JP5675424B2 - Radiotherapy apparatus control apparatus, processing method of radiotherapy apparatus control apparatus, and program thereof - Google Patents

Description

  The present invention relates to a radiation therapy apparatus control apparatus, a radiation therapy apparatus control apparatus processing method, and a program therefor, which detect a marker in a digitally generated radioscopic image by irradiating a subject in which the marker is embedded.

  The position of the affected part is specified by embedding a linear marker mainly made of metal in the vicinity of the affected part of the human body, and then specifying the position of the marker from a radiographic image obtained by irradiating the human body with radiation. Technology is used in radiotherapy equipment. A related technique is disclosed in Patent Document 1.

Japanese Patent No. 3053389

In the technique of the above-described Patent Document 1, a marker is detected from an image by performing template matching using a density normalization cross-correlation method using an image obtained by imaging and a template image. .
However, since this method uses the normalized cross-correlation method, there is a problem that processing time is required.

  Accordingly, an object of the present invention is to provide a radiotherapy apparatus control apparatus, a processing method for the radiotherapy apparatus control apparatus, and a program thereof that can solve the above-described problems.

In order to achieve the above object, the present invention provides a radiotherapy apparatus controller for detecting a marker in a radioscopic image digitally generated by irradiating a subject in which the marker is embedded, wherein the radioscopy is performed. Any of the luminance differences between the luminance of a certain pixel in the image and the luminances of two opposing pixels located on a straight line in a predetermined direction passing through the certain pixel and positioned at a predetermined distance from the certain pixel. One of the luminance differences is calculated for a plurality of different directions passing through the certain pixel, information on the largest luminance difference among them is specified for the certain pixel, and information on the luminance difference is stored in the plurality of fluoroscopic images. a marker edge enhancement unit that creates a pixel for each luminance difference information held about the pixel, the luminance difference in the pixel each brightness difference information in the predetermined direction to the above pixel predetermined threshold A process for identifying a pixel in a predetermined range from the corresponding pixel as a marker appearance pixel, and a predetermined range from the pixel connected in the predetermined direction to a pixel having the luminance difference less than a predetermined threshold in the luminance difference information for each pixel The process of identifying the pixel as a marker non-appearing pixel is repeated a predetermined number of times for each of the plurality of different directions, and noise among pixels whose luminance difference is greater than or equal to a predetermined threshold in the pixel-by-pixel luminance difference information. A noise removing unit that removes pixels to be determined and identifies pixels that are marker candidates, and the pixels that are marker candidates are classified into a plurality of regions based on a group of adjacent pixels, and the region range is equal to or greater than a threshold value A radiotherapy apparatus control apparatus comprising: a marker specifying unit that specifies a pixel belonging to a region as a pixel indicating a marker.

  The present invention further includes a logarithmic conversion unit that logarithmically converts a luminance value of each pixel in the radiographic image in the radiotherapy apparatus control device, wherein the marker edge emphasis unit is the logarithmic conversion unit Of the luminance difference between the luminance of a certain pixel in a fluoroscopic image and the luminance of two opposing pixels located on a straight line in a predetermined direction passing through the certain pixel and positioned at a predetermined distance from the certain pixel Is calculated for a plurality of different directions passing through the certain pixel, information on the largest luminance difference among them is specified for the certain pixel, and information on the luminance difference is included in the radiographic image. The brightness difference information for each pixel stored for the plurality of pixels is created.

  According to the present invention, in the above-described radiotherapy apparatus control apparatus, a plurality of on-marker specifying parts for specifying a plurality of points including end points in the radioscopic image of the marker appearing in the radioscopic image, and a plurality of the markers Template image information including a template range setting unit for setting a template image range including a point, information on the end point, information on the template image range, and information on a position of the template image range in the fluoroscopic image A marker image generation unit that determines a search range including a region where the classified region range is equal to or greater than a threshold, and indicates the pixels in the search range and the template image A correlation value calculated using a luminance difference based on the luminance difference information for each pixel of the pixel is calculated, and the correlation value is equal to or greater than a threshold value. The coordinate of the end point of the marker that appears in the fluoroscopic image is specified using the origin of the position range within the search range and the coordinates of the end point of the marker indicated by the template image in the template image. It is characterized by that.

  Further, the present invention is the above-described radiotherapy device control device, wherein the plurality of on-marker specifying units generate a plurality of points including end points in the radiographic image of markers that appear in the radiographic image at different times. The template range setting unit sets a template image range including a plurality of points of the marker for each of the radioscopic images generated at different times, and the template image generation unit The template image information including the end point information, the template image range information, and the position information of the template image range in the radioscopic image, respectively, the radiographic images generated at different times, respectively. A plurality of pixel values are generated, and a range of pixels close to the template image is determined. A matching target range specifying unit that specifies a matching target range for each of the plurality of template images, and the marker specifying unit determines a search range including a region where the range of the classified region is equal to or greater than a threshold, and the search The matching target range including the range is determined, and the pixel-by-pixel luminance difference information of the pixel in the same position range as the template image range belonging to the determined matching target range in the search range and the pixel indicating the template image When the correlation value calculated using the luminance difference based on is calculated and the correlation value is determined to be equal to or greater than the threshold, the origin of the position range within the search range and the template image belonging to the determined matching target range are Using the coordinates of the end points of the marker to be displayed in the template image. And identifies the endpoint coordinates of the marker to be.

  In the radiotherapy apparatus control device described above, the marker specifying unit may include a plurality of different matching targets including the search range when there are a plurality of matching target ranges including the search range. A range is determined, and for a partial range of the search range belonging to one matching target range among the determined plurality of matching target ranges, a pixel belonging to the partial range and a template belonging to the one matching target range A correlation value calculated using a luminance difference based on the pixel-by-pixel luminance difference information of a pixel indicating an image is calculated, and the origin of the position range in the search range used for determining that the correlation value is equal to or greater than a threshold value and Using the coordinates in the template image of the end point of the marker indicated by the template image belonging to the one matching target range, And identifies the endpoint coordinates of the marker appearing in the serial fluoroscopic images.

  Further, the present invention provides the above-described radiotherapy apparatus control apparatus, wherein information other than pixels corresponding to the line connecting the plurality of points in the radiographic image is determined to be excluded in the subsequent processing, and indicates the target exclusion A template mask setting unit that generates mask information for each pixel corresponding to a line connecting a plurality of points corresponding to the line connecting the plurality of points for each of the radiographic images generated at different times. The marker specifying unit calculates the correlation value without using the information on the luminance difference between the pixels held in the mask information.

Further, the present invention is a processing method for detecting the marker in a radiographic image digitally generated by irradiating a subject in which the marker is embedded, and the luminance of a certain pixel in the radiographic image, The luminance difference of any one of the luminance differences between the luminances of two opposing pixels located on a straight line in a predetermined direction passing through the certain pixel and located at a predetermined distance from the certain pixel is determined as the certain pixel. Per-pixel luminance difference information that is calculated for a plurality of different directions passing through the image, specifies the information on the largest luminance difference among them for the certain pixel, and holds the information on the luminance difference for the plurality of pixels in the radiographic image create a pixel marker occurrence pixel and Japanese in a predetermined range from the pixel of the luminance difference in the pixel each luminance difference information is continuous in the predetermined direction to the above pixel predetermined threshold And a process of identifying a pixel in a predetermined range from the pixel connected in the predetermined direction to a pixel whose luminance difference is less than a predetermined threshold in the luminance difference information for each pixel as a marker non-appearing pixel, a predetermined number of times. And separately repeating for each of the plurality of different directions, and identifying pixels that are marker candidates by removing pixels that are determined to be noise from pixels in which the luminance difference is equal to or greater than a predetermined threshold in the luminance difference information for each pixel. The pixel candidate is classified into a plurality of regions according to a group of adjacent pixels, and a pixel belonging to a region whose region range is equal to or greater than a threshold is specified as a pixel indicating a marker. It is.

  In the processing method described above, the present invention performs logarithmic conversion on the luminance value of each pixel in the radiographic image, and calculates the luminance of a certain pixel and the certain pixel in the radiographic image after the logarithmic conversion. A luminance difference of any one of the luminance differences between two pixels facing each other located on a straight line in a predetermined direction passing through and at a predetermined distance from the certain pixel is a plurality of luminance differences passing through the certain pixel. Calculating for different directions, specifying information on the largest luminance difference among them for a certain pixel, and creating pixel-by-pixel luminance difference information holding the luminance difference information for a plurality of pixels in the fluoroscopic image It is characterized by.

  In the processing method described above, the present invention specifies a plurality of points including end points in the fluoroscopic image of the marker appearing in the fluoroscopic image, sets a template image range including the multiple points of the marker, Generating template image information including information on the end points, information on the template image range, and information on a position of the template image range in the fluoroscopic image, and a region in which the range of the classified region is a threshold value or more A correlation value calculated using a luminance difference based on the luminance difference information for each pixel of the pixel in the search range and the pixel indicating the template image, and the correlation value is equal to or greater than a threshold value In the template image at the origin of the position range in the search range and the end point of the marker indicated by the template image By using the coordinates, and identifies the endpoint coordinates of the marker appearing in the fluoroscopic image.

According to another aspect of the present invention, there is provided a computer of a radiation therapy apparatus control device that detects the marker in a radioscopic image digitally generated by irradiating a subject in which the marker is embedded, with respect to a certain pixel in the radiographic image. The luminance difference of any one of the luminance and the luminance of each of two opposing pixels located on a straight line in a predetermined direction passing through the certain pixel and positioned at a predetermined distance from the certain pixel, For each pixel that is calculated for a plurality of different directions passing through the pixel, specifies the information on the largest luminance difference among the pixels, and holds the information on the luminance difference for the pixels in the fluoroscopic image. marker edge enhancement means for creating a luminance difference information, those of the luminance difference in the pixel each luminance difference information is continuous in the predetermined direction to the above pixel predetermined threshold A process of identifying a pixel in a predetermined range from a pixel as a marker appearing pixel, and a pixel in a predetermined range from the pixel connected in the predetermined direction to a pixel having the luminance difference less than a predetermined threshold in the luminance difference information for each pixel Pixels that are determined as noise among the pixels in which the luminance difference is equal to or greater than a predetermined threshold in the luminance difference information for each pixel, by repeating the process of specifying a non-appearing pixel separately for each of the plurality of different directions. Noise removal means for identifying pixels that are candidate markers by removing the pixels, the pixels that are candidate markers are classified into a plurality of regions according to a group of adjacent pixels, and pixels belonging to regions where the region range is equal to or greater than a threshold , A program that functions as marker specifying means for specifying as a pixel indicating a marker.

  Further, the present invention causes the above-described program to function as a logarithmic conversion unit that logarithmically converts a luminance value for each pixel in the radiographic image, and the marker edge enhancement unit is included in the radiographic image after the logarithmic conversion. Either one of the luminance difference between the luminance of a certain pixel and the luminance of two opposing pixels located on a straight line in a predetermined direction passing through the certain pixel and positioned at a predetermined distance from the certain pixel. Is calculated for a plurality of different directions passing through the certain pixel, information on the largest luminance difference among them is specified for the certain pixel, and the information on the luminance difference is determined for the plurality of pixels in the fluoroscopic image. It is made to function as a means to produce the brightness | luminance difference information for every pixel hold | maintained about.

  Further, the present invention provides the above-described program, a plurality of on-marker specifying means for specifying a plurality of points including end points in the fluoroscopic image of the marker appearing in the fluoroscopic image, a template image including the plurality of markers Template range setting means for setting a range, template image generation for generating template image information including information on the end point, information on the template image range, and information on a position of the template image range in the radiographic image And the marker specifying means determines a search range including a region where the range of the classified region is equal to or greater than a threshold value, and the pixel-by-pixel brightness of pixels within the search range and a pixel indicating the template image A correlation value calculated using a luminance difference based on the difference information is calculated, and it is determined that the correlation value is equal to or greater than a threshold value. As a means for specifying the end point coordinates of the marker appearing in the radioscopic image using the origin of the position range within the search range and the coordinates of the end point of the marker indicated by the template image in the template image It is made to function.

According to the present invention, since the normalized correlation process is not used in generating the marker image, the marker detection speed can be increased.
In addition, logarithmic conversion of the luminance value of each pixel of the fluoroscopic image is performed, thereby performing a process of making the luminance difference between the luminance of each pixel indicating the marker and the luminance of the pixels in other portions uniform. Therefore, it is possible to easily detect a marker using a threshold value of a certain luminance value, thereby increasing the marker detection processing speed.

It is a block diagram which shows the structure of a radiotherapy apparatus control apparatus. It is an image figure of a radiographic image. It is a 1st figure which shows the flow of the pre-processing of a radiotherapy apparatus control apparatus. It is a 1st figure which shows the outline | summary of the preparation process of luminance difference information for every pixel. It is a 2nd figure which shows the outline | summary of the preparation process of luminance difference information for every pixel. It is a figure which shows a template image range. It is a 1st figure which shows a template mask process. It is a 2nd figure which shows a template mask process. It is a 3rd figure which shows a template mask process. It is a figure which shows the outline | summary of a pattern matching pre-processing. It is a figure which shows the processing flow of a marker specific process. It is a figure which shows the process direction on the basis of the object pixel used by a noise removal process. It is a 1st figure which shows the process outline | summary of a noise removal process. It is a 1st figure which shows the process outline | summary of a noise removal process. It is a 2nd figure which shows the process outline | summary of a noise removal process. It is a 3rd figure which shows the process outline | summary of a noise removal process. It is a 4th figure which shows the process outline | summary of a noise removal process. It is a 5th figure which shows the process outline | summary of a noise removal process. It is a figure which shows the outline | summary of a marker candidate specific process. It is a figure which shows the outline | summary of a pattern matching pre-processing. It is a 1st figure which shows the outline | summary of a pattern matching process. It is a figure which shows the outline | summary which calculates the end point of a marker using a rotation template image. It is a 2nd figure which shows the outline | summary of a pattern matching process.

A radiotherapy apparatus control apparatus according to an embodiment of the present invention will be described below with reference to the drawings.
FIG. 1 is a block diagram showing a configuration of a radiotherapy apparatus control apparatus according to the embodiment.
In this figure, reference numeral 1 denotes a radiotherapy apparatus control apparatus. The radiotherapy apparatus control apparatus 1 includes a radiological fluoroscopic image generation unit 101, a template registration control unit 102, a marker multiple point identification unit 103, a logarithmic conversion unit 104, a marker edge enhancement unit 105, a template range setting unit 106, and end point positions. A setting unit 107, a template mask setting unit 108, a template image generating unit 109, a matching target range specifying unit 110, a noise removing unit 111, a marker specifying unit 112, and a storage unit 113 are provided.

  A radiotherapy apparatus (not shown) includes a radiation irradiation apparatus that emits radiation (X-rays) and a radiation detector for detecting the radiation. The radiation detector receives radiation emitted from the radiation irradiating apparatus and transmitted through a subject (human body) (not shown), and generates image data for generating a fluoroscopic image of the subject. The radiotherapy apparatus control apparatus 1 causes the above-described processing units to function by starting a program, such as digitally generating a fluoroscopic image based on the image data obtained from the radiotherapy apparatus. Examples of the radiation detector include an FPD (Flat Panel Detector) and an X-ray II (Image Intensifier).

  In the vicinity of the affected part of the human body, a marker made of Au, Pt or the like that has little influence on the human body and has low X-ray permeability is embedded. For this reason, it is possible to easily identify the affected part position by specifying the marker position in the radioscopic image.

FIG. 2 is an image view of a radiographic image.
FIG. 3 is a first diagram illustrating a pre-processing flow of the radiotherapy apparatus control apparatus.
Next, the details of the processing flow of the radiotherapy apparatus control apparatus according to the present embodiment will be described. In the present embodiment, the marker has a linear shape, and a case where the marker is arranged in a curved shape (a mountain shape) in the subject will be described as an example.

<Pre-processing>
The radiographic image generation unit 101 digitally generates a radioscopic image based on image data obtained from the radiotherapy apparatus. The template registration control unit 102 displays the radioscopic image on the monitor (see FIG. 2). Note that FIG. 2 shows only the marker for convenience, but in addition to the marker, elements constituting the human body such as bones and internal organs are shown in the radiographic image. It is assumed that the X-ray transparency of the marker is lower than that of this human body component.

(Step S1; identification processing of a plurality of points including end points on the marker)
When the radioscopy image is displayed on the monitor, the user uses the input means such as a keyboard or a mouse provided in the radiotherapy device control apparatus 1 to use the width of the marker appearing in the radioscopy image (the x coordinate of the marker). Mark the positions that can represent the distance between the ends of the direction) and the height (the distance between the ends of the marker in the y-coordinate direction). For example, when the shape of the marker is a mountain shape as shown in FIG. 2, a total of three positions of the peak vertex and the two end points are marked. In the present embodiment, as shown in the drawing, an explanation will be given using an example in which the positions of a total of three points, i.e., the peak of a mountain and two end points are marked.

  In FIG. 2, the distance of the x coordinate of the end point is the width of the marker shape, and the distance between the y coordinate of the end point that is further away from the coordinates of the peak of the mountain and the y coordinate of the vertex is the marker shape. It becomes height. In addition, when the shape of a marker shows a straight line, you may make it mark the position of two end points. This is because when the shape of the marker is a straight line, the width and height of the rectangle including the marker can be specified simply by marking the position of the end point. The number of points that the user marks may be any number as long as it is a plurality of points including the end points in the radiographic image of the marker. Then, the marker multiple point specifying unit 103 holds the coordinates of the marked position in the storage unit 113 as the marker temporary coordinate information (in the above example, the coordinates of each of the two end points and the mountain apex). To do. When the user inputs a template registration process start using an input means or the like, a template image generation process included in the template registration process is started.

(Step S2; logarithmic conversion process of luminance value of each pixel of the fluoroscopic image)
A radioscopic image is captured by superimposing markers and elements constituting a human body such as bones and internal organs. The superposition state differs depending on the X-ray irradiation direction (imaging direction) and the marker embedding position of the human body in the radiographic image. For this reason, the case where the brightness | luminance difference of the brightness | luminance of each pixel which shows the marker in a radiographic image, and the brightness | luminance of the pixel of a part other than that occurs in all the pixels showing a marker generate | occur | produces.
In creating the template image, it is necessary to make the marker position clear, and thus it is necessary to make the luminance difference uniform. Therefore, the logarithmic conversion unit 104 performs logarithmic conversion on the luminance of each pixel in the radiographic image using Equation (1).

In the formula (1), i represents the luminance value of the pixel. That is, log ₁₀ (i) in the formula (1) indicates the logarithmic conversion process. Others include an element that compresses 14-bit data to 8 bits and a noise-cut element that cuts low-brightness noise.

FIG. 4 is a first diagram illustrating an outline of the process of creating the luminance difference information for each pixel.
(Step S3; creation process of luminance difference information for each pixel)
Next, the marker edge emphasizing unit 105 performs a process of creating pixel-by-pixel luminance difference information indicating a luminance difference with pixels around the marker appearing in the image, using the radiographic image.
In this process, as shown in FIG. 4, the pixel is located on a straight line in a predetermined direction passing through the certain pixel and the luminance of the certain pixel (target pixel) in the fluoroscopic image, and at a predetermined distance from the certain pixel. The luminance difference of any one of the luminance differences between the two opposing pixels and the luminance difference (in this embodiment, a luminance difference indicating a small value) is calculated for a plurality of different directions passing through a certain pixel. Examples of the predetermined distance include at least half the line width (width direction length) of the marker.

  The marker edge emphasizing unit 105 identifies information on the largest luminance difference among the luminance differences calculated for each direction for the certain pixel. The marker edge emphasizing unit 105 performs a process of calculating information on the luminance difference for all pixels in the radiographic image. In this process, it is not always necessary to perform calculation for all pixels. For example, in the case where the marker position can be estimated in advance, the pixel for which it is determined that the marker is not clearly located in the radiographic image may be omitted. In this case, a process for calculating a plurality of pixels is performed. The same applies to the following processing, but for the sake of convenience, here, a case where calculation is performed for all pixels will be described as an example.

  In other words, the marker edge emphasizing unit 105 performs, for a certain target pixel, the horizontal direction (hereinafter referred to as “a” direction), the upper left and lower right direction (hereinafter referred to as “b” direction), the vertical direction (hereinafter referred to as “c direction”), and the lower left and upper right direction. For each (hereinafter referred to as d direction), a luminance difference with other pixels at a certain distance is calculated, and the smallest luminance difference among the luminance differences is calculated as the intensity of the luminance value of the target pixel. To do. In FIG. 4, the fixed distance to the right and left two pixels on the straight line of a direction is shown about the object pixel (7, 8) of a radiographic image. Here, in this embodiment, the fixed distance is about half of the marker thickness (line width; width length direction length). In FIG. 4, for the convenience of explanation, the length indicating a certain distance appears larger than the line width of the marker.

FIG. 5 is a second diagram showing an outline of the process of creating the luminance difference information for each pixel.
The marker edge emphasizing unit 105 calculates a certain target pixel based on, for example, a luminance difference from a position separated by a certain distance in the a direction. At this time, a smaller value is selected from the luminance difference between the luminance of two opposing pixels (pixel a and pixel b) that are at a fixed distance from the target pixel on a straight line passing through the target pixel and the luminance of the target pixel ( (See FIG. 5A). In FIG. 5, the luminance difference between the target pixel and the pixel a and the luminance difference between the target pixel and the pixel b are smaller than the luminance difference between the pixel a and this luminance difference is adopted.
As shown in FIG. 5A, the luminance transition indicating the marker is in a state where the vicinity of the target pixel position is lower than the surrounding pixels, but the luminance value transition based on the target pixel position is like this. If not, it is highly possible that the target pixel is a pixel that does not represent a marker. For example, the luminance value of one pixel (for example, pixel a) of two pixels (pixel a, pixel b) facing each other at a fixed distance from the target pixel on a straight line passing through the target pixel is substantially the same as the luminance value of the target pixel. It may be lower or lower. In this case, since it is considered that the target pixel does not represent the marker, in order not to detect such a target pixel as the pixel of the marker, two opposing pixels that are at a fixed distance from the target pixel on a straight line passing through the target pixel. A process of selecting a smaller value from the luminance difference between the luminance of one pixel (pixel a, pixel b) and the luminance of the target pixel is performed.
Here, the luminance difference from a position away from a certain distance (about half the thickness of the marker) in a direction for a certain target pixel is the marker when the marker line is reflected in the c direction perpendicular to the a direction. It is effective in detecting

  Note that the luminance difference between the target pixel and a pixel at a certain distance from the target pixel is that the luminance of the target pixel (x, y) is I (x, y) and the constant distance is (dx, dy). The luminance difference I ′ (x, y) from one pixel (pixel a in FIG. 5) is

The luminance difference I ″ (x, y) from the other pixel (pixel b in FIG. 5) is

Can be represented by The smaller one of the luminance differences I ′ (x, y) or the luminance difference I ″ (x, y) is employed.

When calculating the luminance difference in the b direction, the smaller value of the luminance difference between two pixels at a fixed distance from the target pixel is adopted in the pixel direction adjacent to the upper left and lower right of the target pixel. To do.
In the case of calculating the luminance difference in the c direction, a small value of the luminance difference between two pixels at a certain distance from the target pixel is set in the pixel direction (vertical direction) adjacent above and below the target pixel. adopt.
Further, in the case of calculating the luminance difference in the d direction, a small value of the luminance differences between the two pixels at a certain distance from the target pixel is adopted in the pixel direction adjacent to the lower left and upper right of the target pixel. .
Then, the marker edge emphasizing unit 105 calculates the largest luminance difference among the luminance differences calculated for each direction of the a direction, the b direction, the c direction, and the d direction with respect to a certain target pixel. As specified. Then, the marker edge emphasizing unit 105 creates pixel-by-pixel luminance difference information that holds information on the luminance difference specified for all the pixels of the fluoroscopic image, and stores the information in the storage unit 113.

(Step S4; template image range setting process)
FIG. 6 is a diagram showing a template image range.
The template range setting unit 106 reads the marker temporary coordinate information registered in the storage unit 113 by the template registration control unit 102. Then, the template range setting unit 106 specifies a template image range that is a rectangular range including the coordinates of the marker point indicated by the marker temporary coordinate information in the radiographic image. More specifically, the template range setting unit 106 reads the provisional marker coordinate information that holds the coordinates of the peak of the peak marker as shown in FIG. . Of these coordinates, the minimum and maximum X coordinates are x _min (the X coordinate of the left end point) and x _max (the X coordinate of the right end point), respectively, and the minimum and maximum Y coordinates are each y _min (mountain-like). Assuming that the Y coordinate of the vertex) and y _max (Y coordinate of the right end point), the template range setting unit 106, as shown in FIG.
x _min −m ≦ x ≦ x _max + m
y _min −m ≦ y ≦ y _min + m
Is set as the template image range.

Next, the end point position setting unit 107 calculates the relative coordinates of each end point of the marker from the upper left origin of the template image range. The end point of the marker is the end point A (x _a , y _a ) and the end point B (x _b , y _b ) with respect to the coordinates in the respective fluoroscopic images, and the coordinates in the radiographic image of the upper left origin of the template image are (x ₁ , y ₁ ), the relative coordinates of the end point A in the template image are represented by the end point A (x _a −x ₁ , y _a −y ₁ ), and the relative coordinates of the end point B in the template image are the end point B ( _{x b -x 1, y b -y} 1) to become. Then, the end point position setting unit 107 records end point coordinate information holding the coordinates in the template image range of each end point of these markers in the storage unit 113.

(Step S5; mask information generation process)
FIG. 7 is a first diagram showing template mask processing.
FIG. 8 is a second diagram showing template mask processing.
FIG. 9 is a third diagram showing template mask processing.
Next, the template mask setting unit 108 reads the end point coordinate information from the storage unit 113, and specifies pixels on a straight line connecting the coordinates indicated by the end point coordinate information in the template image range (FIG. 7). Then, the identified pixel is identified as not being masked. Further, the template mask setting unit 108 identifies all eight neighboring pixels in the left, right, up, and down diagonal directions of pixels that have already been identified as not subject to masking (see FIG. 8). Then, among the pixels within the template image range, all the pixels that have not been identified as not being masked are identified as mask targets (FIG. 9). Then, the template mask setting unit 108 holds information indicating that all pixels that are not specified as not being masked in the template image range are masked (information indicating that they are not to be processed later, such as flags, for example) Is stored in the storage unit 113 in association with the identification information of the template image.

(Step S6; template image generation process)
The template image generation unit 109 also includes information on each pixel indicated by the template image range (coordinates based on the origin of the radiographic image, coordinates based on the origin of the template image range, luminance difference information for each pixel) and end points. A template image including coordinate information (the coordinates of the two end points of the marker) is generated.
In addition, the radiotherapy device control apparatus 1 digitally generates a fluoroscopic image based on image data obtained from the radiotherapy device a plurality of times at different times (for example, every 0.1 second). Then, using each radiographic image, a template image, mask information, and the like corresponding to each radiographic image are similarly generated by the above-described processes. In the present embodiment, three template images corresponding to each of the three fluoroscopic images are generated. Then, the template registration control unit 102 determines whether or not three template images have been generated. If it is determined that three template images have been generated, the template registration control unit 102 notifies the template image generation unit 109 of template image generation processing. Then, the template registration control unit 102 records the generated three template images in the storage unit 113.

(Step S7; matching condition information registration process)
Further, the user registers matching condition information in the matching process using the template image in the radiation therapy apparatus control apparatus 1.
Here, the template image is used to specify where the affected part is in the human body. The template image is specified (specification of the three template images generated above), the mask information is specified, the template image is Specify the minimum correlation value threshold that indicates whether or not the correlation value of the marker in the newly acquired image is greater than or equal, and cancel the subsequent processing when the correlation value exceeds that value Designation of the value and designation of the upper limit value of the rotation angle when the template image is rotated to perform the matching process are performed, and the information is stored in the radiation therapy apparatus control apparatus 1.

  Then, after the generation of the template image is completed, the radiographic image generation unit 101 digitally generates a new radioscopic image in order to specify the position of the affected part in the human body. The new radioscopic image is hereinafter referred to as an image generated when the affected area is specified for convenience of explanation. When specifying the position of the affected part in the human body, the generated image at the time of specifying the affected part and the template image are compared by pattern matching.

(Step S8; matching target range specifying process)
FIG. 10 is a diagram showing an outline of the pattern matching pre-processing.
Here, as a premise for performing the pattern matching process, the radiotherapy apparatus control apparatus 1 sets further information. The setting of the information is a template corresponding to which matching target range set in which radiographic image according to the range (matching target range) in the image of the marker that will appear in the image generated when the affected area is specified This is a process of setting information specifying whether to perform a matching process using an image. That is, since this process is an information setting process for determining which template image belonging to which matching target range is used to specify the pixel indicating the marker in the image generated when the affected part position is specified, the matching target range specifying process I will call it. Then, this processing is performed by the matching target range specifying unit 110.

  (A1), (a2), and (a3) in FIG. 10 show the positions in the radiographic image of three template images generated based on radiographic images acquired at different timings. As described above, the template images acquired at different timings are images that include the respective ranges of the markers shown at different positions in the radiographic image due to the influence of the patient's respiration and the like.

  In the process of setting information for determining which template image is used to perform the matching process, first, as shown in (b1) of FIG. 10, the matching target range specifying unit 110 first displays each of the radiographic images. The coordinates of the pixels in the template image are read from the template image. Then, the matching target range specifying unit 110 calculates center coordinates in each template image. Here, the three template images are referred to as a template image A, a template image B, and a template image C, respectively. The central pixel of template image A is referred to as pixel a, the central pixel of template image B is referred to as pixel b, and the central pixel of template image B is referred to as pixel c. Any processing for calculating the coordinates of the central pixel of the template image may be used. For example, when the number of pixels serving as the center of the template image is not fixed to one, the coordinates may be calculated using any one of the pixels near the center as the central pixel.

  Then, the matching target range specifying unit 110 calculates the distance from the coordinates of a certain pixel of the radiographic image to the coordinates of the central pixel of each template image, and calculates the template image located at the closest distance from the certain pixel, Calculation is performed for all pixels of the fluoroscopic image. Then, the matching target range specifying unit 110 determines a group of pixels close to the center pixel of a specific template image as one matching target range candidate. That is, in the present embodiment, as shown in FIG. 10B2, the group of pixels close to the central pixel of the template image A is set as the matching target range candidate A and the pixels close to the central pixel of the template image B. Three matching target range candidates are determined with the grouping range as the matching target range candidate B and the grouping range of pixels close to the central pixel of the template image C as the matching target range candidate C.

  Then, the matching target range specifying unit 110 determines the rectangular range that includes all the candidates A for the matching target range as the matching target range A, and determines the rectangular range that includes all the candidates B for the matching target range as the matching target range B. Then, the rectangular range including all the matching target range candidates C is determined as the matching target range C. Thereby, the matching target range specifying unit 110 combines the coordinate information indicated by each matching target range in the image range of the radiographic image and the template image used for specifying the pixel indicating the marker that appears in the matching target range. Is generated for each matching target range, and is stored in the storage unit 113. Specifically, the matching target range information includes, for example, a combination of coordinates indicating the range of the matching target range A (two coordinates on the upper left and lower right) and information on the template image A, and coordinates indicating the range of the matching target range B. Information that holds a combination of (the upper left and two lower coordinates) and the information of the template image B, and a combination of the coordinates indicating the range of the matching target range C (the upper left and the lower right two coordinates) and the information of the template image C It is.

  In the present embodiment, the matching target range specifying unit 110 determines a rectangular range that includes all candidates for the matching target range as the matching target range. However, as shown in FIG. When the target range overlaps with the other matching target range (matching target range B and matching target range C overlap), processing may be performed so as not to generate an overlapping range. For example, each of the overlapping ranges is divided into a range on the side of one matching target range and a range on the side of the other matching target range from the center of the overlapping rectangular range. It may be incorporated into the range and the other matching target range so as not to cause overlapping rectangular ranges.

When the marker that appears in the affected part position-specific generation image appears in the matching target range A, pattern matching is performed with the template image A, and the marker appears in the affected part position-specific generation image. When the marker that appears in the range of the matching target range B is subjected to pattern matching with the template image B, and the marker that appears in the generated image at the time of diseased part position identification is in the range of the matching target range C. When it appears, pattern matching with the template image C is performed.
Alternatively, when a marker that appears in the generated image at the time of specifying the affected area appears in a plurality of matching target ranges, pattern matching is performed using template images corresponding to the plurality of matching target ranges. .

<Marker identification processing>
(Step S11; logarithmic conversion process of luminance value of each pixel of the image generated when the affected area is specified)
FIG. 11 is a diagram illustrating a processing flow of marker specifying processing in the radiotherapy device control apparatus 1.
Next, processing for specifying the position of the marker embedded near the affected part will be described in order.
At this time, the radiotherapy apparatus control apparatus 1 is an image acquired by the radiological fluoroscopic image generation unit 101 based on the image data obtained from the radiotherapy apparatus when the affected part position is specified (the same as the radiographic image). However, the name is changed as described above because it is a new image generated when the affected part position is specified). Then, the logarithmic conversion unit 104 performs logarithmic conversion of the luminance of each pixel in the affected part position specifying generation image using the equation (1), similarly to the above-described processing at the time of template image generation.

(Step S12; creation processing of luminance difference information for each pixel)
When the logarithmic conversion process is completed, the marker edge emphasizing unit 105 uses the generated image at the time of diseased part position identification, similarly to the above-described process at the time of template image generation, to position the target pixel and a position at a certain distance from the target pixel. The luminance difference from the other pixels is calculated for each of the a direction (horizontal direction), b direction (upper left and lower diagonal direction), c direction (up and down direction), and d direction (lower left upper right diagonal direction). The process of calculating the smallest luminance difference among the luminance differences as the intensity from the periphery of the luminance value of the target pixel is performed for all the pixels in the affected part position generation image.
Then, the marker edge emphasizing unit 105 calculates the largest luminance difference among the luminance differences calculated for each direction of the a direction, the b direction, the c direction, and the d direction with respect to a certain target pixel. As specified. Then, the marker edge emphasizing unit 105 creates pixel-by-pixel luminance difference information that holds information on the luminance difference specified for all the pixels of the affected part position-specific generation image, and records the information in the storage unit 113.

  Furthermore, the marker edge emphasizing unit 105 performs a-direction luminance difference information holding information on the luminance difference calculated for the a direction of the target pixel (all pixels in the image) and the b direction of the target pixel (all pixels in the image). B-direction luminance difference information holding information on the luminance difference calculated for, c-direction luminance difference information holding information on the luminance difference calculated for the c direction of the target pixel (all pixels in the image), and the target pixel (image Each of the d-direction luminance difference information holding information on the luminance difference calculated for the d direction of all the pixels in the middle is created and recorded in the storage unit 113.

(Step S13; noise removal processing)
Next, the noise removal unit 111 uses the a-direction luminance difference information generated by the marker edge enhancement unit 105 to identify a pixel whose luminance difference is determined to be less than a predetermined threshold as a pixel indicating a marker candidate, After a pixel determined to have a luminance difference equal to or greater than a predetermined threshold is specified as a pixel other than the marker, noise removal processing is performed.

FIG. 12 is a diagram showing a processing direction based on a certain target pixel used in the noise removal processing.
FIG. 13 is a first diagram showing an outline of the noise removal process.
FIG. 14 is a first diagram showing an outline of the noise removal processing.
FIG. 15 is a second diagram illustrating the outline of the noise removal process.
FIG. 16 is a third diagram illustrating the outline of the noise removal process.
FIG. 17 is a fourth diagram illustrating the outline of the noise removal process.
FIG. 18 is a fifth diagram illustrating the outline of the noise removal process.

The noise removal process will be described with reference to FIGS.
[Process of creating a direction noise removal result data]
First, as shown in FIG. 13A, the noise removing unit 111 reads the a-direction luminance difference information generated by the marker edge emphasizing unit 105, and among the pixels, a pixel having a luminance difference equal to or greater than a threshold (with respect to surroundings). A-direction luminance difference 2 obtained by binarizing the marker appearance intensity for a pixel whose luminance difference is less than the threshold among the pixels to “1” and the marker appearance intensity for a pixel whose luminance difference is less than the threshold among the pixels is “0”. Create value information. In FIG. 13A, hatched pixels are shown as pixels that have been determined to have a marker appearance strength “1”, and pixels that are not hatched have been determined to have been determined to have a marker appearance strength “0”.

Next, the noise removing unit 111 uses the a-direction luminance difference binarization information to perform a graph based on the pixel having the marker appearance strength “1” in the linear direction passing through the pixel having the marker appearance strength “1”. The marker appearance intensities of the two pixels connected in the a direction (the left-right direction of the image) indicated by 12 are all updated to “1”. As a result, as shown in FIG. 13B, the pixel range having the marker appearance intensity “1” expands in the a direction (first expansion process in the a direction using the a direction luminance difference binarization information). .
Further, the noise removing unit 111 is based on the information after the completion of the first expansion process using the a-direction luminance difference binarization information, and is in a straight line direction passing through the pixel having the marker appearance strength “0”. Based on the pixel having the marker appearance strength “0”, the marker appearance strengths of the two pixels connected in the a direction (the left-right direction of the image) shown in FIG. 12 are all updated to “0”. As a result, as shown in FIG. 13C, the range of the pixel having the marker appearance strength “1” contracts in the a direction (first contraction process in the a direction using the a direction luminance difference binarization information). .

Further, the noise removing unit 111 is based on the information after the completion of the first contraction process using the a-direction luminance difference binarization information, and is in a straight line direction passing through the pixel having the marker appearance strength “0”. Based on the pixel having the marker appearance strength “0”, the marker appearance strengths of the two pixels connected in the a direction (the left-right direction of the image) shown in FIG. 12 are all updated to “0”. As a result, as shown in FIG. 14D, the pixel range having the marker appearance intensity “1” is further contracted in the a direction (the second contraction process in the a direction using the a direction luminance difference binarization information). ).
In addition, the noise removing unit 111 is based on the information after the second contraction process, and is in a linear direction passing through the pixel having the marker appearance strength “1”, and is based on the pixel having the marker appearance strength “1”. All marker appearance intensities of two pixels connected in the a direction (the left-right direction of the image) shown in FIG. 12 are updated to “1”. As a result, as shown in FIG. 14E, the pixel range having the marker appearance intensity “1” is expanded in the a direction (second expansion process in the a direction using the a direction luminance difference binarization information). .
The noise removing unit 111 stores a direction marker appearance intensity data using the a direction luminance difference binarization information that holds the result of the second expansion process in the a direction using the a direction luminance difference binarization information. 113.

  As shown in FIG. 12, the two directions between the b direction and the d direction, which are adjacent to each other at an angle of 45 ° with the a direction (horizontal direction), are the e direction (first horizontal / diagonal intermediate direction). ), H direction (second horizontal / diagonal intermediate direction). And the noise removal part 111 performs the 1st expansion process-> 1st contraction process-> 2nd contraction process-> 2nd expansion process performed with respect to a direction also about e direction and h direction.

That is, the expansion process in the e direction (first horizontal / diagonal intermediate direction) using the a-direction luminance difference binarization information is performed in the “e direction (first horizontal / diagonal intermediate direction) expansion of FIG. If the marker appearance intensity value in the a-direction luminance difference binarization information of the pixel indicated by X is “1” as shown in “Description of processing”, each 1 connected to both sides in the a direction on the basis of the X pixel. A process of setting the value of the marker appearance intensity of a pixel indicated by Y, which is indicated by each pixel and each pixel connected in the b direction and one pixel in the outer direction not adjacent to the X pixel, to “1” Is processing for all pixels in the image.
Further, the contraction process in the e direction (first horizontal / diagonal intermediate direction) using the a-direction luminance difference binarization information is performed in the “e direction (first horizontal / diagonal intermediate direction) contraction in FIG. As shown in “Description of Processing”, if the value of the marker appearance intensity in the a-direction luminance difference binarization information of the pixel indicated by M is “0”, each 1 connected to both sides in the a-direction on the basis of the M pixel. A process of setting the value of the marker appearance intensity of a pixel indicated by N, which is indicated by each pixel and each pixel continuous in the b direction and not adjacent to the M pixel, as one pixel, to “0” Is processing for all pixels in the image.

That is, the noise removing unit 111 has already created as shown in FIG. 15F in the expansion process in the e direction (first horizontal / diagonal intermediate direction) using the a-direction luminance difference binarization information. Using the a-direction luminance difference binarization information, the first expansion process in the e-direction (first horizontal / diagonal intermediate direction) as shown in FIG. 15G is performed (a-direction luminance difference binary). E-direction first expansion processing using the conversion information).
Further, the noise removal unit 111 performs the first e as shown in FIG. 15H based on the information after the completion of the first expansion process in the e direction using the a-direction luminance difference binarization information. Shrinkage processing in the direction (first horizontal / diagonal intermediate direction) is performed (first shrinking processing in the e direction using the a-direction luminance difference binarization information).

Further, the noise removing unit 111 performs the second e as shown in FIG. 16I based on the information after the completion of the first contraction process in the e direction using the a-direction luminance difference binarization information. Shrinkage processing in the direction (first horizontal / diagonal intermediate direction) is performed (second shrinking processing in the e direction using the a direction luminance difference binarization information). FIG. 16I shows an example in which the pixel having the marker appearance intensity “1” disappears at this time.
Then, the noise removing unit 111 performs the second e as shown in FIG. 16K based on the information after the end of the second contraction process in the e direction using the a-direction luminance difference binarization information. The expansion process in the direction (first horizontal / diagonal intermediate direction) is performed (the second expansion process in the e direction using the a-direction luminance difference binarization information). FIG. 16I shows an example in which the pixel having the marker appearance strength “1” disappears at the time of (I), and therefore the pixel having the marker appearance strength “1” does not appear even after the expansion process. ing.
The noise removing unit 111 stores e-direction marker appearance intensity data using the a-direction luminance difference binarization information that holds the result of the second expansion process in the e-direction using the a-direction luminance difference binarization information. 113.

Further, the expansion processing in the h direction (second horizontal / diagonal intermediate direction) using the a-direction luminance difference binarization information is omitted, but the a-direction luminance difference binary of the pixel indicated by a certain pixel X is omitted. If the value of the marker appearance intensity in the conversion information is “1”, each pixel connected to both sides in the a direction on the basis of the X pixel, and the pixel is adjacent to the X pixel in the d direction. This is a process of performing the process of setting the value of the marker appearance strength of each pixel connected in the outward direction to “1” for all pixels in the image.
In addition, the h direction (second horizontal / diagonal intermediate direction) contraction processing using the a direction luminance difference binarization information is omitted from illustration, but the a direction luminance difference binary of the pixel indicated by a certain pixel M is omitted. If the value of the marker appearance intensity in the conversion information is “0”, each pixel connected to both sides in the a direction on the basis of the M pixel, and the pixel is adjacent to the M pixel in the d direction. This is a process of performing the process of setting the value of the marker appearance intensity of each pixel connected in the outward direction to “0” for all pixels in the image.

That is, the noise removing unit 111 performs an expansion process in the h direction (second horizontal / diagonal intermediate direction) using the first a direction luminance difference binarization information (the a direction luminance difference binarization information is H-direction first expansion treatment used).
Further, the noise removal unit 111 performs the first h direction (second horizontal / diagonal intermediate) based on the information after the completion of the first expansion process in the h direction using the a-direction luminance difference binarization information. Direction) (first direction shrinking process in the h direction using the a-direction luminance difference binarization information).

Further, the noise removing unit 111 performs the second h direction (second horizontal / diagonal intermediate) based on the information after the completion of the first contraction process in the h direction using the a-direction luminance difference binarization information. (Direction) contraction processing (h direction second contraction processing using a direction luminance difference binarization information).
Then, the noise removing unit 111 performs the second h direction (second horizontal / diagonal intermediate) based on the information after the completion of the second shrinkage process in the h direction using the a-direction luminance difference binarization information. Direction) (second expansion process in the h direction using the a direction luminance difference binarization information).
The noise removing unit 111 stores the h-direction marker appearance intensity data using the a-direction luminance difference binarization information that holds the result of the second expansion process in the h direction using the a-direction luminance difference binarization information. 113.

Through the above processing, the noise removing unit 111 creates a-direction marker appearance strength data, e-direction marker appearance strength data, and h-direction marker appearance strength data using the a-direction luminance difference binarization information. 17A shows a-direction luminance difference binarization information, FIG. 17B shows a-direction marker appearance strength data, FIG. 17C shows e-direction marker appearance strength data, and FIG. 17D shows h-direction marker appearance strength data. ing.
The noise removing unit 111 then selects any one of the pixels indicating the binarized luminance in the created a-direction marker appearance strength data, e-direction marker appearance strength data, and h-direction marker appearance strength data. A pixel indicating a value of at least “1” in the appearance intensity data is adopted as the marker appearance intensity “1”. Then, the noise removing unit 111 sets the pixel appearance strength indicated by the adopted marker appearance strength data as “1” as the marker candidate in the generated image at the time of the affected part position identification, and information on the pixel of the marker candidate. A direction noise elimination result data holding In FIG. 17, there is a pixel having a large value only in the a-direction marker enhancement data shown in (B), and therefore, the marker candidate pixel shown in the a-direction noise removal result data is the a shown in FIG. This is the same as the pixel showing a large value only in the direction marker enhancement data.

  In addition, the noise removing unit 111 creates b-direction noise removal result data, c-direction noise removal result data, and d-direction noise removal result data by the same method as the creation processing of the a-direction noise removal result data.

[Process of creating b-direction noise removal result data]
Although illustration is omitted, in the b-direction noise removal result data creation process, the noise removal unit 111 reads the b-direction luminance difference information generated by the marker edge enhancement unit 105, and the luminance difference of each pixel is equal to or greater than a threshold value. B-direction luminance difference binarization information is generated by binarizing the marker appearance intensity of the pixel of “1” to “1” and the marker appearance intensity of the pixel having a luminance difference of less than the threshold among “0”.

  Then, the noise removing unit 111 performs the first expansion process in the b direction using the b direction luminance difference binarization information, the first contraction process in the b direction using the b direction luminance difference binarization information, and the b direction luminance difference. A second contraction process in the b direction using the binarization information and a second expansion process in the b direction using the b direction luminance difference binarization information are performed. The noise removing unit 111 stores the b-direction marker appearance intensity data using the b-direction luminance difference binarization information that holds the result of the second expansion process in the b direction using the b-direction luminance difference binarization information. 113.

  Also, the two directions between the a direction and the c direction indicating the directions adjacent to each other at an angle of 45 ° with the b direction (upper left and lower right direction) are the e direction (first horizontal / diagonal intermediate direction), f The direction (first up / down / diagonal middle direction) is called. And the noise removal part 111 performs the process similar to 1st expansion | swelling process-> 1st contraction process-> 2nd contraction process-> 2nd expansion process performed with respect to b direction also about e direction and f direction. .

  That is, the noise removing unit 111 performs the first expansion process in the e direction using the b direction luminance difference binarization information, the first contraction process in the e direction using the b direction luminance difference binarization information, and the b direction luminance difference. A second contraction process in the e direction using the binarization information and a second expansion process in the e direction using the b direction luminance difference binarization information are performed. And the noise removal part 111 memorize | stores e direction marker appearance intensity data using b direction luminance difference binarization information holding the result of the 2nd expansion process of e direction using b direction luminance difference binarization information Recorded in the unit 113.

Further, the noise removing unit 111 performs a first expansion process in the f direction using the b-direction luminance difference binarization information, a first contraction process in the f direction using the b-direction luminance difference binarization information, and a b-direction luminance difference. A second contraction process in the f direction using the binarization information and a second expansion process in the f direction using the b direction luminance difference binarization information are performed.
Then, the noise removing unit 111 stores the f-direction marker appearance intensity data using the b-direction luminance difference binarization information that holds the result of the second expansion process in the f direction using the b-direction luminance difference binarization information. 113.

  Further, the noise removing unit 111 performs b-direction marker appearance intensity data using the generated b-direction luminance difference binarization information, e-direction marker appearance intensity data using the b-direction luminance difference binarization information, and b-direction luminance difference. Among the pixels indicating binarized luminance in the f-direction marker appearance intensity data using the binarized information, a pixel indicating a large value in any marker appearance intensity data is set as the marker appearance intensity “1”. adopt. Then, the noise removing unit 111 sets the pixel appearance strength indicated by the adopted marker appearance strength data as “1” as the marker candidate in the generated image at the time of the affected part position identification, and information on the pixel of the marker candidate. B direction noise removal result data is stored.

[C-direction noise removal result data creation process]
Although not shown, in the c direction noise removal result data creation process, the noise removal unit 111 reads the c direction luminance difference information generated by the marker edge enhancement unit 105, and the luminance difference of each pixel is greater than or equal to a threshold value. The c direction luminance difference binarization information is generated by binarizing the marker appearance intensity for the pixel of “1” to “1” and the marker appearance intensity for the pixel having a luminance difference less than the threshold among the pixels as “0”.

Then, the noise removing unit 111 performs the first expansion process in the c direction using the c direction luminance difference binarization information, the first contraction process in the c direction using the c direction luminance difference binarization information, and the c direction luminance difference. A c-direction second contraction process using binarization information and a c-direction second expansion process using c-direction luminance difference binarization information are performed.
The noise removing unit 111 also stores c-direction marker appearance intensity data using the c-direction luminance difference binarization information that holds the result of the second expansion process in the c direction using the c-direction luminance difference binarization information. 113.

  In addition, the noise removing unit 111 has two directions in the middle between the b direction and the d direction indicating the directions adjacent to each other at an angle of 45 ° with the c direction (up and down direction). Direction) and g direction (second up / down / diagonal intermediate direction). And the noise removal part 111 performs the process similar to the 1st expansion process-> 1st contraction process-> 2nd contraction process-> 2nd expansion process performed with respect to the c direction also about the f direction and the g direction. .

That is, the noise removing unit 111 performs the first expansion process in the f direction using the c direction luminance difference binarization information, the first contraction process in the f direction using the c direction luminance difference binarization information, and the c direction luminance difference. A second contraction process in the f direction using the binarization information and a second expansion process in the f direction using the c direction luminance difference binarization information are performed.
And the noise removal part 111 memorize | stores f direction marker appearance intensity data using the c direction brightness | luminance difference binarization information holding the result of the 2nd expansion process of f direction using c direction brightness | luminance difference binarization information Recorded in the unit 113.

Further, the noise removing unit 111 performs a first expansion process in the g direction using the c-direction luminance difference binarization information, a first contraction process in the g direction using the c-direction luminance difference binarization information, and a c-direction luminance difference. A second contraction process in the g direction using the binarization information and a second expansion process in the g direction using the c direction luminance difference binarization information are performed.
The noise removing unit 111 stores the g-direction marker appearance intensity data using the c-direction luminance difference binarization information that holds the result of the second expansion process in the g direction using the c-direction luminance difference binarization information. 113.

  Next, the noise removing unit 111 performs c-direction marker appearance intensity data using the generated c-direction luminance difference binarization information, f-direction marker appearance intensity data using the c-direction luminance difference binarization information, and c-direction luminance difference. Among the pixels indicating the binarized luminance in the g-direction marker appearance intensity data using the binarized information, a pixel indicating a large value in any of the marker appearance intensity data is set as the marker appearance intensity “1”. adopt. Then, the noise removing unit 111 sets the pixel appearance strength indicated by the adopted marker appearance strength data as “1” as the marker candidate in the generated image at the time of the affected part position identification, and information on the pixel of the marker candidate. The c direction noise removal result data that holds is generated.

[D-direction noise removal result data creation process]
Although not shown, in the d-direction noise removal result data creation process, the noise removal unit 111 reads the d-direction luminance difference information generated by the marker edge enhancement unit 105, and the luminance difference among the pixels is determined. The d-direction luminance difference binarization information is generated by binarizing the marker appearance intensity for a pixel equal to or higher than the threshold value to “1” and the marker appearance intensity for a pixel having a luminance difference less than the threshold value among the pixels as “0”. .

Then, the noise removing unit 111 performs a first expansion process in the d direction using the d-direction luminance difference binarization information, a first contraction process in the d direction using the d-direction luminance difference binarization information, and a d-direction luminance difference. A second contraction process in the d direction using the binarization information and a second expansion process in the d direction using the d direction luminance difference binarization information are performed.
The noise removing unit 111 stores d-direction marker appearance intensity data using the d-direction luminance difference binarization information that holds the result of the second dilation processing in the d direction using the d-direction luminance difference binarization information. 113.

  Further, the noise removing unit 111 has two directions in the middle of the c direction and the a direction indicating the directions adjacent to each other at an angle of 45 ° with the d direction (lower left upper right direction) (the second vertical and diagonal directions). Intermediate direction) and h direction (second horizontal / diagonal intermediate direction). The noise removing unit 111 performs the first expansion processing → first contraction processing → second contraction processing using the d-direction luminance difference binarization information performed in the d direction in the g direction and the h direction. → The same process as the second expansion process is performed.

That is, the noise removing unit 111 performs the first expansion process in the g direction using the d-direction luminance difference binarization information, the first contraction process in the g direction using the d-direction luminance difference binarization information, and the d-direction luminance difference. A second contraction process in the g direction using the binarization information and a second expansion process in the g direction using the d direction luminance difference binarization information are performed.
The noise removal unit 111 stores g-direction marker appearance intensity data using the d-direction luminance difference binarization information that holds the result of the second expansion process in the g direction using the d-direction luminance difference binarization information. To record.

Further, the noise removing unit 111 performs a first expansion process in the h direction using the d-direction luminance difference binarization information, a first contraction process in the h direction using the d-direction luminance difference binarization information, and a d-direction luminance difference. A second contraction process in the h direction using the binarization information and a second expansion process in the h direction using the d direction luminance difference binarization information are performed.
The noise removing unit 111 stores, in the storage unit 113, h-direction marker appearance intensity data using the d-direction luminance difference binarization information that holds the result of the second dilation processing in the h direction using the d-direction luminance difference binarization information. Record.

  Then, the noise removing unit 111 uses the created d-direction luminance difference binarization information, d-direction marker appearance intensity data, g-direction marker appearance intensity data using the d-direction luminance difference binarization information, and d-direction luminance difference. Among the pixels indicating the binarized luminance in the h direction marker appearance intensity data using the binarized information, a pixel indicating a large value in any marker appearance intensity data is set as the marker appearance intensity “1”. adopt. Then, the noise removing unit 111 sets the pixel appearance strength indicated by the adopted marker appearance strength data as “1” as the marker candidate in the generated image at the time of the affected part position identification, and information on the pixel of the marker candidate. D-direction noise removal result data is stored.

Through the above processing, the noise removal unit 111 creates a direction noise removal result data, b direction noise removal result data, c direction noise removal result data, and d direction noise removal result data. 18A shows the a-direction noise removal result data, FIG. 18B shows the b-direction noise removal result data, and FIG. 18C shows the c-direction noise removal result data. FIG. 18D shows d-direction noise removal result data.
Then, the noise removal unit 111 identifies pixels that are identified as marker candidates in the a-direction noise removal result data, pixels that are identified as marker candidates in the b-direction noise removal result data, and marker candidates in the c-direction noise removal result data. All the pixels that are identified as marker candidates in the d-direction noise removal result data are identified as pixels that become a marker candidate group, and what information is the pixel that becomes the marker candidate group in the generated image when the affected area is identified Marker candidate group data holding information (the coordinates of pixels included in the marker candidate group in the image) indicating whether or not there is generated.

(Step S14; marker candidate specifying process)
FIG. 19 is a diagram showing an outline of the marker candidate specifying process.
Next, the marker specifying unit 112 specifies marker candidates from the marker candidate group data. For details of the processing, as shown in FIG. 19, first, the marker specifying unit 112 inputs marker candidate group data (FIG. 19A), and the pixels specified as the marker candidate group in the data are adjacent to each other. Labeling (classification) is performed for each of a plurality of regions specified by a group of pixels to be processed. In FIG. 19B, among the two regions P and Q indicating the group of adjacent pixels from the marker candidate group data, labeling “1” is applied to the pixel in region P, and labeling “2” is applied to the pixel in region Q. "Is given to each.

  Then, based on the number of pixels labeled “1” and the number of pixels labeled “2”, the marker specifying unit 112 has a region range in which the region P and the region Q can be matched as markers. It is judged whether it is more than the threshold value. And when it is more than a threshold value, the pixel contained in the area | region is specified as a marker pixel candidate. FIG. 19 shows an example in which the region P is selected assuming that only the region P is a range equal to or larger than the threshold value of the region range suitable as a marker. And the marker specific | specification part 112 produces | generates the marker candidate data which hold | maintain the pixel information specified as the marker candidate in an affected part position specific time generation image (refer FIG.19 (C)), and records it on the memory | storage part 113. FIG.

In addition, when the radiotherapy apparatus control apparatus 1 does not perform the process of specifying the marker by comparing the image generated when the affected area is specified and the template image registered in advance, the pixel candidate pixel obtained at this time is selected. You may make it produce | generate the marker image which pinpoints as a marker and hold | maintains the information of the pixel of the said marker. According to such a process, since the normalized correlation process is not used in generating the marker image, the marker detection speed can be increased.
Further, in the above-described processing, logarithmic conversion of the luminance value of each pixel is performed, and thereby processing for uniformizing the luminance difference between the luminance of each pixel indicating the marker and the luminance of the pixels in other locations. Since this is done, a mechanism is adopted in which a marker can be easily detected using a threshold value of a constant luminance value, thereby increasing the marker detection processing speed.

FIG. 20 is a diagram showing an outline of pattern matching pre-processing.
(Step S15; search range determination process)
The marker specifying unit 112 is a marker candidate indicated by the minimum value xmin and the maximum value xmax of the x coordinate of the pixel specified as the marker candidate in the marker candidate image obtained from the information of the marker candidate pixel indicated by the created marker candidate data. The width W1 is calculated. The marker specifying unit 112 calculates the height H1 of the marker candidate indicated by the minimum value ymin and the maximum value ymax of the y coordinate of the pixel specified as the marker candidate. Then, an area having a width H1 and a height W1 that encompasses the marker candidate pixels is provisionally determined as a search range (see FIG. 20B).

  In addition, the marker specifying unit 112 reads the template image information recorded in advance by the processing of the template registration control unit 102 from the storage unit 113, and coordinates in the radiographic image of the template image stored in the template image information. Based on the information, the width W2 and the height H2 of the template image are detected. Then, the marker specifying unit 112 compares the marker candidate width W1 with the template image width W2. Further, the marker specifying unit 112 compares the height H1 of the marker candidate with the height H2 of the template image. When W1 <W2, the width of the search range is widened (see FIG. 20B). When W1 ≧ W2, the search range does not have to be expanded. Alternatively, the search range may be expanded even when W1 ≧ W2. Whether the search range is expanded or not depends on how much the specific accuracy of the marker is increased.

As the processing for expanding the width of the search range, for example, if the minimum value of the x coordinate in the width direction of the search range is x _min and the maximum value is x _max , the minimum value is x _min −W2 and the maximum value is x _max + W2. Spread to become. By such processing, the width of the search range (distance between both ends of the search range in the x-coordinate direction) widens, for example, approximately twice to three times the width W1 of the search range determined provisionally. When the width of the marker is narrower than the width of the template image, it means that the marker candidate obtained from the image generated when the affected area is specified is detected to be shorter than actual. Therefore, by expanding the search range in this process, it is possible to improve the accuracy of the process of identifying pixels that are considered marker candidates. In addition, when only a part of an actual marker end is specified as a marker candidate, it cannot be specified as a marker candidate on the right or left side of the marker candidate at a distance that is substantially equal to the actual width of the marker. Pixels may be hidden. Accordingly, the distance corresponding to the width of the template image that is considered to be substantially the same as the width of the marker is widened on the right side and the left side of the search range tentatively determined when the search range is expanded.

Similarly, when H1 <H2, the height of the search range is widened (see FIG. 20B). In the case of H1 ≧ H2, the search range does not necessarily have to be expanded. Whether the search range is expanded or not depends on how much the specific accuracy of the marker is increased.
As a process of expanding the height of the search range, for example, if the minimum value of the y coordinate in the height direction of the search range is ymin and the maximum value is ymax, the minimum value is ymin−H2, and the maximum value is ymax + H2. Spread to. By such a process, the height of the search range is expanded to about 2 to 3 times the height H1 of the search range that is provisionally determined. If the height of the marker is narrower than the height of the template image, it means that the marker candidate obtained from the image generated when the affected area is specified is detected to be shorter than the actual position. Therefore, by expanding the search range in this process, it is possible to improve the accuracy of the process of identifying pixels that are considered marker candidates. In addition, when only a part of the actual marker is specified as the marker candidate, it cannot be specified as the marker candidate at a height approximately equal to the actual height of the marker above or below the marker candidate. Pixels may be hidden. Accordingly, the distance corresponding to the height of the template image that is considered to be substantially the same as the height of the marker is widened on the upper and lower sides of the search range tentatively determined when the height of the search range is expanded. The search range is determined by the above processing.

(Step S16; use template image determination process)
Then, the marker specifying unit 112 reads the matching target range information generated by the matching target range specifying unit 110 in advance from the storage unit 113. And the marker specific | specification part 112 is based on the information of the coordinate which shows the range of the matching object range A, B, and C stored in matching object range information, and the area | region of the matching object range A, and the matching object range B A region and a region in the matching target range C are detected. Then, the marker specifying unit 112 detects the position of the generated image at the time of specifying the affected area in the search range determined from the information on the coordinates indicated by the search range, and the position in the generated image at the time of specifying the affected area in the search range is the matching target. It is determined which region of the ranges A, B, and C is included (see FIG. 20C).

Then, when the search range is included in the region of the matching target range A, the marker specifying unit 112 displays information on the template image A associated with the coordinates indicating the range of the matching target range A in the matching target range information. It is determined that the pattern matching process is performed using the template image A.
Further, when the search range is included in the region of the matching target range B, the marker specifying unit 112 reads the information of the template image B linked to the coordinates indicating the range of the matching target range B in the matching target range information. Thus, it is determined that the pattern matching process is performed using the template image B.
In addition, when the search range is included in the region of the matching target range C, the marker specifying unit 112 reads the information of the template image C linked to the coordinates indicating the range of the matching target range C in the matching target range information. Thus, it is determined that the pattern matching process is performed using the template image C.

Alternatively, when the search range is included in both the matching target ranges A and B, the marker specifying unit 112 is a template image associated with coordinates indicating the range of the matching target range A in the matching target range information. It is determined that the pattern matching process is performed using the template image A and the template image B by reading the information of A and the information of the template image B linked to the coordinates indicating the range of the matching target range B. .
Alternatively, when the search range is included in both of the matching target ranges B and C, the marker specifying unit 112 is a template image associated with coordinates indicating the range of the matching target range B in the matching target range information. It is determined that pattern matching processing is performed using the template image B and the template image C by reading the information of B and the information of the template image C linked to the coordinates indicating the range of the matching target range C. .
Here, as shown in FIG. 20D, when the search range extends over a plurality of matching target ranges, a part of the search range included in the matching target range is specified by the matching target range. Pattern matching processing is performed using the template image to be processed.

(Step S17; pattern matching process)
FIG. 21 is a first diagram showing an outline of the pattern matching process.
Now, it is assumed that the search range is included in the matching target range A. In this case, next, the marker specifying unit 112 stores the template image A associated with the coordinates indicating the range of the matching target range A determined based on the position of the search range in the affected part position specifying time generation image. Read from 113. In addition, the marker specifying unit 112 reads the upper limit value of the rotation angle when the matching process is performed by rotating the template image, which is recorded in the storage unit 113 in advance. Then, the marker specifying unit 112 generates two rotated template images obtained by rotating the template image clockwise and counterclockwise using the upper limit value of the rotation angle. For example, as shown in FIG. 21B, if the upper limit value of the rotation angle is 5 °, a rotated template image rotated by 5 ° and a rotated template image rotated by −5 ° are generated. It should be noted that more rotation template images may be created in increments of a predetermined angle up to the upper limit value of the rotation angle.

  Then, the marker specifying unit 112 reads the pixel-by-pixel luminance difference information stored in the information of the template image A and the pixel-by-pixel luminance difference information in the search range of the image generated when the affected part position is specified. Further, the marker specifying unit 112 reads the mask information of the template image A from the storage unit 113. Then, the marker specifying unit 112 uses the pixel-by-pixel luminance difference information of the template image A and the pixel-by-pixel luminance difference information of the search range to superimpose the origin of the search range and the origin of the template image A. In the position range within the search range corresponding to the template image range, the correlation between the luminance difference of each pixel in the position range of the search range and the luminance difference of each pixel of the template image A corresponding to each pixel is calculated. (Pattern matching process). At this time, using the mask information of the template image A, the marker specifying unit 112 does not calculate the correlation between the pixels that are masked in the mask information and the corresponding pixels in the search range. Thereby, the pattern matching processing amount can be reduced.

  The marker specifying unit 112 reads a minimum correlation value threshold value registered in the storage unit 113 as matching condition information in advance and a correlation value at which the processing is terminated (hereinafter referred to as a processing termination correlation value) and correlates with the template image. And whether the current correlation value calculated for the position range within the search range (in the pattern matching process, the origin of the template image) is greater than or equal to the minimum correlation value threshold, and greater than or equal to the processing abort correlation value Judgment is made. If the current correlation value is equal to or greater than the minimum correlation value threshold, the current position range within the search range for which the correlation with the template image is calculated is specified as a marker detection range candidate, and the marker detection is performed. Coordinates indicating the position range within the search range as a range candidate and the correlation value calculated for the position range are associated and recorded in the storage unit 113.

  Then, when the current correlation value is less than the minimum correlation value threshold, the marker specifying unit 112 records, in the storage unit 113, coordinates indicating the position range within the search range that is the specified marker detection range candidate. After that, the position range of the same range as the template image is shifted by one pixel within the search range. Then, the marker specifying unit 112 calculates again the correlation between the luminance difference of each pixel in the position range of the search range and the luminance difference of each pixel of the template image A corresponding to each pixel (pattern matching process). ). And the marker specific | specification part 112 performs a pattern matching process, shifting a position range for every pixel, as shown in FIG.21 (C). In the present embodiment, the method of shifting the position range by one pixel is, for example, shifting in the horizontal direction (right direction). When the position range extends beyond the search range, the position range is shifted by one pixel in the vertical direction (downward). ). Then, after shifting one pixel in the vertical direction, the pixel is shifted in the opposite horizontal direction (left direction). When the position range extends beyond the search range, the pixel is shifted one pixel in the vertical direction (downward) before protruding. Such a process is repeated.

Further, in the process of shifting the position range within the search range, the marker specifying unit 112 matches the coordinate of the diagonal corner with the origin of the template image and the coordinate of the diagonal corner with the origin of the search range. If the position range can no longer be shifted within the search range, the pattern matching process is terminated.
Further, as shown in FIG. 21D, the marker specifying unit 112 performs a similar pattern matching process using the rotation template image, and determines a marker detection range candidate. And the marker specific | specification part 112 determines the position range which shows the highest value among the correlation values about the position range in the search range specified as a marker detection range candidate as a marker detection range. When the current correlation value is greater than or equal to the aborted correlation value, the marker specifying unit 112 specifies the position range as the marker detection range because the correlation value calculated for the current position range is the highest value.

The marker specifying unit 112 then sets the coordinates of the marker detection range in the specified search range and the coordinates of the end points of the markers that appear in the template image A (the end point information stored in the end point coordinate information in the information of the template image A). From the coordinates, the coordinates of the marker in the generated image at the time of specifying the affected area are calculated. For example, the coordinates in the search range corresponding to the origin of the specified marker detection range are (x _t , y _t ), and the coordinates of the two end points of the marker stored in the template image A are (x _e1 , y _e1 ). , (X _e2 , y _e2 ), the coordinates (x _fa , y _fa ) and (x _fb , y _fb ) of the two end points of the marker appearing in the image generated when the affected area is specified are respectively
(X _fa , y _fa ) = (x _t + x _e1 , y _t + y _e1 )
(X _fb , y _fb ) = (x _t + x _e2 , y _t + y _e2 )
And calculate. Thereby, the radiotherapy apparatus control apparatus 1 determines the affected part in the vicinity using the two end points of the marker.

FIG. 22 is a diagram showing an outline of calculating the end point of the marker using the rotation template image.
When calculating the coordinates of the pixel at the end point of the marker that appears in the generated image at the time of diseased part position identification based on the marker detection range determined by the pattern matching process using the rotation template image, as shown in FIG. Processing to rotate the end points of the markers of the template image is required. At this time, the coordinates of the pixel in the search range corresponding to the coordinates of the pixel at the origin of the template image are (x _t , y _t ), the coordinates of the end point of the marker appearing in the template image are (x _e , y _e ), When the rotation angle of the rotation template image is θ, the coordinates (x _f , y _f ) of the end points of the markers appearing in the generated image at the time of diseased part position specification are
(X _f , y _f )
= (X _t + x _e cos θ−y _e sin θ, y _t + x _e sin θ−y _e cos θ)
Calculated by

FIG. 23 is a second diagram showing an outline of the pattern matching process.
As shown in FIG. 23B, when the search range extends over a plurality of matching target ranges, a part of the search range included in the matching target range is identified by the matching target range. The pattern matching process is performed using an image. The pattern matching process in this case will be described.
FIG. 23B shows an example in which the upper portion of the search range is in the region of the matching target range A and the lower portion of the search range is in the region of the matching target range B. Here, the upper part of the search range is called a search range A, and the lower part of the search range is called a search range B.

  In this case, the marker specifying unit 112 uses the luminance difference information for each pixel of the template image A and the luminance difference information for each pixel of the search range A to determine the origin of the search range A and the origin of the template image A. In the position range within the search range A corresponding to the range of the template image A when superimposed, the luminance difference of each pixel in the position range of the search range A and each pixel of the template image A corresponding to each of these pixels The correlation with the luminance difference is calculated. When the search range A is wider than the template image A, the pattern matching process is performed by shifting the position range for each pixel in the same manner as in the above method.

  Similarly, the marker specifying unit 112 superimposes the origin of the search range B and the origin of the template image B using the luminance difference information for each pixel of the template image B and the luminance difference information for each pixel of the search range B. In the position range within the search range B corresponding to the range of the template image B when combined, the luminance difference of each pixel in the position range of the search range B and each pixel of the template image B corresponding to each of these pixels The correlation with the brightness difference is calculated. When the search range B is wider than the template image B, the pattern matching process is performed by shifting the position range for each pixel in the same manner as the above method.

When the correlation value calculated using the pixel-by-pixel luminance difference information of the template image A and the pixel-by-pixel luminance difference information of the search range A is the highest, the marker specifying unit 112 has a marker detection range within the search range. And the coordinates of the end point of the marker appearing in the template image A are calculated in the image at the time of specifying the affected part position.
On the other hand, when the correlation value calculated using the pixel-by-pixel luminance difference information of the template image B and the pixel-by-pixel luminance difference information of the search range B is the highest, the marker specifying unit 112 has a marker detection range within the search range. And the coordinates of the end point of the marker appearing in the template image B, the coordinates in the generated image at the time of specifying the affected part position of the marker are calculated.

  In this way, template images of the markers appearing at different positions in the radiographic image are prepared in advance, and the template image closest to the marker position in the newly generated image at the time of radiation irradiation treatment is used. The pattern matching process is performed. Thereby, the position detection in the image of a marker with high precision can be performed.

  In the above-described processing, a plurality of template images are generated using radiographic images obtained by irradiating radiation from one direction, and radiation is also applied from the same direction for images during radiation irradiation treatment. Pattern matching processing with the template image is performed using the image obtained by doing so. However, in order to detect the marker position even when the radiation irradiation angle on the patient is changed, a plurality of template images are obtained for each irradiation direction using radiographic images obtained by irradiating radiation from a plurality of directions. You may make it produce | generate. In that case, pattern matching processing with a plurality of template images in each direction is performed using an image obtained by irradiating radiation from a plurality of directions with respect to an image at the time of radiation irradiation treatment.

  In addition, the above-mentioned radiotherapy apparatus control apparatus has a computer system inside. Each process described above is stored in a computer-readable recording medium in the form of a program, and the above process is performed by the computer reading and executing the program. Here, the computer-readable recording medium means a magnetic disk, a magneto-optical disk, a CD-ROM, a DVD-ROM, a semiconductor memory, or the like. Alternatively, the computer program may be distributed to the computer via a communication line, and the computer that has received the distribution may execute the program.

  The program may be for realizing a part of the functions described above. Furthermore, what can implement | achieve the function mentioned above in combination with the program already recorded on the computer system, and what is called a difference file (difference program) may be sufficient.

DESCRIPTION OF SYMBOLS 101 ... Radioscopy image generation part 102 ... Template registration control part 103 ... Multiple point specification part 104 on a marker ... Logarithmic conversion part 105 ... Marker edge emphasis part 106 ... Template range setting part 107: end point position setting unit 108 ... template mask setting unit 109 ... template image generating unit 110 ... matching target range specifying unit 111 ... noise removing unit 112 ... marker specifying unit 113 ...・ Storage unit

Claims (12)

A radiotherapy apparatus controller for detecting the marker in a radioscopic image digitally generated by irradiating a subject in which a marker is embedded,
In the radiographic image, a luminance difference between a luminance of a certain pixel and luminances of two opposing pixels located on a straight line in a predetermined direction passing through the certain pixel and positioned at a predetermined distance from the certain pixel. The luminance difference of any one of them is calculated for a plurality of different directions passing through the certain pixel, the information on the largest luminance difference among them is specified for the certain pixel, and the information on the luminance difference is determined as the radiographic image. A marker edge emphasis unit that creates pixel-by-pixel luminance difference information held for a plurality of pixels in the inside,
In the pixel-by-pixel luminance difference information, a process of specifying pixels within a predetermined range from the pixel that is continuous in the predetermined direction to pixels having a luminance difference equal to or greater than a predetermined threshold as marker appearing pixels; A process of identifying a pixel in a predetermined range from the pixel connected in the predetermined direction to a pixel having a luminance difference less than a predetermined threshold as a marker non-appearing pixel is repeated separately for each of the plurality of different directions, respectively. A noise removing unit that removes pixels determined to be noise from among pixels in which the luminance difference is equal to or greater than a predetermined threshold in the pixel-by-pixel luminance difference information, and identifies pixels that are marker candidates;
A marker specifying unit that classifies pixels that are candidates for the marker into a plurality of regions based on a group of adjacent pixels, and specifies pixels that belong to a region having a region range equal to or greater than a threshold value as a pixel indicating a marker;
A radiotherapy apparatus control apparatus comprising: A logarithmic conversion unit that logarithmically converts a luminance value for each pixel in the radiographic image,
The marker edge emphasis unit is located on a straight line in a predetermined direction passing through the certain pixel and the luminance of a certain pixel in the radiographic image after the logarithmic transformation, and is opposed to the certain pixel at a predetermined distance. The luminance difference of any one of the luminance differences between the two pixels is calculated for a plurality of different directions passing through the certain pixel, and the information on the largest luminance difference among them is specified for the certain pixel. 2. The radiotherapy apparatus control apparatus according to claim 1, wherein the brightness difference information for each pixel is generated in which the information on the brightness difference is held for a plurality of pixels in the fluoroscopic image. A plurality of on-marker specifying parts for specifying a plurality of points including end points in the fluoroscopic image of the marker appearing in the fluoroscopic image;
A template range setting unit for setting a template image range including a plurality of points of the marker;
A template image generating unit that generates template image information including information on the end points, information on the template image range, and information on a position of the template image range in the radiographic image;
With
The marker specifying unit determines a search range including a region in which the classified region range is equal to or greater than a threshold, and a luminance difference based on the pixel-by-pixel luminance difference information of a pixel in the search range and a pixel indicating the template image And the origin of the position range in the search range when it is determined that the correlation value is equal to or greater than the threshold, and the coordinates in the template image of the end point of the marker indicated by the template image The end point coordinate of the marker that appears in the radioscopic image is specified using the radiotherapy apparatus control apparatus according to claim 1 or 2. The marker multiple point specifying unit specifies a plurality of points including end points in the radiographic image of the marker appearing in the radiographic image for each of the radiographic images generated at different times,
The template range setting unit sets a template image range including a plurality of points of the marker for each of the radiographic images generated at different times,
The template image generation unit generates template image information including the end point information, the template image range information, and the position information of the template image range in the radiographic image at different times. Generating a plurality of each radiographic image,
A matching target range specifying unit that specifies a matching target range indicating a group of pixels close to the template image for each of the plurality of template images;
The marker specifying unit is
Determining a search range including a region where the range of the classified region is equal to or greater than a threshold;
Determining the matching target range including the search range,
In the search range, a correlation value calculated using a luminance difference based on the pixel-by-pixel luminance difference information of a pixel in the same position range as the range of the template image belonging to the determined matching target range and a pixel indicating the template image. Calculate
Using the origin of the position range in the search range when it is determined that the correlation value is greater than or equal to the threshold, and the coordinates in the template image of the end point of the marker indicated by the template image belonging to the determined matching target range, The radiotherapy apparatus control apparatus according to claim 3, wherein an end point coordinate of a marker that appears in the radioscopic image is specified. When there are a plurality of matching target ranges including the search range, the marker specifying unit determines a plurality of different matching target ranges including the search range,
For a partial range of the search range belonging to one matching target range among the determined plurality of matching target ranges, pixels of the partial range and pixels indicating a template image belonging to the one matching target range Calculating a correlation value calculated using a luminance difference based on the luminance difference information for each pixel;
Using the origin of the position range in the search range used for determining that the correlation value is equal to or greater than the threshold and the coordinates in the template image of the end point of the marker indicated by the template image belonging to the one matching target range, The radiotherapy apparatus control apparatus according to claim 4, wherein end point coordinates of a marker appearing in the radioscopic image are specified. In the radiographic image, pixels other than the pixels corresponding to the line connecting the plurality of points are determined to be excluded in the subsequent processing, and information indicating the object exclusion is set to a plurality of pixels corresponding to the line connecting the plurality of points. A template mask setting unit that generates mask information held for each of the corresponding pixels on the connecting line, for each of the fluoroscopic images generated at different times, and
The said marker specific | specification part calculates the said correlation value, without using the information of the brightness | luminance difference of the pixel hold | maintained with the said mask information. The Claim 1 characterized by the above-mentioned. Radiotherapy device control device. A processing method for detecting the marker in a radioscopic fluoroscopic image generated by irradiating a subject with an embedded marker with radiation,
In the radiographic image, a luminance difference between a luminance of a certain pixel and luminances of two opposing pixels located on a straight line in a predetermined direction passing through the certain pixel and positioned at a predetermined distance from the certain pixel. The luminance difference of any one of them is calculated for a plurality of different directions passing through the certain pixel, the information on the largest luminance difference among them is specified for the certain pixel, and the information on the luminance difference is determined as the radiographic image. Create brightness difference information for each pixel that is stored for multiple pixels in it,
In the pixel-by-pixel luminance difference information, a process of specifying pixels within a predetermined range from the pixel that is continuous in the predetermined direction to pixels having a luminance difference equal to or greater than a predetermined threshold as marker appearing pixels; A process of identifying a pixel in a predetermined range from the pixel connected in the predetermined direction to a pixel having a luminance difference less than a predetermined threshold as a marker non-appearing pixel is repeated separately for each of the plurality of different directions, respectively. In the pixel-by-pixel luminance difference information, a pixel that is determined as noise is removed from pixels in which the luminance difference is equal to or greater than a predetermined threshold, and a pixel that is a marker candidate is specified.
A processing method characterized by classifying pixels as marker candidates into a plurality of regions based on a group of adjacent pixels, and specifying a pixel belonging to a region having a region range equal to or greater than a threshold as a pixel indicating a marker. Logarithmically transform the luminance value for each pixel in the fluoroscopic image,
The luminance of a certain pixel in the radiographic image after the logarithmic conversion, and the luminances of two opposing pixels located on a straight line in a predetermined direction passing through the certain pixel and positioned at a predetermined distance from the certain pixel, Is calculated for a plurality of different directions passing through the certain pixel, information on the largest luminance difference among them is specified for the certain pixel, and information on the luminance difference is calculated. The method according to claim 7, further comprising: creating pixel-by-pixel luminance difference information for a plurality of pixels in the radiographic image. Identify a plurality of points including end points in the fluoroscopic image of the marker appearing in the fluoroscopic image,
Set a template image range including a plurality of points of the marker,
Generating template image information including information on the end points, information on the template image range, and information on a position of the template image range in the fluoroscopic image,
Correlation calculated using a luminance difference based on the pixel-by-pixel luminance difference information of a pixel within the search range and a pixel indicating the template image, by determining a search range including a region where the classified region range is greater than or equal to a threshold value The radiation is calculated using the origin of the position range in the search range when it is determined that the correlation value is equal to or greater than the threshold value, and the coordinates of the end point of the marker indicated by the template image in the template image. The processing method according to claim 7 or 8, wherein an end point coordinate of a marker that appears in a fluoroscopic image is specified. A radiation therapy apparatus controller for detecting a marker in a radioscopic fluoroscopic image generated by irradiating a subject in which a marker is embedded with radiation,
In the radiographic image, a luminance difference between a luminance of a certain pixel and luminances of two opposing pixels located on a straight line in a predetermined direction passing through the certain pixel and positioned at a predetermined distance from the certain pixel. The luminance difference of any one of them is calculated for a plurality of different directions passing through the certain pixel, the information on the largest luminance difference among them is specified for the certain pixel, and the information on the luminance difference is determined as the radiographic image. Marker edge enhancement means for creating pixel-by-pixel luminance difference information held for a plurality of pixels in the inside,
In the pixel-by-pixel luminance difference information, a process of specifying pixels within a predetermined range from the pixel that is continuous in the predetermined direction to pixels having a luminance difference equal to or greater than a predetermined threshold as marker appearing pixels; A process of identifying a pixel in a predetermined range from the pixel connected in the predetermined direction to a pixel having a luminance difference less than a predetermined threshold as a marker non-appearing pixel is repeated separately for each of the plurality of different directions, respectively. A noise removing unit that removes pixels determined to be noise from among pixels in which the luminance difference is equal to or greater than a predetermined threshold in the pixel-by-pixel luminance difference information, and identifies pixels that are marker candidates;
Marker specifying means for classifying the pixels as the marker candidates into a plurality of areas according to a group of adjacent pixels, and specifying pixels belonging to an area whose area range is equal to or greater than a threshold as a pixel indicating a marker,
A program characterized by functioning as Function as a logarithmic conversion means for logarithmically converting the luminance value for each pixel in the radiographic image,
The marker edge enhancement means is positioned on a straight line in a predetermined direction passing through the certain pixel and the luminance of a certain pixel in the radiographic image after the logarithmic transformation, and is opposed to the certain pixel at a predetermined distance. The luminance difference of any one of the luminance differences between the two pixels is calculated for a plurality of different directions passing through the certain pixel, and the information on the largest luminance difference among them is specified for the certain pixel. 11. The program according to claim 10, wherein the program functions as means for creating pixel-by-pixel luminance difference information in which the luminance difference information is stored for a plurality of pixels in the radiographic image. A plurality of on-marker specifying means for specifying a plurality of points including end points in the fluoroscopic image of the marker appearing in the fluoroscopic image;
Template range setting means for setting a template image range including a plurality of points of the marker;
Function as template image generation means for generating template image information including information on the end points, information on the template image range, and information on the position of the template image range in the fluoroscopic image,
The marker specifying unit determines a search range including a region where the range of the classified region is equal to or greater than a threshold, and a luminance difference based on the pixel-by-pixel luminance difference information of a pixel in the search range and a pixel indicating the template image And the origin of the position range in the search range when it is determined that the correlation value is equal to or greater than the threshold, and the coordinates in the template image of the end point of the marker indicated by the template image 12. The program according to claim 10, wherein the program is made to function as means for specifying end point coordinates of a marker appearing in the radioscopic image.

Images