JP2019141717A - Image positioning apparatus and method, and program - Google Patents

Abstract

To provide an image positioning apparatus and method, even when a structure having a high pixel value in one image, represents a low pixel value in another image, performing highly-accurate positioning, and a program.SOLUTION: The image positioning apparatus includes: an image acquisition part 10 for acquiring a plurality of images; a divided region setup part 11 for dividing respective positioning processing objects of the plurality of images to set up divided regions; a pixel value conversion method determination part 12 for determining pixel value conversion methods of the respective divided regions on the basis of the pixel values of the respective divided region of a group for each divided region group set up at same positions of respective positioning processing objects; a pixel value conversion part 13 for performing pixel value conversion to images within the divided regions of the divided region group using the determined pixel value conversion method of the divided region; and a positioning processing part 14 for performing positioning processing to the respective images that are subjected to the pixel value conversion.SELECTED DRAWING: Figure 1

Description

  The present invention relates to an alignment apparatus and method for aligning a plurality of images, and a program.

  Conventionally, perfusion photography has been performed as a technique for measuring blood flow in the myocardium, brain, liver, pancreas, and the like or blood flow in the brain.

  Perfusion imaging is an imaging method in which a contrast medium is injected into a subject, and imaging is performed a plurality of times at different timings using a CT (Computed Tomography) apparatus or the like. Depending on the phase (timing of imaging), an image in which different anatomical structures are contrasted is obtained. Specifically, in the case of the heart, in the initial phase, a contrast medium is injected from a vein, so that the right atrium and the right ventricle are stained with the contrast medium, and then the contrast medium flows in the order of the left ventricle and the left atrium. This changes the contrast of the heart image.

  An image obtained by perfusion photography is used for perfusion analysis. The main purpose of perfusion imaging and analysis is to analyze the blood flow of the target organ and diagnose the disease from the results. For example, the blood flow obtained by myocardial perfusion analysis is used for diagnosis of myocardial infarction.

  As a quantitative analysis method of perfusion (blood flow), there are a Maximum Slope method (for example, see Non-Patent Document 1) and a Deconvolution method (for example, see Non-Patent Document 2). A “time density curve (hereinafter referred to as TDC)” is used for these quantitative analyses. TDC represents a temporal change (a temporal change in contrast agent concentration) of a pixel value (CT value) at a certain position (region) of an organ.

  In order to perform quantitative analysis with high accuracy, it is necessary to accurately create TDC from perfusion images. For this purpose, it is necessary to create a TDC from multiple phase images by referring to pixel values at the same anatomical position (region). For this, it is necessary to accurately align the position of the target organ between phases, and image alignment is performed.

  The image is aligned so that the point x of one image IF (x) (Fixed Image) and the point T (x) of the other image IM (x) (Moving Image) coincide anatomically. This is a problem for obtaining the geometric transformation T. When the geometric transformation T is described by the parameter μ, it becomes a problem for obtaining an optimal solution based on the following equation (1). Geometric transformation T includes rigid body transformation (translation and rotation), Affine transformation and non-rigid transformation (B-Spline and Diffeomorphism). (For example, see Non-Patent Document 3)

Note that S in the above equation (1) is an evaluation function representing the degree of coincidence between the image IF (x) and the image IM (T (•; x)). Here, the smaller the S, the more the two images match. Yes. As an evaluation index of the degree of coincidence, Sum of Squared Difference (SSD), quantifying the statistical dependency between pixel values and mutual information (sum of squares of differences between pixel values at the same position between two images) Various evaluation indexes such as one of the indexes are known.

JP 2011-125431 A

"Measurement of tissue perfusion by dynamic computed tomography.", Miles KA, The British journal of radiology (1991), Vol. 64, No. 761, pp409-412 "Deconvolution-based CT and MR brain perfusion measurement: theoretical model revisited and practical implementation details", Fieselmann, Andreas and Kowarschik, Markus and Ganguly, Arundhuti and Hornegger, Joachim and Fahrig, Rebecca, Journal of Biomedical Imaging (2011), Vol. 2011, pp14 "Survey of medical image registration", Mani, VRS and others, Journal of Biomedical Engineering and Technology (2013), Vol. 1, No. 2, pp.8-25

  Here, when performing image alignment as described above, alignment of images in which different anatomical structures are contrasted becomes a problem. For example, if a structure with a high pixel value in one image shows a low pixel value in the other image, the pixel values in the same region are different, so the degree of coincidence is evaluated low and the alignment is accurate. Can not do it.

  As a method for solving such a problem, a method of performing pixel value conversion on the entire image can be considered. Specifically, a method of converting a high pixel value into a low pixel value can be considered.

  However, in the case of this method, information on the internal structure of the alignment target organ is lost. For example, in the case of an image of myocardial perfusion, in one image between two images, only the right atrium and the right ventricle show high pixel values due to the contrast agent (see FIG. 2I), and the other image is the right atrium and right ventricle. Not only the left atrium and left ventricle may also show high pixel values (see FIG. 2II). In such a case, if the same pixel value conversion is applied to the entire image, the above problem may be solved for the left atrium and the left ventricle that show a high pixel value in only one image. Since the high pixel value indicates, shape information that can be used for alignment is lost, and the position inside the organ cannot be aligned.

  Patent Document 1 proposes a technique related to the alignment of images captured with different modalities, such as a CT image and an ultrasonic image, but does not propose any method for solving the above-described problem. .

  In view of the above circumstances, the present invention can reduce the difference in contrast between images even when a structure having a high pixel value in one image shows a low pixel value in the other image, with high accuracy. An object of the present invention is to provide an image alignment apparatus, method, and program capable of performing alignment.

  The first image alignment apparatus of the present invention includes an image acquisition unit that acquires a plurality of images including alignment processing targets, and a divided region setting that divides each alignment processing target of the plurality of images and sets a divided region And a pixel value conversion method for determining a pixel value conversion method for each divided region based on the pixel value of each divided region of the set for each set of divided regions set at the same position of each alignment processing target A pixel value conversion unit that performs pixel value conversion on an image in a divided region of a set of divided regions using the pixel value conversion method of the divided region determined by the determining unit, the pixel value conversion method determining unit, and a pixel value conversion An alignment processing unit that performs alignment processing on a plurality of images that have undergone pixel value conversion by the unit.

  In the first image alignment device of the present invention, the pixel value conversion method determining unit determines whether or not to perform pixel value conversion of each divided region based on the pixel value of each divided region of the set. can do.

  The second image alignment apparatus of the present invention includes an image acquisition unit that acquires a plurality of images including alignment processing targets, and a divided region setting that divides each alignment processing target of the plurality of images and sets a divided region And an alignment processing unit that performs an alignment process by evaluating the degree of coincidence of each divided area of the set using an evaluation function for each set of divided areas set at the same position of each alignment processing target. The alignment processing unit performs alignment processing for each set of divided regions by changing the evaluation function based on the pixel value of each divided region of the set.

  In the first and second image alignment apparatuses of the present invention, the image acquisition unit acquires an image including a heart image as an alignment processing target, and the divided region setting unit sets the right atrium and the divided regions as divided regions. A first divided region including the right ventricle and a second divided region including the left atrium and the left ventricle can be set.

  In the first and second image alignment apparatuses of the present invention described above, the divided region setting unit includes the first divided region and the second divided region based on the change in the pixel value of the aortic region included in the plurality of images. Can be set.

  In the first and second image alignment apparatuses of the present invention described above, the divided region setting unit may select the left of the plurality of images based on the change in the pixel value of the aortic region included in the plurality of images. The first image in which the pixel values in the right atrium and right ventricle region are higher than the pixel values in the atrial and left ventricular regions and the pixel values in the left atrial and left ventricular regions than the pixel values in the right atrial and right ventricular regions. A second image having a higher pixel value can be identified, the first divided region can be set using the first image, and the second divided region can be set using the second image.

  In the first and second image alignment apparatuses of the present invention, the image acquisition unit acquires an image including a heart image as an alignment processing target, and the divided region setting unit sets the right atrium and the divided regions as divided regions. A first divided region including the right ventricle, a second divided region including the left atrium and the left ventricle, and a third divided region including the myocardium can be set.

  In the first and second image alignment apparatuses of the present invention, the image acquisition unit acquires an image including a heart image as an alignment processing target, and the divided region setting unit sets the right atrium as the divided region. A first divided region including the second divided region including the right ventricle, a third divided region including the left atrium, and a fourth divided region including the left ventricle can be set.

  In the first and second image alignment apparatuses of the present invention, the image acquisition unit acquires an image including a heart image as an alignment processing target, and the divided region setting unit sets the right atrium as the divided region. Setting a first divided region including the second divided region including the right ventricle, a third divided region including the left atrium, a fourth divided region including the left ventricle, and a fifth divided region including the myocardium. Can do.

  In the first and second image alignment apparatuses of the present invention, the image acquisition unit acquires an image including a heart image as an alignment processing target, and the divided region setting unit is perpendicular to the blood flow direction of the heart. A heart image is divided by a simple cross section to set a plurality of divided regions.

  In the first and second image alignment apparatuses of the present invention, the divided region setting unit uses one reference image in which a region corresponding to the divided region is set in advance, as one image among a plurality of images. A divided region can be set for an image other than the one image using the divided region setting result.

  In the first image alignment device of the present invention, the pixel value conversion unit can perform pixel value conversion that lowers the pixel value of the image in the divided area.

  In the first and second image alignment apparatuses of the present invention, the image acquisition unit can acquire a plurality of images taken at different times.

  In the first and second image alignment apparatuses of the present invention, the image acquisition unit acquires an image including a heart image as an alignment processing target, and is captured at the same cardiac phase timing in a plurality of heartbeats. A plurality of images can be acquired.

  The first image alignment method of the present invention acquires a plurality of images including alignment processing targets, divides each alignment processing target of the plurality of images to set a divided region, and sets each of the alignment processing targets. For each set of divided areas set at the same position, a pixel value conversion method for each divided area is determined based on the pixel value of each divided area of the set, and the pixel value conversion method for the determined divided area is used. Thus, pixel value conversion is performed on the images in the divided areas of the set of divided areas, and alignment processing is performed on the plurality of images on which the pixel value conversion has been performed.

  The second image alignment method of the present invention acquires a plurality of images including alignment processing targets, divides each alignment processing target of the plurality of images to set a divided region, and sets each of the alignment processing targets. For each set of divided areas set at the same position, when the matching process is performed by evaluating the degree of coincidence of each divided area of the set using an evaluation function, for each divided area set, each divided area of the set Based on the pixel value, the evaluation function is changed to perform alignment processing.

  According to the first image registration program of the present invention, a computer sets an area by dividing an image acquisition unit that acquires a plurality of images including an alignment processing target and each alignment processing target of the plurality of images. A pixel that determines a pixel value conversion method for each divided region based on the pixel value of each divided region of the set for each set of divided regions set at the same position of each alignment processing target and the divided region setting unit A value conversion method determination unit, a pixel value conversion unit that performs pixel value conversion on an image in a divided region of a set of divided regions, using the pixel value conversion method of the divided region determined by the pixel value conversion method determination unit; It is made to function as an alignment processing unit that performs an alignment process on a plurality of images that have been subjected to pixel value conversion by the pixel value conversion unit.

  The second image registration program of the present invention sets a divided region by dividing a computer with an image acquisition unit that acquires a plurality of images including a registration processing target and a plurality of registration processing targets of the plurality of images. For each set of divided areas set at the same position of each alignment processing target, an alignment processing section that performs alignment processing by evaluating the degree of coincidence of each divided area of the set using an evaluation function The registration processing unit performs registration processing by changing the evaluation function based on the pixel value of each divided region of the set for each set of divided regions.

  According to the first image alignment apparatus, method, and program of the present invention, a plurality of images including alignment processing targets are acquired, and each alignment processing target of the plurality of images is divided to set divided regions. Then, for each set of divided areas set at the same position of each alignment processing target, a pixel value conversion method for each divided area is determined based on the pixel value of each divided area of the set, and the determined division The pixel value conversion is performed on the images in the divided areas of the set of divided areas by using the pixel value conversion method of the area, and the alignment processing is performed on the plurality of images subjected to the pixel value conversion. Thus, by determining the pixel value conversion method for each divided region and performing the pixel value conversion, even if a structure having a high pixel value in one image shows a low pixel value in the other image, the image The contrast difference between them can be reduced, and alignment can be performed with high accuracy.

  According to the second image alignment apparatus, method, and program of the present invention, a plurality of images including alignment processing targets are acquired, and each alignment processing target of the plurality of images is divided to set a divided region. Then, for each set of divided regions set at the same position of each alignment processing target, the degree of coincidence of each divided region of the set is evaluated using an evaluation function, and alignment processing is performed. Each time, the evaluation function is changed based on the pixel value of each divided region of the set, and the alignment process is performed. By performing the alignment process by changing the evaluation function for each divided region in this way, even when a structure having a high pixel value in one image and a low pixel value in the other image, The contrast difference can be reduced and alignment can be performed with high accuracy.

1 is a block diagram showing a schematic configuration of a medical image diagnosis support system using a first embodiment of an image alignment apparatus of the present invention. The figure which shows an example of the image which carried out the perfusion photography of the heart Diagram for explaining an example of perfusion analysis The flowchart for demonstrating the effect | action of the medical image diagnosis assistance system using 1st Embodiment of the image registration apparatus of this invention. The figure which shows an example of the division area which divided and set the heart image by the cross section perpendicular | vertical to the blood flow direction of the heart The figure for demonstrating the method of determining a pixel value conversion method based on the pixel value for every division area shown in FIG. The block diagram which shows schematic structure of the medical image diagnosis assistance system using 2nd Embodiment of the image registration apparatus of this invention. The flowchart for demonstrating the effect | action of the medical image diagnosis assistance system using 2nd Embodiment of the image registration apparatus of this invention.

  Hereinafter, a medical image diagnosis support system using a first embodiment of an image alignment apparatus and method and a program according to the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram showing a schematic configuration of the medical image diagnosis support system of the present embodiment.

  The medical image diagnosis support system according to the present embodiment acquires two or more images taken at different points in time of the heart and the like, aligns these images, and then performs perfusion analysis using the two or more images. Is what you do.

  Specifically, as shown in FIG. 1, the medical image diagnosis support system of the present embodiment includes an image registration device 1, a medical image storage server 2, a display device 3, and an input device 4. .

  The image alignment apparatus 1 is obtained by installing the image alignment program of this embodiment on a computer.

  The image alignment apparatus 1 includes a central processing unit (CPU (central processing unit)), a semiconductor memory, and a storage device such as a hard disk or an SSD (Solid State Drive). The image registration program of this embodiment is installed in the storage device, and when this image registration program is executed by the central processing unit, the image acquisition unit 10, the divided region setting unit 11, and the like shown in FIG. The pixel value conversion method determination unit 12, the pixel value conversion unit 13, the alignment processing unit 14, the perfusion analysis unit 15, and the display control unit 16 operate.

  The image alignment program is recorded and distributed on a recording medium such as a DVD (Digital Versatile Disc) and a CD-ROM (Compact Disc Read Only Memory), and installed on the computer from the recording medium. Alternatively, the image registration program is stored in a state where it can be accessed from the outside with respect to the storage device or network storage of the server computer connected to the network, and is downloaded to the computer and installed upon request.

  The image acquisition unit 10 acquires a three-dimensional image 6 captured in advance. The three-dimensional image 6 is obtained by imaging a patient using, for example, a CT apparatus or an MRI (Magnetic Resonance Imaging) apparatus. The image acquisition unit 10 according to the present embodiment acquires a three-dimensional image 6 captured by so-called perfusion imaging. Two or more phases of heart images obtained by imaging a heart injected with a contrast medium in time series are obtained. get.

  More specifically, the image acquisition unit 10 acquires two or more heart images having an imaging interval of 1 to several heartbeats as the three-dimensional image 6. The two or more heart images are desirably taken in the systole of each heartbeat. There is no big difference in the shape of the heart included in each heart image taken in the systole of each heartbeat, and the shift is about 1 cm. The deviation of about 1 cm is adjusted by the alignment process described later. In this embodiment, the heart images in the systole are acquired. However, the present invention is not limited to this, and heart images in other phases can be acquired as long as the heart images are acquired when the heart shapes substantially match. You may make it do.

  In the present embodiment, the three-dimensional image 6 obtained by imaging the heart into which the contrast medium is injected is acquired. However, the imaging target is not limited to the heart, and the liver, pancreas, brain, or the like into which the contrast medium is injected is imaged. The obtained image may be acquired as the three-dimensional image 6.

  As the three-dimensional image 6, volume data composed of tomographic images such as an axial tomographic image, a sagittal tomographic image, and a coronal tomographic image may be acquired, or a tomographic image alone may be acquired.

  The three-dimensional image 6 is stored in advance together with the patient identification information in the medical image storage server 2, and the image acquisition unit 10 is based on the patient identification information input by the user using the input device 4 or the like. A three-dimensional image 6 having identification information is read from the medical image storage server 2 and temporarily stored.

  The divided area setting unit 11 divides the alignment processing target included in the two or more three-dimensional images 6 acquired by the image acquisition unit 10 and sets divided areas. The alignment processing target in the present embodiment is the heart.

  The divided region setting unit 11 of the present embodiment includes, as divided regions, a first divided region including the right atrium and the right ventricle in the heart region in each three-dimensional image 6, and a first divided region including the left atrium and the left ventricle. Set the area.

  2I to 2III are diagrams illustrating an example of each heart image captured in the systole of each heartbeat. The region R1 in FIGS. 2I to 2III is the right atrial region, the region R2 is the right ventricular region, the region L1 is the left atrial region, and the region L2 is the left ventricular region. In FIG. 2I, since the boundary between the left ventricle and the myocardium is not clear, the boundary between these regions is not shown. A region combining region R1 and region R2 is set as the first divided region, and a region combining region L1 and region L2 is set as the first divided region. The region M is a myocardial region and the region V is an aortic region. These regions will be described in detail later.

  In the divided region setting unit 11, a heart atlas image (corresponding to a reference image of the present invention) is set in advance. The atlas image is obtained by associating a standard three-dimensional image obtained by photographing the heart in advance with a label image in which a region including the right atrium and the right ventricle and a region including the left atrium and the left ventricle are defined. .

  The divided area setting unit 11 sets a first divided area and a second divided area for the three-dimensional image 6 using the above-described atlas image. Specifically, the divided region setting unit 11 first performs the alignment process between one representative three-dimensional image 6 out of two or more three-dimensional images 6 and the above-described atlas image to represent the representative. A first divided area and a second divided area are set for the three-dimensional image 6. As the alignment process, a known rigid body alignment process or a non-rigid alignment process can be used.

  Then, using the setting results of the first divided region and the second divided region of the representative three-dimensional image 6, the first divided region and the second divided region are set for the other three-dimensional image 6. . Specifically, by associating the coordinate values of the first divided region and the second divided region set in the representative three-dimensional image 6 with the other three-dimensional image 6, the first divided region and the second divided region Set the area. In this way, by applying the setting result of the divided area of one representative three-dimensional image 6 to the other three-dimensional image 6, the load of the setting process of the divided area can be reduced, and the processing speed can be increased. Can do.

  In the present embodiment, the first divided region and the second divided region of the other three-dimensional image 6 using the setting results of the first divided region and the second divided region of the representative three-dimensional image 6. However, the present invention is not limited to this, and the first divided region and the second divided region are set using atlas images not only for the representative three-dimensional image 6 but also for other three-dimensional images 6. You may make it do.

  The pixel value conversion method determination unit 12 determines, for each group of divided areas set at the same position of the heart included in each three-dimensional image 6, based on the pixel value of each divided area of the group. It determines the value conversion method. Note that the pixel value conversion method in this specification includes not performing pixel value conversion.

  Generally, since the pixel value of the contrasted portion in the heart image increases to 100 or more, the pixel value conversion method determination unit 12 determines whether each divided region is contrasted using this numerical value, A pixel value conversion method is determined based on the determination result. In the following, in order to make the description easy to understand, the description will focus on the relationship between two three-dimensional images 6 out of two or more three-dimensional images 6.

  Specifically, the pixel value conversion method determination unit 12 calculates the average pixel value of the first divided area of the first three-dimensional image 6 and the average pixel value of the first divided area of the second three-dimensional image 6. Is calculated. When the average pixel value of the first divided area of the first three-dimensional image 6 is 100 or more and the average pixel value of the first divided area of the second three-dimensional image 6 is less than 100 In other words, if the contrast of the first divided areas of the first three-dimensional image 6 and the second three-dimensional image 6 is different, the pixels of the first divided area of the first three-dimensional image 6 Decide to perform pixel value conversion on the value.

  Conversely, the average pixel value of the first divided area of the first three-dimensional image 6 is less than 100, and the average pixel value of the first divided area of the second three-dimensional image 6 is 100 or more. If it is, it is determined that the pixel value conversion is performed on the pixel value of the first divided area of the second three-dimensional image 6.

  On the other hand, when both the average pixel value of the first divided area of the first three-dimensional image 6 and the average pixel value of the first divided area of the second three-dimensional image 6 are 100 or more or less than 100. That is, when the contrasts of both the first divided areas are close, pixel value conversion is not performed on the first divided areas of the first three-dimensional image 6 and the second three-dimensional image 6. decide. When both the average pixel value of the second divided area of the first three-dimensional image 6 and the average pixel value of the second divided area of the second three-dimensional image 6 are 100 or more, or less than 100, Similarly to the first divided area, it is determined that pixel value conversion is not performed on the second divided areas of the first three-dimensional image 6 and the second three-dimensional image 6.

  The pixel value conversion unit 13 uses the pixel value conversion method determined by the pixel value conversion method determination unit 12 to apply pixel values to the first divided areas of the first three-dimensional image 6 and the second three-dimensional image 6. Conversion is performed, and pixel value conversion is performed on the second divided regions of the first three-dimensional image 6 and the second three-dimensional image 6. Specifically, pixel value conversion is performed on the image in the divided area determined to be subjected to pixel value conversion by the pixel value conversion method determination unit 12 based on the following equation (2). In the following formula (2), x is a pixel value, and f (x) is a pixel value conversion function. By performing pixel value conversion based on the following equation (2), the first three-dimensional image 6 and the second three-dimensional image 6 are divided between the first divided regions or the first three-dimensional image 6 and the second three-dimensional image 6. The contrast difference between the second divided areas of the three-dimensional image 6 can be reduced.

The alignment processing unit 14 performs alignment processing on the first three-dimensional image 6 and the second three-dimensional image 6 that have been subjected to pixel value conversion by the pixel value conversion unit 13. As the alignment processing, rigid body alignment processing or non-rigid alignment processing is performed. As the rigid body alignment processing, for example, processing using an ICP (Iterative Closest Point) method or the like can be used, but other known methods may be used. In addition, as the non-rigid body alignment processing, for example, processing using an FFD (Free-Form Deformation) method or processing using a TPS (Thin-Plate Spline) method can be used. You may make it use. Further, after the rigid body alignment process is performed, the non-rigid body alignment process may be performed.

  In the above description, the registration processing of two three-dimensional images of the first three-dimensional image 6 and the second three-dimensional image 6 has been described. However, in reality, a large number (for example, 30) necessary for perfusion analysis is described. The above-described three-dimensional image 6 is subjected to division region setting, pixel value conversion method determination, pixel value conversion, and alignment processing.

  The perfusion analysis unit 15 performs perfusion analysis using two or more three-dimensional images 6 subjected to the alignment process by the alignment processing unit 14. As a method of perfusion analysis, the above-described Maximum Slope method or Deconvolution method is used. Specifically, as shown in FIG. 3I, the region of interest IR is designated on the three-dimensional image 6 displayed on the display device 3 by the user. The perfusion analysis unit 15 calculates an average density value in each region of interest IR of the two or more three-dimensional images 6 subjected to the alignment process, and a time density curve (Time Density Curve) as shown in FIG. 3II. And a perfusion analysis is performed based on the time concentration curve.

  The display control unit 16 includes a display device such as a liquid crystal display, and displays the analysis result by the perfusion analysis unit 15 on the display device 3. The display control unit 16 causes the display device 3 to display two or more three-dimensional images 6 acquired by the image acquisition unit 10 and two or more three-dimensional images 6 after the alignment processing.

  The input device 4 accepts various setting inputs by a user and includes an input device such as a keyboard and a mouse. The input device 4 receives, for example, setting input for patient identification information and setting input for the region of interest IR described above.

  Note that the display device 3 and the input device 4 may be shared by using a touch panel.

  Next, the operation of the medical image diagnosis support system of this embodiment will be described with reference to the flowchart shown in FIG.

  First, based on the input of patient identification information or the like by the user, two or more three-dimensional images 6 obtained by perfusion imaging of the patient's heart are acquired by the image acquisition unit 10 (S10).

  Two or more three-dimensional images 6 acquired by the image acquisition unit 10 are input to the divided region setting unit 11. The divided area setting unit 11 sets a first divided area and a second divided area for two or more three-dimensional images 6 (S12).

  Next, the pixel value conversion method determination unit 12 determines a pixel value conversion method for the first divided region based on the average pixel value of the first divided region of each three-dimensional image 6, and each three-dimensional image 6. A pixel value conversion method for the second divided region is determined based on the average pixel value of the second divided region (S14).

  Then, using the pixel value conversion method determined by the pixel value conversion method determination unit 12, the pixel value conversion unit 13 uses the pixel values for the first divided region and the second divided region of each three-dimensional image 6. Conversion is performed (S16).

  Next, the three-dimensional image 6 on which the pixel value conversion has been performed is input to the alignment processing unit 14, and the alignment processing unit 14 performs alignment processing (S18).

  The three-dimensional image 6 subjected to the alignment process is input to the perfusion analysis unit 15, and the perfusion analysis is performed in the perfusion analysis unit 15 using the input three-dimensional image 6 (S20).

  The perfusion analysis result by the perfusion analysis unit 15 is input to the display control unit 16, and the display control unit 16 causes the display device 3 to display the input perfusion analysis result (S22).

  According to the medical image diagnosis support system of the above-described embodiment, a plurality of images including alignment processing targets are acquired, and each alignment processing target of the plurality of images is divided to set a divided region. Then, for each set of divided areas set at the same position of each alignment processing target, a pixel value conversion method for each divided area is determined based on the pixel value of each divided area of the set, and the determined division The pixel value conversion is performed on the images in the divided areas of the set of divided areas by using the pixel value conversion method of the area, and the alignment processing is performed on the plurality of images subjected to the pixel value conversion. Thus, by determining the pixel value conversion method for each divided region and performing the pixel value conversion, even if a structure having a high pixel value in one image shows a low pixel value in the other image, the image The contrast difference between them can be reduced, and alignment can be performed with high accuracy.

  In the first embodiment, the divided region setting unit 11 sets the first divided region including the right atrium and the right ventricle and the second divided region including the left atrium and the left ventricle as the divided regions. However, the method of setting the divided areas is not limited to this. For example, a first divided region including the right atrium and the right ventricle, a second divided region including the left atrium and the left ventricle, and a third divided region including the myocardium may be set as the divided regions.

  Alternatively, a first divided region including the right atrium, a second divided region including the right ventricle, a third divided region including the left atrium, and a fourth divided region including the left ventricle are set as the divided regions. It may be. Alternatively, as a divided region, a first divided region including the right atrium, a second divided region including the right ventricle, a third divided region including the left atrium, a fourth divided region including the left ventricle, and a first segment including the myocardium. Five divided areas may be set.

  In addition, as described above, instead of setting a segmented region based on the anatomical structure of the heart, in consideration of the direction in which the contrast agent that changes the pixel value flows, the cross section is perpendicular to the blood flow direction of the heart. A divided region may be set by dividing the heart. FIG. 5 is a diagram illustrating an example in which a divided region is set by dividing the heart along a cross section perpendicular to the blood flow direction of the heart. An arrow A1 shown in FIG. 5 indicates a blood flow direction in the right atrium and the right ventricle, an arrow A2 indicates a blood flow direction in the left atrium and the left ventricle, and D1 to D10 indicate blood of the heart. It shows a divided region set by dividing the heart in a cross section perpendicular to the flow direction. Hereinafter, a method for determining a pixel value conversion method for each divided region when the divided regions D1 to D10 are set as illustrated in FIG. 5 will be described.

  The pixel value conversion method determination unit 12 first checks the average pixel value of each divided region along the blood flow direction for both the first three-dimensional image 6 and the second three-dimensional image 6. Specifically, in the case of the example of FIG. 5, after checking the average pixel value of each divided region along the arrow A1, the average pixel value of each divided region is checked along the arrow A2. Then, a divided region in which the average pixel value is equal to or greater than a preset threshold is specified. That is, the divided region where the pixel value is high due to the contrast agent is specified.

  Then, the pixel value conversion method determination unit 12 determines whether the average pixel value of only one of the divided regions at the same position between the first three-dimensional image 6 and the second three-dimensional image 6 is a threshold value. When it is above, it determines as a division area which performs pixel value conversion. On the other hand, if the average pixel value of both divided areas is equal to or greater than the threshold value or less than the threshold value for the divided area at the same position between the first three-dimensional image 6 and the second three-dimensional image 6, the pixel It is determined as a divided area where value conversion is not performed.

  FIG. 6 is a diagram for explaining a determination method of the pixel value conversion method described above. As shown in FIG. 6, for example, the average pixel values of the divided regions D3 to D6 among the divided regions D1 to D10 of the first three-dimensional image 6 are equal to or greater than the threshold value, and the divided regions D1 to D1 of the second three-dimensional image 6 When the average pixel values of the divided areas D5 to D8 in D10 are equal to or larger than the threshold value, the divided areas D3, D4, D7, and D8 are determined as divided areas for pixel value conversion. On the other hand, the divided area D1, the divided area D2, the divided area D5, the divided area D6, the divided area D9, and the divided area D10 are determined as the divided areas where the pixel value conversion is not performed.

  Note that pixel value conversion and alignment processing after determining a pixel value conversion method are the same as those in the first embodiment.

  Next, a medical image diagnosis support system using the second embodiment of the present invention will be described. FIG. 7 is a block diagram illustrating a schematic configuration of the medical image diagnosis support system according to the second embodiment. The medical image diagnosis support system according to the second embodiment differs from the medical image diagnosis support system according to the first embodiment in the configuration of the image registration device 5. About another structure, it is the same as that of the structure of 1st Embodiment.

  The image alignment apparatus 5 according to the second embodiment includes an image acquisition unit 50, a divided region setting unit 51, an alignment processing unit 52, a perfusion analysis unit 53, and a display control unit 54. About the image acquisition part 50, the division area setting part 51, the perfusion analysis part 53, and the display control part 54, the image acquisition part 10, the division area setting part 11, the perfusion analysis part 15, and display control of the said 1st Embodiment. Since it is the same as that of the part 16, description is abbreviate | omitted.

  The alignment processing unit 52 of the second embodiment matches each divided region of the set for each set of divided regions set at the same position in the first three-dimensional image 6 and the second three-dimensional image 6. The degree is evaluated by an evaluation function to perform alignment processing.

  The alignment processing unit 52 is similar to the alignment processing in the first embodiment in that the alignment processing is performed by evaluating the matching degree of each divided region using an evaluation function. Each is different from the alignment process of the first embodiment in that the alignment process is performed by changing the evaluation function based on the pixel value of each divided region of the set.

  Specifically, the alignment processing unit 52 calculates the average pixel value of the first divided region of the first three-dimensional image 6 and the average pixel value of the first divided region of the second three-dimensional image 6. To do. When the average pixel value of the first divided area of the first three-dimensional image 6 is 100 or more and the average pixel value of the first divided area of the second three-dimensional image 6 is less than 100 That is, when the contrast of the first divided areas of the first three-dimensional image 6 and the second three-dimensional image 6 is different, the degree of coincidence is calculated using the evaluation function of the following expression (3). Then, alignment processing is performed.

S1 in the above equation (3) is an evaluation function of the degree of coincidence, and when the pixel values of the corresponding divided areas between the images are different between the images, S1 is smaller as the positions of the divided areas of the images match. Value. N is the total number of pixels to be compared between images (between the fixed image and the moving image), f is the pixel value conversion function of the above equation (2), xi is the coordinate point, and IF (xi) is the fixed image. The pixel value at xi, IM (xi) is the pixel value at xi of the Moving image, T is the geometric transformation function of the coordinate point, and μ is the geometric transformation parameter. One of the first three-dimensional image 6 and the second three-dimensional image 6 is a fixed image, and the other is a moving image.

  Conversely, the average pixel value of the first divided area of the first three-dimensional image 6 is less than 100, and the average pixel value of the first divided area of the second three-dimensional image 6 is 100 or more. In this case, the degree of coincidence is calculated using the evaluation function of the above equation (3) and the alignment process is performed.

  On the other hand, when both the average pixel value of the first divided area of the first three-dimensional image 6 and the average pixel value of the first divided area of the second three-dimensional image 6 are 100 or more or less than 100. That is, when the contrasts of both the first divided regions are close, the degree of coincidence is calculated using the evaluation function of the following expression (4), and the alignment process is performed. Further, when both the average pixel value of the second divided area of the first three-dimensional image 6 and the average pixel value of the second divided area of the second three-dimensional image 6 are 100 or more or less than 100. Similarly to the first divided region, the degree of coincidence is calculated using the evaluation function of the following expression (4) to perform the alignment process.

S2 in the above formula (4) is an evaluation function of the degree of coincidence, and the value becomes smaller as the pixel values coincide. N is the total number of pixels to be compared between images (between the fixed image and the moving image), f is the pixel value conversion function of the above equation (2), xi is the coordinate point, and IF (xi) is the fixed image. The pixel value at xi, IM (xi) is the pixel value at xi of the Moving image, T is the geometric transformation function of the coordinate point, and μ is the geometric transformation parameter.

  For the alignment process itself, for example, the method described in Non-Patent Document 3 described above can be used.

  Next, the operation of the medical image diagnosis support system of the second embodiment will be described with reference to the flowchart shown in FIG.

  First, based on the input of patient identification information and the like by the user, two or more three-dimensional images 6 obtained by perfusion imaging of the patient's heart are acquired by the image acquisition unit 50 (S30).

  Two or more three-dimensional images 6 acquired by the image acquisition unit 50 are input to the divided region setting unit 51. The divided area setting unit 51 sets a first divided area and a second divided area for two or more three-dimensional images 6 (S32).

  Next, the alignment processing unit 52 determines an evaluation function for the first divided region based on the average pixel value of the first divided region of each three-dimensional image 6, and the second function of each three-dimensional image 6. By determining an evaluation function for the second divided region based on the average pixel value of the divided region, and calculating the degree of coincidence using the evaluation functions determined for the first divided region and the second divided region, respectively. Then, alignment processing is performed on each three-dimensional image 6 (S34).

  The three-dimensional image 6 subjected to the alignment process is input to the perfusion analysis unit 53, and the perfusion analysis unit 53 performs perfusion analysis using the input three-dimensional image 6 (S36).

  The perfusion analysis result by the perfusion analysis unit 53 is input to the display control unit 54, and the display control unit 54 displays the input perfusion analysis result on the display device 3 (S38).

  According to the medical image diagnosis support system of the second embodiment, a plurality of images including alignment processing targets are acquired, and each alignment processing target of the plurality of images is divided to set a divided region. Then, for each set of divided regions set at the same position of each alignment processing target, the degree of coincidence of each divided region of the set is evaluated using an evaluation function, and alignment processing is performed. Each time, the evaluation function is changed based on the pixel value of each divided region of the set, and the alignment process is performed. By performing the alignment process by changing the evaluation function for each divided region in this way, even when a structure having a high pixel value in one image and a low pixel value in the other image, The contrast difference can be reduced and alignment can be performed with high accuracy.

  In the second embodiment, the evaluation function of the above equation (3) and the evaluation function of the above equation (4) are switched and used based on the pixel value of each divided region. The function is not limited to this, and the following expression (5) may be used instead of the above expression (3).

S in the above equation (5) is an evaluation function of the degree of coincidence, and the value is smaller as the pixel values are coincident. N is the total number of pixels to be compared between images (between the fixed image and the moving image), xi is the coordinate point, IF (xi) is the pixel value at xi of the fixed image, and IM (xi) is at xi of the moving image. Pixel value, T is a geometric transformation function of coordinate points, μ is a parameter of geometric transformation, and k and σ are constants.

  In the above equation (5), S is an evaluation function of the degree of coincidence as in the above equation (3), and when the pixel values of the corresponding divided regions between images are different, the positions of the divided regions of the images match. The smaller the value, the smaller S is. N is the total number of pixels to be compared between images (between the fixed image and the moving image), xi is the coordinate point, IF (xi) is the pixel value at xi of the fixed image, and IM (xi) is at xi of the moving image. Pixel value, T is a geometric transformation function of coordinate points, μ is a parameter of geometric transformation, and k and σ are constants. S in the above equation (5) is a function such that the closer the pixel value difference is to a predetermined value determined by the constant σ, the smaller the value and the higher the matching degree.

  In the first to third embodiments, the divided region setting unit 11 sets a divided region using an atlas image of the heart, but the divided region setting method is not limited to this. . Hereinafter, other division area setting methods will be described.

  First, since the aorta is connected to the left ventricle, the change in the pixel value due to the contrast agent is close to that of the left atrium and the left ventricle. By utilizing this fact, an image of a phase in which the pixel values of the left atrium and the left ventricle are increased by the contrast agent is specified.

  Specifically, first, an aorta region is extracted from each of the three-dimensional images 6 of two or more phases. As described above, the region V shown in FIGS. 2I to 2III is the region of the aorta. It should be noted that a known image processing can be used as a method for extracting the region of the aorta.

  Next, based on the change in the average pixel value of the region of the aorta in each three-dimensional image 6, the three-dimensional image of the phase in which the pixel values of the left atrium and the left ventricle are higher than the pixel values of the right atrium and the right ventricle Is identified. Specifically, the three-dimensional image 6 in the phase in which the average pixel value of the aortic region is maximum is specified based on the change in the average pixel value of the aortic region of each three-dimensional image. And the area | region of the left atrium and the left ventricle is extracted by specifying the area | region where the pixel value in the specified three-dimensional image 6 is more than the preset threshold value.

  Next, among the three-dimensional images in which the average pixel value of the aortic region is equal to or greater than a preset lower limit value, the three-dimensional image of the phase in which the average pixel value of the aortic region is the minimum is obtained as the right atrial and right ventricular pixels. It is specified as a three-dimensional image of a phase whose value is higher than the pixel values of the left atrium and the left ventricle. And the area | region of the right atrium and the right ventricle is extracted by specifying the area | region where the pixel value in the specified three-dimensional image 6 is more than the preset threshold value.

  Then, the right atrial and right ventricular regions extracted as described above are set as the first divided regions, and the left atrial and right ventricular regions are set as the second divided regions.

  In addition, the setting method of the first divided region and the second divided region is not limited to the above method, and for example, by using an image processing method such as a known contour extraction, the right atrium and right The ventricle region, the left atrium and the left ventricle region may be extracted, and the first divided region and the second divided region may be set using the extraction result.

1, 5 Image registration device 2 Medical image storage server 3 Display device 4 Input device 6 Three-dimensional image 10 Image acquisition unit 11 Division area setting unit 12 Pixel value conversion method determination unit 13 Pixel value conversion unit 14 Registration processing unit 15 Fusion analysis unit 16 Display control unit 50 Image acquisition unit 51 Division region setting unit 52 Registration processing unit 53 Perfusion analysis unit 54 Display control unit A1 Arrow indicating blood flow direction in right atrium and right ventricle A2 In left atrium and left ventricle Arrow D1-D10 indicating blood flow direction Divided region IR Region of interest K Constant L1 Left atrial region L2 Left ventricular region M Myocardial region R1 Right atrial region R2 Right ventricular region V Aortic region

Claims (13)

An image acquisition unit that acquires a plurality of images including an alignment processing target;
A divided region setting unit configured to divide each alignment processing target of the plurality of images and set a divided region;
For each set of divided regions set at the same position of each alignment processing target, an alignment processing unit that performs alignment processing by evaluating the degree of coincidence of each divided region of the set with an evaluation function,
The image alignment apparatus, wherein the alignment processing unit performs the alignment process by changing the evaluation function based on a pixel value of each divided region of the set for each set of the divided regions. The image acquisition unit acquires an image including a heart image as the alignment processing target;
The image alignment according to claim 1, wherein the divided region setting unit sets, as the divided regions, a first divided region including a right atrium and a right ventricle and a second divided region including a left atrium and a left ventricle. apparatus.   The image position according to claim 2, wherein the divided region setting unit sets the first divided region and the second divided region based on a change in a pixel value of an aortic region included in the plurality of images. Alignment device.   The divided region setting unit, based on a change in a pixel value of an aorta region included in the plurality of images, out of the plurality of images, more than the pixel values of the left atrium and left ventricle regions. And a first image having a higher pixel value in the right ventricular region and a second image having a higher pixel value in the left atrial and left ventricular regions than in the right atrial and right ventricular regions. The image registration apparatus according to claim 3, wherein the first divided region is set using the first image, and the second divided region is set using the second image. The image acquisition unit acquires an image including a heart image as the alignment processing target;
The divided region setting unit sets, as the divided regions, a first divided region including the right atrium and the right ventricle, a second divided region including the left atrium and the left ventricle, and a third divided region including the myocardium. The image alignment apparatus according to claim 1. The image acquisition unit acquires an image including a heart image as the alignment processing target;
The divided region setting unit includes, as the divided regions, a first divided region including the right atrium, a second divided region including the right ventricle, a third divided region including the left atrium, and a fourth divided region including the left ventricle. The image alignment apparatus according to claim 1, wherein the region is set. The image acquisition unit acquires an image including a heart image as the alignment processing target;
The divided region setting unit includes, as the divided regions, a first divided region including the right atrium, a second divided region including the right ventricle, a third divided region including the left atrium, and a fourth divided region including the left ventricle. The image registration device according to claim 1, wherein the region and a fifth divided region including the myocardium are set. The image acquisition unit acquires an image including a heart image as the alignment processing target;
The image registration device according to claim 1, wherein the divided region setting unit divides the heart image by a cross section perpendicular to a blood flow direction of the heart to set a plurality of divided regions.   The divided region setting unit sets the divided region for one of the plurality of images using a reference image in which a region corresponding to the divided region is set in advance, and sets the divided region The image alignment apparatus according to claim 1, wherein a divided region is set for an image other than the one image using a result.   The image alignment apparatus according to claim 1, wherein the image acquisition unit acquires a plurality of images taken at different times.   The said image acquisition part acquires the image containing a heart image as said position alignment process object, and acquires the several image image | photographed with the timing of the same cardiac phase in several heartbeats. The image alignment apparatus described in 1. Acquire multiple images including the alignment target,
Dividing each alignment processing target of the plurality of images to set a divided region,
For each group of divided regions set at the same position of each alignment processing target, when performing the alignment process by evaluating the degree of coincidence of each divided region of the set by an evaluation function, for each group of divided regions An image alignment method, wherein the alignment process is performed by changing the evaluation function based on a pixel value of each divided region of the set. An image acquisition unit for acquiring a plurality of images including a registration processing target;
A divided region setting unit configured to divide each alignment processing target of the plurality of images and set a divided region;
For each set of divided areas set at the same position of each alignment processing target, an image position that functions as an alignment processing unit that performs an alignment process by evaluating the degree of coincidence of each divided area of the set using an evaluation function A combination program,
The image alignment program, wherein the alignment processing unit performs the alignment process by changing the evaluation function based on a pixel value of each divided region of the set for each set of the divided regions.