JP2015112150A - Image processing system and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a technology which can, with high accuracy, extract an area where a target part is captured from a biological image obtained by radiography.SOLUTION: By using template matching, of a biological image, a target defined area where at least a part of a target part is captured and an undefined area where a part of the target part may be captured are detected. In so doing, a position of an area which captures a global area part of the target part and a position of an area which captures a local part included in the global part are detected and, according to a result of detection, a target candidate area including the target defined area and the undefined area are set with respect to the biological image. On the basis of at least either a positional relation between the detected global area and the local area or a correlation value between the local area and a template, it is determined whether or not to set a target candidate area in accordance with the result of detection related to the local area. After that, on the basis of information about the target defined area and the undefined area, a target area related to the target part of the biological image is detected.

Description

  The present invention relates to an image processing apparatus and a program.

  In a medical field, an image in which an affected part included in a viscera and a skeleton is captured by imaging using X-rays (also referred to as X-ray imaging), and various examinations and diagnoses using the images are performed.

  In recent years, by applying digital technology, an image (also referred to as a dynamic image) that captures the motion of the affected area by X-ray imaging can be obtained relatively easily. For example, a dynamic image in which a region (also referred to as a target region) including a region (also referred to as a target region) to be inspected and diagnosed is acquired by imaging using a semiconductor image sensor such as an FPD (flat panel detector). Is done. Thereby, a dynamic image can be acquired with a relatively simple configuration without using a contrast agent. And it becomes possible to make a diagnosis by analyzing the movement of the target part, which could not be performed by the diagnosis using the still image obtained by the conventional X-ray imaging.

  For example, in an image that captures the movement of the chest obtained by X-ray imaging (also referred to as a chest X-ray dynamic image), changes in concentration in the lung field are analyzed, so that ventilation and blood flow in the lung field (pulmonary blood flow) Quantitative changes related to the state of (also called). Specifically, for example, by identifying an abnormal portion of ventilation from a quantitative change regarding the state of ventilation, analysis of chronic obstructive disease (COPD) or the like becomes possible. Further, for example, by identifying an abnormal portion of the pulmonary blood flow from a quantitative change regarding the state of the pulmonary blood flow, analysis of pulmonary embolism or the like becomes possible.

  However, in order to perform analysis and diagnosis on the movement of the target part in the diagnosis using the dynamic image, it is necessary to accurately extract a region that captures the target part (also referred to as the target region) from the dynamic image.

  By the way, as a technique for extracting a target region from an X-ray image that captures an affected area, a region that captures a lung as a target region (also referred to as a lung field region) is extracted by binarization processing using a certain threshold value and labeling. Techniques have been proposed (for example, Patent Documents 1 and 2). Further, a low-frequency image is generated from the X-ray image by filtering, a luminance change image indicating a change in luminance is generated from the low-frequency image, and the boundary of the region of interest is based on the pixel values of the luminance change image and the low-frequency image. A technique for selecting a pixel located at is proposed (for example, Patent Document 3).

JP 09-035043 A JP-A-11-151232 JP 2011-255033 A

  However, in capturing a dynamic image that captures the target region, it is necessary to keep the amount of exposure required for capturing each frame low so that the amount of exposure to the affected area caused by X-rays does not increase. It is easy to deteriorate by.

  For this reason, for example, in the techniques of Patent Documents 1 and 2 described above, a simple binarization process is used to extract a lung field region from an X-ray image. It is difficult to accurately extract the target area. Further, in the technique of Patent Document 3, if the luminance change at the boundary portion of the region of interest is small, it is difficult to accurately detect the contour of the region of interest such as the lung region, and other structures such as ribs, clavicles, and breasts. There is a possibility that the contour is erroneously detected as the boundary of the region of interest. That is, there is a possibility that the region of interest is extracted too large or too small. Therefore, it is difficult to accurately extract the region of interest from the image. Such a problem is common to living body images obtained by capturing the inside of a body obtained by X-ray imaging.

  The present invention has been made in view of the above problems, and provides a technique capable of accurately extracting a region in which a target portion is captured from a biological image capturing a body structure obtained by X-ray imaging. The purpose is to do.

  In order to solve the above-described problem, an image processing apparatus according to a first aspect includes an acquisition unit that acquires a biological image in which a structure inside the body is captured by X-ray irradiation, and template matching for the biological image. To detect a target confirmed region in which at least a part of the target part is captured and an undetermined region in which a part of the target part may be captured in the biological image. A detection unit; and a second detection unit that detects a target region from the inside of the target region to an outline based on information on the target determination region and the unconfirmed region, and 1 detection part detects the position of the global region where the global part of the target part in the living body image was captured by template matching using the global template related to the global part. A local region that is included in the global portion of the biological image and in which a local portion smaller than the global portion is captured is detected by template matching using a local template related to the local portion A candidate region setting unit configured to set a target candidate region including the target fixed region and the undetermined region with respect to the biological image according to detection results by the global region detector and the local region detector. And a relative positional relationship between the position of the global region detected by the global region detection unit and the position of the local region detected by the local region detection unit, and template matching in the local region detection unit The candidate region setting is based on at least one of correlation values between the local region and the local template. Having a determination section for determining whether or not to set the target candidate region in accordance with a detection result obtained by the local region detecting unit in section.

  The image processing apparatus according to the second aspect is the image processing apparatus according to the first aspect, wherein the global area is selected by the candidate area setting unit for the area where the local part of the biological image is captured. The undetermined region set in the candidate region setting unit according to the detection result by the local region detection unit is narrower than the undetermined region set in accordance with the detection result by the region detection unit.

  An image processing device according to a third aspect is the image processing device according to the first or second aspect, wherein the candidate region setting unit responds to a detection result by the global region detection unit and the local region detection unit. Then, a non-target fixed region where the target part is not captured is set for the biological image.

  An image processing apparatus according to a fourth aspect is the image processing apparatus according to any one of the first to third aspects, wherein the local part is a part where a difference in a periodic state changes between individuals , And at least one of the portions where the shape is different between individuals.

  An image processing device according to a fifth aspect is the image processing device according to the fourth aspect, wherein the target portion is a structure relating to a lung field, and the local portion includes an aortic arch, a clavicle, a heart, It includes at least one portion of the diaphragm and heel angle.

  An image processing device according to a sixth aspect is the image processing device according to any one of the first to fifth aspects, wherein the determination unit detects the global area in at least one preset direction. If the amount of deviation between the position of the global region detected by the unit and the position of the local region detected by the local region detection unit is out of a preset value range, the candidate region setting unit Setting of the target candidate region according to the detection result by the local region detection unit is prohibited.

  An image processing apparatus according to a seventh aspect is the image processing apparatus according to any one of the first to sixth aspects, wherein the determination unit determines that the correlation value is out of a preset range. For example, the setting of the target candidate region according to the detection result by the local region detection unit in the candidate region setting unit is prohibited.

  The program according to the eighth aspect is executed by a control unit included in the image processing apparatus, thereby causing the image processing apparatus to function as the image processing apparatus according to any one of the first to seventh aspects.

  Whether or not the image processing apparatus according to any one of the first to seventh aspects sets the target fixed region and the undefined region related to the target part corresponding to the local region according to the detection result of the local region. Therefore, the region where the target part is captured can be extracted with high accuracy from the biological image capturing the internal structure obtained by X-ray imaging.

  According to the image processing apparatus according to the second aspect, the undetermined region corresponding to the local region is set to be narrower, so that the region where the target region is captured is detected excessively and the target region is captured. It is difficult for undetected areas to be detected too narrowly. That is, a region in which a target part is captured from a biological image can be stably and accurately extracted.

  In the image processing apparatus according to any of the fourth and fifth aspects, a local region is detected with respect to a portion where a shift is likely to occur between the position of the global portion captured in the biological image and the position of the detected global region. Therefore, a region where the target part is captured from the biological image can be extracted with high accuracy.

  In the image processing apparatus according to any of the sixth and seventh aspects, since the setting of the target definite region and the undetermined region related to the target region corresponding to the local region with low detection reliability is suppressed, A region that captures the target part can be extracted with high accuracy.

  According to the program according to the eighth aspect, the same effects as those of the image processing apparatus according to the first to seventh aspects can be obtained.

It is a block diagram which shows the structure of the imaging | photography system which concerns on one Embodiment. It is a block diagram which shows the functional structure of the image processing apparatus which concerns on one Embodiment. It is a figure which illustrates the dynamic image obtained by X-ray imaging. It is a figure for demonstrating the characteristic of the frame image which comprises a dynamic image. It is a figure for demonstrating the characteristic of the frame image which comprises a dynamic image. It is a figure for demonstrating the characteristic of the frame image which comprises a dynamic image. It is a figure for demonstrating the process which correct | amends an image. It is a figure for demonstrating the process which correct | amends an image. It is a figure for demonstrating the process which correct | amends an image. It is a figure for demonstrating the process which correct | amends an image. It is a figure for demonstrating the process which correct | amends an image. It is a figure for demonstrating the process which correct | amends an image. It is a figure for demonstrating the process which correct | amends an image. It is a figure for demonstrating the process which correct | amends an image. It is a figure for demonstrating the process which correct | amends an image. It is a figure for demonstrating the process which correct | amends an image. It is a figure which illustrates the frame image used as the object of rough detection. It is a figure which shows an example of the detection result by rough detection typically. It is a block diagram which shows the functional structure of a rough detection part. It is a figure which shows the example of a setting of a global region and a local region. It is a figure which shows typically the one aspect | mode of the change of the size in the vertical direction of a lung field. It is a figure which shows typically the one aspect | mode of the change of the size in the vertical direction of a lung field. It is a figure which shows typically the one aspect | mode of the change of the size in the vertical direction of a lung field. It is a figure which shows typically the one aspect | mode of the individual difference in the shape of a diaphragm and a heart. It is a figure which shows typically the one aspect | mode of the individual difference in the shape of a diaphragm and a heart. It is a figure which shows typically the one aspect | mode of the individual difference in the shape of a diaphragm and a heart. It is a figure which shows typically the one aspect | mode of the individual difference in the shape of a diaphragm and a heart. It is a figure for demonstrating the production | generation of a diaphragm template. It is a figure for demonstrating the production | generation of a diaphragm template. It is a figure for demonstrating the production | generation of a diaphragm template. It is a figure for demonstrating the production | generation of a diaphragm template. It is a figure for demonstrating the production | generation of partial division information. It is a figure for demonstrating the production | generation of partial division information. It is a figure for demonstrating the production | generation of partial division information. It is a figure for demonstrating the production | generation of partial division information. It is a figure for demonstrating the production | generation of partial division information. It is a figure which shows an example of the area | region division information produced | generated by integration of partial division information. It is a figure which illustrates the frame image used as the object of precise detection. It is a figure which shows typically an example of the detection result by a precise detection. It is a flowchart which shows the operation | movement flow in a target area | region extraction process. It is a flowchart which shows the operation | movement flow in a target area | region extraction process.

  Hereinafter, an embodiment of the present invention will be described with reference to the drawings. In the drawings, parts having the same configuration and function are denoted by the same reference numerals, and redundant description is omitted in the following description. Further, the drawings are schematically shown, and the correlation between the sizes and positions of the images respectively shown in the different drawings is not necessarily accurately described. 4 and 16 are attached with an XY coordinate system in which the upper left of the image is the origin, the right direction is the X direction, and the lower direction is the Y direction.

<(1) Overview of shooting system>
FIG. 1 is a block diagram illustrating a configuration of an imaging system 100 according to an embodiment. The imaging system 100 captures an image (also referred to as a biological image) in which a structure in the body of an animal including a human being is taken as an object, and performs various image processing on the biological image. . In addition, as a structure in a body, the skeleton containing the organ contained in the area | region of the chest containing lungs and indirect etc. is mentioned, for example.

  In the present embodiment, for example, the imaging system 100 is a system for a human body, and a region (also referred to as a target region) that is a main target of imaging is a person (also referred to as a subject) M1 that is an inspection target. Any structure related to the human lung field may be used. In the present embodiment, for example, radiation is X-rays, and a living body image is obtained by imaging the human body of the subject M1 for medical treatment or medicine, or measuring the human body of the subject M1. It is only necessary that the image has been processed (also referred to as a medical image).

  As shown in FIG. 1, the imaging system 100 includes an imaging device 1, an imaging control device (imaging console) 2, and an image processing device (diagnosis console) 3. The imaging device 1 and the imaging control device 2 and the imaging control device 2 and the image processing device 3 are connected to each other via a communication line so as to be able to transmit and receive data. The communication line connecting the image capturing apparatus 1 and the image capturing control apparatus 2 may be a wired line such as a communication cable or a wireless line. Further, the communication line connecting the imaging control apparatus 2 and the image processing apparatus 3 may be either a wired line or a wireless line, and may be a network line NTW such as a LAN (Local Area Network) line.

  In the imaging system 100, for example, a medical image format and a communication protocol between devices that handle images conform to a standard called DICOM (Digital Image and Communications in Medicine).

<(2) Shooting device>
The imaging apparatus 1 is configured by, for example, an apparatus (also referred to as an X-ray imaging apparatus) that acquires a medical image capturing a body structure by irradiation with X-rays as radiation. In the imaging device 1, for example, imaging (also referred to as dynamic imaging) is performed on the dynamics of the structure in the chest of the subject M <b> 1 associated with breathing or the like. In the dynamic imaging, for example, a two-dimensional distribution of X-rays transmitted through the subject M1 is detected in time sequence while X-rays are repeatedly irradiated to the chest of the subject M1. Thereby, a plurality of images (also referred to as frame images) are acquired in time sequence. These series of a plurality of frame images show how the physical state (geometric shape, blood flow concentration, etc.) in the region including the lung of the subject M1 (also referred to as the lung field region) changes over time. A dynamic image (moving image) captured sequentially is constructed.

  As shown in FIG. 1, an imaging apparatus 1 includes an irradiation unit (X-ray source) 11, an irradiation control device 12, an imaging unit (X-ray detection unit) 13, a reading control device 14, a cycle detection sensor 15, and a cycle detection device. 16 is provided.

  The irradiation unit 11 irradiates the subject M1 with X-rays according to the control of the irradiation control device 12. The irradiation control device 12 is connected to the imaging control device 2 and controls the irradiation unit 11 based on information indicating the X-ray irradiation conditions input from the imaging control device 2.

  The imaging unit 13 is configured by, for example, a semiconductor image sensor such as an FPD (flat panel detector), and digitally radiates X-rays irradiated from the irradiation unit 11 to the subject M1 and transmitted through the subject M1. Convert to signal.

  The reading control device 14 is connected to the imaging control device 2 and controls the switching unit of each pixel of the imaging unit 13 based on the reading condition input from the imaging control device 2. Thereby, the reading control device 14 sequentially reads digital signals corresponding to the charges accumulated in the respective pixels of the imaging unit 13, and corresponds to the two-dimensional intensity distribution of the X-rays received on the imaging unit 13. Image data is acquired. Then, the reading control device 14 outputs the acquired image data (also simply referred to as an image) to the imaging control device 2 as a frame image constituting a dynamic image. The reading conditions may be, for example, the frame rate, pixel size, image size, and the like. The frame rate may be the number of frame images acquired per second.

  Here, the irradiation control device 12 and the reading control device 14 are electrically connected to each other, and by transmitting and receiving a synchronization signal to each other, the irradiation operation of the X-ray by the irradiation unit 11 and the reading control device 14 are performed. Synchronize with image data reading operation.

  The cycle detection sensor 15 is configured by, for example, a digital camera or the like, and acquires a dynamic image in which the movement of the chest of the subject M1 is captured by photographing the appearance of the chest of the subject M1.

  The cycle detection device 16 analyzes the dynamic image acquired by the cycle detection sensor 15 to detect the breathing cycle time of the subject M1 and outputs the detected time to the control unit 21 of the imaging control device 2. In accordance with this information, for example, an aspect in which the imaging control device 2 changes the irradiation condition, the reading condition, and the like can be considered.

<(3) Shooting control device>
The imaging control device 2 controls the X-ray imaging and frame image reading operations of the imaging device 1 by outputting information indicating the irradiation condition, the reading condition, and the like to the imaging device 1. In the imaging control device 2, for example, an image acquired by the imaging device 1 is appropriately displayed on the display unit 24. Thereby, the positioning including the position and posture of the subject M1 and the suitability for image diagnosis can be confirmed by the imaging technician.

  As shown in FIG. 1, the imaging control device 2 includes a control unit 21, a storage unit 22, an operation unit 23, a display unit 24, and a communication unit 25 that are connected to each other via a bus 26 so as to be able to exchange data. Configured.

  The control unit 21 includes, for example, a processor and a memory. The processor is, for example, a CPU (Central Processing Unit), and the memory may be, for example, a volatile RAM (Random Access Memory). For example, the processor reads out a program stored in the storage unit 22 in accordance with an operation of the operation unit 23, expands the program in the memory, and executes various processes according to the expanded program. The operation of each part of the apparatus 1 and the imaging control apparatus 2 is controlled.

  The storage unit 22 includes, for example, a nonvolatile semiconductor memory or a hard disk. The storage unit 22 stores various programs executed by the control unit 21 and various data necessary for execution of processing. Here, the various data includes data indicating parameters and the like necessary for execution of processing according to the program, and data generated at least temporarily as a result of the arithmetic processing.

  The operation unit 23 includes, for example, a pointing device such as a keyboard and a mouse. In the operation unit 23, a signal (also referred to as an instruction signal) input in response to an operation on a keyboard, a mouse, or the like is output to the control unit 21. The operation unit 23 may employ a configuration such as a touch panel.

  The display unit 24 includes a monitor such as an LCD (Liquid Crystal Display). In the display unit 24, in accordance with a signal input from the control unit 21, the content of an instruction input from the operation unit 23 and various data are displayed.

  The communication unit 25 includes, for example, a LAN adapter, a modem, or a terminal adapter (TA). The communication unit 25 controls transmission / reception of data to / from each device connected to the network line NTW.

<(4) Image processing device>
<(4-1) Schematic Configuration of Image Processing Device>
The image processing device 3 is an information processing device that performs information processing, acquires a dynamic image obtained by photographing in the photographing device 1 via the photographing control device 2, and performs image processing on the dynamic image, thereby obtaining a doctor. An image suitable for interpretation and diagnosis by the like is generated. The image is appropriately displayed on the display unit 34. In the present embodiment, the image processing device 3 captures a target site to be diagnosed from a plurality of frame images constituting a dynamic image in which a dynamic aspect of the chest mainly including the heart and both lungs is captured. Lung field regions as regions (also referred to as target regions) are respectively detected. In addition, the information processing apparatus should just have the functional structure similar to the structure of a personal computer (personal computer) or a personal computer, for example.

  As shown in FIG. 1, the image processing apparatus 3 includes a control unit 31, a storage unit 32, an operation unit 33, a display unit 34, and a communication unit 35 that are connected to each other via a bus 35 so as to be able to exchange data. Configured.

  The control unit 31 includes, for example, a processor 31a and a memory 31b. The processor 31a is, for example, a CPU (Central Processing Unit), and the memory 31b may be, for example, a volatile RAM (Random Access Memory). For example, the processor 31a reads out the program P1 stored in the storage unit 32 in accordance with the operation of the operation unit 33, expands it in the memory 31b, and executes various processes according to the expanded program P1. Thus, the operation of each unit of the image processing apparatus 3 is controlled. That is, the program P1 is executed by the control unit 31 included in the information processing apparatus, thereby causing the information processing apparatus to function as the image processing apparatus 3.

  The storage unit 32 includes, for example, a nonvolatile semiconductor memory or a hard disk. The storage unit 32 stores a program P1 executed by the control unit 31 and various data D1 necessary for execution of processing. Here, the various data D1 includes data indicating parameters and the like necessary for execution of processing according to the program, data generated at least temporarily as a result of the arithmetic processing, and the like. The program P1 includes, for example, a program for executing image processing. The program P1 is stored in the storage unit 32 in the form of readable program code, and the control unit 31 sequentially executes operations according to the program code.

  The operation unit 33 includes, for example, a pointing device such as a keyboard and a mouse. In the operation unit 33, a signal (also referred to as an instruction signal) that is input in response to an operation on the keyboard and the mouse is output to the control unit 31. The operation unit 33 may employ a configuration such as a touch panel.

  The display unit 34 includes, for example, a monitor such as an LCD (Liquid Crystal Display). In the display unit 34, in accordance with a signal input from the control unit 31, the contents of instructions input from the operation unit 33, various data, and various images are displayed.

  The communication unit 35 includes, for example, a LAN adapter, a modem, or a terminal adapter (TA). The communication unit 35 controls transmission / reception of data to / from each device connected to the network line NTW.

<(4-2) Functional Configuration of Image Processing Device>
FIG. 2 is a block diagram showing a functional configuration realized by the control unit 31 when the processor 31a and the like operate according to the program P1 in the image processing apparatus 3, together with other configurations. As shown in FIG. 2, the control unit 31 mainly includes an image acquisition unit 311, an image correction unit 312, a rough detection unit 313 as a first detection unit, a precision detection unit 314 as a second detection unit, and a target region. An extraction unit 315 is provided.

  In the image processing apparatus 3 according to the present embodiment, information processing (also referred to as target area extraction processing) including mainly the following processes A1 to A5 is executed. Thereby, the area | region which caught the object site | part can be accurately extracted from the series of frame images I0 (refer FIG. 3) by which the dynamic aspect of the internal structure obtained by X-ray imaging was caught.

  [Process A1] The image acquisition unit 311 acquires a series of a plurality of frame images I0 from the imaging control device 2.

  [Process A2] The frame image IC0 after correction (FIGS. 9 and 10) is performed by reducing the noise and the influence of the structure other than the target part by correcting the frame image I0 by the image correction unit 312. Reference) is obtained.

  [Process A3] The rough detection unit 313 performs a process (also referred to as a rough detection process) in which the position of the target portion is roughly detected from each corrected frame image IC0 using template matching. Note that a template normally used in template matching is, for example, an average image simply generated from a plurality of images, or an image (also referred to as a model image) expressed by a simple model expressed in black and white. For this reason, in template matching for X-ray images, if the number of templates used is small, due to the difference in shape such as the size of the lung field and heart between individuals and the difference in shape due to fluctuations in respiration and heartbeat, The lung field region is not accurately detected up to the contour. Therefore, the detection result by template matching is used as information indicating the approximate position of the target portion in each frame image IC0.

  [Process A4] Information regarding the position of the roughly detected target part is set as an initial value, and the precise detection unit 314 detects the position of the target part in detail from each frame image IC0 (also referred to as a precision detection process). ) Is performed.

  [Process A5] Based on the information regarding the position of the target part detected in detail, the target area is extracted from the frame image I0 by the target area extraction unit 315.

  Hereinafter, each part 311 to 315 will be sequentially described.

<(4-2-1) Image acquisition unit>
The image acquisition unit 311 acquires a biological image in which a structure inside the body is captured by X-ray irradiation. In this embodiment, the structure of the lung field is captured in the biological image. Then, the image acquisition unit 311 acquires image data related to a plurality of biological images constituting a dynamic image (also referred to as an X-ray dynamic image) obtained by X-ray imaging in the imaging apparatus 1 via the imaging control apparatus 2. The Specifically, for example, the image acquisition unit 311 acquires a plurality of image data stored in the storage unit 22 of the imaging control apparatus 2 via the communication unit 25, the network line NTW, and the communication unit 35 in order. As a result, the image acquisition unit 311 acquires a dynamic image in which a structure in the lung field as a structure inside the body is captured by X-ray irradiation.

  FIG. 3 is a diagram schematically showing an example of an X-ray dynamic image in which the dynamics of the structure accompanying breathing in the lung field of the subject M1 is captured. FIG. 3 illustrates a frame image I0 as a series of a plurality of biological images obtained by sequentially capturing images at a fixed capturing timing in one period related to the breathing cycle. The plurality of frame images I0 shown in FIG. 3 are a plurality of frame images I1, I2, I3,... Obtained by shooting at the shooting timing at time T of t1, t2, t3,. I10. Of the series of frame images I1 to I10, the first half frame images I1 to I5 capture the dynamics of expansion of the lung field due to the operation of lung inspiration, and the second half frame images I6 to I10 include The dynamics of the lung field contracting due to the exhalation of the lungs are captured. Note that the frame image I0 is obtained by converting a two-dimensional intensity distribution of X-rays into, for example, a two-dimensional distribution of pixel values (also referred to as gray values) indicating black and white.

<(4-2-2) Image Correction Unit>
The image correction unit 312 obtains a corrected frame image IC0 by performing correction on each frame image I0 constituting the X-ray dynamic image acquired by the image acquisition unit 311 and obtains the corrected frame image. IC0 is output to the coarse detection unit 313.

  4 to 6 are views for explaining the characteristics of the frame image I0. FIG. 4 shows a frame image I0, and FIG. 5 shows grayscale values in an area HR1 (refer to FIG. 4) in which the background of the subject M1 in the frame image I0 is captured. A histogram showing the frequency is shown. In the histogram, the horizontal axis indicates the gray value, and the vertical axis indicates the frequency of the gray value. Further, FIG. 6 shows a graph showing the gray value distribution in the vertically long region VR2 (see FIG. 4) including the lung field as the target site in the frame image I0. In the graph, the horizontal axis indicates the coordinate in the Y direction, and the vertical axis indicates the average of the gray value at each Y coordinate in the region VR2. As shown in FIGS. 4 to 6, the frame image I0 has the following problems Pm1 to Pm3.

  [Problem Pm1] As shown in FIG. 5, in the frame image I0, there is a lot of noise also in the background region HR1 related to the part where the subject M1 does not exist, and it is compared with an X-ray image as a general still image. Thus, the S / N ratio in the entire frame image I0 tends to be low. This is because the amount of exposure required for photographing each frame image is kept low so that the amount of exposure of the subject M1 due to X-rays does not increase in the acquisition of the X-ray dynamic image capturing the target region. If there is a lot of noise in this way, it becomes difficult to accurately extract the lung field region as the target region where the lung field as the target region is captured.

  [Problem Pm2] As shown in FIG. 6, in the frame image I0, the grayscale value related to the structure other than the lung (structure such as ribs and blood vessels) is superimposed on the grayscale value related to the lung. Since there is a variation in the gray value relating to, the outline of the lung field region as the target region is unclear. Thus, if the outline is unclear, it is difficult to accurately extract the lung field region as the target region where the lung field as the target region is captured.

  [Problem Pm3] In the frame image I0, the gray value in the vicinity of the region where the lateral angle is captured is closer to the gray value in the surrounding region captured outside the lung field than in the lung field, and the region where the heart is captured and the lungs. The border with the field is unclear. If the boundary is unclear in this way, it becomes difficult to accurately extract the lung field region as the target region where the lung field as the target region is captured. Note that the region where the heel lateral angle is captured corresponds to the region where both ends of the diaphragm are captured.

  In order to reduce the problems Pm1 to Pm3, as a process of reducing noise in the frame image I0 and reducing the influence of structures other than the lungs on the frame image I0, a correction process for clarifying the outline of the lung field region is performed. It is applied to the frame image I0.

  FIGS. 7 to 10 are diagrams for explaining the content of the correction process performed on the frame image I0 in the image correction unit 312. FIG.

  FIG. 7 shows a frame image I0. FIG. 8 shows the frame image I1 after the processing for reducing noise existing in the entire image (also referred to as noise reduction processing) is performed on the frame image I0. 9 and 10 show the frame image IC0 after the enhancement processing is performed on the frame image I1. FIG. 9 shows the frame image IC0 after the enhancement processing (also referred to as overall enhancement processing) is performed on the entire frame image I1. FIG. 10 shows the frame image IC0 after an enhancement process (also referred to as an in-area enhancement process) is performed on a preset area (also referred to as a predetermined area) of the frame image I1. 8 to 10, histograms are shown in which the horizontal axis is the gray value (pixel value) and the vertical axis is the frequency (unit:%) of the gray value (pixel value). .

  In the present embodiment, in the enhancement processing in the image correction unit 312, noise reduction processing and enhancement processing are performed in this order.

  Here, the contents of the noise reduction process and the enhancement process will be briefly described.

<(4-2-2-1) Noise reduction processing>
Examples of noise reduction processing include reduction processing. In the reduction process, for example, a selected area (also referred to as a selection area) is replaced with a representative pixel value (also referred to as a representative pixel value) related to a plurality of pixel values (light / dark values) constituting the selection area. The representative pixel value may be, for example, an average pixel value (gray value) in the selection area, or a minimum pixel value in the selection area. When the representative pixel value is an average pixel value in the selected area, the reduction process is realized by, for example, a filter process using an average value filter.

  11 to 15 are diagrams for explaining an example of the reduction process. FIG. 11 shows a state where one selection area R3 is set in the frame image I0. FIG. 12 shows an enlarged view of the selection area R3 in the frame image I0, and shows a pixel Pm having the smallest pixel value in the selection area R3. Here, a reduction process using the pixel value of the pixel Pm as the representative pixel value is performed for each selection area R3. Thereby, the pixel value (that is, noise) on the maximum value side in the lung field region captured in the image I0 is reduced.

  For example, FIG. 13 shows a state where an area R4 is set in the frame image I0. FIG. 14 shows an enlarged view of the area R4 in the frame image I0, and FIG. 15 shows an enlarged view of the area R4 in the frame image I1 after the reduction processing is performed on the frame image I0. Yes. As shown in FIGS. 14 and 15, by applying the enhancement process to the frame image I0, a small pixel value in the lung field region is enhanced, and noise in the lung field region is reduced. Note that smoothing may be performed when noise on the minimum pixel value side is enlarged.

  Note that the noise reduction process is not limited to the above reduction process, and other processes such as a process of cutting a high-frequency component of the frame image I0 using a bandpass filter or the like may be employed.

<(4-2-2) Enhancement processing>
In the overall enhancement process, for example, pixel values that are equal to or greater than a preset threshold value are replaced with relatively low pixel values that are equal to or less than the threshold value in accordance with a preset rule. Thereby, the influence of structures other than the lungs on the frame image I0 is reduced.

  As shown in FIG. 8 and FIG. 9, in the frame image IC0 subjected to the overall enhancement processing, the white region having a large gray value is relatively less than the frame image I1 subjected to the noise reduction processing, There are relatively many black areas with small gray values. That is, the whole area including the lung field area in the frame image I1 is reduced by the whole enhancement process, and the white area corresponding to the structure other than the lung is reduced, and the black area is emphasized.

  The threshold value in the enhancement process may be a fixed value set in advance or may be changed as appropriate. For example, a mode in which the threshold value is changed in accordance with the state of the frame image I1 that is the processing target can be considered. In this aspect, for example, regarding the histogram of the gray value of the frame image I1, the frequency is accumulated in order from the smaller gray value, and the gray value at which the accumulated frequency reaches a predetermined value is set as the threshold value. In addition, when the cumulative value of the frequency is expressed as a percentage, the predetermined value may be 80%, for example. Here, for example, a mode in which a threshold is set for each frame image I1 with respect to a series of a plurality of frame images I1 is conceivable. In addition, for example, a threshold is set for a series of a plurality of frame images I1 based on the first frame image I1, and the temporarily set threshold is applied to enhancement processing for the other frame images I1. Conceivable. If such an aspect is adopted, the time required for image processing by reducing the amount of calculation can be shortened.

  By the way, in the in-area enhancement process, the enhancement process is performed on a predetermined region in the frame image I1. For example, the predetermined region may be a desired fixed region, but may be changed according to the distribution (profile) of the gray value in the frame image I1 as the processing target in order to improve the robustness. .

  FIG. 16 is a diagram for explaining an example of a predetermined area setting method in the in-area enhancement process. In the right part of FIG. 16, the relationship between the Y coordinate and the accumulated value of the gray value (also referred to as the gray value accumulated value) at each Y coordinate of the frame image I1 is shown as a vertical profile. Further, in the lower part of FIG. 16, the relationship between the X coordinate and the accumulated value of the gray value at each X coordinate of the frame image I1 (light gray value) is shown as a profile in the horizontal direction.

  Here, for example, points (also referred to as change points) Px1, Px2, Py1, and Py2 at which the tendency of the accumulated gray value changes greatly in each profile are extracted. A rectangular region R5 in which the X coordinate is in the range from the change point Px1 to the change point Px2 and the Y coordinate is in the range from the change point Py1 to the change point Py2 is set as the predetermined region.

<(4-2-3) Coarse detection unit>
The coarse detection unit 313 uses the template matching for each frame image IC0 (see FIG. 17) as a biological image, and among the frame image IC0, the target confirmed region PR11 (see FIG. 18) and the undetermined region PR12. (See FIG. 18). Examples of template matching methods include known SAD (Sum of Absolute Differences) methods, NCC (Normalized Cross Correlation) methods, POC (Phase-Only Correlation) methods, and RIPOC (Rotation Invariant Phase-Only Correlation) methods. Is adopted. As a result, the target candidate region PR1 (see FIG. 18) including the target confirmed region PR11 and the undetermined region PR12 is detected. Here, the target confirmed region PR11 is a region in which at least a part of the lung field as the target region is captured in the frame image IC0, and the undetermined region PR12 is the target region in the frame image IC0. This is a region where a part of the lung field may be captured.

  In the present embodiment, the coarse detection unit 313 also detects a region PR2 in the lung image IC0 where a lung field as a target region is not captured (also referred to as a non-target determination region). Thereby, as shown in FIG. 18, information (also referred to as region division information) IC1 for dividing the frame image IC0 into the target fixed region PR11, the undetermined region PR12, and the non-target fixed region PR2 is generated. .

  By the way, assuming the next precise detection process, in the rough detection process, the target fixed region PR11 and the non-target fixed region are usually from the frame image IC0 to the portion as close as possible to the contour of the structure that is difficult to detect. It suffices if PR2 is detected. That is, it is only necessary that the undetermined region PR12 becomes narrow. Thereby, in the precision detection process, it is difficult to cause excessive detection of the target region beyond the unclear outline of the structure. In addition, as a structure where it is difficult to detect the contour, for example, a structure where the contour is unclear, or a structure that is divided by the presence of another structure, or the like can be considered.

  However, if the undetermined region PR12 detected by the rough detection process is narrow, the structure may be lost due to a lesion, the shape may change due to a periodic motion difference, or the structure may be hidden due to overlapping of other structures. Due to this problem, the target determination region PR11 may be detected excessively. For this reason, excessive detection of the target area exceeding the outline of the structure is likely to occur in the subsequent precision detection processing.

  Therefore, in the coarse detection processing according to the present embodiment, the position of the global region where the global part is captured is detected by template matching using the global template related to the global part with respect to the frame image IC0. Further, the position of the local region where the local portion is captured is detected by template matching using the local template related to the local portion with respect to the frame image IC0. At this time, in accordance with the reliability of the position of the detected local region, it is determined whether or not the information related to the position of the detected local region is used for generating the region segment information IC1.

  Here, the local part is a local part of the target part, and is a part that is included in the global part and occupies a relatively narrow area than the global part. The local portion only needs to include, for example, at least one of a portion in which a difference in a periodic state varies between individuals and a portion in which a shape difference occurs between individuals. Such a local portion only needs to include at least one portion of, for example, the aortic arch, the clavicle, the heart, the diaphragm, and the transverse angle. As a result, a local region is detected in a portion where a shift is likely to occur between the position of the actual global portion captured in the frame image IC0 and the position of the detected global region. For this reason, the area | region which caught the object site | part from the biological image can be detected and extracted with a sufficient precision. Further, the local portion may be, for example, a portion whose outline is unclear and difficult to detect. This suppresses detection of an excessively small target region in which the contours of other structures such as ribs, collarbones, and breasts are erroneously detected as the contours of the region of interest.

  The global portion is a portion that occupies a relatively large region of the target portion, and includes one or more local portions. The global template is a template related to the global region, and the local template is a template related to the local region.

  With respect to the portion where the position of the local area detected from the local viewpoint is low from the frame image IC0 by the rough detection process, the position of the global area detected from the frame image IC0 from the relatively macroscopic viewpoint is determined. The area division information IC1 is generated based on the information shown. Thereby, it becomes difficult to produce the excessive detection of the object area | region exceeding the unclear outline of a structure. As a result, a region that captures a lung field as a target region can be accurately extracted from a series of frame images I0 that capture a dynamic aspect of a body structure obtained by X-ray imaging.

  FIG. 19 is a block diagram illustrating a functional configuration realized by the coarse detection unit 313 together with other configurations. As illustrated in FIG. 19, the coarse detection unit 313 mainly includes a global region detection unit 3131, a local region detection unit 3132, a region setting determination unit 3133, and a candidate region setting unit 3134.

  Hereinafter, setting of the global region and the local region, a template generation method, and each unit 3131 to 1133 of the rough detection unit 313 will be described.

<(4-2-3-1) Setting of global region and local region>
FIG. 20 is a diagram illustrating a setting example of the global region and the local region. FIG. 20 illustrates the positions of the global region and the local region in a typical X-ray image (also referred to as a model image) MI0 in which the lung field is captured.

  As shown in FIG. 20, in the model image MI0 related to the lung field, a region (also referred to as a diaphragm region) R1 including the entire diaphragm, a region including the left diaphragm (also referred to as a left diaphragm region) R1L, and a right A region including the diaphragm (also referred to as a right diaphragm region) R1R is included. The left diaphragm region R1L includes a region including a port lateral angle (also referred to as a port lateral angle region) R1L1 and a region relating to the heart (also referred to as a heart region) R1L2, and the right diaphragm region R1R includes a starboard A region including a lateral angle (also referred to as a starboard lateral angle region) R1R1 is included. Further, in the model image MI0 related to the lung field, a region including the left lung apex (also referred to as the left lung apex region) R2L and a region including the right lung apex (also referred to as the right lung apex region) R2R include The left apex region R2L includes a region (also referred to as an aortic arch region) R2L1 that includes the aortic arch.

  For this reason, for example, when the diaphragm region R1 is a global region, the left diaphragm region R1L and the right diaphragm region R1R are local regions. Further, the left lateral lateral region R1L1 and the heart region R1L2 are local regions with respect to the left diaphragm region R1L, and the right lateral lateral region R1R1 is a local region with respect to the right diaphragm region R1R. When the left lung apex region R2L is a global region, the aortic arch region R2L1 is a local region.

<(4-2-3-2) Template types>
As shown in FIG. 19, the various data D1 stored in the storage unit 32 includes information relating to the global template (also referred to as global template information) D11 and information relating to the local template (also referred to as local template information) D12. It is.

  As shown in FIG. 20, the global template information D11 includes information indicating the diaphragm template T1, the left lung apex template T2L, and the right lung apex template T2R, respectively. The diaphragm template T1 is a template that captures the entire diaphragm corresponding to the diaphragm region R1. The left lung apex template T2L is a template in which the left lung apex corresponding to the left lung apex region R2L is captured. The right lung apex template T2R is a template in which the right lung apex corresponding to the right lung apex region R2R is captured.

  The local template information D12 includes information indicating the left diaphragm template T1L, the right diaphragm template T1R, the left lateral lateral angle template T1L1, the heart template T1L2, the right lateral lateral angle template T1R1, and the aortic arch template T2L1, respectively. The left diaphragm template T1L is a template in which the left diaphragm corresponding to the left diaphragm region R1L is captured. The right diaphragm template T1R is a template in which the right diaphragm corresponding to the right diaphragm region R1R is captured. The port side horizontal angle template T1L1 is a template in which the port side horizontal angle corresponding to the port side horizontal angle region R1L1 is captured. The heart template T1L2 is a template in which the heart corresponding to the heart region R1L2 is captured. The starboard lateral angle template TlR1 is a template in which the starboard lateral angle corresponding to the starboard lateral angle region R1R1 is captured. The aortic arch template T2L1 is a template in which the aortic arch corresponding to the aortic arch region R2L1 is captured.

  Although FIG. 20 shows a setting in which the local area corresponding to the local template is completely included in the global area corresponding to the global template, the present invention is not limited to this. For example, a setting may be adopted in which a part of the local area related to the local template protrudes from the global area related to the global template.

  The global template may be an average image of a plurality of global regions cut out from a plurality of sample images, for example. Here, each pixel value of the average image is calculated, for example, by dividing a cumulative value of pixel values related to pixels at the same position in a plurality of global regions by the number of sample images. In addition, the extraction of the global area from each sample image may be performed manually via the operation unit 33, or is realized by template matching for each sample image in which one global area is a template. Also good. Further, the residence template may be an average image of a plurality of local areas respectively cut out from a plurality of sample images, for example, like the global template.

  However, with regard to the global portion and the local portion, the periodic shape change and the shape difference between individuals are large. For example, as shown in FIGS. 21 to 23, when the longitudinal size and shape of the lung field and diaphragm change periodically due to respiration, and as shown in FIGS. The case where the individual difference in a shape arises is considered. For this reason, for example, a plurality of templates may be prepared for each global portion and each local portion in consideration of a change in shape and a difference in shape in each portion. Further, in each portion, a structure may be lost due to a lesion, an abnormality in a periodic shape change, or the like may occur. For this reason, for example, a template may be prepared for each case in consideration of omission and abnormality of a structure portion due to a lesion.

  For example, the shape of the diaphragm differs depending on the size of the lung field and the heart, and the position of the diaphragm differs on the left and right. For this reason, in the present embodiment, a plurality of templates that are in a relationship of being enlarged and reduced in the vertical direction and the horizontal direction (also referred to as a relationship of vertical / horizontal scaling) with respect to the difference in the shape of the diaphragm are employed. In addition, different templates are employed for the left diaphragm and the right diaphragm for differences in the left and right positions of the diaphragm. Further, for example, the shape of the contour of the lung field along the heart differs depending on the difference in the size of the heart due to cardiac hypertrophy as a specific case. For this reason, in the present embodiment, a plurality of templates having a relationship of being enlarged and reduced in the vertical direction and the horizontal direction with respect to the difference in the shape of the heart are employed.

<(4-2-3-3) Template generation method>
FIG. 28 to FIG. 31 are diagrams showing a method of generating a plurality of diaphragm templates T1 having a relationship of vertical / horizontal scaling. A sample image SI0 is schematically shown on the left part of FIGS. 28 to 31, and diaphragm templates T1a to T1d (T1) each representing the shape of the diaphragm of the sample image SI0 are shown on the right part of FIGS. It is shown schematically.

  As shown in the sample images SI0 of FIGS. 28 to 31, there are large individual differences in the shape of the diaphragm, and the shape of the diaphragm varies greatly depending on the size of the heart. For this reason, for example, the sample images SI0 are classified into a plurality of groups according to the size of the heart and the like. Here, it is shown that the shapes of the heart are classified into four types of groups: small, general, medium and large. Next, as shown in FIGS. 28 to 31, a diaphragm region R1 including the diaphragm in the sample image SI0 (region surrounded by a broken line in the figure) is designated and cut out. Designation of the diaphragm region R1 for each sample image SI0 may be performed manually via the operation unit 33, or each sample image SI0 in which one diaphragm region R1 designated by manual operation or the like is used as a template. It may be realized by targeted template matching. Then, four types of diaphragm templates T1a to T1d (T1) are generated by generating an average image of images cut out from the diaphragm region R1 for each of the four types of groups classified in advance. That is, the diaphragm template T1 as the global template includes four types of diaphragm templates T1a to T1d. Here, each pixel value of the average image is calculated, for example, by dividing a cumulative value of pixel values related to pixels at the same position in an image cut out from a plurality of diaphragm regions R1 by the number of diaphragm regions R1. .

  Other global templates and local templates are also generated, for example, by the same method as the diaphragm template T1 as the global template. Note that the left apex template T2L and the right apex template T2R are classified into a plurality of groups from the viewpoint of the size of the apex and the inclination of the clavicle, for example, and templates for each group are generated. It should be done. That is, each global template includes a plurality of types of global templates, and each local template includes a plurality of types of local templates.

<(4-2-3-4) Global Area Detection Unit>
The global area detection unit 3131 detects the position of the global area where the global part of the frame image IC0 as a biological image is captured by template matching using a global template related to the global part. Here, for each global portion, a plurality of types of global templates are used, and an area having the highest degree of correlation (also referred to as a degree of correlation) with the global template in the frame image IC0 is detected as the global area. In the present embodiment, template matching is performed using the diaphragm template T1, the left lung apex template T2L, and the right lung apex template T2R as the global template. Accordingly, the position of the diaphragm region R1 as the global region related to the diaphragm, the position of the left lung apex region R2L as the global region related to the left lung apex, and the right lung apex region as the global region related to the right lung apex The position of R2R is detected respectively. Information Rt11 indicating the result detected by the global region detection unit 3131 is sent from the global region detection unit 3131 to the candidate region setting unit 3134.

<(4-2-3-5) Local region detection unit>
The local region detection unit 3132 uses the local template related to the local portion as the position of the local region that is included in the global portion and captures the local portion smaller than the global portion in the frame image IC0 as the biological image. Detect by matching. Here, for each local portion, a plurality of types of local templates are used, and a region having the highest degree of correlation with the local template in the frame image IC0 is detected as a local region. In this embodiment, template matching using the left diaphragm template T1L, the right diaphragm template T1R, the left lateral lateral template T1L1, the cardiac template T1L2, the right lateral lateral template T1R1, and the aortic arch template T2L1 as local templates is performed. Done. Accordingly, the position of the left diaphragm region R1L as the local region related to the left diaphragm and the position of the right diaphragm region R1R as the local region related to the right diaphragm are detected. Further, the position of the port side lateral angle region R1L1 as a local region related to the port lateral angle, the position of the heart region R1L2 as the local region related to the heart, and the starboard lateral angle region as the local region related to the starboard lateral angle The position of R1R1 is detected. Further, the position of the aortic arch region R2L1 as a local region related to the aortic arch is detected.

  Information Rt12 indicating the result detected by the local region detection unit 3132 is sent from the local region detection unit 3132 to the region setting determination unit 3133.

  Here, for example, if information such as a missing structure due to a lesion such as lung resection is given in advance, the amount of calculation required for image processing is reduced by reducing the types of templates used in template matching. Is reduced. That is, the image processing can be speeded up.

<(4-2-3-6) Region Setting Determination Unit>
The region setting determination unit 3133 determines whether or not the candidate region setting unit 3134 sets the target candidate region PR1 according to the detection result by the local region detection unit 3132.

  This determination is performed based on information (also referred to as determination criterion information) D13 that is a criterion for determination included in various data D1 stored in the storage unit 32. Specifically, this determination is performed based on a value (also referred to as a correlation value) indicating the degree of correlation between the local region and the local template calculated by template matching in the local region detection unit 3132. The determination is performed based on, for example, the relative positional relationship between the position of the global region detected by the global region detection unit 3131 and the position of the local region detected by the local region detection unit 3132.

  Thus, it is determined whether or not the target confirmed region PR11 and the undetermined region PR12 related to the target portion corresponding to the detected local region are set according to the reliability of the position of the detected local region. For this reason, the target area | region which caught the target site | part from the frame image IC0 as a biological image which caught the structure inside the body obtained by X-ray imaging is detected and extracted with high precision.

  In the present embodiment, the determination is based on both the correlation value between the local region and the local template and the relative positional relationship between the position of the detected global region and the position of the detected local region. However, the present invention is not limited to this. For example, the determination is performed based on at least one of a correlation value between the local region and the local template and a relative positional relationship between the position of the detected global region and the position of the detected local region. It ’s fine.

  Here, the information Rt13 indicating the determination result in the region setting determination unit 3133 is sent from the region setting determination unit 3133 to the candidate region setting unit 3134.

<Determination based on (4-2-3-6-1) correlation value>
For example, if the correlation value between the local region and the local template is out of the preset value range, the setting of the target candidate region PR1 according to the detection result by the local region detection unit 3132 in the candidate region setting unit 3134 is set. It is forbidden. Thereby, the setting of the target confirmed region PR11 and the undetermined region PR12 related to the target portion corresponding to the local region with low reliability is suppressed. As a result, the target region in which the target part is captured is detected and extracted with high accuracy from the frame image IC0 as a biological image.

  Here, the correlation value may be a value indicating the degree of coincidence between the local template and the local region calculated by a technique such as SAD, NCC, POC, and RIPOC. In addition, when the local template and the local region completely match, the correlation value becomes 1, and when the degree of matching between the local template and the local region decreases, the correlation value approaches 0. The value range that has been set may be, for example, a value range that exceeds a preset threshold and is less than 1. The threshold value may be set to about 0.8 to 0.9, for example.

<(4-2-3-6-2) Determination by relative positional relationship>
For example, a deviation amount between the position of the global region detected by the global region detection unit 3131 and the position of the local region detected by the local region detection unit 3132 in at least one preset direction is set in advance. If it is out of the range, setting of the target candidate region PR1 according to the detection result by the local region detection unit 3132 in the candidate region setting unit 3134 is prohibited. Thereby, the setting of the target confirmed region PR11 and the undetermined region PR12 related to the target portion corresponding to the local region with low reliability is suppressed. As a result, the target region in which the target part is captured is detected and extracted with high accuracy from the frame image IC0 as a biological image.

  Here, an aspect in which at least one direction is, for example, the Y direction is conceivable. In this aspect, for example, if the port lateral angle detected by the local region detection unit 3132 is located above the diaphragm detected by the global region detection unit 3131, it cannot occur in a normal healthy person. State. For this reason, in such a state, it is considered that the reliability of the position of the detected port side lateral region R1L1 is low. Therefore, setting of the target candidate region PR1 according to the position of the port side lateral region R1L1 as a detection result by the local region detection unit 3132 is prohibited.

  More specifically, when the Y coordinate of the predetermined position of the diaphragm region R1 as the global region is c1 and the Y coordinate d1 of the predetermined position of the port lateral region R1L1 as the local region, the following formula (1 ) Is established, it is determined that the reliability of the position of the detected port lateral area R1L1 is low.

    c1-d1> 0 (1).

  Here, a mode is conceivable in which the predetermined position of the diaphragm region R1 is the position of the lower end of the diaphragm region R1, and the predetermined position of the port side lateral angle region R1L1 is the position of the lower end of the port lateral angle region R1L1. . However, depending on the setting of the predetermined positions of the diaphragm region R1 and the port side lateral angle region R1L1, or the setting of the diaphragm template T1 and the port side lateral angle template T1L1, the predetermined value of the port side lateral angle region R1L1 is more than the predetermined position of the diaphragm region R1. Even if the position is located above, it may be allowed. If such a case is assumed, the Y coordinate c1 of the predetermined position of the diaphragm region R1, the Y coordinate d1 of the predetermined position of the port lateral angle region R1L1, and a preset threshold th1 If the following equation (2) is satisfied, it may be determined that the reliability of the position of the detected port side angle region R1L1 is high. That is, a certain allowable range may be set for the amount of deviation between the detected global region and the detected local region.

    | C1-d1 | <th1 (2).

  Note that the threshold value th1 may be appropriately set in advance according to the setting status of the diaphragm template T1 and the port side lateral angle template T1L1, for example.

<(4-2-3-7) Candidate area setting unit>
The candidate area setting unit 3134 selects the target candidate area PR1 including the target confirmed area PR11 and the undetermined area PR12 in the frame image IC0 as a biological image in accordance with the detection results by the global area detector 3131 and the local area detector 3132. Set. In the present embodiment, the candidate region setting unit 3134 is a non-target determination region in which the target part is not captured in the frame image IC0 as a biological image, according to the detection results by the global region detection unit 3131 and the local region detection unit 3132. PR2 is also set.

  As illustrated in FIG. 19, the candidate area setting unit 3134 includes a partial segment information setting unit 3134a and an area integration unit 3134b. Here, the partial segment information setting unit 3134a and the region integration unit 3134b will be described.

<(4-2-3-7-1) Partial section information setting unit>
The partial segment information setting unit 3134a is based on the partial segment information D14 of the various data D1 stored in the storage unit 32 with respect to the global region and the local region detected by the global region detection unit 3131 and the local region detection unit 3132. Partial segment information Rt14 is set for each.

  In the partial segment information D14, a global template, partial segment information corresponding to the correlation value between the global template and the global region, and partial segment information corresponding to the local template and the correlation value between the local template and the local region are shown. Contains information. Here, the partial division information is information for dividing each global region or each local region into a target defined region PR11, an undefined region PR12, and a non-target defined region PR2.

  For this reason, the partial segment information setting unit 3134a obtains, from the partial segment information D14, the global template when the global region is detected and the partial segment information corresponding to the correlation value between the global template and the global region. Thereby, the partial division information Rt14 is set for the global region. Further, the partial segment information setting unit 3134a obtains from the partial segment information D14 the local template when the local region is detected and the partial segment information corresponding to the correlation value between the local template and the local region. Thereby, the partial segment information Rt14 is set for the local region.

  It should be noted that the partial classification information Rt14 is not set for the local region where the setting of the target candidate region PR1 according to the detection result by the local region detection unit 3132 in the candidate region setting unit 3134 is prohibited by the region setting determination unit 3133. Also good. As a result, processing speed can be increased by reducing computation.

  Here, a method of generating the partial segment information D14 will be described.

  32 to 36 are diagrams for explaining a method of generating the partial segment information D14.

  As described above, for example, when the global template and the local template are generated, the sample images SI0 are classified into a plurality of groups, and for each group, the global region and the local region are cut out from the plurality of sample images SI0 and averaged. An image is generated.

  At this time, first, for each of the extracted global region and each local region, there is also data (also referred to as correct data) indicating whether the target region is a target region where the target region is captured or a non-target region where a portion other than the target region is captured Generated. FIG. 32 exemplifies correct data A0 in which a target area is expressed in white and a non-target area is expressed in black for a sample image SI0. FIG. 33 shows the right lung apex region R2R as the global region and the correct data A2R (A0) related to the right lung apex region R2R, the left lung apex region R2L as the global region, and the left lung apex region. The correct data A2L (A0) related to R2L is illustrated. FIG. 34 shows correct data A1L (A0) related to the left diaphragm region R1L as the local region and the left diaphragm region R1L, and correct data A1R (A0) related to the right diaphragm region R1R and the right diaphragm region R1R as the local region. ) Is illustrated.

  Next, a plurality of correct data A0 is overlapped for each group for each global region and each local region, and the number of times correct data A0 is overlapped for each pixel is divided by the number of correct answer data A0. Thus, the probability (also referred to as a cumulative rate) that the correct data A0 overlap is calculated. The accumulation rate may be shown in a form such as a percentage such as 0 to 100%. In the lower left of FIG. 35, an image (also referred to as a cumulative rate image) Am1L showing the distribution of the cumulative rate for a plurality of correct answer data A1L (A0) related to the left diaphragm region R1L is illustrated. In the lower right of FIG. 35, an accumulation rate image Am1R showing the distribution of accumulation rates for a plurality of correct answer data A1R (A0) related to the right diaphragm region R1R is illustrated. In the upper left of FIG. 35, a cumulative rate image Am2L showing a cumulative rate distribution for a plurality of correct answer data A2L (A0) related to the left lung apex region R2L is illustrated. In the upper right of FIG. 35, a cumulative rate image Am2R showing the distribution of cumulative rates for a plurality of correct answer data A2R (A0) related to the right diaphragm region R2R is illustrated.

  Here, for a certain group, it is certain that an area with a cumulative rate of 100% corresponds to a target area that captures the target part, and a region with a cumulative rate of 0% does not capture the target part. It is certain to correspond to the target area. On the other hand, it can be said that an area having an accumulation rate of 1 to 99% is an undetermined area that corresponds to a target area or a non-target area. However, for example, for each group for each global region, a region with a higher accumulation rate corresponds to the target region as the correlation value between the global template and the global region detected by the global region detection unit 3131 increases. The possibility increases. Similarly, for each group for each local region, a region having a higher accumulation rate can correspond to the target region as the correlation value between the local template and the local region detected by the local region detection unit 3132 increases. Increases nature.

  Therefore, for each global template, partial segment information corresponding to the correlation value between the global template and the global region is generated based on the corresponding cumulative rate image. Further, for each local template, partial segment information corresponding to the correlation value between the local template and the local region is generated based on the corresponding cumulative rate image. Thereby, partial division information D14 is generated.

  For example, for partial segment information for a local template, if a range of correlation values is set according to a threshold value related to a correlation value in determination by the region setting determination unit 3133, and partial segment information is generated for each range of correlation values good. For example, if the threshold is 0.8, the correlation value has a large value range (0.99 <correlation value ≦ 1), the correlation value has a medium value range (0.9 <correlation value ≦ 0.99), and the correlation value is A mode in which a small value range (0.8 <correlation value ≦ 0.9) is set is conceivable.

  In this aspect, for a range with a large correlation value (0.99 <correlation value ≦ 1), for example, a region with a cumulative rate of 0% is a non-target defined region, and a cumulative rate exceeds 0% and is 50% or less. A region is an unconfirmed region, and a region having an accumulation rate of more than 50% and not more than 100% is a target confirmed region. An example of the partial segment information PV1 in this case is shown in the left part of FIG. In addition, for an intermediate correlation value range (0.9 <correlation value ≦ 0.99), for example, an area with an accumulation rate of 0% is a non-target determination area, an accumulation ratio exceeds 0% and is less than 70% Is an undetermined area, and an area where the accumulation rate exceeds 70% and is equal to or less than 100% is a target confirmed area. An example of the partial segment information PV2 in this case is shown in the center of FIG. Furthermore, for a range where the correlation value is small (0.8 <correlation value ≦ 0.9), for example, a region where the accumulation rate is 0% is a non-target determination region, and the accumulation rate exceeds 0% and is 99% or less. A region is an unconfirmed region, and a region where the accumulation rate exceeds 99% and is 100% or less is a target confirmed region. An example of the partial segment information PV3 in this case is shown in the right part of FIG.

  For each global template, as in the case of each local template, partial classification information corresponding to the correlation value between the global template and the global area is generated based on the corresponding cumulative rate image. However, the partial classification information may be generated only when the correlation value between the global template and the global region is small to some extent so that a region where the detection result by the rough detection unit 313 cannot be obtained does not occur.

  Here, for the same part in the body, it is sufficient that the undetermined area in the partial classification information related to the local template is set to be narrower than the undetermined area in the partial classification information related to the global template. . Thereby, in the candidate area setting unit 3134, the area in which the local part of the frame image IC0 is captured is set in accordance with the detection result by the local area detection unit 3132 rather than the detection result by the global area detection unit 3131. The region PR12 is narrower. As a result, the undetermined region corresponding to the local region is set to be narrower, so that over-detection in which the region where the target region is captured is detected too broadly and undetected where the region where the target region is captured is detected too narrowly. It becomes difficult to occur. That is, the target region that captures the target portion can be stably and accurately extracted from the frame image IC0 as a biological image.

<(4-2-3-7-2) region integration unit>
The area integration unit 3134b integrates the partial segment information Rt14 set for each global area and each local area. At this time, as a general rule, the partial section information Rt14 related to the local region is adopted for the portion where the local region and the global region overlap. For example, the partial segment information Rt14 related to the local region is overwritten on the partial segment information Rt14 related to the global region. However, in the region setting determination unit 3133, for the local region where the setting of the target candidate region PR1 according to the detection result by the local region detection unit 3132 in the candidate region setting unit 3134 is prohibited, the partial segment information Rt14 related to the global region is Adopted. As a result, the setting of the target candidate region PR1 and the non-target determined region PR2 including the target determined region PR11 and the undetermined region PR12 in the candidate region setting unit 3134 is completed. FIG. 37 is a diagram illustrating an example of the region classification information IC1 indicating the target candidate region PR1 and the non-target determination region PR2 set by the region integration unit 3134b.

<(4-2-4) Precision detection unit>
Based on the information on the target confirmed region PR11 and the undetermined region PR12 set by the candidate region setting unit 3134, the precision detection unit 314 selects a target region from the frame image IC0 as a biological image to the contour from the inside of the target part. Performs precise detection processing to detect. In the present embodiment, it is determined whether the undetermined region PR12 is inside or outside the lung field region as the target region.

  FIG. 38 is a diagram illustrating an example of the frame image IC0 input from the image correction unit 312 to the precision detection unit 314. FIG. 39 is a diagram illustrating an example of a lung field region RT1 as a target region detected from the frame image IC0 in the precision detection unit 314. As shown in FIG. 37 to FIG. 39, the precise detection unit 314 receives the frame image IC0 input from the image correction unit 312 based on the region classification information IC1 as a result of the rough detection process input from the rough detection unit 313. And the lung field region RT1 as the target region is detected.

  The precision detection unit 314 detects the lung field region RT1 as the target region by a detection method that does not depend on the shape of the target part. Examples of detection methods include methods that use edge information such as level set and snake, methods that use region information such as region growing, and graph cuts (Graph Cut). ) And the like using edge and area information.

  In the Snake process, for example, an initial value of a closed curve is defined so as to surround a region to be detected, and the closed curve is deformed so as to satisfy the following conditions [i] to [iii] by repeated calculation. When the shape no longer changes, the deformation of the closed curve is terminated. By this processing, the contour ED1 of the lung field region RT1 as the target region is detected. Regarding the processing contents of Snake, for example, various known information (Hiro Mosquito, “Chapter 11“ Regional Processing ””, Kyoto Sangyo University, Faculty of Computer Science and Engineering, Department of Network Media, “Spring Semester, Tuesday, Three Times” [Heisei Searched November 18, 2013], Internet <www.cc.kyoto-su.ac.jp/~kano/pdf/course/chapter11.pdf>, and representative: Nobuhiko Yajima "Processing Cases | sliceOmatic" Area / Volume Examples of area definition processing for measurement ”disclosed on Image Laboratories website [searched on November 18, 2013], Internet <URL: http://www.imagelabo.com/Appli_sliceO.html>, etc.)

  [I] The shape of the closed curve itself tends to be continuous and smooth.

  [Ii] If the closed curve is on an edge, it tries to stay there.

  [Iii] The closed curve tends to be small.

  Here, in the method using edge information such as level set and snake, if there is information on the target defined region PR11 and the undefined region PR12 in the region classification information IC1, A contour can be detected. Therefore, in the candidate area setting unit 3134, the target candidate area PR1 including the target confirmed area PR11 and the undefined area PR12 is included in the frame image IC0 according to the detection results by the global area detector 3131 and the local area detector 3132. It only has to be set.

  In the region growing process, for example, first, a pixel that satisfies a condition relating to a pixel value is manually determined, and a single label is attached to the seed point. And the area | region is expanded by repeating the process in which the same label is attached | subjected to the pixel which satisfy | fills the conditions regarding a pixel value in the point of the seed point vicinity. In accordance with such processing, if the target confirmed region PR11 is used as a seed point and the region is expanded in the undetermined region PR12, the lung field region RT1 as the target region can be detected.

  Regarding the processing of graph cut, for example, publicly known information (Hiroshi Ishikawa “Theory and Application of Graph Cut”, 14th Image Sensing Symposium SS1108, [retrieved on November 18, 2013], Internet < URL: http://www.f.waseda.jp/hfs/SSIITutorial.pdf>).

  Information Rt2 indicating the result detected by the precision detection unit 314 is sent from the precision detection unit 314 to the target region extraction unit 315.

<(4-2-5) Target area extraction unit>
The target region extraction unit 315 extracts an image of the lung field region as the target region from the frame image I0 based on the target region detected by the precision detection process in the precision detection unit 314. Here, an image of the lung field region as the target region may be extracted from the frame image IC0 instead of the frame image I0.

<(5) Operation flow of target area extraction processing>
40 and 41 are flowcharts illustrating the operation flow of the target area extraction processing in the image processing apparatus 3. This operation flow is realized by the function of the control unit 31, for example. Here, for example, this operation flow is started in accordance with the operation of the operation unit 33 by the operator of the image processing apparatus 3, and the process proceeds to step St1 in FIG.

  In step St1, the image acquisition unit 311 acquires a dynamic image composed of a series of a plurality of frame images I0.

  In step St2, the image correction unit 312 corrects each of the plurality of frame images I0 acquired in step St1. As a result, noise is reduced and the influence of the structure other than the target part is reduced on each frame image I0, thereby generating a corrected frame image IC0.

  In step St <b> 3, the control unit 31 sets a number N (N is a natural number) for identifying the frame image IC <b> 0 to be processed among the plurality of frame images IC <b> 0 constituting the dynamic image to 1. The number N may be any number indicating that it is the Nth frame image IC0 constituting the dynamic image, for example.

  In step St4, the coarse detection unit 313 performs a coarse detection process. In this rough detection process, a process along the operation flow shown in FIG. 41 is executed.

  In step St41 of FIG. 41, the global region detection unit 3131 detects one or more global regions.

  In step St42, the partial segment information setting unit 3134a sets partial segment information Rt14 for each global area detected in step St41.

  In step St43, the local region detection unit 3132 detects one or more local regions. At this time, for example, a number M (M is a natural number) is assigned to each local region.

  In step St44, the control unit 31 sets the number M for identifying the local region to be processed among the one or more local regions detected in step St43 to 1.

  In step St45, the region setting determination unit 3133 determines whether or not to set the target candidate region PR1 according to the local region corresponding to the number M. If it is determined that the target candidate region PR1 is set according to the local region, the process proceeds to step St46, and if it is determined that the target candidate region PR1 is not set according to the local region, the process proceeds to step St47. .

  In step St46, the partial segment information setting unit 3134a sets the partial segment information Rt14 corresponding to the local region corresponding to the number M.

  In step St47, the region setting determination unit 3133 prohibits the setting of the partial segment information Rt14 corresponding to the local region corresponding to the number M.

  In step St48, the control unit 31 determines whether or not the processing for all local regions detected in step St43 has been completed. If processing for all local regions has not been completed, the process proceeds to step St49. If processing for all local regions has been completed, the process proceeds to step St50.

  In step St49, 1 is added to the number M by the control part 31, and it progresses to step St45. That is, the processes in steps St45 to St49 are repeated until the processes for all local regions are completed.

  In step St50, the region integration unit 3134b integrates the partial segment information Rt14 related to the global region set in step St42 and the partial segment information Rt14 related to the local region set in step St46, thereby obtaining a region segmentation. Information IC1 is generated.

  In step St5 of FIG. 40, the precision detection unit 314 performs precision detection processing. Here, the target area is detected from the frame image IC0 associated with the number N based on the area division information IC1 generated in step St50.

  In step St6, the control unit 31 determines whether or not the processing for all the frame images IC0 in the dynamic image has been completed. Here, if the process for all the frame images IC0 is not completed, the process proceeds to step St7, and if the process for all the frame images IC0 is completed, the process proceeds to step St8.

  In step St7, the control unit 31 adds 1 to the number N for specifying the frame image IC0 to be processed among the plurality of frame images IC0 constituting the dynamic image, and the process proceeds to step St4. That is, the processes of steps St4 to St7 are repeated until the processes for all the frame images IC0 are completed.

  In step St8, the target region extraction unit 315 extracts a lung field region image as the target region from each frame image I0 based on the target region related to each frame image IC0 detected in step St5.

  In step St9, the control unit 31 stores an image of the lung field region as the target region extracted in step St8 in the storage unit 32. At this time, on the display unit 34, images of lung field regions related to the plurality of frame images I0 may be displayed in a dynamic manner.

<Summary of (6) One Embodiment>
As described above, in the image processing device 3 according to this embodiment, based on at least one of the relative positional relationship between the detected global region and the local region and the correlation value between the local region and the local template, It is determined whether or not to set the target candidate area according to the local area. That is, according to the detection result of the local region, it is determined whether or not the target confirmed region PR11 and the undetermined region PR12 related to the target portion corresponding to the local region are set. For this reason, whether to set the target confirmed region PR11 and the undetermined region PR12 according to the local region is determined according to the reliability of the detected local region. As a result, it is possible to accurately extract a region in which the target part is captured from the frame image IC0 as a living body image capturing the internal structure obtained by X-ray imaging.

<(7) Modification>
Note that the present invention is not limited to the above-described embodiment, and various modifications and improvements can be made without departing from the gist of the present invention.

  For example, in the above-described embodiment, the correction by the image correction unit 312 is performed on the frame image I0, but is not limited thereto. For example, the correction to the frame image I0 by the image correction unit 312 may not be performed.

  In the above-described embodiment, the functional configuration of the control unit 31 illustrated in FIGS. 2 and 19 is realized by executing the program P1, but is not limited thereto, and is realized by, for example, dedicated hardware. May be.

  In the above embodiment, the heart template T1L2 related to the heart region R1L2 included in the left diaphragm region R1L is set. However, the present invention is not limited to this. For example, a heart template may be set for a region related to the heart included in the right diaphragm region R1R. That is, other global regions and global templates, and local regions and local templates may be set.

  Moreover, in the said one Embodiment, although the object site | part was the whole lung field, it is not restricted to this. For example, the target site may be a left lung field and a right lung field that are a part of the structure of the lung field, or may be another structure related to other organs, skeletons, and the like.

  Needless to say, all or a part of each of the above-described embodiment and various modifications can be combined as appropriate within a consistent range.

DESCRIPTION OF SYMBOLS 1 Imaging device 2 Imaging control device 3 Image processing device 31 Control unit 32 Storage unit 100 Imaging system 311 Image acquisition unit 312 Image correction unit 313 Rough detection unit 314 Precision detection unit 315 Target region extraction unit 3131 Global region detection unit 3132 Local region detection Unit 3133 region setting determination unit 3134 candidate region setting unit 3134a partial segment information setting unit 3134b region integration unit I0, I1 to I10, IC0 frame image IC1 region segment information P1 program PR1 target candidate region PR11 target defined region PR12 undefined region PR2 non Target fixed area

Claims (8)

An acquisition unit for acquiring a biological image in which a structure inside the body is captured by irradiation with X-rays;
Using template matching for the biological image, there is a possibility that, in the biological image, a target fixed region in which at least a part of the target part is captured and a part of the target part is captured. A first detection unit for detecting an undefined region;
A second detection unit that detects a target region from the inside of the target part to an outline based on information on the target fixed region and the unconfirmed region;
The first detection unit is
A global region detection unit that detects a position of a global region in which the global part of the target portion in the biological image is captured by template matching using a global template according to the global part;
A local region detection unit that detects a position of a local region that is included in the global portion of the biological image and that captures a local portion smaller than the global portion by template matching using a local template related to the local portion. When,
A candidate region setting unit that sets a target candidate region including the target fixed region and the unconfirmed region with respect to the biological image according to a detection result by the global region detection unit and the local region detection unit;
Calculated by the relative positional relationship between the position of the global region detected by the global region detection unit and the position of the local region detected by the local region detection unit, and template matching in the local region detection unit Based on at least one of correlation values between the local region and the local template, the candidate region setting unit determines whether to set the target candidate region according to the detection result by the local region detection unit A determination unit to perform,
An image processing apparatus. The image processing apparatus according to claim 1,
The candidate area setting unit is set to an area in which the local part of the biological image is captured, rather than the undetermined area set in the candidate area setting unit according to the detection result by the global area detection unit. The image processing apparatus according to claim 1, wherein the undetermined region set in accordance with a detection result by the local region detection unit is narrower. The image processing apparatus according to claim 1 or 2,
The candidate area setting unit
An image processing apparatus that sets a non-target fixed region where the target part is not captured for the biological image according to detection results by the global region detection unit and the local region detection unit. An image processing apparatus according to any one of claims 1 to 3, wherein
The local portion is
An image processing apparatus including at least one of a portion in which a change in a periodic state varies between individuals and a portion in which a shape varies between individuals. The image processing apparatus according to claim 4,
The target site is
It is a structure related to the lung field,
The local portion is
An image processing apparatus including at least one portion of an aortic arch, a clavicle, a heart, a diaphragm, and a transverse angle. The image processing apparatus according to any one of claims 1 to 5, wherein:
The determination unit is
A deviation amount between the position of the global region detected by the global region detection unit and the position of the local region detected by the local region detection unit in at least one preset direction is set in advance. An image processing apparatus that prohibits setting of the target candidate region according to a detection result by the local region detection unit in the candidate region setting unit if it is out of the range. The image processing apparatus according to any one of claims 1 to 6, wherein
The determination unit is
An image processing apparatus that prohibits setting of the target candidate area according to a detection result by the local area detection unit in the candidate area setting unit if the correlation value is out of a preset range.   A program for causing the image processing apparatus to function as the image processing apparatus according to any one of claims 1 to 7 when executed by a control unit included in the image processing apparatus.

Images