TWI425963B - Image collating apparatus, patient positioning apparatus, and image collating method - Google Patents

Description

Image checking device, patient positioning device and image checking method

The present invention relates to a spectroscopic device using CT (computed tomography) image data in a radiotherapy apparatus that irradiates radiation to X-rays, γ-rays, particle beams, and the like to an affected part of a patient; and uses this image check The device is positioned to position the patient to a patient positioning device that illuminates the location of the radiation exposure of the radiation.

In recent years, in the field of radiation therapy devices for cancer treatment, cancer treatment devices using particle beams such as protons and heavy ions have been developed and constructed (herein, referred to as particle beam therapy devices). It is well known that particle beam therapy using a particle beam is more effective than conventional radiation therapy using X-rays, gamma rays, etc., to focus on the affected part of the cancer, that is, to match the shape of the affected part with a pinpoint. The beam illuminates the particle beam and can be treated without affecting normal cells.

In particle beam therapy, it is important to irradiate the particle beam with high precision to an affected part such as cancer. Therefore, in the particle beam treatment, the patient is fixed by a fixture or the like so as not to be displaced from the treatment table of the treatment room (irradiation chamber). In order to accurately position the affected part such as cancer in the radiation irradiation range, a setting such as a rough fixation of a patient such as a laser pointer is performed, and the patient is then subjected to an X-ray image or the like. Precise positioning of the affected part.

Patent Document 1 proposes that the same number of plural images are not specified in the portraits of the reference image (X-ray fluoroscopic image) and the current image (the image captured by the X-ray receiver). The same position, but a two-stage pattern matching to generate a positioning device for positioning information for driving the treatment table and a positioning method thereof. The one-time pattern matching is to set a second setting area which is approximately the same size as the first setting area for the two-dimensional current image (where the first setting area includes the isocenter (beam irradiation center) set for the two-dimensional reference image a region), and then sequentially moving the second set region in the region of the two-dimensional current image, and comparing the two-dimensional reference image in the first set region with the second set region in each position of the second set region The two-dimensional present image extracts a second set region of the two-dimensional current portrait that is most similar to the two-dimensional reference image of the first set region. The quadratic pattern matching system compares the two-dimensional current image in the second set region extracted in the one-time pattern matching with the two-dimensional reference image in the first setting region to make the two images most consistent. The way to match the pattern.

(previous technical literature)

(Patent Literature)

(Patent Document 1) Japanese Patent No. 3748433 (Sections 0007 to 0009, 0049, 8 and 9)

Since the shape of the affected part is a three-dimensional three-dimensional shape, the positioning of the affected part at the position of the affected part at the time of the treatment plan can achieve a higher accuracy than the case of using the two-dimensional image. In general, when a treatment plan data is created, an X-ray CT (Computed Tomography) image is used to determine the three-dimensional shape of the affected part. In recent years, it has been desired to provide an X-ray CT apparatus in a treatment room to perform positioning using an X-ray CT current image taken by an X-ray CT apparatus and an X-ray CT image at the time of treatment planning using a treatment. Requirements. This is because: in the X-ray fluoroscopic portrait, the affected part of the soft tissue usually does not appear very clear, so basically the relative position with the bone is used to perform the alignment, and vice versa, the X-ray CT image is used. If the positioning is not necessary, the alignment of the affected parts taken in the X-ray CT image can be directly performed.

Therefore, it is considered to expand the reference image and the current image into a three-dimensional image in the conventional two-stage pattern matching. The three-dimensional reference image and the three-dimensional current image include a plurality of tomographic images (slice images) taken by an X-ray CT apparatus. Since there are concerns about the amount of X-ray exposure, etc., it is assumed that the number of images included in the three-dimensional current image is small. Therefore, when comparing, it is necessary to perform a three-dimensional reference image having detailed image information and a simpler than the three-dimensional reference image. Comparison of three-dimensional portraits of portrait information. The traditional two-stage pattern matching can compare the two-dimensional reference image with the same density of the portrait information and the two-dimensional current image, but compares the three-dimensional reference image with the density of the image information and the three-dimensional current image. In addition, simply changing the portrait dimension of traditional technology from two-dimensional to three-dimensional has the problem of not being able to achieve two-stage pattern matching. In other words, there is a problem that it is impossible to realize the simple matching of the three-dimensional reference image in the first setting region set to the three-dimensional current image in the second setting region, and the second extraction is simply performed. The three-dimensional current image portion in the setting area is compared with the three-dimensional reference image in the first setting area to match the pattern in which the two images are most consistent.

Therefore, the object of the present invention is to achieve high-precision two-stage pattern matching even when the number of tomographic images of the three-dimensional present image is smaller than that of the three-dimensional reference image when the patient is subjected to radiotherapy (two-stage collation) ).

The image collation apparatus of the present invention includes a three-dimensional image input unit that reads a three-dimensional reference image taken during a treatment plan for radiation therapy and a three-dimensional current image taken during treatment; The collation processing unit checks the three-dimensional reference image and the three-dimensional current image, and calculates the posture correction amount that matches the position and posture of the affected part in the three-dimensional current image with the position and posture of the affected part in the three-dimensional reference image. The check processing unit has a primary check unit that performs primary pattern matching from the three-dimensional reference image with respect to the three-dimensional current image, and a secondary check portion that performs a three-dimensional reference image or three-dimensional image from the result of matching according to the primary pattern. The predetermined sample area generated by one of the current portraits is a predetermined search object generated from the other of the three-dimensional reference image or the three-dimensional current image different from the basis of the generation of the predetermined template region with respect to the result of the one-pattern matching. Secondary pattern matching of the area.

In the image collation apparatus of the present invention, the pattern matching from the three-dimensional reference image matching to the three-dimensional current image is first performed, and then a predetermined template region and a predetermined retrieval target region are generated based on the result of the primary pattern matching, and then the retrieval target is executed. Since the area is matched with the quadratic pattern of the template area, even if the number of tomographic images of the three-dimensional current image is smaller than that of the three-dimensional reference image, high-precision two-stage pattern matching can be realized.

Embodiment 1

Fig. 1 is a view showing the configuration of an image collating apparatus and a patient positioning apparatus according to Embodiment 1 of the present invention. Fig. 2 is a view showing the overall configuration of a machine relating to the image collating apparatus and the patient positioning apparatus of the present invention. In Fig. 2, 1 shows a CT simulation room for performing a treatment plan performed before radiotherapy, in which there is a CT gantry 2 and a top plate 3 of a CT portrait photography bed, and the patient 4 is lying on the side. Above the top plate 3, CT image data for a treatment plan including the affected part 5 is taken. On the other hand, 6 denotes a treatment room for performing radiation therapy, in which there is a CT pylon 7 and a rotating treatment table 8, the upper portion of the rotary treatment table 8 has a top plate 9, and the patient 10 is placed on the top plate 9, and The CT image data for the positioning of the affected part 11 at the time of treatment is taken.

Here, the positioning means calculating the position of the patient 10 and the affected part 11 at the time of treatment from the CT image data for the treatment plan, and calculating the position correction for the position of the patient 10 and the affected part 11 in accordance with the treatment plan. The amount is then carried out in such a manner that the affected part 11 at the time of treatment comes to the radiation irradiation center 12 of the radiation therapy. The alignment is achieved by driving control of the rotary treatment table 8 in a state where the patient 10 is lying on the top plate 9 to move the position of the top plate 9. The rotary treatment table 8 can perform translation correction of six degrees of freedom of translation and rotation, and rotates the top plate 9 of the rotary treatment table 8 by 180 degrees, and can also rotate the top plate 9 of the rotary treatment table 8 from the CT photographing position. (indicated by a solid line in Fig. 2), it moves to a certain treatment position (indicated by a broken line in Fig. 2) of the irradiation head 13 that emits radiation. In the second figure, although the CT imaging position and the treatment position have a relative positional relationship of 180 degrees, the present invention is not limited to this arrangement, and the positional relationship between the two may be different from other angles such as 90 degrees. form.

The CT image data for the treatment plan and the CT image data for positioning are transmitted to the positioning computer 14. The CT image data for the treatment plan is a three-dimensional reference image, and the CT image data for positioning is a three-dimensional current image. The image collating device 29 and the patient positioning device 30 in the present invention are related to the computer software existing in the positioning computer 14, and the image collating device 29 calculates the posture correction amount (translation amount, rotation amount), and the patient. The positioning device 30 includes not only the image collation device 29 but also parameters for controlling the respective drive shafts for controlling the rotary treatment table 8 (hereinafter referred to simply as the treatment table 8 as appropriate) based on the posture correction amount. The function of the person. The patient positioning device 30 controls the treatment table 8 based on the result of the comparison (check result) made by the image verification device 29, thereby guiding the target affected portion of the particle beam treatment to the beam irradiation center 12 of the treatment device.

The positioning made in the conventional radiotherapy is to check the DRR (Digitally Reconstructed Radiography) image generated from the CT image data of the treatment plan or the X-ray fluoroscopic image taken at the same time as the DRR image, and the treatment during treatment. The X-ray perspective image taken by the room is used to calculate the position offset. In the X-ray fluoroscopic portrait, the affected part of the soft tissue is usually not very clear, so basically the relative position with the bone is used to perform the alignment. The positioning using the CT image data described in the present embodiment is characterized in that the CT pylon 7 is provided in the treatment room 6, and the CT image data before treatment and the CT image data for the treatment plan are performed. It is the right position, so you can directly depict the affected part and you can do it in the affected part.

Next, the calculation procedure of the posture correction amount in the image verification device 29 and the patient positioning device 30 of the present embodiment will be described. Fig. 1 shows the relationship between the data processing units constituting the image collating apparatus and the patient positioning apparatus. The image collating apparatus 29 has a three-dimensional image input unit 21 for reading CT image data, a collation processing unit 22, and a collation result display. The unit 23 and the verification result output unit 24. The patient positioning device 30 is added to the treatment table control parameter calculation unit 26 in addition to the image verification device 29.

As described above, the three-dimensional reference image is a data obtained for the treatment plan at the time of the treatment plan, and is characterized in that the information on the affected part (the shape of the affected part, etc.) of the affected part which is the target of the particle beam treatment is used. Enter by hand. The three-dimensional current image is a document taken for patient positioning at the time of treatment, and is characterized in that the number of tomographic images (also referred to as slice images) is small from the viewpoint of suppressing the amount of X-ray exposure.

The present invention is configured to first perform a pattern matching from a three-dimensional reference image matching to a three-dimensional current image, and then generate a predetermined template region and a predetermined retrieval target region based on the result of the primary pattern matching, and then use the predetermined template. The composition of the two-stage pattern matching of the quadratic pattern matching in the same direction or in the opposite direction. The two-stage pattern matching can achieve high speed or high precision of processing by making the comparison parameters when the primary pattern is matched and the matching parameters when the secondary pattern is matched. For example, there is a pattern matching for a wider range with a coarser resolution, and then using the found template area or the search object area to perform quadratic pattern matching on the reduced range with finer resolution. The method.

Next, the three-dimensional image input unit 21 will be described. The three-dimensional image input unit 21 is an image group (a group of sliced images) in the form of DICOM (Digital Imaging and Communications in Medicine), which is imaged by an X-ray CT apparatus and configured as a plurality of tomographic images. It is read as a three-dimensional volume data. The CT image data for the treatment plan is a three-dimensional volume data for the treatment plan, that is, a three-dimensional reference image. The CT image data for positioning is a three-dimensional volume data at the time of treatment, that is, a three-dimensional current portrait. Moreover, the CT image data is not limited to the DICOM format, and may be in other forms.

The collation processing unit 22 collates the three-dimensional reference image and the three-dimensional current image, and calculates the posture correction amount for matching the position and posture of the affected part in the three-dimensional current image with the position and posture of the affected part in the three-dimensional reference image. The collation result display unit 23 is a result of the collation processing unit 22 collating the image (the amount of the body posture correction described later and the three-dimensional current image that has been moved by the posture correction amount on the three-dimensional reference image). Displayed on the monitor screen of the positioning computer 14. The collation result output unit 24 is a correction amount obtained when the collation processing unit 22 collates the three-dimensional reference image and the three-dimensional current image, that is, the posture correction amount (translation amount, rotation amount) calculated by the collation processing unit 22 Output. The treatment table control parameter calculation unit 26 converts the output value (translation three axes [ΔX, ΔY, ΔZ], and the rotation three axes [ΔA, ΔB, ΔC] total six degrees of freedom) of the verification result output unit 24 into The parameters of the axes of the treatment table 8 are controlled, that is, the parameters for controlling the axes of the treatment table 8 are calculated. The treatment table 8 drives the drive devices of the respective axes of the treatment table 8 based on the treatment table control parameters calculated by the treatment table control parameter calculation unit 26. In this way, the amount of posture correction for which the position of the affected part coincides with the treatment plan can be calculated, and then the affected part 11 at the time of treatment can be aligned to the beam irradiation center 12 of the radiation therapy.

The collation processing unit 22 includes a position and posture converting unit 25, a primary collating unit 16, a secondary collating unit 17, and a reference template area generating unit 18. The position/posture conversion unit 25 changes the position and posture of the target data when the primary pattern matching or the secondary pattern matching is performed. The primary collating unit 16 performs a pattern matching of the three-dimensional reference image with respect to the three-dimensional current image. The secondary collating unit 17 is a predetermined template region generated from one of the three-dimensional reference image or the three-dimensional current image based on the result of the primary pattern matching, and is generated from the predetermined template region with respect to the result of the primary pattern matching. A predetermined type of search target region generated by the other three-dimensional reference image or the three-dimensional current image having different foundations is subjected to quadratic pattern matching.

The verification processing unit 22 will be described in detail below using the third to ninth drawings. Fig. 3 is a view showing a three-dimensional reference image and a reference image template area in the first embodiment of the present invention. Fig. 4 is a view showing a three-dimensional present portrait in the first embodiment of the present invention. Fig. 5 is a view for explaining a primary pattern matching method in the first embodiment of the present invention. Fig. 6 is a view for explaining the relationship between the reference image template area and the slice image in the primary pattern matching method of Fig. 5. Fig. 7 is a view showing a single extraction region of a slice image extracted by the primary pattern matching method in the first embodiment of the present invention. Fig. 8 is a view for explaining a secondary pattern matching method in the first embodiment of the present invention. Fig. 9 is a view for explaining the relationship between the reference image template area and the slice image in the quadratic pattern matching method of Fig. 8.

The reference template area generating unit 18 of the collation processing unit 22 generates the reference image template area 33 from the three-dimensional reference image 31 by using the shape of the affected part (affected part information) which is input during the treatment plan. The three-dimensional reference image 31 is composed of a plurality of slice images 32. In the third drawing, an example in which the three-dimensional reference image 31 is composed of five slice images 32a, 32b, 32c, 32d, and 32e is displayed so as not to complicate the description. The shape of the affected part is input as a ROI (Region of Interest) 35 in a form in which each sliced image is a closed contour surrounding the affected part. The area including the aforementioned closed contour may be set to, for example, an circumscribed square 34, and the cubic area including each circumscribed square 34 may be set as a template area. This template area is used as the reference image template area 33. The primary collating unit 16 of the collation processing unit 22 performs primary pattern matching in which the reference image template area 33 is matched to the three-dimensional current image 36. The three-dimensional present image 36 shown in Fig. 4 is an example of three slice images 37a, 37b, and 37c. The present portrait area 38 shown in Fig. 5 is represented as a cube including three slice images 37a, 37b, 37c. As shown in Fig. 5, the reference image template area 33 (33a, 33b, 33c) is moved in a raster scan manner in the current image area 38, and the correlation value with the three-dimensional current image 36 is calculated. In terms of correlation values, any correlation value used in image comparison (portrait reconciliation) can be used using standardized cross-correlation values and the like.

The reference image template area 33a is moved in a progressive scan shape along the scanning path 39a on the slice image 37a. Similarly, the reference image template area 33b is scanned in a progressive scan along the scanning path 39b on the slice image 37b, and the reference image template area 33c is scanned in a progressive scan along the scanning path 39c on the slice image 37c. . Among them, the scanning paths 39b and 39c are simplified in order not to complicate the drawing.

When the pattern matching is performed once, as shown in Fig. 6, each of the slice images 53 constituting the reference image template area 33 is subjected to image collation with the slice image 37 constituting the current image area 38. The slice image 53 is an image cut out by the reference image template area 33 in the slice image 32 of the three-dimensional reference image 31. The reference image template area 33 is composed of five slice images 53a, 53b, 53c, 53d, and 53e corresponding to the five slice images 32a, 32b, 32c, 32d, and 32e in the three-dimensional reference image. Therefore, in the case of the pattern matching, the image is collated with the five slice images 53a, 53b, 53c, 53d, 53e of the reference image template region 33 with respect to the slice image 37a of the three-dimensional current image 36. In the same manner, the image is collated in the same manner as the slice images 37b and 37c of the three-dimensional current image 36.

The primary collating unit 16 extracts the primary extraction region 43 from each slice image 37 of the three-dimensional current image 36 by one-time pattern matching (this primary extraction region 43 includes correlation values between the current image region 38 and the reference image template region 33). The highest area). As shown in Fig. 7, the extraction area 43a is extracted once from the slice image 37a of the three-dimensional current image 36. Further, the extracted regions 43b and 43c are extracted from the slice images 37b and 37c of the three-dimensional current image 36. Then, it is generated that the one-time extraction area 43a, 43b, 43c is included in the image area 42 as the search target area to be used in the quadratic pattern matching. In this manner, the primary collating unit 16 generates a one-time drawing of the search target region to be used in the quadratic pattern matching in the portrait region 42.

However, in the state before the positioning, the three-dimensional reference image 31 does not coincide with the posture (rotation three axes) of the three-dimensional current image 36. Therefore, as in the simple progressive scan of FIG. 5, the number of slices in the three-dimensional current image 36 is small. In the case of the high-precision matching in which the angular offset is also detected, there is no problem in extracting the primary extraction region 43 required for the secondary pattern matching. Therefore, in the one-time pattern matching, the angular offset is not detected and the correlation value is calculated, and the subsequent quadratic pattern matching is performed with the high-precision matching which is detected by the joint angle shift.

Next, the secondary pattern matching will be explained. In the quadratic pattern matching, the position and posture conversion unit 25 of the collation processing unit 22 generates the position and posture conversion template area 40 obtained by converting the position and orientation of the reference image template area 33 generated from the three-dimensional reference image 31. . As shown in the eighth and ninth figures, when the quadratic pattern matching system is matched, the posture change amount (rotation three axes) of the reference image template area 33 is added as a parameter. The second collation unit 17 is formed by the position and posture conversion template region 40 obtained by the position and posture conversion unit 25, and the three-dimensional current image 36 having a small number of slice images is present between the image regions 42. Perform a high-precision match that also includes the angular offset factor inside. In this way, a high-precision two-stage pattern matching in which the angular offset factor is also included can be realized. By using a narrower range including the region found in the one-time pattern matching as the search range of the quadratic pattern matching, it is possible to use a wider range including a rougher resolution. The one-time extraction area 43 found by performing the pattern matching as the object is once drawn in the portrait area 42 to perform the quadratic pattern matching with a finer resolution, and the time required for the pattern matching can be shortened.

The one-time drawing shown in Fig. 8 appears in the image area 42, and is represented as a cube containing three primary extraction regions 43a, 43b, 43c. The position and orientation conversion template area 40a, which is the reference image template area which has been converted into the position and posture, is moved in a progressive scanning manner along the scanning path 39a in the primary extraction area 43a of the slice image 37a. In the same manner, the position and posture change template area 40b which has been converted by the position and posture is moved in a progressive scanning manner along the scanning path 39b in the primary extraction region 43b of the slice image 37b, and the position and posture are changed after the position and posture are converted. The change template area 40c is moved in a progressive scan manner along the scan path 39c in the primary extraction area 43c of the slice image 37c. Among them, the scanning paths 39b and 39c are simplified in order not to complicate the drawing.

In the case of the quadratic pattern matching, as shown in Fig. 9, the cross section 41 of the position and posture changing template area 40 by the secondary collating unit 17 and the one-time drawing area 43 of the slice image 37 which is formed in the image area 42 at one time are drawn. Check the portraits between them. Further, the slice image 55 (the image cut out in the slice image 37 of the three-dimensional current image 36 in the image region 42 in one shot) may be subjected to image collation with the cross section 41. The section 41 of the position and orientation conversion template area 40 is generated from a plurality of slice images 32 of the three-dimensional reference image 31. For example, the data of the section 41 is cut out from a plurality of slice images 32 constituting the three-dimensional reference image 31. In general, the data density of the section 41 of the position and orientation change template area 40 may be different from the data density of the one-time extraction area 43 of the three-dimensional present image 36, but the correlation value may be calculated for each pixel of the section 41. Further, the section 41 of the position/posture conversion template area 40 may be included in the same manner as the data density of the one-time extraction area 43 of the three-dimensional current image 36 is increased.

Here, the two-stage pattern matching method of the first embodiment will be described. First, the reference template region generating unit 18 of the collation processing unit 22 generates the reference image template region 33 from the three-dimensional reference image 31 (reference image template region generating step). Then, the primary collating unit 16 performs a pattern matching of the reference image template region 33 with respect to the three-dimensional present portrait 36 (primary pattern matching step). The primary pattern matching system constitutes each slice image 53 of the reference image template area 33, and is collated with the image of the slice image 37 constituting the current image area 38. The primary collating unit 16 scans the reference image template area 33, and calculates a correlation value between the current image area 38 and the reference image template area 33 for each scanning position (correlation value calculation step), which is extracted by one pattern matching. The primary extraction area 43 is extracted by the correlation value between the current image area 38 and the reference image template area 33 so as to include the highest area (primary extraction area extraction step). Then, the primary collating unit 16 generates a single extraction region 43 in such a manner that each of the slice images 37 constituting the current image region 38 is generated, and a single extraction of the search target region to be used in the quadratic pattern matching occurs in the image region. 42 (search object generation step). The two-stage pattern matching method of the first embodiment includes a reference image template region generation step, a primary pattern matching step, and a secondary pattern matching step to be described later. The one-time pattern matching step includes a correlation value calculation step, a single extraction region extraction step, and a retrieval object generation step.

Then, the secondary collation unit 17 of the collation processing unit 22 changes the position and posture of the reference image template region 33 from the position and posture conversion unit 25, and the position and posture conversion template region 40 is present in the image with respect to the three-dimensional current image 36. The region 42 performs a quadratic pattern matching (secondary pattern matching step). The quadratic pattern matching generates a plurality of sections 41 (section generation steps) of the position and posture transformation template area 40 which are transformed to a predetermined position and posture, and for each section 41, a one-time drawing appears in the portrait area. The first extraction area 43 or the slice image 55 of the sliced image 37 of 42 is subjected to image collation with the section 41. The secondary collating unit 17 scans the position and orientation transformation template area 40, and calculates the correlation value of the plurality of sections 41 appearing in the image area 42 and the position and posture transformation template area 40 for each scanning position (correlation value calculation step) ). Then, the position/posture conversion unit 25 converts the position and posture into a position and posture (position and posture conversion step) different from the previous position and posture, and the secondary collation unit 17 generates a plurality of position and posture conversion template regions 40 in the position and posture. Section 41 (section generation step), and the position and orientation transformation template area 40 is scanned, and the correlation value of the plurality of sections 41 appearing in the image area 42 and the position and posture transformation template area 40 is calculated for each scanning position. (Relevant value calculation step). The secondary collating unit 17 of the collation processing unit 22 selects the positional relationship (position and posture information) of the highest three-dimensional reference image and the three-dimensional current image among the calculated correlation values as the optimal solution (optimal deselection step). ). In this way, it is possible to match the three-dimensional reference image and the three-dimensional portrait of the three-dimensional current image. The quadratic pattern matching step includes a section generation step, a correlation value calculation step, a position and orientation transformation step, and an optimal deselection step.

After the pattern matching is completed, the collation processing unit 22 calculates the position and orientation of the position and posture conversion template region 40 which is the highest among the calculated correlation values, and calculates the collation of the three-dimensional reference image 31 with the three-dimensional current portrait 36. Position correction amount (translation amount, rotation amount) (position correction amount calculation step). Then, the collation result display unit 23 displays the posture correction amount and an image displayed by superimposing the three-dimensional current image on which the posture correction amount has been superimposed on the three-dimensional reference image on the monitor screen of the computer 14. Further, the collation result output unit 24 outputs the posture correction amount (translation amount, rotation amount) when the three-dimensional reference image 31 and the three-dimensional current image 36 are collated by the collation processing unit 22 (body position correction amount outputting step). Then, the treatment table control parameter calculation unit 26 converts the output value (translation three axes [ΔX, ΔY, ΔZ], and the rotation three axes [ΔA, ΔB, ΔC] total six degrees of freedom) of the verification result output unit 24 into The parameters of the respective axes of the treatment table 8 are controlled, that is, the parameters for controlling the axes of the treatment table 8 (the treatment table control parameter calculation step) are calculated. Then, the treatment table 8 drives the drive means of each axis of the treatment table 8 in accordance with the treatment table control parameter calculated by the treatment table control parameter calculation unit 26 (the treatment table driving step).

In the image collating apparatus 29 of the first embodiment, the pattern matching of the three-dimensional reference image 31 with respect to the three-dimensional current image 36 is performed first, and then a predetermined quadratic pattern is generated from the three-dimensional reference image 31 based on the result of the primary pattern matching. a template area for matching, that is, a position and orientation change template area 40, and a predetermined search target area in which the one-time extraction area 43 is included for the quadratic pattern matching is generated from the three-dimensional current image 36, that is, Since the number of tomographic images (the number of slice images) of the three-dimensional current image 36 is smaller than that of the three-dimensional reference image 31, high-precision two-stage pattern matching can be realized.

In the case of the image collation device 29 of the first embodiment, even when the number of tomographic images (the number of slice images) of the three-dimensional current image 36 is smaller than that of the three-dimensional reference image 31, high-precision two-stage pattern matching can be realized, so that the pair can be reduced. The number of tomographic images of the three-dimensional current image 36 captured by the X-ray CT apparatus at the time of the position can reduce the amount of X-ray exposure that the X-ray CT apparatus receives for the patient during the alignment.

In addition, the image collating apparatus 29 of the first embodiment generates a single-draw appearance in the image area 42 based on the result obtained by performing the pattern matching of the three-dimensional reference image 31 with respect to the three-dimensional current image 36, and the portrait image 42 is present. This one-time drawing of the narrow region of the region 38 appears as the retrieval object in the portrait region 42, so that the pattern matching can be performed with a relatively wide resolution for a wider range, and then the one-time extraction region found in the one-time pattern matching is used. A single draw of 43 occurs in the portrait area 42 to perform quadratic pattern matching with a finer resolution, and the time required for pattern matching can be shortened.

The patient positioning device 30 according to the first embodiment can match the position and posture of the patient with the position and posture of the treatment plan based on the posture correction amount calculated by the image verification device 29. Since the position and posture of the patient can be made to coincide with the position and posture of the treatment plan, it is possible to align the affected part 11 at the time of treatment to the beam irradiation center 12 of the radiation therapy.

Further, in the patient positioning device 30 of the first embodiment, the position and posture conversion unit 25 can generate the number of the tomographic images (the number of slice images) from the reference image template region 33 obtained from the three-dimensional reference image 31 to be compared with the three-dimensional reference image. The three-dimensional three-dimensional present image 36 is suitable for the position and orientation change template area 40, and a two-stage type matching that also incorporates the angular offset factor into the high precision can be realized.

The image collating apparatus 29 according to the first embodiment includes a three-dimensional reference image 31 that is captured when the treatment plan for radiation therapy is taken, and a three-dimensional image input unit that reads the three-dimensional current image 36 that is taken during the treatment. 21; the three-dimensional reference image 31 and the three-dimensional current image 36 are collated, and the collation processing unit 22 that calculates the posture correction amount of the affected part in the three-dimensional current image 36 and the position and posture of the affected part in the three-dimensional reference image 31 is calculated, and the collation processing is performed. The part 22 has a primary matching unit 16 that performs one-time matching of the three-dimensional reference image 31 with respect to the three-dimensional current image 36; and generates one of the three-dimensional reference image 31 or the three-dimensional current image 36 based on the result of the primary pattern matching. The predetermined template area (position and orientation change template area 40) is different from the three-dimensional reference image 31 or three-dimensional now different from the basis of the generation of the predetermined template area (position and orientation transformation template area 40) according to the result of the one-type pattern matching. The secondary matching portion 17 of the quadratic pattern matching of the predetermined search target region 42 generated by the other of the image 36 is Even now the three-dimensional tomographic number 36 Portrait Portrait Portrait case 31 less than the reference three-dimensional, it is possible to achieve high precision two-phase samples match.

According to the patient positioning device 30 of the first embodiment, the image collating device 29 is provided, and the treatment table control parameters for controlling the parameters of the respective axes of the treatment table 8 are calculated based on the posture correction amount calculated by the image collation device 29. The image forming device 29 includes a three-dimensional reference image 31 that is captured when the treatment plan for radiotherapy is taken, and a three-dimensional image input unit that reads the three-dimensional current image 36 that is taken during the treatment. The three-dimensional reference image 31 and the three-dimensional current image 36 are collated, and the collation processing unit 22 that calculates the posture correction amount of the position and posture of the affected part in the three-dimensional current image 36 and the position and posture of the affected part in the three-dimensional reference image 31 is calculated. The collation processing unit 22 includes a primary collating unit 16 that performs one-time matching of the three-dimensional reference image 31 with respect to the three-dimensional current image 36, and one of the three-dimensional reference image 31 or the three-dimensional current image 36 based on the result of the primary pattern matching. The generated predetermined template area (position and orientation change template area 40) is different from the generation basis of the predetermined template area (position and orientation transformation template area 40) with respect to the result of the primary pattern matching, or a three-dimensional reference image 31 or three-dimensional In the case where the number of the tomographic images of the three-dimensional current image 36 is smaller than that of the three-dimensional reference image 31, the number of the three-dimensional reference images 31 can be changed even if the number of the three-dimensional reference images 31 is smaller than the three-dimensional reference image 31. High precision alignment.

The image collation method according to the first embodiment is a method for collating the three-dimensional reference image 31 taken during the treatment plan for the radiotherapy treatment and the three-dimensional current image 36 taken during the treatment, and includes: performing the three-dimensional reference a pattern matching step of the pattern 31 with respect to the one-time matching of the three-dimensional current portrait 36; and a predetermined template region (position) generated from one of the three-dimensional reference image 31 or the three-dimensional current image 36 according to the result of the one-time pattern matching The posture change template area 40) is generated from the other of the three-dimensional reference image 31 or the three-dimensional current image 36 different from the basis of the generation of the predetermined template region (position and orientation transformation template region 40) based on the result of the primary pattern matching. Since the secondary pattern matching step of the quadratic pattern matching of the predetermined search target region 42 is performed, even if the number of tomographic images of the three-dimensional current image 36 is smaller than that of the three-dimensional reference image 31, a high-precision two-stage pattern can be realized. match.

Embodiment 2

The two-stage pattern matching of the second embodiment is performed by first matching the three-dimensional reference image 31 to the one-dimensional pattern matching of the three-dimensional current image 36, and then generating a predetermined second from the three-dimensional current image 36 based on the result of the primary pattern matching. In the template area for the pattern matching (the present image template area 44), the posture conversion reference image area 47 obtained by converting the position and posture of the three-dimensional reference image 31 is used as a search target, and the current image template area 44 is changed with respect to the posture. The secondary pattern matching of the reference image area 47. The quadratic pattern matching is a pattern matching in the opposite direction to the one-type matching.

Fig. 10 is a view for explaining a primary pattern matching method in the second embodiment of the present invention, and Fig. 11 is for explaining a relationship between a reference image template area and a slice portrait in the primary pattern matching method of Fig. 10. Figure. In the primary pattern matching according to the second embodiment, the primary collating unit 16 performs the search including the rotation three axes to obtain the posture change amount.

The present portrait area 38 shown in Fig. 10 is represented as a cube containing three slice images 37a, 37b, 37c. The position and orientation conversion template areas 40a, 40b, and 40c of the reference image template area in the second embodiment are the areas in which the position and posture are converted by the position and posture conversion unit 25. The initial position and posture is a default state, for example, a state in which the parameters of the three axes of rotation are all zero. The position and posture conversion template area 40a obtained by converting the position and posture of the reference image template area is moved in a progressive scan manner along the scanning path 39a on the slice image 37a. In the same manner, the position and posture change template area 40b obtained by changing the position and posture is moved in a zigzag manner along the scan path 39b on the slice image 37b, and the position and posture change template area 40c is obtained by converting the position and posture. The slice image 37c is moved in a progressive scan along the scanning path 39c. Among them, the scanning paths 39b and 39c are simplified in order not to complicate the drawing.

The correlation between the slice images 37a, 37b, 37c of the three-dimensional current image 36 and the position and posture conversion template region 40 is performed while changing the position and posture. For example, the respective axes of the three axes of rotation are changed by a predetermined amount of change or rate of change, and the correlation calculation is performed, and then moved to the next scanning position, and then the correlation calculation is performed. As shown in Fig. 11, the primary collating unit 16 collates the image between the section 41 of the position and orientation conversion template area 40 and the slice image 37 constituting the present portrait area 38. The cross-section 41 of the position-and-posture conversion template area 40 is a section obtained by cutting the position-and-posture conversion template area 40 on a plane parallel to the sliced image 32 of the three-dimensional reference image 31 in the first position and posture. The generator of the plurality of slice images 32 of the three-dimensional reference image 31 (section generation step). The section 41 can be produced by, for example, the method described in the first embodiment. That is, the data of the section 41 can be set to be cut out from the plurality of slice images 32 constituting the three-dimensional reference image 31. Further, the section 41 of the position/posture conversion template area 40 may be included in a material that is added in such a manner that the data density thereof is the same as the data density of the three-dimensional current portrait 36.

Then, the primary collating section 16 produces the present portrait template area 44 to be used in the quadratic pattern matching. The primary matching unit 16 obtains the result of the search including the rotation three axes from the slice images 37a, 37b, and 37c, for example, and obtains the section 41 of the position and posture conversion template area 40 having the highest correlation value. The position and posture change the posture change amount of the template region 40 and the extraction region of the slice image 37 corresponding to the cross section 41. The primary matching unit 16 generates a current image template area 44 including the extraction area of the three-dimensional current image having the highest correlation value from among the extracted extraction areas of the respective slice images. The portrait template area 44 is now a two-dimensional image.

Then, as shown in FIG. 12, the posture and posture conversion unit 25 of the collation processing unit 22 converts the posture of the entire three-dimensional reference image 31 by the posture change amount obtained when the current image template region 44 is generated, and generates The three-dimensional posture transformation reference image 45 after the posture conversion, that is, the posture transformation reference image region 47 is generated. Fig. 12 is a view showing a three-dimensional reference image after the posture conversion in the second embodiment of the present invention. The slice images 46a, 46b, 46c, 46d, and 46e are slice images in which the postures of the slice images 32a, 32b, 32c, 32d, and 32e are changed by the posture change amount.

Then, as shown in Fig. 13, the secondary collating unit 17 causes the current image template area 44 to be progressively scanned along the scanning path 49 in the three-dimensional posture conversion reference image 45 after the posture conversion, that is, the posture conversion reference image area 47. When the shape is moved and compared, the shift of only the translation can be detected at high speed. Figure 13 is a view for explaining a quadratic pattern matching method in the second embodiment of the present invention. The posture conversion reference image area 47 which is transformed by the posture is expressed as a cube including five slice images 46a, 46b, 46c, 46d, and 46e. The collation execution surface 48 is a portrait surface corresponding to a posture (this posture is a posture in which the pose corresponding to the slice image 37 of the three-dimensional current image 36 has the highest correlation value). In other words, the posture in the posture transformation reference image region 47 and the posture corresponding to the slice image 37 of the three-dimensional current image 36 have the same posture. The secondary collating unit 17 generates a predetermined collation execution surface 48 (check execution surface generation step) in the posture conversion reference image region 47 from the plurality of slice images 46 of the three-dimensional posture conversion reference image 45. The collation execution surface 48 can be generated by, for example, the method described in the first embodiment. That is, the data of the collation execution surface 48 can be set as a plurality of slice image cutouts constituting the three-dimensional posture transformation reference image 45. Further, the collation execution surface 48 may be a data included in a manner that allows the data density to be increased in the same manner as the data density of the current image template area 44.

The two-stage pattern matching method of Embodiment 2 will be summarized. First, the position/posture conversion template area 40 obtained by the position and posture transformation is generated from the three-dimensional reference image 31 by the position/posture conversion unit 25 of the collation processing unit 22 (position and posture conversion template area generation step). Then, the primary collating unit 16 of the collation processing unit 22 performs the pattern matching of the position and orientation conversion template region 40 with respect to the three-dimensional present image 36 (primary pattern matching step). The primary pattern matching is performed for each of the slice images 37 constituting the current image region 38, and each time the position and posture of the position and posture conversion template region 40 is changed (each time the position and posture transformation step is performed), the position and posture transformation template region is generated. The section 41 of the 40 (the section generating step) performs the image collation between the section 41 of the position and orientation conversion template area 40 and the section image 37 constituting the current portrait area 38.

Further, the primary collating unit 16 calculates a correlation value appearing in the portrait region 38 and the position and posture transformation template region 40 each time the position and posture of the position and posture transformation template region 40 is changed (correlation value calculation step). And, the primary collating unit 16 calculates the correlation value appearing in the image area 38 and the position and posture transformation template area 40 each time the position and posture transformation template area 40 is scanned, and the first pattern matching is used to generate the current portrait. The extracted area of the positional posture changing template area 40 in which the correlation value of the area 38 and the position and posture changing template area 40 is the highest is included in the present portrait template area 44 (the present image template area generating step).

Then, the collation processing unit 22 converts the posture of the entire three-dimensional reference image 31 by the posture change amount obtained when the current image template region 44 is generated by the position and orientation conversion unit 25, and generates a three-dimensional posture transformation reference after the posture conversion. The image 45, that is, the posture conversion reference image region 47 (posture conversion reference image region generation step) is generated. Then, the secondary collating unit 17 performs the quadratic pattern matching (secondary pattern matching step) of the present portrait template region 44 with respect to the posture transformation reference image region 47. The quadratic pattern matching generates a collation execution surface 48 in the collation execution surface generation step, and then performs an image collation of the collation execution surface 48 generated in the collation execution surface generation step and the current image template region 44. When the image is checked, the correlation value between the collation execution surface 48 and the current image template area 44 is calculated while the current image template area 44 is rotated (non-rotation) (correlation value calculation step).

In the quadratic pattern matching, the secondary collating unit 17 of the collation processing unit 22 sets the correlation value to the position and posture relationship (position) of the highest three-dimensional posture transformation reference image 45 and the current image template region 44 among the calculated correlation values. Posture information) is selected as the best solution (best deselection step). In this way, the pattern matching of the three-dimensional portraits of the three-dimensional reference image 31 and the three-dimensional current image 36 can be achieved by two-stage matching. The two-stage pattern matching method of the second embodiment includes a position and orientation transformation template region generation step, a primary pattern matching step, a posture transformation reference image region generation step, and a quadratic pattern matching step. The one-time pattern matching step includes a section generation step, a correlation value calculation step, a position and orientation transformation step, and a current portrait template area generation step. The quadratic pattern matching step includes a check execution surface generation step, a correlation value calculation step, and an optimal deselection step.

After the pattern matching is completed, the collation processing unit 22 calculates the position and orientation of the high correlation value region in the three-dimensional posture transformation reference image 45 which is the highest among the calculated correlation values, and calculates the three-dimensional reference image 31 and the three-dimensional present. The posture correction amount (translation amount, rotation amount) at the time of the 36-phase collation of the portrait (the posture correction amount calculation step). Then, the collation result display unit 23 displays the posture correction amount and an image displayed by superimposing the three-dimensional current image on which the posture correction amount has been superimposed on the three-dimensional reference image on the monitor screen of the computer 14. In addition, the collation result output unit 24 outputs the posture correction amount (translation amount, rotation amount) when the collation processing unit 22 collates the three-dimensional reference image 31 and the three-dimensional current image 36 (the posture correction amount output step). Then, the treatment table control parameter calculation unit 26 converts the output value (translation three axes [ΔX, ΔY, ΔZ], and the rotation three axes [ΔA, ΔB, ΔC] total six degrees of freedom) of the verification result output unit 24 into The parameters of the respective axes of the treatment table 8 are controlled, that is, the parameters for controlling the axes of the treatment table 8 (the treatment table parameter calculation step) are calculated. Then, the treatment table 8 drives the drive means of each axis of the treatment table 8 in accordance with the treatment table control parameter calculated by the treatment table control parameter calculation unit 26 (the treatment table driving step).

In the image collating apparatus 29 of the second embodiment, the position and orientation conversion template area 40 of the three-dimensional reference image 31 is matched with the one-dimensional pattern of the three-dimensional current image 36 including the three-axis rotation image, and then the primary type is matched. As a result of the matching, the template area for the quadratic pattern matching is generated from the three-dimensional current image 36, that is, the current image template area 44, so that the number of tomographic images (the number of slice images) of the three-dimensional current image 36 is larger than the three-dimensional reference image. In the case of 31, it is also possible to achieve high-precision two-stage pattern matching.

Further, in the image collating apparatus 29 of the second embodiment, the three-dimensional posture conversion reference image 45 (three-dimensional reference image after the posture conversion) is generated from the three-dimensional reference image 31, that is, the posture conversion reference image region 47 is generated. Direct pattern matching is achieved by using the two-dimensional present portrait template area 44 so as to be moved in parallel with the posture-changing reference image area 47 without rotational movement. Further, in the quadratic pattern matching, only the correlation value for each parallel movement is calculated, so that the speed of the quadratic pattern matching can be achieved as compared with the case of calculating the correlation value of each rotation movement and parallel movement.

Embodiment 3

In the third embodiment, the reference image region 33 for the primary pattern matching in the first embodiment or the reference image based on the position and posture conversion template region 40 in the second embodiment is generated using the atlas model. The points of the template area 33 are different from those of the first and second embodiments. Fig. 14 is a view showing the configuration of a portrait collating apparatus and a patient positioning apparatus according to a third embodiment of the present invention. The image collating apparatus 29 of the third embodiment differs from the image collating apparatus 29 of the first and second embodiments in that it has a human body library input unit 50 and an average template area generating unit 51. The patient positioning device 30 according to the third embodiment includes an image collation device 29 and a treatment table control parameter calculation unit 26.

The human body database input unit 50 acquires an atlas model from a memory device such as a database device. The average template region generating portion 51 cuts out the average template region 54 from the organ portions of the human body database corresponding to the affected portions 5, 11 of the patients 4, 10. The reference template region generating unit 18 of the collation processing unit 22 automatically generates the reference image template region 33 (reference image template region generating step) by pattern matching the average template region 54 with the three-dimensional reference image 31.

The two-stage pattern matching in the first embodiment or the two-stage pattern matching in the second embodiment is performed using the above-described reference image template area 33. In this way, even if the information indicating the affected part (the shape of the affected part or the like) is not prepared in advance in the three-dimensional reference image, the two-stage pattern matching can be realized.

Further, it is also conceivable that the average template region generating portion 51 cuts out a two-dimensional average template region from the organ portions of the human body database corresponding to the affected portions 5, 11 of the patients 4, 10. In the case of the two-dimensional average template region 54, a plurality of two-dimensional average template regions can be cut out, and then a plurality of two-dimensional average template regions are collectively output to the verification processing portion 22. The reference template region generating unit 18 of the collation processing unit 22 automatically generates the reference image template region 33 by patterning the plurality of two-dimensional average template regions with the three-dimensional reference image 31.

1. . . CT simulation room

2,7. . . CT tube rack

3,9. . . roof

4,10. . . patient

5,11. . . Suffering department

6. . . Treatment room

8. . . Rotary treatment table

12. . . Beam irradiation center

13. . . Illuminating head

14. . . computer

16. . . One check department

17. . . Secondary checkup

18. . . Reference template area generation unit

twenty one. . . 3D portrait input section

twenty two. . . Check processing department

twenty three. . . Check result display

twenty four. . . Check result output

25. . . Position and posture transformation unit

26. . . Treatment table control parameter calculation unit

29. . . Image checking device

30. . . Patient positioning device

31. . . 3D reference image

32a~32e. . . Slice portrait

33, 33a, 33b, 33c. . . Benchmark image template area

34. . . External quadrilateral

35. . . ROI (attention area)

36. . . 3D present portrait

37a~37c. . . Slice portrait

38. . . Current portrait area

39a~39c. . . Scanning path

40, 40a, 40b, 40c. . . Position and posture transformation template area

41. . . Section

42. . . One draw appears in the portrait area

43. . . One extraction area

44. . . Present model area

45. . . 3D posture transformation reference image

46a~46e. . . Slice portrait

47. . . Posture transformation reference image area

48. . . Check execution surface

49. . . Scanning path

50. . . Human body database input

51. . . Average template area generation department

53. . . Slice portrait

55. . . Slice portrait

Fig. 1 is a view showing the configuration of an image collating apparatus and a patient positioning apparatus according to Embodiment 1 of the present invention.

Fig. 2 is a view showing the overall configuration of a machine relating to the image collating apparatus and the patient positioning apparatus of the present invention.

Fig. 3 is a view showing a three-dimensional reference image and a reference image template area in the first embodiment of the present invention.

Fig. 4 is a view showing a three-dimensional present portrait in the first embodiment of the present invention.

Fig. 5 is a view for explaining a primary pattern matching method in the first embodiment of the present invention.

Fig. 6 is a view for explaining the relationship between the reference image template area and the slice image in the primary pattern matching method of Fig. 5.

Fig. 7 is a view showing a primary extraction region of a slice image extracted by the primary pattern matching method in the first embodiment of the present invention.

Fig. 8 is a view for explaining a secondary pattern matching method in the first embodiment of the present invention.

Fig. 9 is a view for explaining the relationship between the reference image template area and the slice image in the quadratic pattern matching method of Fig. 8.

Fig. 10 is a view for explaining the primary pattern matching method in the second embodiment of the present invention.

Fig. 11 is a view for explaining the relationship between the reference image template area and the slice portrait in the primary pattern matching method of Fig. 10.

Fig. 12 is a view showing a three-dimensional reference image after the posture conversion in the second embodiment of the present invention.

Figure 13 is a view for explaining a quadratic pattern matching method in the second embodiment of the present invention.

Fig. 14 is a view showing the configuration of a portrait collating apparatus and a patient positioning apparatus according to a third embodiment of the present invention.

16. . . One check department

17. . . Secondary checkup

18. . . Reference template area generation unit

twenty one. . . 3D portrait input section

twenty two. . . Check processing department

twenty three. . . Check result display

twenty four. . . Check result output

25. . . Position and posture transformation unit

26. . . Treatment table control parameter calculation unit

29. . . Image checking device

30. . . Patient positioning device

Claims (16)

An image collation apparatus includes: a three-dimensional image input unit that reads a three-dimensional reference image taken during a treatment plan for radiation therapy and a three-dimensional current image taken during treatment; and a check processing unit And collating the three-dimensional reference image and the three-dimensional current image, and calculating a posture correction amount for causing the position and posture of the affected part in the three-dimensional current image to match the position and posture of the affected part in the three-dimensional reference image, the collation processing unit having: a check a first type matching of the three-dimensional reference image from the three-dimensional current image; and a secondary collation unit for performing the three-dimensional reference image or the three-dimensional current image from the result of the matching of the one-time pattern The predetermined template region generated by one of the predetermined search results generated from the other of the three-dimensional reference image or the three-dimensional current image different from the generation basis of the predetermined template region based on the result of the matching of the one-time pattern. A quadratic pattern match of the object area. The image matching device according to the first aspect of the invention, wherein the verification processing unit includes a reference image template region in which a three-dimensional region is generated from the three-dimensional reference image based on the affected part information prepared in the three-dimensional reference image. The reference template area generation unit. The image collation apparatus according to claim 1, wherein the human body database input unit is configured to obtain a human body database from the database device; and the average template region generating unit is configured from the human body database. An organ portion corresponding to the affected part of the patient is used to generate an average template region, and the collation processing portion has a pattern matching from the average template region with respect to the three-dimensional reference image, and is matched according to the result of the pattern matching. A reference template region generating portion of the reference image template region of the three-dimensional region is generated from the three-dimensional reference image. The image collation apparatus according to claim 2, wherein the primary collating unit matches the pattern of the three-dimensional current image from the reference image template region when the primary pattern is matched. The image collation apparatus according to claim 4, wherein the primary collating unit generates, as the search target, a region in which the correlation value with the reference image template region is the highest from the three-dimensional current image. A single draw of the area appears in the portrait area. The image collation processing unit according to claim 5, wherein the collation processing unit includes a position and posture converting unit that converts a position and a posture of the three-dimensional image, and the position and posture converting unit generates the reference image template area. The position and orientation change to the position and orientation change template area of the predetermined position and posture, and the secondary check unit performs the position and posture change template area as the predetermined template area with respect to the aforementioned A pattern matching appears in the portrait area at a time. The image collation apparatus according to claim 6, wherein the second collation unit generates a cross section of the position and orientation conversion template region, and then performs the type of the first extraction between the image region and the cross section. Like matching. The image collation processing unit according to claim 2, wherein the collation processing unit includes a position and posture converting unit that converts a position and a posture of the three-dimensional image, and the position and posture converting unit generates the reference image. The position and orientation of the template area is changed to a position and posture change template area of a predetermined position and posture, and the primary check unit performs pattern matching of the three-dimensional current image from the position and posture change template region when the one-time pattern matching is performed. . The image collation apparatus according to claim 8, wherein the first collating section generates a cross section of the position and orientation conversion template region, and then performs pattern matching between the three-dimensional current image and the cross section. And executing: determining a high correlation section of the section having the highest correlation value among the plurality of sections, calculating a posture change amount of the position and posture transformation template area, and corresponding to the high correlation section in the three-dimensional current image Extraction of the extracted area. The image collation apparatus according to claim 9, wherein the primary collating unit generates a current portrait in which the extracted region extracted when the one-time pattern is matched is included in the predetermined template region. In the template region, the position and orientation conversion unit generates a three-dimensional posture conversion reference image region in which the position and orientation of the three-dimensional reference image is converted only by the posture change amount of the extraction region corresponding to the current image template region, and the secondary collation portion is In the case of the above-described quadratic pattern matching, the pattern matching of the three-dimensional posture conversion reference image region as the search target region from the current image template region is performed. The image collation apparatus according to claim 10, wherein the secondary collation unit generates a collation execution surface of a cross section belonging to the posture transformation reference image region, and then performs the current portrait template region and the collation execution. The pattern between the faces matches. A patient positioning device comprising: an image collation device according to any one of claims 1 to 3, 5 to 7, and 9 to 11; and a posture calculated according to the image reconciliation device The correction amount is used to calculate a treatment table control parameter calculation unit for controlling parameters of each axis of the treatment table. A method for checking a portrait, which is a three-dimensional reference image taken at the time of the treatment plan for radiation therapy and a three-dimensional current image taken at the time of treatment, including: performing the three-dimensional reference image from the aforementioned three-dimensional current portrait a pattern matching matching one-time pattern matching step; and performing a predetermined template region generated from one of the aforementioned three-dimensional reference image or the aforementioned three-dimensional current image based on the result of the aforementioned one-time pattern matching, with respect to matching according to the aforementioned one-time pattern As a result, a quadratic pattern matching step of matching the secondary pattern of the predetermined search target region generated by the other of the three-dimensional reference image or the three-dimensional current image different from the generation basis of the predetermined template region is performed. The image collation method according to claim 13, comprising: generating a reference image template region generating step of the reference image template region of the three-dimensional region from the three-dimensional reference image, and performing the one-time pattern matching step The pattern of the reference image template is matched with the one-time pattern of the aforementioned three-dimensional present image. A patient positioning device comprising: an image collation device according to item 4 of the patent application scope; and a treatment table for calculating parameters of each axis of the treatment table according to the posture correction amount calculated by the image reconciliation device Control parameter calculation unit. A patient positioning device comprising: a portrait collation device according to item 8 of the patent application scope; and a treatment table for calculating parameters of each axis of the treatment table according to the posture correction amount calculated by the image reconciliation device Control parameter calculation unit.