JP6732593B2 - Image processing apparatus and image processing method - Google Patents

Description

The present invention relates to an image processing device and an image processing method.

In medical care, it is important for a doctor to compare a plurality of different images to make a diagnosis. Examples of different images include images with different imaging devices (modalities), images with different imaging times (date and time), and images with different postures (prone position, supine position, etc.). At the time of comparison, at least one of the captured images may be deformed to align the images. In addition, a difference image between images may be generated using the result of alignment and used for diagnosis.

Depending on the deformation alignment method and scale, it may take a long calculation time of several hours to several days. Therefore, in order to apply the same deformation to different images, it is preferable to store not only the deformation result image but also the deformation parameter.

Patent Document 1 discloses a technique in which displacement amount information of neighboring voxels is represented by one voxel, targeting a displacement vector field, which is the simplest and most versatile form of a deformation parameter. As a result, the amount of data to be saved in a disk is reduced.

JP, 2013-141603, A

However, in the method of Patent Document 1, the number of voxels that represent the displacement amount by one voxel is fixed at a predetermined value. Therefore, for example, when a complex deformation is included, the expression capability of the displacement vector field may be insufficient.

The present invention has been made in view of the above problems. An object of the present invention is to provide a technique for reducing the amount of data while maintaining the expression capability of deformation when performing alignment between a plurality of images.

The present invention employs the following configurations. That is, obtaining means for obtaining a first displacement vector field having a first resolution for transforming the first image to align it with the second image, and the first displacement having the first resolution. Deciding means for deciding the second resolution based on an error in the amount of displacement between the vector field and the second displacement vector field having a resolution different from the first resolution; and the deciding means decided by the deciding means. An image processing device, comprising: a generation unit configured to generate a third displacement vector field with a resolution of 2.

The present invention also employs the following configurations. That is, an acquisition step of acquiring a displacement vector field of a first resolution for deforming the first image and aligning it with the second image; the displacement vector field of the first resolution; A determination step of determining a second resolution based on an error in the amount of displacement of the displacement vector field having a resolution different from one resolution; and generating the displacement vector field at the determined second resolution. An image processing method comprising: a generation step.

According to the present invention, it is possible to provide a technique for reducing the amount of data while maintaining the expression capability of deformation when aligning a plurality of images.

The figure which shows the apparatus structure of the image processing apparatus which concerns on 1st Embodiment. Flowchart showing the processing procedure in the first embodiment Flowchart showing the processing procedure in the second embodiment The schematic diagram which shows the displacement vector field produced|generated in 3rd Embodiment. Flowchart showing the processing procedure in the third embodiment The figure which shows the apparatus structure of the image processing apparatus which concerns on 4th Embodiment. Flowchart showing the overall processing procedure in the fourth embodiment

Hereinafter, preferred embodiments of the present invention will be described with reference to the drawings. However, the dimensions, materials, shapes, and their relative arrangements of the components described below should be appropriately changed depending on the configuration of the apparatus to which the invention is applied and various conditions. Therefore, the scope of the present invention is not intended to be limited to the following description.

TECHNICAL FIELD The present invention relates to a technique of imaging a subject and imaging and displaying a region of interest of the subject. Therefore, the present invention can be understood as an image processing apparatus or an image processing system, a control method thereof, an image processing method, a signal processing method, or an information processing method. The present invention can also be understood as a subject information acquisition apparatus or a control method thereof, or a subject information acquisition method. The present invention can also be regarded as a program that causes an information processing apparatus including hardware resources such as a CPU and a memory to execute each of the above methods, and a computer-readable non-transitory storage medium that stores the program.

The device according to the present invention may have a function of acquiring image data of a subject, or may have only a function of processing image data acquired by another device. The image data of the subject is obtained as a two-dimensional or three-dimensional characteristic information distribution based on the characteristic information at each position. The characteristic information may be obtained not as numerical data but as distribution information of each position in the subject.

Various image capturing devices (modalities) can be used as the three-dimensional image data source of the subject to be processed in the present invention. For example, photoacoustic tomography apparatus (PAT), nuclear magnetic resonance imaging apparatus (MRI), X-ray computed tomography apparatus (CT), ultrasonic echo apparatus (US), positron emission tomography apparatus (PET), single photon emission There are a tomographic imaging device (SPECT), an optical coherence tomographic imaging device (OCT), and the like. Image data indicating distribution information of characteristic information corresponding to each modality is obtained in the three-dimensional region of the subject by imaging (measurement) in each modality.

The present invention is typically used when performing a temporal comparison between images of the same patient taken at the same modality at different times. It is also used when comparing images taken by a plurality of modalities of the same patient at the same time. For example, it is used when comparing the image data obtained by each of the photoacoustic tomography apparatus and the nuclear magnetic resonance imaging apparatus. The photoacoustic tomography apparatus is an apparatus that receives an acoustic wave generated in a subject by irradiating the subject with light and acquires characteristic information of the subject as image data. As the characteristic information, the initial sound pressure in the subject, the light absorption coefficient, the constituent concentration, the oxygen saturation, and the like can be acquired. The nuclear magnetic resonance imaging apparatus is an apparatus for imaging the inside of a subject by utilizing a nuclear magnetic resonance phenomenon generated by applying a high frequency magnetic field to the subject. A nuclear magnetic resonance imaging device can mainly draw information about hydrogen in the subject.
However, the combination of modalities is not limited to this and is arbitrary.

A typical subject is a living breast. In addition, each part of the living body such as hands, feet, head, chest, abdomen, animals other than humans such as mice, and non-living organisms that are easily deformed can be measurement targets.

<First Embodiment>
(Device configuration)
FIG. 1 is a diagram showing an example of an image processing system including the image processing apparatus according to the present embodiment. The image processing system includes an image processing device 100, a data server 150, an operation unit 160, and a display unit 170. The image processing apparatus 100 also includes a data acquisition unit 101, a positioning unit 102, a resolution conversion unit 103, a modified image generation unit 110, and a display control unit 111.

The image processing apparatus according to this embodiment is an apparatus that performs deformation alignment between a first image and a second image of a subject. That is, an image (deformed image) in which the first image is deformed (that is, deformed and aligned with the second image) so that the shape of the subject is the same in the first image and the second image. Is a device for generating.

The image processing apparatus according to the present embodiment outputs not only the deformed image but also the displacement vector field representing the deformation information between the first image and the second image as a result of the deformation alignment. The displacement vector field has a reduced resolution depending on the characteristics of the deformation. In this way, the reduced resolution displacement vector field is stored in the data server in association with which image and which image was used for registration. At this time, the displacement vector field stored in the data server can be used when generating a deformed image in which the third image captured in the same position and orientation as the first image is deformed and aligned with the second image. .. The image processing device uses the displacement vector field to generate a transformed image obtained by transforming the third image, and displays the transformed image on the display unit. In addition, in the state where the position and orientation are the same, a state in which the position and the orientation of the subject in the two images have an error in a range that can be considered to be the same in view of the prerequisite device performance and the purpose of diagnosis. Including.

Here, the displacement vector field is volume data in which the displacement amount in the three-axis directions at that position is stored in each voxel in the three-dimensional image space. Details of the displacement vector field will be described later in the description of the alignment unit 102.

A feature of the image processing apparatus according to the present embodiment is that the resolution of the displacement vector field is reduced based on the alignment result between the first image and the second image. In Patent Document 1 described above, how much the resolution of the displacement vector field is reduced is fixed to a predetermined value. On the other hand, in the present invention, the resolution of the displacement vector field is adaptively reduced so that the error from the modified representation between the two original images is within the allowable range. In this way, by adaptively determining the resolution of the displacement vector field, even if the deformation obtained by inter-image registration is complicated, the amount of data of the displacement vector field can be appropriately reduced without losing the deformation information. it can.

In the present embodiment, the first image and the second image are three-dimensional tomographic images obtained by imaging the breast of the same patient under different imaging conditions (different modality, imaging mode, date and time, posture, etc.).

As described above, it is assumed in the present embodiment that the third image is an image captured in the same position and orientation as the first image. As described above, as an example of different images of the same subject imaged in the same position and orientation, there is an MRI image composed of a plurality of sequences imaged at approximately the same time. In addition, a plurality of images simultaneously captured by a plurality of modalities such as PET/CT are also examples of the first image and the third image.

As shown in FIG. 1, the image processing apparatus 100 is electrically connected to the data server 150, the operation unit 160, and the display unit 170. The data server 150 holds image data and information indicating which set of image data has the same position and orientation. Further, it is possible to hold the displacement vector field obtained as a result of the alignment for each subject. A data holding format in the data server 150 and a method for the image processing apparatus 100 to use the data in the data server 150 will be described later in step S2010. The first, second, and third images and the displacement vector field are input to the image processing apparatus 100 via the data acquisition unit 101 in response to a request from the image processing apparatus 100. The displacement vector field obtained as a result of the alignment is stored in the data server 150 via the resolution conversion unit 103 according to an instruction from the image processing apparatus 100.

As the image processing apparatus 100, an information processing apparatus such as a PC or a workstation, which includes a CPU, a storage device, a communication device, and the like, and operates according to an instruction of a program loaded in the memory can be used. A plurality of information processing devices may cooperate to function as the image processing device 100. The data acquisition unit 101, the alignment unit 102, the resolution conversion unit 103, the modified image generation unit 110, and the display control unit 111, which are functional blocks of the image processing apparatus, are each considered as a module of a program for operating the information processing apparatus. May be. The information processing apparatus operates as the image processing apparatus 100 by each program module performing information processing using the arithmetic resources of the information processing apparatus.

The operation unit 160 operates a mouse, a keyboard, and a GUI on the display unit 170. As a result, the instructions regarding the start and end of the alignment process, the designation of which image the displacement vector field is applied to, and the instructions regarding the start and end of the modified image generation are acquired and input to the image processing apparatus 100.
The display unit 170 displays the display image generated by the image processing apparatus 100. The display unit 170 may also display a GUI for inputting an instruction from the user.

As the operation unit 160, a user interface such as a mouse, a keyboard, a touch panel, or the like, which is connected to the information processing device configuring the image processing device 100, is suitable. Various instructions are given by a doctor or a technician operating these user interfaces. For the display unit 170, any display device such as a liquid crystal panel or a plasma panel can be used.

The data acquisition unit 101 acquires various information input to the image processing apparatus from the data server 150. When the image processing apparatus 100 performs registration between images to calculate the displacement vector field, the first image and the second image are acquired. On the other hand, when the image processing apparatus 100 applies the displacement vector field to generate the deformed image, the image processing apparatus 100 acquires the deformation source image (third image) and the displacement vector field to be applied to the third image. The data acquisition unit corresponds to the image acquisition means of the present invention.

The alignment unit 102 aligns the first image and the second image to calculate the alignment result for deforming the first image and aligning it with the second image. Positioning will be described later in the description of step S2030. The alignment result is output as a displacement vector field. The displacement vector field is volume data in which displacement amounts in the three axial directions are recorded for each voxel in the image.

In the displacement vector field defined on the three-dimensional space, for example, the displacement vector at a certain voxel position pr={xr, yr, zr} in the three-dimensional image space is given by Tpr={txr, tyr, tzr}. .. Thus, the displacement vector field at a certain voxel position is composed of three pieces of data. In order to store these data, three volume data of X-axis direction displacement amount, Y-axis direction displacement amount, Z-axis direction displacement amount are prepared, and each axis is used as a scalar value of the voxel position of each volume data. The displacement amount in the direction is stored.

Thus, in a general displacement vector field, since the displacement amount is stored in each voxel in the three-dimensional image space in which the deformation is defined, the resolution of the displacement vector field is equal to the resolution of the three-dimensional image space. Become. The displacement vector field in the three-dimensional space is composed of three volume data having displacement amounts on the X-axis, the Y-axis, and the Z-axis, respectively. Therefore, the required data amount is the data amount required by the three-dimensional image space. It will be 3 times.

The resolution conversion unit 103 calculates the resolution of the displacement vector field. Then, the displacement vector field is resolution-converted with the resolution calculated in the step described later to generate a displacement vector field with reduced resolution. Then, the generated low resolution displacement vector field is stored in the data server 150. The generation process of the low resolution displacement vector field will be described later in the description of step S2040. The resolution conversion unit corresponds to the determination means, the generation means and the storage means of the present invention.

The deformed image generation unit 110 reads the displacement vector field acquired by the data acquisition unit 101 and the image to which the deformation is applied, and generates the deformed image. Details of the method of generating the deformed image will be described later in the description of step S2050. The modified image generation unit corresponds to the modified image generation means of the present invention.

The display control unit 111 performs display control for displaying the modified image output from the modified image generation unit 110 on the display unit 170. The display control unit corresponds to the display control means of the present invention.

Next, the operation of the image processing system including the image processing apparatus according to this embodiment will be described. FIG. 2 is a flowchart showing a processing procedure performed by the image processing apparatus 100.

(Processing flow)
(Acquisition of subject information: Step S2000)
In step S2000, the data acquisition unit 101 receives an instruction from the user and acquires subject information indicating which image is to be aligned with which image from the operation unit 160. That is, information on the floating image and the reference image is acquired. Then, the data acquisition unit 101 stores the subject information in a storage unit (not shown).

(Distribution processing as to whether or not alignment is required: step S2010)
In step S2010, the image processing system 100 performs a process of deciding whether the alignment needs to be performed or the existing displacement vector field can be applied based on the information of the subject acquired in step S2000. To do.

In order to explain the processing of this step, a concrete example of the data holding format of the data server 150 will be shown. Now, for example, it is assumed that the data server 150 holds three image data of the first, second, and third images. Here, it is assumed that the first image and the third image are registered in the data server 150 as a “set of image data having the same position and orientation”. In the following description, the process of aligning the first image (floating image) with the second image (reference image) is performed, and then the third image (floating image) is aligned with the second image (reference image). The processing procedure performed by the image processing apparatus 100 will be described by taking as an example the case of performing the alignment processing. When the alignment between the first image and the second image is performed, the displacement vector field (first displacement vector field) that deforms the first image to align it with the second image is the data server. Held at 150. At this time, since the first image and the third image have the same spatial orientation, the first displacement vector field can be applied to the process of deforming the third image into the second image. .. In this way, it is possible to determine whether the displacement vector field is applicable by combining the information of the set of image data having the same orientation and the information of the displacement vector field.

As a result of the above processing, when it is determined that there is no applicable displacement vector field and the alignment needs to be performed (“Yes” in S2010), the image processing apparatus 100 advances the processing to step S2020. On the other hand, if there is an applicable displacement vector field and it is determined that alignment is not necessary (“No” in S2010), the process proceeds to step S2070.

As a result of the above processing, when there is no applicable displacement vector field (first displacement vector field) and it is determined that the alignment needs to be performed (“Yes” in S2010), the image processing apparatus 100 performs the processing. To Step S2020. On the other hand, if there is an applicable displacement vector field (first displacement vector field) and it is determined that alignment is not required (“No” in S2010), the process proceeds to step S2070. In the above example, when the first image is aligned with the second image, the process proceeds to step S2020. On the other hand, when aligning the third image with the second image, the process proceeds to step S2070.

(Image data acquisition process: Step S2020)
In step S2020, the data acquisition unit 101 acquires the first image, which is a floating image, and the second image, which is a reference image, from the data server 150. Then, the data acquisition unit 101 outputs the first image and the second image to the alignment unit 102, and outputs the first image to the modified image generation unit 110.

(Alignment process: step S2030)
In step S2030, the alignment unit 102 performs deformation alignment (deformation estimation) between the first image (floating image) and the second image (reference image). Then, as a result of the deformation estimation, the calculated displacement vector field is output to the resolution conversion unit 103. The result of deformation registration shall be output as a displacement vector field. Various known methods can be used as the deformation alignment method as long as the alignment result is output as a displacement vector field. For example, a method such as Demons algorithm or LDDMM (Large Deformation Diffeomorphic Metric Mapping) can be used.

The displacement vector field used in the present embodiment is assumed to be a displacement field that “represents the displacement from the second image to the first image”. That is, this displacement field represents which coordinate position in the first image corresponds to all voxel positions on the second image. It is common to use a "reverse displacement vector field" from the second image to the first image to deform the first image to match the second image. This is because what is needed to generate a transformed image obtained by transforming the first image is "a displacement in the opposite direction," which is "a luminance value at a position on the first image corresponding to a voxel in the second image". It is because it is "information". Thus, since the displacement vector field is defined in the direction of “second image to first image”, the image space in which the displacement vector field is defined matches the image space in which the second image is defined. To do.

(Generation/storing process of displacement vector field with reduced resolution: step S2040)
In step S2040, the resolution conversion unit 103 generates a displacement vector field with the coarsest resolution. At this time, there is a constraint condition that the difference between the resolution of the generated displacement vector field and the resolution of the displacement vector field (original displacement vector field) acquired in step S2030 is within a predetermined allowable amount. The resolution of the original displacement vector field corresponds to the first resolution. The original displacement vector field corresponds to the first displacement vector field.

The predetermined allowable amount of difference from the original displacement vector field will be described in detail. In this embodiment, the allowable amount is determined based on the resolution of the image space expressing the deformation (this is the same value as the resolution of the original displacement vector field). When the voxel size in each axial direction in the space where the displacement vector field is defined (the space of the second image) is sx, sy, sz, the permissible amounts in each axial direction are 2/sx, 2/sy, 2 respectively. /Sz, that is, a value that is half the voxel size in the image space.

The reason for setting the voxel size in this way will be described. The displacement vector represents which voxel in one image is associated with which voxel in the other image by alignment. Even if a difference in displacement vector value is caused by reducing the resolution, if the difference is smaller than half the voxel size, the voxel to which a voxel in one image is connected does not change.

A point to be noted in the calculation of the displacement vector field with the coarse resolution is that the resolution calculation is independently performed for each axis in the displacement vector field. As described above, the displacement vector field consists of three volume data corresponding to the displacement amounts of the X axis, Y axis, and Z axis. The resolution of these three volume data is calculated independently. Further, even one volume data representing the displacement amount of a certain axis has resolution in the X-axis/Y-axis/Z-axis direction. In the processing of this step, the resolution in each of these axial directions is also calculated independently.

ここで、ｓｘ，ｓｙ，ｓｚはそれぞれ、元の変位ベクトル場における各軸方向のボクセルサイズ、Ｎｘ，Ｎｙ，Ｎｚはそれぞれ、元の変位ベクトル場における各軸方向のボクセルの数である。 Now, it is assumed that the displacement amounts in the respective axial directions are stored in Vx, Vy, and Vz. At this time, for example, the displacement vector field when the resolution is ½ is expressed by the following equation.

Here, sx, sy, and sz are voxel sizes in each axial direction in the original displacement vector field, and Nx, Ny, and Nz are the number of voxels in each axial direction in the original displacement vector field.

In the equations (1) and (2), when the resolution is reduced to 1/r, in order to generate a displacement vector field with reduced resolution, skip each r-1 voxel in the original displacement vector field. It is shown to sample to generate a coarse resolution displacement vector field. Here, "r" is an integer of 2 or more. The method of calculating the displacement vector field with reduced resolution is not limited to the method of sampling every r-1 skips. It is possible to use various known methods in which voxel values in a plurality of voxels are represented by one voxel. For example, the maximum value or the median value may be acquired, or linear interpolation or spline function interpolation may be used. That is, in this embodiment, by increasing the sampling interval, it is possible to generate a displacement vector field with a coarse resolution.

ここで、ｐｔは、画像空間Ｓを一定間隔でサンプリングした際の点群Ｓｔの夫々の位置を表す。上式は、サンプリング位置において、解像度を低減した変位ベクトル場の変位量（第２の変位量）と元の解像度の変位ベクトル場の変位量（第１の変位量）との差分を、全サンプリング位置にわたって加算することを意味する。このようにして算出された残差が、あらかじめ設定された許容量以内である場合は、ｒの値を１大きくする。即ち、解像度を１段階粗くする。一方、残差が許容量よりも大きい場合は、その時点で処理を終了し、一つ前の解像度を採用する。即ち現在の解像度がｒ＝２である場合は、ｒ＝１を採用する。これにより第２の解像度が取得される。 As described above, the displacement vector field with the resolution reduced to 1/r is calculated using the equation (1). For the displacement vector field whose resolution is reduced to 1/r, the residual difference from the original displacement vector field is calculated using the formula shown below.

Here, pt represents each position of the point group St when the image space S is sampled at constant intervals. In the above equation, the difference between the displacement amount of the displacement vector field with reduced resolution (second displacement amount) and the displacement amount of the displacement vector field of the original resolution (first displacement amount) is sampled at all sampling positions. Means to add over position. When the residual calculated in this way is within the preset allowable amount, the value of r is increased by 1. That is, the resolution is reduced by one step. On the other hand, when the residual is larger than the allowable amount, the process is terminated at that point and the previous resolution is adopted. That is, when the current resolution is r=2, r=1 is adopted. As a result, the second resolution is acquired.

It should be noted that the sampling point group St must not be the same as the sampling position group (equation (2)) used to generate the reduced resolution displacement vector field. This is because at the sampling position used to generate the displacement vector field, the difference from the original displacement field is always 0 and does not reflect the actual residual. Furthermore, it is desirable that the sampling point group St be more finely sampled than the sampling position group represented by the equation (2). In the simplest case, all voxel positions in the space where the displacement vector field is defined can be the sampling point group St. Alternatively, the sampling point group St may be obtained by randomly sampling the space in which the displacement vector field is defined so that the number of sampling points reaches a specified number.

As described above, in the present embodiment, the case where the difference between the displacement vector field with reduced resolution and the original displacement vector field is used as the allowable amount has been described as an example, but various other various criteria may be used. Is possible.

For example, there is a method of calculating the allowable amount based not on the displacement vector field but also on a deformed image generated using the displacement vector field. As a specific example of this method, the brightness difference between the deformed image generated by applying the original displacement vector field and the deformed image generated by using the displacement vector field with reduced resolution is calculated, and the brightness difference is equal to or less than a certain value. If there is a way to allow. Further, a feature amount other than the brightness may be used. Any other method may be used as long as it is a method of comparing images and acquiring a difference.

As another specific example, when the deformation source image is a plurality of images whose positions and orientations match each other, such as an MRI image having a plurality of sequences, a deformation image is generated for each image and the brightness difference is calculated. There is a way. In this case, since it is possible to consider a plurality of images having different distributions of brightness values depending on the presence or absence of a contrast agent, the resolution of the displacement vector field can be calculated more accurately.

The generated displacement vector field (third displacement vector field) is output to the deformed image generation unit 110, and at the same time, information indicating that "the first image and the second image are aligned" is added. The data is associated and stored in the data server 150.

(Transformed image generation process: step S2050)
In step S2050, the deformed image generation unit 110 deforms the floating image (first image or third image) to generate a deformed image aligned with the reference image (second image). Here, the first image is acquired in step S2020 and the third image is acquired in step S2070 by the data acquisition unit 101. In addition, the displacement vector field (third displacement vector field) is acquired from the resolution conversion unit 103 when it is determined in step S2010 that “it is necessary to perform alignment” (Yes). On the other hand, if it is determined in step S2010 that “there is no need to perform alignment” (No), the data is acquired from the data acquisition unit 101.

ここで、Ｉ１は第１の画像（または第３の画像）、Ｓは第２の画像が定義されている画像空間である。上式は、画像空間中のある位置ｐにおける輝度値は、変位ベクトル場における位置ｐに保持されている変位量を参照して得られる位置の輝度値を用いることを意味している。 The generation of the deformed image is represented by the following equation.

Here, I1 is the first image (or the third image), and S is the image space in which the second image is defined. The above expression means that the brightness value at a certain position p in the image space uses the brightness value at the position obtained by referring to the displacement amount held at the position p in the displacement vector field.

The deformed image generation unit 110 deforms the first image (or the third image) and generates a deformed image aligned with the second image, instead of the deformed image and the second image. It may be configured to generate a difference image. In this case, the modified image generation unit 110 transmits the difference image to the display control unit 111 instead of the modified image. Further, it may be configured to generate and output both the modified image and the difference image.

(Display process of alignment result: step S2060)
In step S2060, the display control unit 111 controls to display any image of the first, second, and third images and the modified image (or the difference image) on the display unit 170 according to the user's operation. .. In addition, when the display control unit 111 receives an instruction input of “end image confirmation” from the user via the operation unit 160 after displaying the alignment image, the process of the image processing apparatus 100 ends.

(Acquisition of image data: Step S2070)
In step S2070, the data acquisition unit 101 acquires the third image, which is a floating image, from the data server 150. Then, the data acquisition unit 101 outputs the image data to the modified image generation unit 110.

(Displacement Vector Field Acquisition Processing: Step S2080)
In step S2070, the data acquisition unit 101 acquires the displacement vector field (third displacement vector field) applied for image deformation from the data server 150. The acquired displacement vector field is output to the deformed image generation unit 110.

Steps S2070 to S2080 are processing executed when it is determined in step S2010 that "it is not necessary to perform alignment" (S2010="No"). That is, the first image having the same position and orientation as the third image and the second image have been aligned in the past, and the displacement of the alignment result is required to deform the third image. This is the processing when the vector field is available.

Here, the first image and the third image are the same subject and have the same spatial positional relationship, for example, like MRI images of different sequences taken at close times, the same. It is assumed that the images are different. In this case, in order to align the third image and the second image, the alignment result of the first image and the second image can be applied as it is. Therefore, when the alignment between the first image and the second image has been performed and the displacement vector field (third displacement vector field) is stored in the data server, the information of the displacement vector field is acquired. To do.

By using the image processing apparatus according to the present embodiment, for example, as (Process 1), the alignment result between certain images can be stored in the data server as a displacement vector field. After that, as (Processing 2), the retained displacement vector field can be used for transformation between images of a combination different from the combination of Processing 1. In this case, the process 1 corresponds to the case of (S2010=“Yes”), and the process 2 corresponds to the (S2010=“No”). The process 1 and the process 2 do not necessarily need to be continuously performed, and the process 2 may be performed at an arbitrary timing after the process 1 is completed.

According to this embodiment, if the images can be applied with the same deformation, the image deformation process can be executed without re-adjusting the position. That is, the alignment process and the image transformation process using the alignment result can be executed at different timings. In that case, the amount of data consumed by the displacement vector field can be reduced.

<Modification 1 of the first embodiment>
In the first embodiment, the case where the alignment result is acquired as the displacement vector field in the alignment process of step S2030 has been described as an example. However, instead of the displacement vector field which is a general-purpose format, a configuration of "information about which deformation model to use" and "deformation parameter for expressing deformation in the model" may be acquired. Even with such deformation information that is a combination of the deformation model and the deformation parameters, the deformation alignment can be defined at a plurality of resolutions.

When the deformation model is used, the deformation alignment between the first image and the second image can be performed by various configurations in which various known deformation models and cost functions are combined. For example, a configuration using FFD (Free Form Deformation) as a deformation model and a similarity evaluation scale between images such as normalized cross-correlation as a cost function can be used. In this case, the alignment result is output as the FFD deformation parameter. In addition, various well-known deformation models and cost functions such as RBF (Radial Basis Function) and dynamic models as deformation models and normalized mutual information and distance values between anatomical corresponding points as cost functions can be used. Available.

ここで、関数Ｄは、変形パラメータベクトルがｖである変形モデルＭに基づいて、位置ｐにおける変位量を３次元ベクトルとして算出する関数である。式（５）及び式（６）は
、第１の実施形態におけるオリジナル変位ベクトル場の代わりに、変形モデルによる変形表現を用いることを意味する。 When the modified model is used, the equations (1) and (3) in the first embodiment are replaced with the following equations (5) and (6), respectively.

Here, the function D is a function that calculates the displacement amount at the position p as a three-dimensional vector based on the deformation model M whose deformation parameter vector is v. Expressions (5) and (6) mean that a modified expression by a modified model is used instead of the original displacement vector field in the first embodiment.

Note that the vector indicating the displacement amount obtained by using the deformation model instead of the original displacement vector field in the first embodiment can also be called the first displacement vector field. Further, the vector indicating the displacement amount obtained by making the sampling interval larger by using the deformation model can be said to be the second displacement vector field. Further, the displacement vector field obtained by sampling the first displacement vector field obtained by using the deformation model at a larger sampling interval can also be called the second displacement vector field. Further, the third displacement vector field may be generated and stored by using the sampling interval determined based on the error between the first displacement vector field and the second displacement vector field and the deformation model.

According to this modification, the registration result is stored as the displacement vector field regardless of which deformation model is used for the deformation registration between the first image and the second image. Therefore, it is not necessary to know which deformation model was used when a deformation image is generated by using the alignment result later, and thus it is possible to perform general-purpose processing.

<Second Embodiment>
An example of an embodiment of the present invention will be described in detail with reference to the drawings. The configuration of this embodiment is the same as that of the first embodiment shown in FIG. However, the processing of the resolution conversion unit 103 is different.

In the first embodiment, when the resolution conversion unit 103 generates the displacement vector field with the reduced resolution in step S2040, the entire image space in which the first image is defined is taken into consideration. On the other hand, the image processing apparatus according to the present embodiment is characterized in that the resolution is calculated by considering only the local area having a large deformation complexity, not the entire image space in which the first image is defined. It is expected that it will be difficult to reproduce the original deformation in the region where the deformation complexity is high when the resolution of the displacement vector field is lowered. Then, the resolution calculated in this way considering only a part of the space is applied to the entire image space. This means that the residual of the displacement vector field calculated by the equation (3) is calculated using only a part of the image space instead of the entire image space.

Next, the operation of the image processing system including the image processing apparatus according to this embodiment will be described with reference to FIG. In the processing procedure, steps S3000 to S3030 and steps S3070 to S31000 are the same as steps S2000 to S2030 and steps S2050 to S2080 in FIG.

(Calculation of Complexity at Each Sampling Position: Step S3040)
In step S3040, the resolution conversion unit 103 samples the image space in which the first image is defined at regular intervals, and calculates the “complexity” of the deformation represented by the displacement vector field at each sampling position. Then, the sampling position whose complexity exceeds a predetermined condition is recorded in a storage unit (not shown). The predetermined condition can be arbitrarily set according to the calculation time and the accuracy of deformation. In addition, general conditions may be used at first, and can be changed by the user's operation using the operation unit 160.

In this embodiment, bending energy is used as the complexity. This is a known method for quantifying the distortion of deformation around the sampling position. The complexity calculation method is not limited to bending energy, and any known method calculated based on a displacement vector field can be used. For example, it is a method using the Jacobian of the displacement vector field. In this case, a local volume change around the sampling position is calculated as the complexity.

(Calculation of local area based on complexity: S3050)
In step S3050, the resolution conversion unit 103 reads the high-complexity sampling position recorded in step S3040 from a storage unit (not shown), and calculates a resolution of the displacement vector field centered on the sampling position. Set the area. This local area is a rectangular parallelepiped area centered on the sampling position, and the length of each side of the rectangular parallelepiped is the same as the sampling interval in step S3040. By doing so, the local areas defined at each sampling position do not overlap, and by combining the local areas at all sampling positions, the entire image space can be covered. In this step, the resolution converter also functions as the local area acquisition unit of the present invention.

(Generation of displacement vector field with reduced resolution: S3060)
In step S3060, the resolution conversion unit 103 calculates the resolution of the displacement vector field using the local area calculated in step S3050. The calculation method is the same as the method of step S2040 in the first embodiment. However, in the calculation of the residual of the displacement vector field in Expression (3), the entire image space is used in the first embodiment, whereas the local region calculated in step S3050 is used in the present embodiment. ..

According to the present embodiment, since the number of sample positions used in the calculation for reducing the resolution of the displacement vector field is reduced, there is an effect that the calculation time can be reduced as compared with the method of the first embodiment. ..

<Third Embodiment>
An example of an embodiment of the present invention will be described in detail with reference to the drawings. The configuration of this embodiment is the same as that of the first and second embodiments shown in FIG. However, the processing of the resolution conversion unit 103 is different.

In the second embodiment, when the resolution conversion unit 103 generates the displacement vector field with the reduced resolution as the process of step S3060, only the local region having a high deformation complexity, that is, the influence of reducing the resolution is large. I was considering. Then, the resolution calculated using the local area is applied to the displacement vector field of the entire image space. However, in this case, since the high resolution considering the local area is applied to the area where the deformation complexity is small and the effect of reducing the resolution is small, the data amount may increase. there were.

On the other hand, in the image processing apparatus according to the present embodiment, displacement vector fields having a plurality of resolutions are combined, and a high resolution displacement vector field is applied to a local area that is greatly affected by reducing the resolution, and displacement vector fields are applied to other areas. It is characterized by applying a low resolution displacement vector field. That is, in this embodiment, a multi-resolution displacement vector field having a plurality of resolutions is used.

An outline of a plurality of displacement vector fields having different resolutions generated in this step will be described with reference to FIG. The image targeted by the present invention is a three-dimensional image, and the displacement vector field is also three-dimensional volume data, but here it is shown as a two-dimensional image for simplicity. This displacement vector field is assumed to have two resolutions, resolution 1 and resolution 2.

The resolution 1 displacement vector field is the highest resolution displacement vector field. The displacement vector field having a resolution of resolution 1 can represent the deformation of the entire image space with an error within an allowable range (within an allowable amount). On the other hand, even in a displacement vector field with a coarser resolution, the deformation can be represented by the error within the allowable range for the area where the error is within the allowable range (the colored area in the displacement vector field of resolution 1 in FIG. 4). .. With respect to a region that can be expressed even with such a coarse resolution, the resolution 1 is an unnecessarily high resolution.

When the displacement vector field has multiple resolutions as in this embodiment, a region to which the displacement vector field of a certain resolution is not applied is called an "invalid region" for that resolution. For example, in the upper side of FIG. 4 (displacement vector field with resolution 1), the colored area is an invalid area. The information about the invalid area can be saved by, for example, storing a value indicating “invalid” in the invalid area.

The displacement vector field of resolution 2 is defined for the area to which the displacement vector field having the resolution of resolution 2 is applied among the invalid areas in resolution 1. However, in the case where the resolution is the coarsest resolution, if an invalid region exists, a region to which the displacement vector field of any resolution is not applied is generated. Therefore, it is assumed that the displacement vector field does not have the invalid region. The displacement vector field of resolution 2 is a displacement vector field of coarse resolution by 2 times in each axial direction as compared with the displacement vector field of resolution 1. As described above, in the present embodiment, it is assumed that the larger the resolution 1, the resolution 2, the resolution 3,...

The voxel size of the displacement vector field when the resolution value is r does not necessarily have to be r times the original resolution (resolution 1). It may have a coefficient such as 2r times, or the voxel size may double as the value of r increases by 1.

In the present embodiment, it is assumed that which voxel in the image space is assigned with the displacement vector field of which resolution is held in data called “resolution volume”. The resolution volume has the same resolution as the image space. Each voxel of the resolution volume holds the value of the resolution of the displacement vector field for calculating the displacement of that voxel (for example, resolution 1, resolution 2,... ). Since the resolution volume is temporary data for resolution calculation, it is not stored in the data server. Therefore, the resolution volume does not consume the data amount of the data server.

Next, the operation of the image processing system including the image processing apparatus according to this embodiment will be described with reference to FIG. In the processing procedure, steps S5000 to S5030 and steps S5110 to S5130 are the same as steps S2000 to S2030 and steps S2060 to S2080 in FIG.

(Set the current displacement vector field resolution to 1: step S5040)
In step S5040, the resolution conversion unit 103 sets the resolution of the current displacement vector to 1 as a pre-stage of the displacement vector field generation processing in which the resolution is reduced. As described in the explanation of the equation (3) in the first embodiment, the resolution 1 means that the resolution of the displacement vector field is the image space in which the displacement vector field is defined (that is, the second image is defined). (Space) means that the resolution is the same. In other words, the resolution 1 is the highest resolution that represents the deformation of the image space. The current resolution value is stored in a storage unit (not shown) because it needs to be referred to in the subsequent steps.

(Calculation of processing target area: step S5050)
In step S5050, the resolution conversion unit 103 calculates an area to be processed in the processing of the current resolution in the image space. The calculated processing target area is held in a storage unit (not shown) as a set of voxel coordinate positions.

A method of calculating the processing target area in this embodiment will be specifically described. First, when the current resolution is 1 (rc=1), the processing target area is the entire area of the image space. On the other hand, when the current resolution is 2 or more, the area is classified as the "temporary invalid area" in the process for the previous resolution (rc-1). The temporary invalid area will be described later in the description of step S5060. In other words, only the areas that were not processed in the higher resolution are processed in the lower resolution.

(Distribution of the processing target area into the resolution fixed area and the temporary invalid area: step S5060)
In step S5060, the resolution conversion unit 103 allocates all voxels in the processing target area calculated in step S5050 to either the “resolution determination area” or the “temporary invalid area”. The resolution-determining area is an area in which the deformation is expressed by a displacement vector field having the previous resolution (rc-1). On the other hand, the temporary invalid region is a region in which the error is within the allowable range even when the displacement vector field whose resolution is coarser than the current resolution (rc) is used. With respect to the temporary invalid region, it is undecided at this point how far the resolution of the displacement vector field can be roughened, but it is decided through the process when the current resolution is further roughened.

ここで、ｒは現在処理対象としている解像度である。式（７）で算出される値が、あらかじめ決めた許容範囲内である場合「仮無効領域」に振り分け、許容範囲を超えている場合「解像度確定領域」に振り分ける。 The method of allocating the resolution fixed area and the temporary invalid area is as follows. The displacement vector field having a resolution of rc is given by the equation (3) in the first embodiment. Then, assuming that the coordinates of the voxel position are ps, the error with respect to the original displacement vector field is given by the following equation.

Here, r is the resolution currently being processed. When the value calculated by the equation (7) is within the predetermined allowable range, it is distributed to the "temporary invalid area", and when it exceeds the allowable range, it is distributed to the "resolution fixed area".

Incidentally, if the current resolution is 1, that is the same as the resolution of the image space, since V in Equation _{(7) (p s) (} r) and V (p _s) is the same value, the total area In, the calculation result of the equation (7) becomes 0. In other words, the entire area is assigned to the “temporary invalid area”.

After distributing all voxels in the processing target area to the “resolution determination area” or the “temporary invalid area”, the distribution result is recorded in the “resolution volume”. The value of the previous resolution (rc-1) is recorded in the voxel allocated to the resolution fixed area, and the value indicating "invalid" is recorded in the voxel allocated to the temporary invalid area.

(Processing for Determining Whether or Not Resolution of All Areas of Image Space is Determined: Step S5070)
In step S5070, the resolution conversion unit 103 performs a determination process as to whether or not the resolution of the entire image space area has been determined. Then, when it is determined that the resolution has been determined (“Yes” in step S5070 of FIG. 5), the process proceeds to step S5090. On the other hand, if it is determined that there is a voxel whose resolution has not been determined yet (“No” in step S5070 of FIG. 5), the process proceeds to step S5080.

The determination process as to whether or not the resolution of the entire area of the image space is fixed is performed using the resolution volume. When all voxels in the resolution volume are scanned and resolution values are recorded in all voxels, it is determined that "resolution has been fixed", and even one voxel that has been recorded as "invalid" is recorded. Is present, it is determined that “resolution is not determined”.

(Increase the resolution of the current displacement vector field by 1: step S5080)
In step S5080, the resolution conversion unit 103 increases the resolution of the current displacement vector field by 1. For example, if the current displacement vector field resolution is rc=1, then the resolution rc=2. As described above in the explanation of the equation (2) in the first embodiment, when the resolution is rc, the resolution of the original displacement vector field having the same resolution as the image space is 1/rc in each axial direction. Means that.

(Generation of Displacement Vector Field with Reduced Resolution: Step S5090)
In step S5090, the resolution conversion unit 103 generates a displacement vector field with reduced resolution. Through the processes from step S5050 to step S5080, the coarsest resolution (rm) of the displacement vector field is calculated and recorded in the storage unit (not shown). The input in the processing of this step is the displacement vector field of resolution 1 calculated in step S5030, the resolution volume calculated through the processing of steps S5050 to S5080, and rm. By inputting these pieces of information, a plurality of displacement vector fields corresponding to respective resolutions from 1 to rm are generated.

The method of generating the displacement vector field corresponding to a certain resolution rc is as follows. First, all voxels in the resolution volume are scanned to obtain all voxels assigned to the resolution rc. Now, when the resolution of the displacement vector field is rc, the voxel size in each axial direction is {sx×rc, sy×rc, sz×rc}. Then, a circumscribed rectangular parallelepiped having a voxel size of {sx×rc, sy×rc, sz×rc} and including all voxels assigned to the resolution rc is defined. Then, for each voxel of the circumscribed rectangular parallelepiped, the displacement vector field is calculated using the equation (3) in the first embodiment. In this way, a displacement vector field of resolution rc is generated.

When there is no voxel assigned to the resolution rc currently being processed, the displacement vector field corresponding to the relevant resolution is not generated. Such a situation occurs when the deformation of the entire region in the image space can be represented by a displacement vector field having a resolution of 2 or more and a displacement vector field having a resolution of 1 is not required. Since the displacement vector field with unnecessary resolution is not generated, the amount of data required to store the displacement vector field can be suppressed.

Further, as described above, the displacement vector field of a certain resolution rc is not defined over the entire area of the image space, but is defined in the circumscribed rectangular parallelepiped area including all the voxels assigned to the resolution rc. As described above, since the displacement vector field of each resolution is defined only in the necessary area, the amount of data required to store the displacement vector field can be suppressed.

(Transformed image generation process: step S5100)
In step S5100, the deformed image generation unit 110 generates a deformed image using the displacement vector field having the plurality of resolutions generated in step S5090. What is different from the deformed image generation process (step S2050) in the first embodiment is which resolution of the displacement vector field is used when acquiring the displacement amount from the displacement vector field in order to generate the deformed image. This is a point that needs to be selected.

The displacement amount acquisition from the displacement vector field having a plurality of resolutions is performed by the following procedure. Now, assume that the coordinate value for acquiring the displacement amount is ps. First, for a displacement vector field of resolution 1, the displacement amount at ps is acquired. If the displacement amount is a valid value, that value is adopted. When the value representing the "invalid area" is stored, the displacement vector field of the next resolution (resolution 2) is searched. In this way, the resolution 1, the resolution 2, and so on are sequentially searched, and the processing is performed until the effective resolution is reached.

The process of generating a deformed image from the displacement amount thus acquired is the same as the process of step S2050 in the first embodiment.

According to this embodiment, a high-resolution displacement vector field is assigned to a local area that is greatly affected by reducing the resolution, and a low-resolution displacement vector field is assigned to a local area that is less affected by reducing the resolution. Then, these displacement vector fields are combined to represent the deformation of the entire image space. Therefore, as compared with the case where the resolution of the displacement vector field of the entire image space is determined according to the region where the effect of reducing the resolution is large as in the second embodiment, the expression capability of the displacement vector field is impaired. There is an effect that the amount of data can be reduced.

<Fourth Embodiment> (The first image and the third image do not have to spatially match)
An example of an embodiment of the present invention will be described in detail with reference to the drawings. FIG. 6 is a diagram showing an example of an image processing system including the image processing apparatus according to the present embodiment. The image processing system includes an image processing device 600. The image processing device 600 includes a first alignment unit 602 and a second alignment unit 609. The function of each block of the image processing device 600 can be realized by the information processing device operating according to a program, as in the case of FIG. The description of the same configurations, functions, and operations as those of the first to third embodiments, such as the data acquisition unit 601, the resolution conversion unit 603, the modified image generation unit 610, and the display control unit 611, will be simplified and mainly described. Differences from the present embodiment will be described.

In the first to third embodiments, the displacement vector field obtained as a result of the alignment between the first image and the second image is stored in the data server 150, and then the spatial vector field is spatially separated from the first image. The case where the displacement vector field is applied to the third image having the same position and orientation has been described. In these embodiments, it is assumed that the first image used for the alignment and the third image that is the generation source of the deformed image have the same spatial position and orientation.

On the other hand, the image processing device 600 according to the present embodiment relaxes this premise, and the displacement vector field is calculated even if the spatial position and orientation between the first image and the third image do not completely match. It is characterized in that a deformed image can be generated by applying it. This is because the fourth image in which the third image is aligned with the first image is generated prior to step S2050 (the process of applying the displacement vector field to the third image) in the first embodiment. Will be realized.

Examples of the first image and the third image include, for example, MRI images of different sequences captured at the same time. In this case, since the first image and the third image are captured at approximately the same time, the displacement occurring between the first image and the third image is a small displacement that can be approximated by rigid body transformation. It is believed that there is. Besides, it is also possible to use a morphological image and a functional image that are simultaneously captured, such as a PET-CT image, as the first image and the third image.

As described above, in the present embodiment, it is assumed that the displacement between the first image and the third image is a displacement that is small enough to be approximated by rigid body transformation, and manual rigid body alignment is performed. Then, the third image is aligned with the first image. Then, an image obtained by displacing the third image and matching it with the first image is referred to as a fourth image. In the present embodiment, the second alignment unit 609 aligns the first image and the third image. Then, the third image is displaced to generate the fourth image aligned with the first image.

The deformed image generation unit 610 applies the displacement vector field acquired from the data acquisition unit 601 to the fourth image acquired from the second alignment unit 609 to generate a deformed image.

Next, the operation of the image processing system including the image processing apparatus according to this embodiment will be described. FIG. 7 is a flowchart showing a processing procedure performed by the image processing device 600. In addition, comparing the processing procedure in the first embodiment shown in FIG. 2, "rigid body alignment (step S
7090)” is added, and is the same as the first embodiment.

(Rigid body alignment processing: Step S7090)
In step S7080, the second alignment unit 609 aligns the transformation source image (third image) acquired from the data obtaining unit 601 with the transformation source image (first image) in step S2030. Then, a second modified image, which is a modification of the third image, is generated based on the alignment result, and is output to the modified image generation unit 610.

Positioning is performed by obtaining an instruction from the user. The user visually confirms the first image, the third image, and the image in which these images are superimposed and displayed, and moves/rotates the third image to align with the first image. Perform processing. The user issues a position alignment instruction via the operation unit 160. Note that the present embodiment may have a configuration in which the alignment between the first image and the third image is performed automatically, instead of obtaining the user's instruction and performing the alignment manually. Various known methods can be used for automatic alignment.

According to the present embodiment, even when the first image and the third image do not spatially match, as in the case of the first embodiment, the resolution retained in the data server 150 is reduced. A displacement vector field can be used.

As described above, according to the image processing device and the image processing method according to the embodiments of the present invention, even when the resolution of the images for which the inter-image registration is performed is high, the displacement vector is not lost and the displacement vector is not lost. The amount of data in the field can be reduced.

<Other embodiments>
The present invention can also be implemented by a computer (or a device such as a CPU or MPU) of a system or apparatus that realizes the functions of the above-described embodiments by reading and executing a program recorded in a storage device. Further, for example, the present invention can be implemented by a method including steps executed by a computer of a system or apparatus that realizes the functions of the above-described embodiments by reading and executing a program recorded in a storage device. .. It can also be realized by a circuit (for example, ASIC) that realizes one or more functions. For this purpose, the program is stored in the computer through a network or from various types of storage media that can serve as the storage device (that is, a computer-readable storage medium that holds data non-temporarily). Provided to. Therefore, the computer (including devices such as CPU and MPU), the method, the program (including program code and program product), and the computer-readable recording medium that holds the program non-temporarily are all the present book. It is included in the category of invention.

100, 600: image processing device, 102: alignment unit, 103, 603: resolution conversion unit

Claims (20)

Acquisition means for acquiring a first displacement vector field of a first resolution for deforming the first image and aligning it with the second image;
Determining means for determining the second resolution based on an error in the amount of displacement between the first displacement vector field having the first resolution and the second displacement vector field having a resolution different from the first resolution. When,
An image processing apparatus, comprising: a generation unit configured to generate a third displacement vector field at the second resolution determined by the determination unit. The image processing apparatus according to claim 1, further comprising storage means for storing the third displacement vector field generated at the second resolution. The image processing apparatus according to claim 1, wherein the acquisition unit acquires the first displacement vector field of the first resolution by using a deformation model. 4. The image processing according to claim 1, wherein the first image and the second image are images obtained by imaging a subject under different imaging conditions. apparatus. The image processing apparatus according to claim 4, wherein the imaging condition is an imaging modality. Image acquisition means for acquiring the first image and the second image;
A deformed image generating unit that generates a deformed image obtained by deforming the first image using the third displacement vector field of the second resolution, and generates a difference image between the deformed image and the second image. When,
The image processing apparatus according to any one of claims 1 to 5, further comprising: Image acquisition means for acquiring the first image or a third image in which the position and orientation of the subject are substantially the same as the first image;
Deformed image generation means for generating a deformed image obtained by deforming the first image or the third image using the third displacement vector field of the second resolution,
The image processing apparatus according to any one of claims 1 to 5, further comprising: Image acquisition means for acquiring the first image and a third image in which the subject is the same and has different positions and orientations;
The third image is deformed to generate a deformed image aligned with the first image, and the deformed image obtained by deforming the third image is converted into the third displacement vector field of the second resolution. Deformed image generation means for further deforming using
The image processing apparatus according to any one of claims 1 to 5, further comprising: The determining means determines an error between a displacement amount due to the first displacement vector field having the first resolution and a displacement amount due to the second displacement vector field having a resolution lower than the first resolution to be a predetermined value. 9. The image processing apparatus according to claim 1, wherein a resolution that is lower than the first resolution is determined as the second resolution when it is within the allowable amount of. The determining means is
Sampling a plurality of voxels included in the first displacement vector field of the first resolution to obtain the second displacement vector field of a resolution lower than the first resolution,
In each of the sampled voxels, a first displacement amount by the first displacement vector field having the first resolution and a second displacement vector field by the second displacement vector field having a resolution lower than the first resolution. 2 displacement amount is acquired, and a difference between the first displacement amount and the second displacement amount is acquired,
The image processing apparatus according to claim 9, wherein a value obtained by adding the difference in each of the sampled voxels is compared with the predetermined allowable amount. It said determining means before Symbol first and the deformed image of said first using the displacement vector field to deform said first image resolution, the first resolution said reduced resolution than the second resolution difference obtained using the displacement vector field by comparing the deformed image obtained by deforming the first image, if it is within the allowable amount of Jo Tokoro, which is lower than the first resolution of the image processing apparatus according to any one of claims 1 to 8, wherein the determining a second resolution. Further having a local area acquisition means for acquiring a local area having a large degree of deformation complexity,
12. The image processing apparatus according to claim 1, wherein the determining unit determines the second resolution of the third displacement vector field based on information of the local area. .. The third displacement vector field is composed of a plurality of displacement vector fields having different resolutions, and which of the plurality of displacement vector fields is applied to each voxel of the third displacement vector field. 13. The image processing apparatus according to claim 1, wherein is set. Acquisition means for acquiring a first displacement vector field for deforming the first image and aligning it with the second image using the deformation model;
Wherein a first displacement vector field, and determining means wherein the first displacement vector field based on the error of the amount of displacement of the second displacement of vector field sampling interval are different, determines the sampling interval,
An image processing apparatus comprising: a generation unit configured to generate a third displacement vector field at the sampling interval determined by the determination unit. The image processing apparatus according to claim 14, further comprising a storage unit that stores the third displacement vector field. Computer
An acquisition step of acquiring a displacement vector field of a first resolution for deforming the first image to align it with the second image;
A determining step of determining a second resolution based on an error in the amount of displacement between the displacement vector field having the first resolution and the displacement vector field having a resolution different from the first resolution;
An image processing method characterized by executing a generation step of generating the displacement vector field in said second resolution determined by the determining step. Computer
The image processing method according to claim 16, characterized by further executing a storage step of storing the displacement vector field generated by the second resolution by the generation step. Computer
An acquisition step of acquiring a first displacement vector field for deforming the first image and aligning it with the second image using the deformation model;
A determining step of determining a sampling interval based on an error in displacement amount between the first displacement vector field and a second displacement vector field having a sampling interval different from the first displacement vector field;
A generating step of generating a third displacement vector field at the sampling interval determined by the determining step;
An image processing method characterized by executing . Computer
The image processing method according to claim 18, characterized by further executing a storage step of storing said third displacement vector field. A program for causing a computer to execute the image processing method according to any one of claims 16 to 19.

Images