JP2023156934A - Medical image processing device, medical image processing method, and medical image processing program - Google Patents

Abstract

To provide a medical image processing device capable of improving accuracy of non-rigid body positioning of a plurality of pieces of medical image data.SOLUTION: A medical image processing device includes a processing unit. The processing unit acquires first medical image data and second medical image data composed of two-dimensional or three-dimensional pixels indicating a subject, executes non-rigid body positioning processing for executing non-rigid body positioning between the first medical image data and the second medical image data by deforming the second medical image data for the fixed first medical image data, and displays the first medical image data and the second medical image data subjected to the non-rigid body positioning in a display unit. The non-rigid body positioning processing includes processing for generating deformation information on the deformation of the second medical image data in a region at least including a reachable region of the second medical image data, and processing for deforming the second medical image data on the basis of the deformation information. The deformation information includes moving information on a movement of at least one pixel included in the second medical image data in a region that is not included in a space region of the first medical image data.SELECTED DRAWING: Figure 3A

Description

The present disclosure relates to a medical image processing device, a medical image processing method, and a medical image processing program.

Conventionally, non-rigid registration has been performed for registration of two pieces of medical image data. For example, it is known to perform 3D PET-CT registration of the chest (see Non-Patent Document 1). Furthermore, it is known to non-rigidly align medical volume data acquired at different times or different types of volume data (see Non-Patent Document 2).

David Mattes, David R. Haynor, Hubert Vesselle, Thomas K. Lewellen, William Eubank, "Nonrigid multimodality image registration", Medical Imaging 2001 Yuichiro Tajima, Koichi Ito and Takafumi Aoki, “A Non-Rigid Registration Method for Medical Volume Data Using 3D Phase-Only Correlation” 21st International Conference on Pattern Recognition (ICPR 2012)

In conventional non-rigid registration, the accuracy of non-rigid registration of multiple pieces of medical image data is insufficient, and there is room for improvement.

The present disclosure has been made in view of the above circumstances, and provides a medical image processing device, a medical image processing method, and a medical image processing program that can improve the precision of non-rigid positioning of a plurality of medical image data.

One aspect of the present disclosure is a medical image processing apparatus, which includes a processing section, and the processing section is configured to process first medical image data and second medical image data that are composed of two-dimensional or three-dimensional pixels indicating a subject. By acquiring medical image data and transforming the second medical image data with respect to the fixed first medical image data, the first medical image data and the second medical image data are transformed. Execute a non-rigid alignment process to perform rigid body alignment, display the non-rigid aligned first medical image data and second medical image data on a display unit, and display the second medical image data on a display unit. The spatial region includes a portion of the first medical image data that is not included in the spatial region, and the non-rigid alignment process determines a reachable region that is a region that the second medical image data can reach by deformation. A process of generating deformation information regarding deformation of the second medical image data in at least a region including the deformation information, and a process of deforming the second medical image data based on the deformation information, wherein the deformation information is , a medical image processing apparatus including movement information regarding movement of at least one pixel included in the second medical image data in an area not included in the spatial area of the first medical image data.

One aspect of the present disclosure includes a step of acquiring first medical image data and second medical image data that are composed of two-dimensional or three-dimensional pixels indicating a subject, and a step of acquiring the first medical image data that is fixed. performing non-rigid registration of the first medical image data and the second medical image data by deforming the second medical image data; displaying the first medical image data and the second medical image data on a display unit, wherein the spatial region of the second medical image data is the same as the spatial region of the first medical image data. The step of performing the non-rigid alignment includes the non-rigid alignment of the second medical image data in an area that includes at least a reachable area that is an area that the second medical image data can reach by deformation. the step of generating deformation information regarding deformation; and the step of deforming the second medical image data based on the deformation information, the deformation information being included in a spatial region of the first medical image data. The medical image processing method includes movement information regarding movement of at least one pixel included in the second medical image data in a region where the second medical image data does not move.

One aspect of the present disclosure is a medical image processing program for causing a computer to execute the above medical image processing method.

According to the present disclosure, it is possible to improve the accuracy of non-rigid alignment of multiple pieces of medical image data.

A block diagram showing an example of the hardware configuration of a medical image processing device in the first embodiment Block diagram showing an example of functional configuration of a medical image processing device A diagram showing an example of the shape of the deformed mesh and volume data before and after deformation of the fixed side volume and deformed side volume. Diagram showing an example of movement of each point before and after deformation of the deformed volume A diagram showing a first example of a deformed mesh in the first embodiment A diagram showing a second example of the deformed mesh in the first embodiment A diagram showing an example of a convex hull region including a deformed volume and a fixed volume. Diagram showing an image of the arterial phase Diagram showing an image of the delayed phase Flowchart showing a first operation example during alignment of the medical image processing device Flowchart showing a second operation example during alignment of the medical image processing device Flowchart showing a second operation example during alignment of the medical image processing device Diagram for explaining a second operation example during alignment of the medical image processing device A diagram showing an example of the alignment result of a wide area including the abdomen of a subject in a comparative example. A diagram showing an example of the alignment result of the abdominal region of a subject in a comparative example. A diagram showing an example of the alignment result of a wide area including the abdomen of a subject in the first embodiment A diagram showing an example of alignment results of the abdominal region of a subject in the first embodiment Diagram showing an example of setting a deformed mesh in a comparative example

Embodiments of the present disclosure will be described below with reference to the drawings.

(How one embodiment of the present disclosure was obtained)

In Non-Patent Documents 1 and 2, a deformed mesh MSX is used in non-rigid positioning. FIG. 12 is a diagram for explaining non-rigid positioning using a conventional deformed mesh MSX.

The deformed mesh MSX is generated based on one piece of medical image data out of a plurality of pieces of medical image data. This medical image data is the medical image data to be aligned, and is not deformed and has a fixed shape, so it is also referred to as fixed image data VFX. The size of the deformed mesh MSX is basically the same as the size of the fixed image data VFX, and one or two rows of fixed points are added to the outer periphery depending on the interpolation formula used. Furthermore, the medical image data to be modified is also referred to as modified image data VMX. Each pixel of the deformed image data VMX corresponds to a movement vector expressed by interpolating each node NDX of the deformed mesh MSX. Therefore, as the node NDX of the deformed mesh MSX moves, the medical image processing apparatus can move and deform the pixels of the deformed image data VMX corresponding to this node NDX.

Here, if the deformed image data VMX is larger than the fixed image data VFX in the three-dimensional space (virtual three-dimensional space), the deformed image data VMX does not fit within the range of the fixed image data VFX, and is outside the range of the fixed image data VFX. There are cases where nodes NDX of the deformed mesh MSX corresponding to pixels of the deformed image data VMX are absent. In this case, an area in the deformed image data VMX where the node NDX of the corresponding deformed mesh MSX is absent cannot be deformed. Therefore, at least a portion of the deformed image data VMX may not be deformable, and in this case, the precision of non-rigid positioning of the plurality of medical image data becomes insufficient.

Note that, conventionally, the area that includes both the fixed image data VFX and the modified image data VMX has been an important target of interest, and it seems that there has been no interest in the accuracy of non-overlapping areas. In addition, in PET-CT registration, which is the main application of non-rigid body alignment, it is considered that areas outside the imaging range of CT data are not of interest (in the case of Non-Patent Document 1, the area outside the imaging range of CT data is Both arms of the patient were not imaged by the CT machine, but this is not considered a problem). Furthermore, in the case of phase-to-phase registration of 4D (four-dimensional) data, which is the main application of non-rigid registration, it is thought that this did not pose a problem because the area of all volumes was the same (Non-patent Document Case 2).

In the following embodiments, a medical image processing apparatus, a medical image processing method, and a medical image processing program that can improve the accuracy of non-rigid alignment of a plurality of medical image data will be described.

(First embodiment)
FIG. 1 is a block diagram showing a configuration example of a medical image processing apparatus 100 according to the first embodiment. The medical image processing apparatus 100 includes a port 110, a UI 120, a display 130, a processor 140, and a memory 150.

A CT device 200 is connected to the medical image processing device 100 . The medical image processing apparatus 100 acquires volume data from the CT apparatus 200 and processes the acquired volume data. The medical image processing apparatus 100 may be configured by a PC and software installed in the PC.

The CT apparatus 200 irradiates a subject with X-rays and captures an image (CT image) by utilizing differences in the absorption of X-rays by tissues within the body. The subject may include a living body, a human body, an animal, or the like. The CT apparatus 200 acquires a sinogram from an X-ray detector, and generates a tomographic image (also referred to as a slice image or slice data) of the subject by image reconstruction based on the sinogram. The CT apparatus 200 generates volume data based on the slice data, for example, by layering the slice data. The slice data and volume data include information on any location inside the subject. The CT apparatus 200 transmits volume data as a CT image to the medical image processing apparatus 100 via a wired line or a wireless line. When taking a CT image, imaging conditions regarding CT imaging and contrast conditions regarding administration of a contrast agent may be taken into consideration. Note that contrast imaging may be performed on the digestive organs, bile ducts, etc. in addition to blood vessels. Contrast imaging may be performed multiple times at different timings depending on the characteristics of the organ.

A port 110 in the medical image processing apparatus 100 includes a communication port, an external device connection port, a connection port to an embedded device, etc., and acquires volume data obtained from a CT image. The acquired volume data may be immediately sent to the processor 140 for various processing, or may be stored in the memory 150 and then sent to the processor 140 for various processing when necessary. Further, the volume data may be acquired via a recording medium or a recording medium. Further, the volume data may be obtained in the form of intermediate data, compressed data, sinogram, slice data, or the like. Further, the volume data may be obtained from information from a sensor device attached to the medical image processing apparatus 100. The port 110 functions as an acquisition unit that acquires various data such as volume data.

UI 120 may include a touch panel, pointing device, keyboard, or microphone. The UI 120 accepts any input operation from the user of the medical image processing apparatus 100. Users may include doctors, radiologists, students, or other paramedic staff.

The UI 120 accepts various operations. For example, operations such as specifying a region of interest (ROI) and setting brightness conditions in volume data or images based on volume data (for example, three-dimensional images and two-dimensional images described later) are accepted. Regions of interest may include regions of various tissues (eg, blood vessels, bronchi, organs, organs, bones, brain). Tissues may include diseased tissue, normal tissue, tumor tissue, and the like.

Display 130 may include, for example, an LCD, and displays various information. The various information may include three-dimensional images and two-dimensional images obtained from volume data. The three-dimensional image may include a volume rendering image, a surface rendering image, a virtual endoscopic image, a virtual ultrasound image, a CPR image, or the like. The volume rendering image may include a RaySum image, a MIP image, a MinIP image, an average value image, a raycast image, or the like. The two-dimensional image may include an axial image, a sagittal image, a coronal image, an MPR image, or the like.

The memory 150 includes various types of ROM and RAM primary storage devices. The memory 150 may include a secondary storage device such as an HDD or an SSD. The memory 150 may include a tertiary storage device such as a USB memory or an SD card. Memory 150 stores various information and programs. The various information may include volume data acquired by the port 110, images generated by the processor 140, setting information set by the processor 140, and various programs. Memory 150 is an example of a non-transitory recording medium in which programs are recorded.

Processor 140 may include a CPU, DSP, or GPU. The processor 140 functions as a processing unit 160 that performs various processes and controls by executing a medical image processing program stored in the memory 150.

FIG. 2 is a block diagram showing an example of the functional configuration of the processing unit 160.

The processing section 160 includes a region processing section 161, an image generation section 162, an alignment processing section 163, and a display control section 165. The processing section 160 controls each section of the medical image processing apparatus 100. The processing unit 160 performs, for example, processing related to alignment of a plurality of pieces of medical image data (for example, volume data). Note that each unit included in the processing unit 160 may be realized as different functions by one piece of hardware, or may be realized as different functions by a plurality of pieces of hardware. Further, each unit included in the processing unit 160 may be realized by a dedicated hardware component.

The region processing unit 161 acquires volume data of the subject via the port 110, for example. The area processing unit 161 extracts an arbitrary area included in the volume data. The region processing unit 161 may automatically specify a region of interest and extract the region of interest based on, for example, voxel values of volume data. The region processing unit 161 may manually specify a region of interest via the UI 120, for example, and extract the region of interest.

The image generation unit 162 generates various images. The image generation unit 162 generates a three-dimensional image, a two-dimensional image, or a tomographic image based on at least part of the acquired volume data (for example, volume data of the extracted region). The image generation unit 162 may generate an image by performing various types of rendering (for example, volume rendering or surface rendering).

The alignment processing unit 163 aligns a plurality of pieces of medical image data. The medical image data may include, for example, volume data, slice data, three-dimensional images or two-dimensional images based on volume data or slice data, or the like. Further, the volume data may be a plurality of volume data created in chronological order. Therefore, alignment may include alignment in a three-dimensional space (virtual three-dimensional space) or alignment in a two-dimensional plane (virtual two-dimensional plane). The alignment here includes at least non-rigid registration, and may further include rigid registration.

Rigid body alignment is alignment in which the shape of a region such as an organ in medical image data does not change, and may include alignment by linear transformation using translation or rotation, for example. Rigid body registration is applied to bones, brains, etc., for example. The alignment processing unit 163 may perform rigid body alignment using a known method.

Non-rigid registration is registration in which the shape of the region of the organ within the medical image data changes. Non-rigid alignment may include, for example, alignment by linear transformation using not only translation or rotation but also enlargement or reduction, or alignment by non-linear transformation instead of linear transformation. For example, the alignment processing unit 163 may express the nodes on the deformed mesh as finite elements, and may express the movement of pixels by the movement of the nodes. The alignment processing unit 163 may express the movement of pixels by interpolating between nodes according to the B-spline method. The alignment processing unit 163 may perform non-rigid alignment by performing alignment using non-linear transformation. The alignment processing unit 163 may perform non-rigid alignment by large deformation simulation using the finite element method.

Note that the image generation unit 162 may generate an image by performing rendering based on volume data as a plurality of medical image data aligned by the alignment processing unit 163. Further, the alignment processing unit 163 may perform alignment, including rigid body alignment and non-rigid body alignment, between images as a plurality of medical image data generated by the image generation unit 162.

The display control unit 165 causes the display 130 to display various data, information, or images. The image is an image representing a part of the tissue inside the subject, and may include, for example, an image generated by the image generation unit 162 or a tomographic image of a predetermined tomogram. For example, the display control unit 165 displays an image based on a plurality of aligned (eg, non-rigid aligned) medical image data.

Next, processing related to alignment will be described in detail.
In this embodiment, the medical image data is mainly exemplified as volume data. Further, alignment of volume data in three-dimensional space will be mainly illustrated.

The alignment processing unit 163 takes measures against misalignment that occurs when a moving volume is larger than a fixed volume in non-rigid alignment. The deformable volume VM is volume data whose shape can be deformed. The fixed volume VF is volume data whose shape does not change (fixed). The alignment processing unit 163 may manually or automatically set which volume data is to be set as the deformed volume VM or the fixed volume VF. For example, it may be determined via the UI 120 which volume data is to be used as the deformable volume VM or the fixed volume VF based on a user operation. For example, it may be determined which volume data is to be used as the deformed volume VM or the fixed volume VF based on the characteristics of the organ, organ, or tissue included in the volume data. For example, organs and blood vessels may be used as the deformed volume VM, and bones may be used as the fixed volume VF.

In non-rigid alignment, the alignment processing unit 163 generates and sets a deformed mesh MS corresponding to a region including the fixed volume VF and the deformed volume VM in the three-dimensional space. Even if there is some positional deviation between the volume data, the alignment processing unit 163 can suppress the positional deviation by deforming the deformable volume VM with respect to the fixed volume VF and performing non-rigid alignment. Therefore, the displayed three-dimensional image or two-dimensional image is an image in which positional deviations between volume data are suppressed, resulting in a natural view for the user and increasing options for treatment and diagnosis.

Next, non-rigid positioning using the deformed mesh MS will be explained.

FIG. 3A is a diagram showing an example of the shapes of the deformed mesh MS and volume data before and after deformation of the fixed volume VF and the deformed volume VM. Note that in the figure, the fixed volume VF is shown as "fixed" and the modified volume VM is shown as "moving". Furthermore, in FIG. 3A, it is imagined that volume data is rendered and displayed as an image on a two-dimensional plane.

The deformation-side volume VM undergoes non-rigid deformation during non-rigid alignment. Non-rigid deformation is expressed as deformation of the deformed mesh MS. The deformed mesh MS is installed corresponding to the coordinates defined by the fixed side volume VF. The deformed mesh MS includes a plurality of nodes ND. A deformed mesh MS is formed by arranging a plurality of nodes ND in a three-dimensional space where non-rigid alignment is performed. The plurality of nodes ND are arranged, for example, in a grid in a three-dimensional space, but may be arranged in other shapes.

The alignment processing unit 163 generates deformation information regarding the deformation of the deformation side volume VM that is deformed using the deformation mesh MS. The deformation information includes movement information regarding movement of each voxel of the deformation volume VM corresponding to each node ND of the deformation mesh MS. That is, the deformation information includes movement information indicating which voxel before deformation (current state) in the deformation side volume VM moves to (corresponds to) which voxel in the fixed side volume. That is, the deformation information indicates information on the correspondence between each point before and after the deformation of the deformed volume VM. Further, the deformation information may include information on the displacement from the initial state in each voxel of the deformable volume VM with reference to the coordinates of the fixed volume VF.

The alignment processing unit 163 sets each node of the deformed mesh MS in correspondence with the coordinates of the fixed volume VF, generates the deformed mesh MS, and generates deformation information. In FIG. 3A, the deformed mesh MS is set so that each node ND forms a grid with respect to the fixed volume VF. Note that the fixed volume VF is not deformed, so in FIG. 3A, as an example, the shape before and after deformation remains triangular and is not deformed. On the other hand, before deformation, the deformable volume VM has a shape different from that of the fixed volume VF, and in FIG. 3A, it has a distorted triangular shape. Therefore, the deformed mesh MS before deformation has a distorted shape when compared with the reference shape (the shape based on the coordinates of the fixed volume VF).

The alignment processing unit 163 moves each node ND of the deformed mesh MS based on the deformation information, moves the corresponding voxel of the deformed volume VM, and changes the deformed volume VM to the shape of the fixed volume VF. , and perform non-rigid alignment. Therefore, the alignment processing unit 163 obtains the deformed volume VM after deformation (post-deformed volume data VMA) from the deformed volume VM before deformation (pre-deformed volume data VMB) based on the deformation information. As a result, as shown in FIG. 3A, the deformable volume VM is deformed into a triangular shape similar to the fixed volume VF.

Further, nodes of the finely deformed mesh MS may not be provided corresponding to all voxels in the fixed volume VF. In this case, the transformation information may include voxels for which movement information does not exist in the transformation volume VM. In this case, the alignment processing unit 163 interpolates (for example, (linear interpolation).

For example, the grid of the deformed mesh MS may be in units of 64 voxels (one for every 64 voxels). Furthermore, the arrangement density of each node ND in the deformed mesh MS does not have to be uniform, and may differ depending on the position (region) within the deformed mesh MS. For example, in a region that can be finely deformed or that is desired to be finely deformed in the deformation side volume, the density of the nodes ND of the deformation mesh MS can be increased to obtain a large amount of deformation information (movement information of voxels within the region) in this region. You can do it like this. For example, in a region in the deformation side volume where a large movement is desired, the density of the nodes ND of the deformation mesh MS may be made small so that the deformation information (movement information of voxels within the region) in this region is stabilized. In this way, the alignment processing unit 163 deforms the deformed volume VM based on the movement information of the voxels of the deformed volume VM that correspond to the coordinate points defined in the fixed volume VF.

FIG. 3B is a diagram showing an example of movement of each point before and after the transformation of the transformation volume VM. In FIG. 3B, as indicated by each arrow, the movement of the node ND is expressed by a vector for each node ND (vertex) on the deformed mesh MS. The movement information of each voxel of the deformation side volume VM corresponding to each node ND of such a deformation mesh MS is included in the deformation information. In FIG. 3B, it can be seen that voxel VXB corresponding to node NDB before transformation has moved to voxel VXA corresponding to node NDA after transformation.

FIG. 4 is a diagram showing a first example of the deformed mesh MS in this embodiment.

The alignment processing unit 163 generates a deformed mesh MS so as to include at least the region of the deformed volume VM, and generates deformation information. Specifically, a deformed mesh MS is generated that includes a reachable region VMR that is a region that the deformed volume VM can reach through deformation. The reachable area VMR includes, for example, the pre-transformation volume data VMB and the post-transformation volume data VMA, and may further include an area of the transformation volume VM that is currently being transformed. For example, when transforming the transformation volume VM (pre-transformation volume data VMB) to match the fixed volume VF, the area that includes at least the fixed volume VF and the transformation volume VM in FIG. 4 is the reachable area VMR. becomes.

Furthermore, the alignment processing unit 163 may generate the deformed mesh MS so as to include both the fixed volume VF and the reachable region VMR of the deformed volume VM. Information about the area of the fixed volume VF and the reachable area VMR of the deformable volume VM may be stored in the memory 150 in advance, or may be obtained from an external device via the port 110. Further, the deformation information includes movement information regarding movement of at least one pixel (voxel) included in the deformation side volume VM in an area not included in the area of the fixed side volume VF.

As a result, the node ND of the deformed mesh MS exists corresponding to any voxel in the pre-deformed volume data VMB, and also exists corresponding to any voxel in the post-deformed volume data VMA. There is. Therefore, the end of the deformation side volume VM will not be located outside the deformation mesh MS. Therefore, the medical image processing apparatus 100 can suitably deform the deformable volume VM up to the end thereof, and can prevent the deformation from becoming impossible.

FIG. 5 is a diagram showing a second example of the deformed mesh MS in this embodiment.

In the example of FIG. 5, the deformed mesh MS is made smaller than that of FIG. Specifically, the node ND of the deformed mesh MS is absent in a region where the deformed volume VM cannot move, that is, outside the reachable region VMR. Therefore, the alignment processing unit 163 generates a deformed mesh of the minimum size necessary for deforming the deformed volume VM. Therefore, if nodes ND are arranged at the same density (fineness) in the deformed mesh MS in FIGS. 4 and 5, the number of nodes ND in FIG. 5 is greater than that in the deformed mesh MS in FIG. It becomes less. Therefore, the amount of calculation involved in deriving the movement information at each voxel of the deformation side volume VM corresponding to each node ND of the deformation mesh MS is reduced, the amount of calculation involved in deriving the deformation information is reduced, and the amount of calculation involved in deriving the deformation information is reduced. The amount of calculation related to deformation is reduced. In this way, the medical image processing apparatus 100 can reduce the deformed mesh MS, and can also reduce the amount of calculations related to non-rigid positioning using the deformed mesh MS.

In FIG. 5, the deformed mesh MS may be larger by one square than the reachable region VMR, that is, the deformed mesh MS may be made larger by one at the node ND. The reason why the deformed mesh MS is set to be slightly larger than the reachable region VMR is to serve as a boundary for spline interpolation processing (for example, B-spline interpolation) and to enable smooth execution of spline interpolation processing. Note that, also in FIG. 4, the deformed mesh MS may be made larger by at least one square than the reachable region VMR. Furthermore, an immovable fixed point may be installed outside the reachable region VMR. This fixed point can be used to facilitate the spline interpolation process.

Further, when the alignment processing unit 163 reduces the number of nodes ND by reducing the deformed mesh MS, it may set immovable fixed points in place of the reduced nodes ND. Thereby, the medical image processing apparatus 100 can also reduce the amount of calculations related to non-rigid body alignment. Furthermore, the alignment processing unit 163 may interpolate the movement of each voxel of the deformed volume VM between the fixed point and the node ND. Note that the fixed point and the movement of each voxel obtained by interpolation are not considered to be included in the deformed mesh MS.

FIG. 6 is a diagram showing an example of a convex hull region VC including a deformable volume VM and a fixed volume VF. The convex hull region VC is an example of a containing region including a deformable volume VM and a fixed volume VF.

When omitting the deformed mesh MS as being absent outside the reachable region VMR as shown in FIG. ) may generate a convex hull region VC. The alignment processing unit 163 may generate the deformed mesh MS in accordance with the size of the convex hull region VC. In this case, the deformed mesh MS may be made larger by one square than the convex hull region VC, and interpolation may be taken into account, as described above. Then, corresponding to this convex hull region VC, each voxel of the deformed volume VM is deformed based on the deformation information, and non-rigid alignment is performed with each voxel of the fixed volume VF. In FIG. 6, since alignment is performed from the deformed volume VM to the fixed volume VF, the convex hull region VC includes the reachable region VMR.

Further, when assuming a lattice-like deformed mesh MS that includes a deformed volume VM and a fixed volume VF, a circumscribed rectangular parallelepiped area that circumscribes the deformed volume VM and the fixed volume VF may be generated. The circumscribed rectangular parallelepiped area is an example of a containing area. Then, the deformed mesh MS may be generated in accordance with the size of this circumscribed rectangular parallelepiped region. Since the deformation side volume VM is aligned with the fixed side volume VF, this circumscribed rectangular parallelepiped area includes the reachable area VMR.

Next, a specific example of the deformable volume VM and the fixed volume VF will be described. The two-dimensional data corresponding to the deformed volume VM and the fixed volume VF will be described as a deformed image and a fixed image.

Here, it is assumed that three-dimensional data (for example, volume data) and two-dimensional data can be acquired not only by the CT apparatus 200 but also by various modality apparatuses via the port 110, for example. The various modality devices include, for example, an MRI (Magnetic Resonance Imaging) device, a PET (Positron Emission Tomography) device, an angiography device, or other modality devices. Further, it is assumed that the size of the deformable volume VM in the three-dimensional space is larger than the size of the fixed volume VF in the three-dimensional space.

The first specific example is as follows. The fixed side volume VF is obtained by imaging in a narrow range because the timing (time) at which it can be imaged is short in the arterial phase (for example, early arterial phase). The deformation side volume VM is a delayed phase, and the imaging time is longer than that of the arterial phase, so it is obtained by imaging a wider range than the arterial phase. Here, since the arterial phase shows finer blood vessels and is mainly used in diagnosis, the user would like to observe the image as it is captured in the arterial phase. In other words, the deformable volume VM is larger in size in three-dimensional space than the fixed volume VF.

FIG. 7A is a diagram showing an image G31 of the arterial phase. FIG. 7B is a diagram showing a delayed phase image G32. The arterial phase image G31 has higher definition than the image G32, and is an image based on a CT image captured by the CT apparatus 200 in a narrower imaging range than the image G32. The delayed phase image G32 has lower definition than the image G31, and is an image based on a CT image captured by the CT apparatus 200 in a wider imaging range than the image G31. In this case, since the arterial phase image G31 serves as the diagnostic standard, it is preferable not to deform it. Furthermore, the size of the image (data) in the three-dimensional space is larger for the image G32 than for the image G31. When performing non-rigid alignment of the arterial phase image G31 and the delayed phase image G32, in the first specific example, the alignment processing unit 163 sets the arterial phase image T31 to the fixed side volume VF, and uses the delayed phase image G32 as the fixed side volume VF. Let be the modified volume VM.

The second specific example is as follows. The fixed volume VF is a CT image captured by the CT apparatus 200. The deformed volume VM is volume data as an MRI image captured by an MRI apparatus. Since the MRI apparatus sometimes captures images obliquely with respect to the axial plane, the MRI image may protrude from the region of the three-dimensional space corresponding to the CT image. Here, since the CT image has a higher resolution and is mainly used, the user wants to observe the CT image as it is captured. In this case, the deformable volume VM is larger in size in three-dimensional space than the fixed volume VF. Furthermore, MRI images are often obtained with distortion due to the magnetic field. Even in such a case, the medical image processing apparatus 100 can smoothly connect and draw the fixed side volume VF and the deformed side volume VM by using the non-rigid positioning of this embodiment, and can obtain a highly reliable image. .

The third specific example is as follows. The fixed side image is a 2D moving image of a single cross section or a 2D moving image of multiple cross sections obtained by imaging with an MRI apparatus. Further, the deformed volume VM is a CT image captured by the CT apparatus 200. For example, a single cross-section video or a 2D video with multiple cross-sections is an image for diagnosing cardiac function. The CT image is an auxiliary image for measuring positional relationships and ventricular volumes. Therefore, the user would like to view the 2D video of a single cross section or the 2D video of multiple cross sections as it is captured. In this case, the alignment processing unit 163 uses, as the fixed-side image, a phase of the 2D video of multiple sections corresponding to the cardiac phase used to capture the CT image by electrocardiographic synchronization.

The fourth specific example is as follows. The fixed volume VF is a CT image captured by the CT apparatus 200 with a narrowed FOV (Field of View). The FOV corresponds to an imaging range on a slice (on the same cross section). The deformed side volume VM is volume data as an XA image captured by an XA device (angiography X-ray diagnostic device, C-arm machine) in a wider imaging range than the above-mentioned fixed side volume VF. The XA device is small and lightweight, and can be used during surgery. Further, the XA device can capture a 2DT image by capturing a moving image with the arm of the XA device fixed. Furthermore, the XA device can also capture 4D (four-dimensional) data by rotating the arm and capturing a moving image. CT images taken before surgery have higher resolution, and are used to formulate a surgical plan. Therefore, users want to observe CT images as they are captured.

The fifth specific example is as follows. The fixed side image is two-dimensional data (two-dimensional image) as an X-ray image captured by a simple X-ray device. Further, the fixed side image may be an MIP image generated based on a CT image captured by the CT apparatus 200. The deformed side image is a 2D moving image captured by the XA device. X-ray images taken before surgery have higher resolution, and are used to formulate a surgical plan. Therefore, users want to observe X-ray images as they are taken.

The sixth specific example is as follows. The fixed volume VF is 4D (four-dimensional) data of the heart obtained in electrocardiographic synchronization, and is data imaged in a narrow imaging range. The 4-dimensional data is data in which a plurality of volume data, slice data, etc. as 3-dimensional data are arranged in time series, that is, it is a moving image of 3-dimensional data. The deformed volume VM is chest data that includes the heart, and is three-dimensional data that has a wider imaging range than the above-described fixed volume VF. The 4D data of the heart is data for diagnosing cardiac function. Chest data is an auxiliary image for confirming the catheter route, etc. Therefore, the user wants to observe the 4D data as it is captured.

Next, an example of the operation of the medical image processing apparatus 100 will be described.

There are two types of operation examples during alignment by the medical image processing apparatus 100: a first operation example and a second operation example.

In the first operation example, the alignment processing unit 163 generates a deformed mesh MS (large deformed mesh MS) corresponding to the inclusion area including the deformed side volume VM and the fixed side volume VF, and uses this deformed mesh MS. The deformation side volume VM and the fixed side volume VF are aligned.

In the second operation example, the alignment processing unit 163 generates a first deformed mesh MS1 corresponding to the fixed side volume VF as in the past, and uses the first deformed mesh MS1 to connect the deformed side volume VM and the fixed side. Perform alignment with volume VF. After that, the alignment processing unit 163 generates a second deformed mesh MS2 corresponding to at least a part of the deformed volume VM that was not covered by the first deformed mesh MS1, and uses the second deformed mesh MS2. Then, the deformable volume VM and the fixed volume VF are aligned. Then, the image generation unit 162 connects the image aligned using the first deformed mesh MS1 and the image aligned using the second deformed mesh MS2 to generate an image that is entirely aligned.

FIG. 8 is a flowchart illustrating a first operation example during alignment of the medical image processing apparatus 100.

First, the port 110 acquires, for example, volume data of a subject (for example, a patient) in an early arterial phase (arterial phase) with a small imaging range as a fixed side volume VF (S11). Further, the port 110 acquires volume data of the subject having a large imaging range in the venous phase (delayed phase) as the deformation-side volume VM before deformation (that is, pre-deformation volume data VMB) (S11).

The alignment processing unit 163 rigidly aligns the acquired fixed volume VF and the pre-deformation volume data VMB (S12). The alignment processing unit 163 generates a deformed mesh MS corresponding to the inclusion area that includes the fixed volume VF and the pre-deformed volume data VMB (S13). The alignment processing unit 163 performs non-rigid alignment of the fixed volume VF and the pre-deformation volume data VMB, and adds the post-deformation volume to the deformation mesh MS from the pre-deformation volume data VMB corresponding to each node ND of the deformation mesh MS. Movement information regarding movement of each voxel to the data VMA is recorded (S14A). Then, the alignment processing unit 163 generates post-transformation volume data VMA from the pre-transformation volume data VMB using the movement information recorded in the deformation mesh MS (S14B). In the deformed mesh MS, when deforming an area that is not included in either the fixed side volume VF or the pre-deformed volume data VMB, or is included only in either of them, the alignment processing unit 163 deforms the area according to a spongy tissue such as a lung. The node ND may be set to follow the deformation of the node ND included in the pre-deformation volume data VMB. In the deformed mesh MS, when deforming a region that is not included in either the fixed side volume VF or the pre-deformed volume data VMB, or is included only in either, the alignment processing unit 163 uses Poisson's ratio of 0 and microelasticity. A finite element node may be set to follow the deformation of the node ND included in the pre-deformation volume data VMB.

The display control unit 165 causes the display 130 to display (eg, superimpose display) an image based on the aligned fixed side volume VF and the post-transformation volume data VMA (S15). Note that the display control unit 165 may blend the fixed volume data and the transformed volume data at a predetermined blend ratio (mixing ratio), and display an image based on the blended data. As a result, the boundary between the fixed volume and the transformed volume data becomes less noticeable, and discontinuity is further suppressed.

Thereby, the medical image processing apparatus 100 can perform non-rigid positioning using the deformed mesh MS corresponding to the size of the convex hull region VC that includes both the fixed volume VF and the deformed volume VM. Therefore, the medical image processing apparatus 100 can cope with the deformation of the deformation side volume VM taking into account the reachable region VMR by performing non-rigid positioning once, and can suitably perform non-rigid positioning.

9A and 9B are flowcharts showing a second operation example during alignment of the medical image processing apparatus 100. In FIGS. 9A and 9B, descriptions of processes similar to those in FIG. 8 are omitted or simplified. FIG. 9C is a diagram for explaining the second operation example.

First, the port 110 acquires volume data of a subject having a small imaging range obtained by imaging with the CT apparatus 200 as the fixed volume VF (S21). In addition, the port 110 acquires volume data of a subject having a large imaging range obtained by imaging with a PET apparatus as pre-transformation volume data VMB (S21).

The alignment processing unit 163 rigidly aligns the acquired fixed volume VF and the pre-deformation volume data VMB (S22). The alignment processing unit 163 generates a first deformed mesh MS1 that includes the fixed volume VF (S23). The alignment processing unit 163 performs non-rigid alignment of the fixed volume VF and the pre-deformation volume data VMB, and sets the pre-deformation volume corresponding to each node ND of the first deformation mesh MS1 to the first deformation mesh MS1. Movement information regarding movement of each voxel from the data VMB to the post-transformed volume data VMA is recorded (S24). This movement information also includes information on the position of the volume data before transformation (information on the initial position before transformation) with respect to the fixed volume VF.

Note that, since the first deformed mesh MS1 is a conventional deformed mesh MSX, the reachable region VMR of the deformed volume VM cannot be completely covered, and the reachable region VMR is outside the area of the first deformed mesh MS1. may be located. Therefore, at the time of steps S24A and S24B, non-rigid alignment is insufficient.

Proceed to Figure 9B. The alignment processing unit 163 calculates a difference region VS by subtracting the spatial region of the fixed side volume VF from the region of the pre-transformed volume data VMB (the region where the pre-transformed volume data VMB exists in three-dimensional space, also referred to as a spatial region). do. Then, a second deformed mesh MS2 corresponding to the difference region VS (including the difference region VS) is generated (S25). As non-rigid alignment, the alignment processing unit 163 uses the first deformed mesh MS1 as a reference to adjust the second deformed mesh so that the second deformed mesh MS2 smoothly connects to the first deformed mesh MS1. Movement information regarding movement of each voxel from the pre-deformation volume data VMB to the post-deformation volume data VMA corresponding to each node ND of the second deformation mesh MS2 is recorded in MS2 (S26). For example, the alignment processing unit 163 uses the movement of the node ND of the first deformed mesh MS1 of the boundary portion BP of the first deformed mesh MS1 and the second deformed mesh MS2 as a boundary condition, and adjusts the second deformed mesh of the boundary portion BP. The second deformed mesh MS2 is deformed by moving the node ND of the deformed mesh MS2 and setting nodes ND similar to spongy tissues such as lungs for other parts of the second deformed mesh MS2. Thereby, the first deformed mesh MS1 and the second deformed mesh MS2 are smoothly connected. For example, if the movement of the node ND of the first deformed mesh MS1 of the boundary portion BP between the first deformed mesh MS1 and the second deformed mesh MS2 is set as a boundary condition, the node ND of the second deformed mesh MS2 of the boundary portion BP is The second deformed mesh MS2 may be deformed by moving the second deformed mesh MS2 and setting finite element nodes with a Poisson's ratio of 0 and minute elasticity for other parts of the second deformed mesh MS2. Thereby, the first deformed mesh MS1 and the second deformed mesh MS2 are smoothly connected. As a result, output results equivalent to those of the first operation example can be obtained while suppressing the amount of calculation.

Therefore, information regarding deformation using the first deformation mesh MS1 corresponds to deformation information regarding deformation of the deformation volume VM in the spatial region of the fixed volume VF. Further, the information regarding the deformation using the second deformed mesh MS2 corresponds to the deformation information regarding the deformation of the deformation side volume VM in the spatial region of the difference region VS.

The alignment processing unit 163 connects the first deformed mesh MS1 and the second deformed mesh MS2 to create a third deformed mesh MS3, and by creating the third deformed mesh MS3, the pre-deformed volume Non-rigid positioning is performed for the combined area of the spatial area of the data VMB and the spatial area of the fixed side volume VF. (S27A). That is, the alignment processing unit 163 performs non-rigid alignment of the fixed side volume VF and the pre-deformation volume data VMB in the above-mentioned combined area, and uses the first deformed mesh MS1 and the second deformed mesh MS2. , the movement information regarding the movement of each voxel from the pre-deformation volume data VMB to the post-deformation volume data VMA corresponding to each node ND of the third deformation mesh MS3 is recorded in the third deformation mesh MS3. The alignment processing unit 163 generates post-transformation volume data VMA from the pre-transformation volume data VMB using the movement information recorded in the third deformation mesh MS3 (S27B).

The display control unit 165 displays (for example, superimposes) an image based on the fixed volume VF and the transformed volume data VMA on the display 130 (S28).

As a result, the medical image processing apparatus 100 performs non-rigid positioning using the first deformed mesh MS1 based on the size of the fixed side volume VF, and then calculates the deformed side volume that could not be covered by the first deformed mesh MS1. Parts of the VM can be non-rigidly registered using the second deformed mesh MS2. Therefore, the medical image processing apparatus 100 can compensate for the insufficient portion with the same amount of calculation as conventional non-rigid body alignment.

(Comparison of images obtained in the comparative example and this embodiment)
FIG. 10A is a diagram illustrating an example of the alignment result of a wide area including the abdomen of a subject in a comparative example. FIG. 10B is a diagram illustrating an example of the alignment result of the abdominal region of the subject in the comparative example. In FIG. 10A, the body surface of the subject is not drawn. The images shown in FIGS. 10A and 10B are images obtained when non-rigid alignment is performed, for example, by the conventional method of Non-Patent Document 1 or Non-Patent Document 2.

In FIG. 10A, a positional shift between the volume data occurs in the pelvis, and appears as a discontinuous line L11 that crosses the pelvis in the image G11 based on the volume data that has been non-rigidly aligned in the comparative example. In FIG. 10B, the positional deviation between the volume data occurs in the abdomen, and appears as a discontinuous line L12 that crosses the abdomen in the image G12 based on the volume data that has been non-rigidly aligned in the comparative example.

FIG. 11A is a diagram illustrating an example of the alignment result of a wide area including the abdomen of a subject in this embodiment. FIG. 11B is a diagram illustrating an example of the alignment result of the abdominal region of the subject in this embodiment. In FIG. 11A, the body surface of the subject is not drawn. The region of the object shown in FIG. 11A is similar to the region of the object shown in FIG. 10A. The region of the object shown in FIG. 11B is similar to the region of the object shown in FIG. 10B.

The alignment in FIGS. 11A and 11B is performed by the above-mentioned non-rigid alignment, and may also be performed by rigid alignment. In FIG. 11A, the positional deviation between the volume data is suppressed, and the positional deviation shown in FIG. 10A is eliminated. Therefore, in the image G21 based on the volume data subjected to non-rigid registration in this embodiment, the discontinuous line L11 shown in FIG. 10A does not exist. In FIG. 11B, the positional deviation between the volume data is suppressed, and the positional deviation shown in FIG. 10B is eliminated. Therefore, in the image G22 based on the volume data subjected to non-rigid alignment in this embodiment, the discontinuous line L12 shown in FIG. 10B does not exist.

Next, variations of this embodiment will be explained.

In transforming the transformation volume VM, the alignment processing unit 163 may directly transform the pre-transformation volume data VMB to derive the post-transformation volume data VMA. In this case, the pre-transformation volume data VMB may not be held in the memory 150, and only the post-transformation volume data VMA may be held in the memory 150. Furthermore, the alignment processing unit 163 may create new post-transformation volume data VMA based on the pre-transformation volume data VMB. In this case, both the pre-transformation volume data VMB and the new post-transformation volume data VMA may be held in the memory 150. Furthermore, the alignment processing unit 163 may refer to the voxel values of the pre-transformation volume data VMB when calculating the voxel values of the post-transformation volume data VMA without immediately creating the post-transformation volume data VMA.

The alignment processing unit 163 may perform non-rigid alignment of at least part of the volume data (regions of interest). In other words, a part of the fixed volume VF and a part of the deformable volume VM may be aligned in a non-rigid manner. For example, for a chest image (chest region in volume data), if it is sufficient to align only the left ventricle, the chest region of the fixed side volume VF and the chest region of the deformed side volume VM can be moved to non-rigid positions. May be combined. In this case, the alignment processing unit 163 may deform the outside of the region of interest using the method of the second operation example.

Further, a plurality of deformed meshes MS may be prepared and held in the memory 150. The deformed mesh MS may have multiple types (multi-resolution) of grid fineness of the deformed mesh MS. Further, the size of the deformed mesh MS may be different depending on the type of the deformed mesh MS, and the area covered by the deformed mesh MS may be different. Further, the deformed mesh MS may be configured in three dimensions when the medical image data is three-dimensional data, and may be configured in three dimensions when the medical image data is two-dimensional data.

Furthermore, when the medical image data is a combination of three-dimensional data and two-dimensional data, the deformed mesh MS may be configured in either three-dimensional or two-dimensional form. In addition, in the case of a combination of the fixed side two-dimensional image and the deformed volume VM, the alignment processing unit 163 performs non-rigid alignment of the cross section of the fixed side two-dimensional image in the deformed side volume VM when deforming the deformed side volume VM. After this, the deformation side volume VM may be deformed by extrapolating a similar deformation in the depth direction of the cross section. In addition, in the case of a combination of a two-dimensional image of a plurality of fixed-side cross sections and a deformation-side volume VM, the alignment processing unit 163 generates a two-dimensional image of each cross-section of a two-dimensional image of a plurality of fixed-side cross-sections in the deformation of a deformation-side volume VM. The deforming volume VM may be deformed by performing non-rigid positioning for the corresponding cross section in the deforming volume VM, and then interpolating deformations between a plurality of cross sections in the depth direction of the cross section.

Further, the alignment processing unit 163 may repeatedly perform non-rigid alignment. For example, after performing non-rigid positioning using a deformed mesh MS with a rough grid, non-rigid positioning may be performed using a deformed mesh MS with a fine grid. Further, the fineness of the deformed mesh MS is prepared in multiple stages, and the alignment processing unit 163 starts with non-rigid positioning using the deformed mesh MS with a rough grid, and gradually uses the deformed mesh MS with finer grids. Non-rigid alignment may be performed sequentially.

In this way, the medical image processing apparatus 100 of the present embodiment generates the deformation mesh MS in a range that includes the reachable region VMR that the deformation side volume VM can reach by deformation, so that the deformation side volume VM to be deformed is There is no region that cannot be deformed. Therefore, the deformable side volume VM can be suitably deformed and non-rigid positioning can be performed with respect to the fixed side volume VF, so that the accuracy of the non-rigid positioning can be improved.

Although various embodiments have been described above with reference to the drawings, it goes without saying that the present disclosure is not limited to such examples. It is clear that those skilled in the art can come up with various changes or modifications within the scope of the claims, and these naturally fall within the technical scope of the present disclosure. Understood.

Further, the medical image processing apparatus 100 may include at least a processor 140 and a memory 150. The port 110, UI 120, and display 130 may be external to the medical image processing apparatus 100.

Further, the volume data as a captured CT image is transmitted from the CT apparatus 200 to the medical image processing apparatus 100, as an example. Alternatively, the volume data may be sent to a server on a network (for example, an image data server (PACS) (not shown)) and stored so that the volume data is temporarily stored. In this case, the port 110 of the medical image processing apparatus 100 may acquire volume data from a server or the like via a wired line or wireless line, or via an arbitrary storage medium (not shown) when necessary. It's okay.

Further, the volume data as a captured CT image is transmitted from the CT apparatus 200 to the medical image processing apparatus 100 via the port 110, as an example. This also includes the case where the CT apparatus 200 and the medical image processing apparatus 100 are substantially combined into one product. It also includes a case where the medical image processing apparatus 100 is treated as a console of the CT apparatus 200.

Moreover, although the CT apparatus 200 captures images and generates volume data including information inside the object, it is also possible to capture images using other devices and generate volume data. Other devices include various modality devices. Additionally, the PET device may be used in combination with other modality devices.

Further, the operation of the medical image processing apparatus 100 can be expressed as a defined medical image processing method. Further, it can be expressed as a program for causing a computer to execute each step of the medical image processing method.

(Summary of the above embodiment)
As described above, the medical image processing apparatus 100 of the embodiment described above includes the processing section 160. The processing unit 160 acquires first medical image data (for example, fixed side volume VF) and second medical image data (for example, deformed side volume VM) composed of two-dimensional or three-dimensional pixels indicating the subject. . The processing unit 160 performs non-rigid positioning of the first medical image data and the second medical image data by deforming the second medical image data with respect to the fixed first medical image data. Execute alignment processing. The processing unit 160 causes the display 130 (an example of a display unit) to display the first medical image data and the second medical image data that have been subjected to non-rigid alignment. The spatial region of the second medical image data includes a portion that is not included in the spatial region of the first medical image data. The non-rigid registration process includes a process of generating deformation information regarding deformation of the second medical image data in an area that includes at least the reachable region VMR, which is a region that the second medical image data can reach by deformation; and processing to transform the second medical image data based on the information. The deformation information includes movement information regarding movement of at least one pixel included in the second medical image data in a region that is not included in the spatial region of the first medical image data.

Thereby, the medical image processing apparatus 100 deforms the deformation volume VM based on deformation information regarding deformation of the deformation volume VM in the range including the reachable region VMR. Therefore, there is no region in which the deforming volume VM to be deformed cannot be deformed. Therefore, the medical image processing apparatus 100 can suitably deform the deformable volume VM and perform non-rigid positioning with respect to the fixed volume VF, thereby improving the accuracy of non-rigid positioning.

Further, the deformation information includes deformation information regarding deformation of the second medical image data in the convex hull region VC of the sum region of the spatial region of the first medical image data and the spatial region of the second medical image data. good.

As a result, the medical image processing apparatus 100 obtains deformation information regarding the deformation of the entire convex hull region VC at one time as deformation information, and deforms the deformation side volume VM in the entire convex hull region VC to form a non-rigid body. This can be done by positioning. Therefore, the time required for non-rigid positioning can be reduced.

Further, the deformation information includes first deformation information regarding deformation of the second medical image data in the spatial region of the first medical image data, and first deformation information regarding deformation of the second medical image data in the spatial region of the second medical image data before deformation. and second deformation information regarding deformation of the second medical image data in the difference region VS excluding the spatial region of the image data.

As a result, the medical image processing apparatus 100 deforms the deformation side volume VM in the area covering the fixed side volume VF, and then deforms the deformation side volume VM in the remaining area to be deformed, as in the conventional case. conduct. Therefore, after performing non-rigid alignment using the conventional method, the medical image processing apparatus 100 can also perform deformation and non-rigid alignment of portions that were insufficient using the conventional method.

Further, the first medical image data and the second medical image data may be at least part of volume data imaged by a CT apparatus. Thereby, the medical image processing apparatus 100 can improve the accuracy of non-rigid alignment even when performing non-rigid alignment of CT images captured by the same modality device.

Further, the first medical image data may be at least a portion of the first volume data imaged by the CT apparatus. The second medical image data may be at least a portion of the second volume data captured by the MRI apparatus. Thereby, the medical image processing apparatus 100 can improve the accuracy of non-rigid alignment even when non-rigid alignment is performed between a CT image and an MRI image captured by different modality devices.

Further, the first medical image data may be one or more two-dimensional medical image data. The second medical image data may be three-dimensional medical image data. Thereby, the medical image processing apparatus 100 can improve the accuracy of non-rigid alignment even when performing non-rigid alignment of two-dimensional medical image data and three-dimensional medical image data.

The present disclosure is useful for a medical image processing device, a medical image processing method, a medical image processing program, etc. that can improve the accuracy of non-rigid positioning of multiple pieces of medical image data.

100 Medical image processing device 110 Port 120 User interface (UI)
130 Display 140 Processor 150 Memory 160 Processing unit 161 Area processing unit 162 Image generation unit 163 Alignment processing unit 165 Display control unit 200 CT device MS Deformed mesh MS1 First deformed mesh MS2 Second deformed mesh ND Node VC Convex hull region VF Fixed side volume VS Differential area VM Transformation side volume VMA Volume data after transformation VMB Volume data before transformation

Claims (8)

A medical image processing device,
Equipped with a processing section,
The processing unit includes:
Obtaining first medical image data and second medical image data composed of two-dimensional or three-dimensional pixels indicating the subject,
a non-rigid position where the first medical image data and the second medical image data are aligned in a non-rigid manner by deforming the second medical image data with respect to the fixed first medical image data; Execute the matching process,
Displaying on a display unit an image based on the first medical image data and the second medical image data that are non-rigidly aligned;
The spatial region of the second medical image data includes a portion not included in the spatial region of the first medical image data,
The non-rigid alignment process includes:
a process of generating deformation information regarding deformation of the second medical image data in an area that includes at least a reachable area that is an area that the second medical image data can reach by deformation;
a process of transforming the second medical image data based on the transformation information,
The deformation information includes movement information regarding movement of at least one pixel included in the second medical image data in a region not included in the spatial region of the first medical image data.
Medical image processing device. The deformation information includes deformation information regarding deformation of the second medical image data in a convex hull region of a sum region of a spatial region of the first medical image data and a spatial region of the second medical image data. ,
The medical image processing device according to claim 1. The deformation information is
first deformation information regarding deformation of the second medical image data in the spatial domain of the first medical image data;
second transformation information regarding transformation of the second medical image data in a difference region obtained by removing the spatial region of the first medical image data from the spatial region of the second medical image data before transformation; ,
The medical image processing device according to claim 1. The first medical image data and the second medical image data are at least a part of volume data imaged by a CT device.
The medical image processing device according to claim 2 or 3. The first medical image data is at least part of first volume data imaged by a CT device,
The second medical image data is at least a part of second volume data imaged by an MRI apparatus.
The medical image processing device according to claim 2 or 3. The first medical image data is one or more two-dimensional medical image data,
the second medical image data is three-dimensional medical image data;
The medical image processing device according to claim 2 or 3. acquiring first medical image data and second medical image data composed of two-dimensional or three-dimensional pixels indicating the subject;
performing non-rigid registration of the first medical image data and the second medical image data by deforming the second medical image data with respect to the fixed first medical image data; ,
displaying on a display unit an image based on the first medical image data and the second medical image data that are non-rigidly aligned;
has
The spatial region of the second medical image data includes a portion not included in the spatial region of the first medical image data,
The step of performing the non-rigid alignment includes:
generating deformation information regarding deformation of the second medical image data in an area that includes at least a reachable area that is an area that the second medical image data can reach by deformation;
deforming the second medical image data based on the deformation information,
The deformation information includes movement information regarding movement of at least one pixel included in the second medical image data in a region not included in the spatial region of the first medical image data.
Medical image processing method. A medical image processing program for causing a computer to execute the medical image processing method according to claim 7.

Images