduction process (Steps S2 to S4).
　より具体的には、スキンライン強調画像生成部３１が、縮小処理後のデジタルマンモグラフィ画像に対して１次微分フィルタであるソーベルフィルタ処理を行うことにより乳房領域Ｅ１のスキンラインＳＬを強調したスキンライン強調画像（図２）を生成する（ステップＳ２）。また、このとき、スキンライン強調画像生成部３１が、スキンライン強調画像（図２）に対して二値化処理を行い、スキンラインＳＬだけではなく、乳房領域Ｅ１内の構造物も強調されたスキンライン強調画像（図３）を生成する。 More specifically, the skin line-enhanced image generation unit 31 performs a Sobel filter process that is a first-order differential filter on the digital mammography image after the reduction process, thereby enhancing the skin line SL of the breast region E1. A line-enhanced image (FIG. 2) is generated (step S2). At this time, the skinline enhanced image generation unit 31 performs binarization processing on the skinline enhanced image (FIG. 2), and not only the skinline SL but also the structure in the breast region E1 is enhanced. A skinline enhanced image (FIG. 3) is generated.
　その後、構造物強調画像生成部３２が、縮小処理後のデジタルマンモグラフィ画像に対して２次微分フィルタであるラプラシアンフィルタ処理を行うことにより乳房領域Ｅ１内の構造物（乳腺、脂肪等）を強調した構造物強調画像（図４）を生成する（ステップＳ３）。 Thereafter, the structure-enhanced image generating unit 32 emphasizes structures (mammary gland, fat, etc.) in the breast region E1 by performing Laplacian filter processing that is a second-order differential filter on the digital mammography image after reduction processing. A structure emphasized image (FIG. 4) is generated (step S3).
　その後、強調画像合成部３３が、スキンライン強調画像（図３）と構造物強調画像（図４）とを合成することによりエッジ強調画像（図５）を生成する（ステップＳ４）。 Thereafter, the enhanced image composition unit 33 synthesizes the skinline enhanced image (FIG. 3) and the structure enhanced image (FIG. 4) to generate an edge enhanced image (FIG. 5) (step S4).
　次いで、基準画像生成部４が、エッジ強調画像（図５）に対してクロージング処理を行うことにより基準画像（図６）を生成する（ステップＳ５）。 Next, the reference image generation unit 4 generates a reference image (FIG. 6) by performing a closing process on the edge enhanced image (FIG. 5) (step S5).
　次いで、乳房領域検出部５が、基準画像（図６）に基づいて乳房領域を検出する（ステップＳ６～Ｓ８）。 Next, the breast region detection unit 5 detects a breast region based on the reference image (FIG. 6) (steps S6 to S8).
　より具体的には、乳房領域検出部５が、デジタルマンモグラフィ画像に対して濃度値の閾値を変更し、基準画像（図６）に最も近い二値化画像となる閾値を探索する（ステップＳ６）。例えば、乳房領域検出部５が、濃度値の閾値を順次増加させ、それにより得られる二値化画像と基準画像とを順次比較して、重複しない画素数が最も少なくなる閾値を探索する。この処理の例として図１４にフローチャートを示す。ステップＳ１０３において、基準画像であるｉ１と閾値ｎにおける二値化画像であるｉ２を画素毎にＸＯＲすると異なる領域にのみビットが立つので、「ビットが立つ領域が多い＝基準画像に似ていない」、「ビットが立つ領域が少ない＝基準画像に似ている」ということになる。 More specifically, the breast region detection unit 5 changes the threshold value of the density value for the digital mammography image, and searches for a threshold value that becomes a binary image closest to the reference image (FIG. 6) (step S6). . For example, the breast region detection unit 5 sequentially increases the threshold value of the density value, sequentially compares the binarized image obtained thereby and the reference image, and searches for the threshold value that minimizes the number of non-overlapping pixels. As an example of this processing, a flowchart is shown in FIG. In step S103, when the OR of the reference image i1 and the binarized image i2 at the threshold value n is XORed for each pixel, a bit is set only in a different area, so that “there are many areas where bits are set = not similar to the reference image”. , “There are few areas where bits are set = similar to the reference image”.
　その後、乳房領域検出部５が、探索した閾値に基づく二値化画像を乳房領域候補画像（図７）として取得する（ステップＳ７）。 Thereafter, the breast region detection unit 5 acquires a binarized image based on the searched threshold value as a breast region candidate image (FIG. 7) (step S7).
　その後、乳房領域検出部５が、当該取得した乳房領域候補画像（図７）に対してラベリング処理を行い、最も面積の大きい領域Ｅ１１を乳房領域として検出する（ステップＳ８）。 Thereafter, the breast region detection unit 5 performs a labeling process on the acquired breast region candidate image (FIG. 7), and detects the region E11 having the largest area as a breast region (step S8).
　本発明の実施形態に係る乳房領域検出方法によれば、濃度値のヒストグラムを使用せずに乳房領域を検出するようにしているので、撮影条件等による影響を抑えて、乳房領域Ｅ１をより正確に検出することができる。 According to the breast region detection method according to the embodiment of the present invention, since the breast region is detected without using the histogram of density values, the influence of the imaging condition or the like is suppressed, and the breast region E1 is more accurately detected. Can be detected.
　また、本発明の実施形態に係る乳房領域検出方法によれば、自動的に作成可能な基準画像（図６）に基づいて乳房領域を検出するようにしているので、人為的な検出ミスを抑えることができる。 In addition, according to the breast region detection method according to the embodiment of the present invention, since a breast region is detected based on a reference image (FIG. 6) that can be automatically created, an artificial detection error is suppressed. be able to.
　なお、前記では、スキンライン強調画像（図３）と構造物強調画像（図４）とを合成することによってエッジ強調画像（図５）を生成するようにしたが、本発明はこれに限定されない。例えば、スキンライン強調画像（図３）において、後続のクロージング工程で乳房領域Ｅ１内を特定の色で埋められる程度に、乳房領域Ｅ１の構造物が十分に表示されている場合には、当該スキンライン強調画像をエッジ強調画像としてもよい。同様に、構造物強調画像（図４）において、後続の工程で乳房領域Ｅ１内を特定の色で埋められる程度に、乳房領域Ｅ１の構造物が十分に表示されている場合には、当該構造物強調画像をエッジ強調画像としてもよい。但し、２種類以上の強調画像を合成してエッジ強調画像（図５）を生成するようにすることで、撮影条件等の影響を受けて適切なエッジ強調画像が得られないという事態が発生することをより確実に抑えることができる。 In the above description, the edge-enhanced image (FIG. 5) is generated by synthesizing the skinline-enhanced image (FIG. 3) and the structure-enhanced image (FIG. 4). However, the present invention is not limited to this. . For example, in the skinline enhanced image (FIG. 3), when the structure of the breast region E1 is sufficiently displayed to the extent that the breast region E1 is filled with a specific color in the subsequent closing process, the skin is displayed. The line enhanced image may be an edge enhanced image. Similarly, in the structure-emphasized image (FIG. 4), when the structure of the breast region E1 is sufficiently displayed to the extent that the breast region E1 is filled with a specific color in the subsequent process, the structure is displayed. The object enhanced image may be an edge enhanced image. However, when an edge-enhanced image (FIG. 5) is generated by combining two or more types of emphasized images, a situation occurs in which an appropriate edge-enhanced image cannot be obtained due to the influence of shooting conditions and the like. This can be suppressed more reliably.
　また、前記では、ソーベルフィルタ処理及びラプラシアンフィルタ処理を行うことによってエッジ強調画像（図５）を生成するようにしたが、本発明はこれに限定されない。エッジ強調画像は、後続の工程で乳房領域Ｅ１内を特定の色で埋められる程度に、乳房領域Ｅ１の構造物が十分に表示されている画像であればよい。すなわち、他の１種類以上の画像処理を行うことによって、エッジ強調画像（図５）を生成するようにしてもよい。例えば、ソーベルフィルタ処理に替えて他の１次微分フィルタ処理（例えば、プレヴィット（Prewitt）フィルタ処理等）を行ってもよい。また、ラプラシアンフィルタ処理に替えて他の２次微分フィルタ処理を行ってもよい。 In the above description, the edge-enhanced image (FIG. 5) is generated by performing the Sobel filter process and the Laplacian filter process, but the present invention is not limited to this. The edge-enhanced image may be an image in which the structure of the breast region E1 is sufficiently displayed so that the breast region E1 is filled with a specific color in the subsequent process. That is, the edge-enhanced image (FIG. 5) may be generated by performing one or more other types of image processing. For example, instead of the Sobel filter process, another first-order differential filter process (for example, a Prewitt filter process or the like) may be performed. Further, instead of the Laplacian filter process, another secondary differential filter process may be performed.
　また、前記では、図６に示すように、クロージング処理によって乳房領域Ｅ１を特定の色で完全に埋めるものとしたが、実際には、図１２に示すように、クロージング処理では乳房領域Ｅ１を特定の色で完全に埋められないことも想定される。この場合、他の処理によって、又はクロージング処理と他の処理とを併用することによって、乳房領域Ｅ１を特定の色で完全に埋めるようにしてもよい。 In the above description, as shown in FIG. 6, the breast region E1 is completely filled with a specific color by the closing process. However, actually, as shown in FIG. 12, the breast region E1 is specified by the closing process. It is also assumed that it is not completely filled with the color. In this case, the breast region E1 may be completely filled with a specific color by another process or by using a closing process and another process in combination.
　また、前記では、基準画像（図６）に最も近い二値化画像を乳房領域候補画像としたが、本発明はこれに限定されない。例えば、乳房領域内に図１２に示すような穴（特定の色で完全に埋められていない部分）が無いなどの所定の基準（例えば、後述するエラーチェックの内容）を満たせば、基準画像を乳房領域候補画像としてもよい。すなわち、エラーチェックを図１１のステップＳ５の後に行って、所定の基準を満たせば、基準画像から直接的に乳房領域を検出するようにしてもよい。 In the above description, the binarized image closest to the reference image (FIG. 6) is the breast region candidate image, but the present invention is not limited to this. For example, if a predetermined standard (for example, the contents of error check described later) such as the absence of a hole (a part not completely filled with a specific color) as shown in FIG. It may be a breast region candidate image. That is, an error check may be performed after step S5 in FIG. 11 to detect a breast region directly from the reference image if a predetermined reference is satisfied.
　なお、乳房領域内の病変が非常に微細である場合、乳房領域の検出の正確性が十分ではないと、読影医が当該病変を見落としてしまう可能性がある。このため、図１１を用いて説明したステップＳ８の後に、検出した乳房領域が所定の基準を満たしているか否かをチェックするエラーチェックを行うことが好ましい。この処理の例として図１５にエラーチェック処理の設定画面を示す。 If the lesion in the breast region is very fine, the interpretation doctor may miss the lesion if the accuracy of detection of the breast region is not sufficient. For this reason, it is preferable to perform an error check after step S8 described with reference to FIG. 11 to check whether the detected breast region satisfies a predetermined criterion. As an example of this processing, FIG. 15 shows a setting screen for error check processing.
　「所定の基準」とは、例えば、乳房領域内の穴の有無（図１２参照）、乳房領域のサイズ、乳房領域の垂直位置のずれ、乳房領域の複雑度である。 The “predetermined reference” is, for example, the presence or absence of a hole in the breast region (see FIG. 12), the size of the breast region, the deviation of the vertical position of the breast region, and the complexity of the breast region.
　すなわち、乳房領域内に穴がある場合、乳房領域の検出が不正確であるため、エラーと判定する。また、乳房領域の面積が画面の総面積に対して所定の割合以上である場合、乳房領域のサイズが大きすぎるため、エラーと判定する。また、乳房領域は画面の中央側に存在すべきであるので、乳房領域の重心と画面の中心とが垂直方向に離れる距離が所定の距離以上である場合、エラーと判定する。また、乳房領域の複雑度（＝輪郭の長さ×輪郭の長さ／乳房領域の面積）が大きい場合、乳房領域の輪郭が滑らかでないので、エラーと判定する。 That is, if there is a hole in the breast region, it is determined that there is an error because the detection of the breast region is inaccurate. Further, when the area of the breast region is equal to or larger than a predetermined ratio with respect to the total area of the screen, the size of the breast region is too large, and it is determined as an error. Since the breast region should be present on the center side of the screen, if the distance between the center of gravity of the breast region and the center of the screen in the vertical direction is equal to or greater than a predetermined distance, it is determined as an error. Further, when the complexity of the breast region (= contour length × contour length / breast region area) is large, the contour of the breast region is not smooth, and it is determined as an error.
　前記エラーチェックによってエラーと判定した場合には、例えば、画面にエラー表示する、或いは、画面に元のデジタルマンモグラフィ画像を表示する。これにより、乳房領域の検出の正確性が十分ではないことを読影医に知らせることができる。 If the error check determines that an error has occurred, for example, an error is displayed on the screen, or the original digital mammography image is displayed on the screen. As a result, it is possible to inform the interpretation doctor that the accuracy of detection of the breast region is not sufficient.
　なお、本実施形態に係る乳房領域検出システム１の各部２～５の機能は、ソフトウェアで実現されてもよい。当該ソフトウェアは、ダウンロード等により提供されてもよい。また、このソフトウェアは、ＣＤ-ＲＯＭなどのコンピュータ可読記憶媒体に記録されて提供されてもよい。なお、このことは、本明細書における他の実施形態についても該当する。なお、本実施形態に係る乳房領域検出システム１を実現するソフトウェアは、以下のようなプログラムである。すなわち、当該プログラムは、コンピュータを、デジタルマンモグラフィ画像に基づいて乳房領域のエッジ強調画像を生成するエッジ強調画像生成部と、前記エッジ強調画像に対して乳房領域を背景領域とは異なる色で埋める処理を行うことにより基準画像を生成する基準画像生成部と、前記基準画像に基づいて乳房領域を検出する乳房領域検出部として機能させるためのプログラムである。 Note that the functions of the units 2 to 5 of the breast region detection system 1 according to the present embodiment may be realized by software. The software may be provided by downloading or the like. The software may be provided by being recorded on a computer-readable storage medium such as a CD-ROM. This also applies to other embodiments in this specification. In addition, the software which implement | achieves the breast area | region detection system 1 which concerns on this embodiment is the following programs. That is, the program includes an edge-enhanced image generation unit that generates an edge-enhanced image of a breast region based on a digital mammography image, and a process of filling the edge-enhanced image with a color different from a background region. Is a program for functioning as a reference image generation unit that generates a reference image by performing and a breast region detection unit that detects a breast region based on the reference image.
　本発明は、添付図面を参照しながら好ましい実施の形態に関連して充分に記載されているが、この技術に熟練した人々にとっては種々の変形や修正は明白である。そのような変形や修正は、添付した請求の範囲による本発明の範囲から外れない限りにおいて、その中に含まれると理解されるべきである。 Although the present invention has been fully described in connection with preferred embodiments with reference to the accompanying drawings, various changes and modifications will be apparent to those skilled in the art. Such changes and modifications are to be understood as being included therein, so long as they do not depart from the scope of the present invention according to the appended claims.
　本発明は、乳房領域をより正確に検出することができるので、例えば、デジタルマンモグラフィ画像から乳房領域を検出する乳房領域検出システム、乳房領域検出方法、及びプログラムに有用である。 Since the breast region can be detected more accurately, the present invention is useful for, for example, a breast region detection system, a breast region detection method, and a program that detect a breast region from a digital mammography image.
　　１　　乳房領域検出システム
　　２　　前処理部
　　３　　エッジ強調画像生成部
　　４　　基準画像生成部
　　５　　乳房領域検出部
　３１　　スキンライン強調画像生成部
　３２　　構造物強調画像生成部
　３３　　強調画像合成部
　ＳＬ　　スキンライン
　Ｅ１　　乳房領域
　Ｅ２　　背景領域 DESCRIPTION OF SYMBOLS 1 Breast area detection system 2 Pre-processing part 3 Edge emphasis image generation part 4 Reference image generation part 5 Breast area detection part 31 Skin line emphasis image generation part 32 Structure emphasis image generation part 33 Emphasis image composition part SL Skin line E1 Breast area E2 Background area

Claims (16)

  1. 　デジタルマンモグラフィ画像に基づいて乳房領域のエッジ強調画像を生成するエッジ強調画像生成部と、
    　前記エッジ強調画像に対して乳房領域を背景領域とは異なる色で埋める処理を行うことにより基準画像を生成する基準画像生成部と、
    　前記基準画像に基づいて乳房領域を検出する乳房領域検出部と、
    　を備える、乳房領域検出システム。     An edge-enhanced image generation unit that generates an edge-enhanced image of a breast region based on a digital mammography image;
    A reference image generation unit that generates a reference image by performing a process of filling a breast region with a color different from a background region on the edge-enhanced image;
    A breast region detection unit for detecting a breast region based on the reference image;
    A breast region detection system comprising:
  2. 　前記乳房領域検出部は、前記デジタルマンモグラフィ画像に対して濃度値の閾値を変更して、前記基準画像に最も近い二値化画像を乳房領域候補画像として取得し、当該乳房領域候補画像に基づいて乳房領域を検出する、請求項１に記載の乳房領域検出システム。 The breast region detection unit changes a threshold value of the density value with respect to the digital mammography image, acquires a binarized image closest to the reference image as a breast region candidate image, and based on the breast region candidate image The breast region detection system according to claim 1, wherein the breast region is detected.
  3. 　前記乳房領域検出部は、前記乳房領域候補画像に対してラベリング処理を行い、最も面積の大きい領域を乳房領域として検出する、請求項２に記載の乳房領域検出システム。 The breast region detection system according to claim 2, wherein the breast region detection unit performs a labeling process on the breast region candidate image and detects a region having the largest area as a breast region.
  4. 　前記エッジ強調画像生成部は、
    　前記乳房領域のスキンラインを強調したスキンライン強調画像を生成するスキンライン強調画像生成部と、
    　前記乳房領域内の構造物を強調した構造物強調画像を生成する構造物強調画像生成部と、
    　前記スキンライン強調画像と前記構造物強調画像とを合成することにより前記エッジ強調画像を生成する強調画像合成部と、
    　を備える、請求項１～３のいずれか１つに記載の乳房領域検出システム。     The edge-enhanced image generation unit
    A skinline-enhanced image generation unit that generates a skinline-enhanced image that emphasizes the skinline of the breast region;
    A structure-enhanced image generating unit that generates a structure-enhanced image in which the structure in the breast region is emphasized;
    An enhanced image combining unit that generates the edge enhanced image by combining the skinline enhanced image and the structure enhanced image;
    The breast region detection system according to any one of claims 1 to 3, further comprising:
  5. 　前記スキンライン強調画像生成部は、前記デジタルマンモグラフィ画像に対してソーベルフィルタ処理を行うことにより前記スキンライン強調画像を生成する、請求項４に記載の乳房領域検出システム。 The breast region detection system according to claim 4, wherein the skinline enhanced image generation unit generates the skinline enhanced image by performing a Sobel filter process on the digital mammography image.
  6. 　前記構造物強調画像生成部は、前記デジタルマンモグラフィ画像に対してラプラシアンフィルタ処理を行うことにより前記構造物強調画像を生成する、請求項４又は５に記載の乳房領域検出システム。 The breast region detection system according to claim 4 or 5, wherein the structure emphasized image generation unit generates the structure emphasized image by performing a Laplacian filter process on the digital mammography image.
  7. 　前記基準画像生成部は、前記エッジ強調画像に対してクロージング処理を行うことにより前記基準画像を生成する、請求項１～６のいずれか１つに記載の乳房領域検出システム。 The breast region detection system according to any one of claims 1 to 6, wherein the reference image generation unit generates the reference image by performing a closing process on the edge enhanced image.
  8. 　コンピュータを、
    　デジタルマンモグラフィ画像に基づいて乳房領域のエッジ強調画像を生成するエッジ強調画像生成部と、
    　前記エッジ強調画像に対して乳房領域を背景領域とは異なる色で埋める処理を行うことにより基準画像を生成する基準画像生成部と、
    　前記基準画像に基づいて乳房領域を検出する乳房領域検出部、
    　として機能させるためのプログラム。     Computer
    An edge-enhanced image generation unit that generates an edge-enhanced image of a breast region based on a digital mammography image;
    A reference image generation unit that generates a reference image by performing a process of filling a breast region with a color different from a background region on the edge-enhanced image;
    A breast region detection unit for detecting a breast region based on the reference image;
    Program to function as.
  9. 　前記乳房領域検出部は、前記デジタルマンモグラフィ画像に対して濃度値の閾値を変更して、前記基準画像に最も近い二値化画像を乳房領域候補画像として取得し、当該乳房領域候補画像に基づいて乳房領域を検出する、請求項８に記載のプログラム。 The breast region detection unit changes a threshold value of the density value with respect to the digital mammography image, acquires a binarized image closest to the reference image as a breast region candidate image, and based on the breast region candidate image The program according to claim 8, wherein a breast region is detected.
  10. 　前記乳房領域検出部は、前記乳房領域候補画像に対してラベリング処理を行い、最も面積の大きい領域を乳房領域として検出する、請求項９に記載のプログラム。 The program according to claim 9, wherein the breast region detection unit performs a labeling process on the breast region candidate image and detects a region having the largest area as a breast region.
  11. 　前記エッジ強調画像生成部は、
    　前記乳房領域のスキンラインを強調したスキンライン強調画像を生成するスキンライン強調画像生成部と、
    　前記乳房領域内の構造物を強調した構造物強調画像を生成する構造物強調画像生成部と、
    　前記スキンライン強調画像と前記構造物強調画像とを合成することにより前記エッジ強調画像を生成する強調画像合成部と、
    　を備える、請求項８～１０のいずれか１つに記載のプログラム。     The edge-enhanced image generation unit
    A skinline-enhanced image generation unit that generates a skinline-enhanced image that emphasizes the skinline of the breast region;
    A structure-enhanced image generating unit that generates a structure-enhanced image in which the structure in the breast region is emphasized;
    An enhanced image combining unit that generates the edge enhanced image by combining the skinline enhanced image and the structure enhanced image;
    The program according to any one of claims 8 to 10, comprising:
  12. 　前記スキンライン強調画像生成部は、前記デジタルマンモグラフィ画像に対してソーベルフィルタ処理を行うことにより前記スキンライン強調画像を生成する、請求項１１に記載のプログラム。 12. The program according to claim 11, wherein the skinline emphasized image generation unit generates the skinline emphasized image by performing Sobel filter processing on the digital mammography image.
  13. 　前記構造物強調画像生成部は、前記デジタルマンモグラフィ画像に対してラプラシアンフィルタ処理を行うことにより前記構造物強調画像を生成する、請求項１１又は１２に記載のプログラム。 The program according to claim 11 or 12, wherein the structure emphasized image generation unit generates the structure emphasized image by performing a Laplacian filter process on the digital mammography image.
  14. 　前記基準画像生成部は、前記エッジ強調画像に対してクロージング処理を行うことにより前記基準画像を生成する、請求項８～１３のいずれか１つに記載のプログラム。 14. The program according to claim 8, wherein the reference image generation unit generates the reference image by performing a closing process on the edge enhanced image.
  15. 　デジタルマンモグラフィ画像に基づいて乳房領域のエッジ強調画像を生成し、
    　前記エッジ強調画像に対して乳房領域を背景領域とは異なる色で埋める処理を行うことにより基準画像を生成し、
    　前記基準画像に基づいて乳房領域を検出する、
    　ことを含む、乳房領域検出方法。     Generate an edge-enhanced image of the breast region based on the digital mammography image,
    A reference image is generated by performing processing for filling the breast region with a color different from the background region on the edge-enhanced image,
    Detecting a breast region based on the reference image;
    A method of detecting a breast region.
  16. 　前記基準画像に基づいて乳房領域を検出することは、
    　前記デジタルマンモグラフィ画像に対して濃度値の閾値を変更して、前記基準画像に最も近い二値化画像を乳房領域候補画像として取得し、
    　前記乳房領域候補画像に基づいて乳房領域を検出する、
    　ことを含む、請求項１５に記載の乳房領域検出方法。     Detecting a breast region based on the reference image comprises:
    By changing a threshold value of density value for the digital mammography image, a binary image closest to the reference image is obtained as a breast region candidate image,
    Detecting a breast region based on the breast region candidate image;
    The breast region detection method according to claim 15, further comprising:
PCT/JP2017/029125 2016-08-10 2017-08-10 Breast region detecting system, breast region detecting method, and program WO2018030519A1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP2018533566A JP7010225B2 (en) 2016-08-10 2017-08-10 Breast region detection system, breast region detection method, and program

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2016157603 2016-08-10
JP2016-157603 2016-08-10

Publications (1)

Publication Number Publication Date
WO2018030519A1 true WO2018030519A1 (en) 2018-02-15

Family

ID=61162214

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2017/029125 WO2018030519A1 (en) 2016-08-10 2017-08-10 Breast region detecting system, breast region detecting method, and program

Country Status (2)

Country Link
JP (1) JP7010225B2 (en)
WO (1) WO2018030519A1 (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP7452046B2 (en) 2020-01-31 2024-03-19 株式会社Ｊｖｃケンウッド Display control device, image display device, control method and control program

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH1099306A (en) * 1996-09-30 1998-04-21 Fuji Photo Film Co Ltd Method and device for detecting abnormal shading candidates
JP2004283281A (en) * 2003-03-20 2004-10-14 Fuji Photo Film Co Ltd Image processor and image processing method
JP2011104149A (en) * 2009-11-18 2011-06-02 Toshiba Corp Mammography apparatus

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2010017370A (en) * 2008-07-11 2010-01-28 Konica Minolta Medical & Graphic Inc Density adjusting device, density adjusting method and program
JP2013000370A (en) * 2011-06-17 2013-01-07 Fujifilm Corp Radiological image radiographing apparatus and method

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH1099306A (en) * 1996-09-30 1998-04-21 Fuji Photo Film Co Ltd Method and device for detecting abnormal shading candidates
JP2004283281A (en) * 2003-03-20 2004-10-14 Fuji Photo Film Co Ltd Image processor and image processing method
JP2011104149A (en) * 2009-11-18 2011-06-02 Toshiba Corp Mammography apparatus

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP7452046B2 (en) 2020-01-31 2024-03-19 株式会社Ｊｖｃケンウッド Display control device, image display device, control method and control program

Also Published As

Publication number Publication date
JPWO2018030519A1 (en) 2019-06-13
JP7010225B2 (en) 2022-01-26

Similar Documents

Publication Publication Date Title
US7720266B2 (en) Ultrasound image enhancement and speckle mitigation method
JP5645399B2 (en) X-ray image processing apparatus, X-ray image processing method, and computer program
KR101899866B1 (en) Apparatus and method for detecting error of lesion contour, apparatus and method for correcting error of lesion contour and, apparatus for insecting error of lesion contour
JP6564018B2 (en) Radiation image lung segmentation technology and bone attenuation technology
US9953441B2 (en) Processing metal artifact in CT image
CN110610498A (en) Mammary gland molybdenum target image processing method, system, storage medium and equipment
US20160180526A1 (en) Image processing apparatus, image processing method, image processing system, and non-transitory computer-readable storage medium
US20080080768A1 (en) Machine learning based triple region segmentation framework using level set on pacs
US9762773B2 (en) Image processing apparatus and method for increasing sharpness of images
JP2008194239A (en) Image processing apparatus and method for the same
JP6492553B2 (en) Image processing apparatus and program
CN113808054B (en) Method for repairing optic disc region of fundus image and related product
WO2018030519A1 (en) Breast region detecting system, breast region detecting method, and program
Shi et al. Tracing vocal fold vibrations using level set segmentation method
KR20200046775A (en) Surgical assist device and method for 3D analysis based on liver cancer area in CT image
US11311259B2 (en) Image processing apparatus, image processing method, and computer-readable medium
CN102073991A (en) Method and system for enhancing contrast of spatially-localized phenomena in mammography image
CN114299081B (en) Maxillary sinus CBCT image segmentation method, maxillary sinus CBCT image segmentation device, maxillary sinus CBCT storage medium and electronic equipment
Mazlan et al. Enhancement of retinal images for microaneurysms detection in diabetic retinopathy
JP7019104B2 (en) Threshold learning method
CN112884699B (en) Method and image processing device for segmenting image data and computer program product
JP2022181493A (en) Image processing apparatus, image processing method, and program
Eskandari et al. A semi-automated algorithm for segmentation of the left atrial appendage landing zone: application in left atrial appendage occlusion procedures
Stolojescu-Crisan et al. The improvement of the robust exemplar-based inpainting algorithm for SONAR images
Li et al. Fast and robust clinical triple-region image segmentation using one level set function

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17839583

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2018533566

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 17839583

Country of ref document: EP

Kind code of ref document: A1