JP4681358B2 - Ultrasonic diagnostic apparatus and volume data processing method - Google Patents

Description

  The present invention relates to an ultrasonic diagnostic apparatus, and more particularly to a data processing technique for volume data acquired from a three-dimensional space in a living body.

  A polyp is generally a proliferated and raised mucosa on the inner wall of a tissue, and is likely to occur in the gallbladder, large intestine, stomach, uterus, esophagus and the like. For example, benign polyps such as cholesterol polyps, adenomyomatosis polyps, inflammatory polyps, and malignant polyps as gallbladder cancer are known as gallbladder polyps. Examples of the shape of the polyp include warts, mushrooms, and other protrusions.

  In the diagnosis of the polyp, an ultrasonic diagnostic apparatus is used in addition to an X-ray CT apparatus and an endoscope. For example, for gallbladder polyps, diagnosis by oral endoscope is currently impossible, and it is difficult to sufficiently identify the properties of polyps on X-ray CT images. The In that case, on the two-dimensional tomographic image of the gallbladder, the position and posture of the ultrasonic probe in contact with the living body are artificially adjusted so that the cross section of the gallbladder polyp becomes the largest. A freeze operation is performed when a good two-dimensional tomographic image can be displayed, and the polyp diameter is measured using a marker on the two-dimensional tomographic image as a still image in that state.

  Patent Document 1 below discloses an ultrasonic diagnostic apparatus that extracts a contour of a tissue. Patent Literature 2 below discloses an ultrasonic diagnostic apparatus that extracts a tissue and calculates its volume. Patent Document 3 below discloses a system for extracting a tumor and quantitatively evaluating the surface shape thereof. Patent Document 4 discloses a technique for extracting a cavity in the heart. However, those patent documents do not disclose a technique for automatically separating and extracting a protrusion such as a polyp from a tissue.

JP 7-194597 A JP-A-8-299341 JP 2000-126178 A Japanese Patent Laid-Open No. 2004-159997

  Recently, three-dimensional measurement using ultrasonic waves has been put into practical use. For example, the technique is used when calculating the volume of the left ventricle in the circulatory organ region, and the technique is used when measuring the size of the fetus in the uterus in the obstetric region. Such measurement techniques generally extract spatially independent parts, and even if such techniques are used as they are, it is difficult to extract continuous polyps and other protrusions from the inner wall of the tissue. .

  An object of the present invention is to enable automatic extraction of protrusions protruding from the inner wall of tissue.

  Another object of the present invention is to make it possible to identify and display a protrusion protruding from the inner wall of a tissue on a three-dimensional image.

  Another object of the present invention is to make it possible to accurately calculate the volume of a protrusion protruding from the inner wall of a tissue.

  Another object of the present invention is to make it possible to quantitatively evaluate the surface shape of a protrusion protruding from the inner wall of a tissue.

(1) The present invention relates to a transmitting / receiving means for acquiring volume data by transmitting / receiving ultrasonic waves to / from a three-dimensional space in a living body, and a target tissue in the three-dimensional space based on the volume data. Boundary surface extracting means for extracting a boundary surface between the tissue part and the cavity part, and a protrusion extracting means for extracting a protrusion protruding from the body part of the tissue part toward the cavity part based on the shape of the boundary surface It is characterized by including these.

  According to the above configuration, volume data is acquired by transmitting and receiving ultrasonic waves to and from a living body. Volume data corresponds to a set of voxels (voxel data) in a three-dimensional space. The three-dimensional space includes a tissue part and a cavity part, that is, the volume data includes data corresponding to the tissue part and data corresponding to the cavity part. The tissue portion includes a main body portion and a protrusion, and the protrusion is a portion protruding from the main body portion toward the cavity. Since the main body portion and the protrusion are continuous, it is impossible to separate and extract only the protrusion as it is. Therefore, first, a boundary surface between the tissue part and the cavity part is extracted. For example, binarization processing, three-dimensional boundary extraction processing, etc. are performed on the volume data. Next, the surface shape of each position on the boundary surface, particularly desirably, the variation in the shape of the surface is referred to, and based on this, a separation surface that crosses the base portion (base portion) of the protrusion is set. This separation surface makes it possible to separate the main body portion and the protrusion in the tissue portion. In other words, on the boundary surface, by utilizing the fact that the shape change is more severe in the base part of the protrusion than in the other parts, the base part is automatically recognized, and the protrusion is separated in the data processing based on it. .

  The above process may be executed on the entire transmission / reception space, or the above process may be executed on a part thereof. In the latter case, it is desirable to use a three-dimensional region of interest. The target tissue is desirably a gallbladder, but may be another organ having a cavity inside. The protrusion is preferably a polyp, but the method of the present invention can be similarly applied to other protrusions protruding from the wall surface. The size and surface shape of the polyp are information for determining whether the polyp is malignant or benign. Therefore, it is desirable to provide a function capable of performing such analysis. Further, it is desirable that the polyp can be identified from other tissues and observed on the ultrasonic image.

  Desirably, the protrusion extraction means is a shape change degree calculating means for calculating a shape change degree at each position on the boundary surface, and a separation across the root of the protrusion based on the shape change degree at each position on the boundary surface. Separation surface setting means for setting a surface, and separation means for separating the protrusion from the main body portion of the tissue part by the separation surface. Each position on the boundary surface may be the position of each voxel extracted as the boundary surface. In that case, it is desirable to calculate the degree of shape change for all of the voxel groups corresponding to the boundary surface, but in order to reduce the amount of calculation, the shape change degree is calculated only for a plurality of representative voxels after thinning or resampling. You may make it do. The degree of shape change indicates the degree of shape variation at each position. A part having a large shape change degree can be recognized as the root of the protrusion. In other words, the degree of change in shape is relatively small on the inner wall of the organ and the main surface of the polyp (other than the root), but the degree of change in shape is relatively large at the root. The tendency can be used to identify the root and set a separation plane across it. The separation surface is preferably a flat surface, but may be a curved surface as long as it is necessary for more accurately separating the protrusions and the surface shape can be easily defined. The separation surface may be a plane extending over the entire three-dimensional space, or may be a surface having a finite extent enough to cut the base of the protrusion sufficiently. As the shape, it is desirable to select a desired shape such as a rectangle or a circle according to various conditions.

  Preferably, the shape change degree calculating means is configured as the shape change degree based on a normal line calculation unit that calculates a normal line at each position on the boundary surface and a normal line calculated at each position on the boundary surface. A dispersion value calculation unit for calculating a normal dispersion value at each position on the boundary surface, and a separation surface calculation for calculating the separation surface based on a normal dispersion value calculated at each position on the boundary surface Part. The normal represents the slope or slope of the surface at the location through which it passes. When a plurality of normal lines face in different directions at a certain portion, there is a high possibility that this is the root of the protrusion. The normal corresponds to the normal and vertical lines of the surface. The length of the normal line may be normalized, and the shape change degree may be obtained based on each coordinate axis component of the normal line. The dispersion value of the normal line does not need to be a mathematically exact dispersion value, and may be any value that indicates the degree of variation in the direction of the normal line.

  Preferably, the normal line calculation unit sets each position on the boundary surface as a target position, sets a unit sphere centering on the target position, and a tissue part element or cavity included in the unit sphere A center-of-gravity calculation unit that calculates the element centroid of the partial element, and a unit vector generation unit that generates a unit vector that faces the direction connecting the position of interest and the element centroid as the normal. The unit vector corresponds to a normal having a standardized length. The unit sphere defines a reference region around the target position (target voxel). The direction of the normal can be defined using the center of gravity of the tissue part or cavity in the interior. A normal may be defined from the surface inclination itself without using a unit sphere, or the surface inclination rate or the like may be used as the degree of shape change.

  Preferably, the separation surface calculation unit includes a plurality of candidate positions corresponding to the roots of the protrusions from among a plurality of positions on the boundary surface based on a variance value of normals at the respective positions on the boundary surface. A group extraction unit that extracts a ring-shaped group; and a function calculation unit that calculates a function representing the separation plane so that the ring-shaped group is divided into two ring-shaped subgroups by the separation plane. . It is desirable that the selection conditions for candidate positions can be changed manually or automatically according to various situations.

  Preferably, evaluation value calculating means for calculating a shape evaluation value for the surface of the protrusion based on a dispersion value of a normal line at each position on the surface of the protrusion. Preferably, the boundary surface extraction means extracts the boundary surface based on a means for identifying the tissue part and the cavity part based on the volume data, and a result of identification between the tissue part and the cavity part. Means. Preferably, the apparatus includes means for setting a three-dimensional region of interest with respect to the three-dimensional space, and the tissue portion and the cavity portion are identified in the three-dimensional region of interest.

(2) Further, the present invention provides a volume data processing method for processing volume data obtained by transmitting and receiving ultrasonic waves to and from a three-dimensional space in a living body, and based on the volume data, the three-dimensional A step of calculating boundary surface data representing a boundary surface between a tissue portion and a cavity portion in a target tissue in the space, and a shape variation degree representing a shape variation degree for each position on the boundary surface based on the boundary surface data A cutting plane that represents a cutting plane that crosses the root of the protrusion that protrudes from the main body portion of the target tissue toward the cavity based on the step of calculating data and the shape change degree data calculated for each position on the boundary surface A step of calculating data, and a step of specifying projection data representing the projection based on the cut surface data.

  The volume data processing method described above can be realized as hardware on the ultrasonic diagnostic apparatus or as a software function. The volume data processing method described above may be executed on a computer that processes data obtained from the ultrasonic diagnostic apparatus.

  As described above, according to the present invention, it is possible to automatically extract the protrusions protruding from the inner wall of the tissue. In addition, if necessary, the protrusions can be identified and displayed on the three-dimensional image, and the volume of the protrusions can be calculated with high accuracy, or the surface shape of the protrusions can be quantitatively evaluated.

  DESCRIPTION OF EXEMPLARY EMBODIMENTS Hereinafter, preferred embodiments of the invention will be described with reference to the drawings.

  First, the protrusion extraction method according to the present embodiment will be described with reference to FIGS. This protrusion extraction method is executed in the medical ultrasonic diagnostic apparatus in this embodiment, but the protrusion extraction method can also be executed on a computer that has obtained data from the ultrasonic diagnostic apparatus.

  FIG. 1 shows a flowchart of the protrusion extraction method according to this embodiment. The protrusion extraction method will be described according to the flowchart shown in FIG. 1 with reference to FIGS.

  In S101, ultrasonic waves are transmitted / received to / from a three-dimensional region in the living body, thereby acquiring volume data. Volume data is composed of echo data acquired at each address in the three-dimensional space. Here, each address is positioned as a voxel, and each echo data corresponds to voxel data. The volume data can be acquired, for example, by using an ultrasonic probe having a 2D array transducer, or can be acquired by an ultrasonic probe that mechanically scans the 1D array transducer.

  In S102, binarization processing is performed on the volume data. That is, a tissue part and a cavity part are included in the three-dimensional space, and binarization processing is executed to discriminate them in terms of data. For example, echo data having a value larger than a predetermined echo level is regarded as tissue echo data, and smaller echo data is regarded as cavity (liquid portion) echo data. . In that case, it is desirable to appropriately variably set the reference level. In the present embodiment, a value of 0 is given to the echo data of the cavity and a value of 1 is given to the echo data of the tissue. The reverse is also possible. Prior to the binarization processing, processing such as noise removal or smoothing may be performed on the volume data.

  FIG. 2A is a schematic diagram showing the result of performing binarization processing on the volume data acquired in the three-dimensional space 10. Reference numeral 12 represents a tissue part (organ), and reference numeral 14 represents a cavity in the tissue part 12. For example, body fluid or blood is present in the cavity 14. The level of echo data from the liquid is generally small, whereas the level of echo data from the tissue part 12 is generally large, so that the tissue part 12 and the cavity part 14 can be distinguished by the above binarization process. It is. The tissue portion 12 includes a main body portion 12A and a protrusion 12B protruding from the inner wall surface to the cavity portion 14 side. Here, the tissue part 12 is, for example, a gallbladder, and the protrusion 12B is a polyp existing in the gallbladder. This polyp has various shapes such as warts or mushrooms, but generally the root portion, that is, the root portion, rises from the inner wall of the tissue or has a constricted shape. A polyp or the like that rises in a mountain shape is an object of protrusion extraction according to the present invention as long as the root can be determined.

  The boundary surface extraction process described below is executed for the volume data after the binarization process, but prior to that, noise removal or smoothing process may be performed. For example, such processing is also described in the above Japanese Patent Application Laid-Open No. 2004-159997.

  In S103 shown in FIG. 1, the boundary surface between the tissue part and the cavity part is extracted. That is, since the data of the tissue part and the data of the cavity part are identified as a result of performing the binarization process in S102, the boundary surface can be extracted based on the data. Specifically, as shown in FIG. 2B, the boundary surface 16 between the tissue part 12 and the cavity part 14 is extracted by data processing in the three-dimensional space. In this case, a known three-dimensional spatial filter can be used. Examples of such filters include a Sobel filter that extracts horizontal / vertical edges, a Prewitt filter that extracts horizontal or vertical edges, a Laplacian filter using second derivative (difference), and a target voxel. A gradient edge enhancement filter that extracts edges in a specific direction among eight directions, a shift difference edge extraction filter that extracts edges in the horizontal / vertical / diagonal directions, and the like are known.

  In this embodiment, boundary detection is performed using a three-dimensional differentiation method on the assumption that each data is normalized to one of binary values. Specifically, a three-dimensional filter 20 as shown in FIG. 3 is used. When the value of the center voxel 22 is H, that is, 1 and the value L, that is, 0 exists in any one of the six neighboring voxels around the filter 20, the filter 20 determines the center voxel on the boundary surface. It is considered to be a voxel. By scanning the filter 20 as shown in FIG. 3 with respect to the volume data and performing the filtering process on each data, a value 1 is finally given to each voxel on the boundary surface, For other voxels, the value 0 is given. Thereby, the boundary surface is extracted.

  In S104 of FIG. 1, a unit vector as a normal is calculated for each voxel on the boundary surface. That is, as shown in FIG. 2C, a normal corresponding to a perpendicular line or a vertical line is obtained for each voxel on the boundary surface 16 between the tissue part 12 and the cavity part 14, and more specifically, details will be given later. As will be described, a unit vector corresponding to the normal is obtained. The boundary surface can be roughly divided into three regions. That is, a relatively smooth region corresponding to the inner wall of the organ, a region having a relatively small uneven shape according to the disease as the surface of the protrusion 12B, that is, the polyp, and a region corresponding to the root 12C as the rising portion of the protrusion 12B , Can be classified. Here, in the region corresponding to the base 12C, the surface shape changes relatively remarkably, that is, the base of the protrusion 12B can be automatically recognized from the tendency of the surface shape to change.

  In this embodiment, a unit vector is calculated for each voxel on the boundary surface as will be described below with reference to FIGS. FIG. 4 shows a unit sphere 26 for calculating a unit vector. The unit sphere 26 is set for each voxel on the boundary surface, and the radius of the unit sphere 26 is a predetermined value (for example, 10 voxels). The unit sphere 26 includes a tissue part element 34 belonging to the tissue part and a cavity part element 36 belonging to the cavity part. As described above, the unit sphere 26 is set for each voxel, and reference numeral 28 represents a target voxel that forms the origin of the unit sphere. In the present embodiment, the center of gravity is calculated for the tissue part element 34, and the center of gravity is represented by reference numeral 30. A normal line can be defined as a direction connecting the center of gravity 30 and the target voxel 28, and a unit vector U can be defined as a normal line having a predetermined unit length.

  5 and 6 show the unit vector calculation principle. The origin O shown in FIG. 5 represents the origin of the position vector. The point A corresponds to, for example, the center of gravity 30 described above, and the point B corresponds to the target voxel 28 described above. The vector AB can be defined as shown by the vector OB and the vector OA. Then, as shown in FIG. 6, a unit vector U having a length of 1 can be defined based on the vector AB. Here, kXc, kYc, and kZc are the x component, y component, and z component of the unit vector U, respectively. As will be described later in detail, it is possible to easily obtain the degree of unit vector variation by using each component.

  FIG. 7 shows a unit sphere 26 set for each voxel on the boundary surface 16, and a unit vector is defined for each voxel by each unit sphere 26. Since the unit vector is basically a normal line, it is possible to quantitatively grasp the degree of change in the inclination of the minute surface on the boundary surface by observing the change tendency of the direction of the unit vector.

  In S105 of FIG. 1, for each voxel on the boundary surface, a reference sphere different from the above unit sphere is defined, the unit vector for the voxel belonging to the inside is referred to, and the variation in the direction of these unit vectors A variance value representing the degree is calculated. In an embodiment described later, a standard deviation is obtained for each of the x component, the y component, and the z component, and the magnitude of the shape change degree is determined based on the standard deviation.

  FIG. 8 shows a plurality of unit vectors set on the boundary surface 16. Here, the distribution state of the unit vectors is shown for the three regions 40, 42, and 44 in particular. In the region 40 corresponding to the inner wall of the main body portion 12A in the tissue part, the unit vector groups tend to be substantially aligned. Further, even in the region 44 corresponding to the surface of the protrusion 12B, even if there is a minute unevenness, the vector group tends to be aligned. On the other hand, in the region 42 corresponding to the root of the protrusion 12B, the plurality of unit vectors are directed in different directions, and the dispersion value or the shape change degree related to the unit vector becomes extremely large. That is, it is possible to automatically recognize the root portion using such a property. Of course, for example, a malignant polyp may have relatively large irregularities on its surface shape, or the polyp may be distorted. However, the discrimination criteria or criteria for the degree of dispersion or shape change are set appropriately. By doing so, it is possible to accurately extract the root portion.

  In an embodiment described later, the standard deviation for each component, that is, the degree of component variation is used as the variance value of the unit vector. FIG. 9 shows, for example, a variation in the size of the x component as a histogram for the three regions 40, 42, and 44 shown in FIG. (A) shows a representative example of the histogram for the region 40 and the region 44, and (B) shows a representative example of the histogram for the region 42. As described above, since the shapes of the histograms are greatly different, it is possible to specify the root portion using the difference in the shape of the histogram, that is, the difference in the standard deviation. In the above description, the change in the direction of the unit vector on the two-dimensional plane is shown, but the actual processing is executed in the three-dimensional space, that is, the degree of three-dimensional variation is evaluated.

  In S106 of FIG. 1, a candidate voxel group corresponding to the root portion is specified based on the variance value for each voxel calculated as described above. Usually, the voxel population has a ring-like shape. In S106, subsequently, a separation plane is set so as to divide the candidate voxel group into two ring-shaped partial groups, and the projection is separated from the inner wall of the tissue part using the separation plane.

  That is, as shown in FIG. 2 (D), a separation surface 18 that traverses the root portion 12C specified in the three-dimensional space is automatically set, and using it, (E) As shown, the protrusion 12B is separated as a three-dimensional isolated region. In FIG. 2D, the minimum necessary separation surface 18 is shown. Actually, however, a separation surface extending over the entire three-dimensional space having a certain extent is set, which is shown in FIG. Has been. The separation surface 46 is configured as a plane that crosses the base 12C of the protrusion 12B. However, the separation surface 46 may be defined as a curved surface. A specific method for setting the separation plane based on the ring-shaped candidate voxel population will be described in detail later with reference to FIGS.

  As described above, by separating the protrusions spatially, it is possible to perform a coloring process only on the protrusions in a three-dimensional image or a two-dimensional tomographic image, or to calculate the volume or surface shape of the protrusions. It becomes possible. That is, for example, a polyp protruding from the inner wall of an organ on a three-dimensional image can be identified and displayed with a color different from that of the organ, and there is an advantage that its form can be recognized more easily. The volume of the polyp and its surface shape are one index for determining whether the polyp is benign or malignant, but such an index can be automatically calculated and is objective. There is an advantage that useful information can be provided in disease diagnosis.

  Next, the configuration of the ultrasonic diagnostic apparatus according to the present embodiment will be described with reference to FIGS. 11 and 12. This ultrasonic diagnostic apparatus has a function of executing the protrusion extraction method described above.

  FIG. 11 is a block diagram showing the overall configuration of the ultrasonic diagnostic apparatus, and FIG. 12 is a block diagram showing a specific configuration example of the projection processing unit 56 shown in FIG. Each data processing function described below can be configured as dedicated hardware, or can be realized as a function of a processor that performs a program operation.

  In FIG. 11, a 3D probe 50 is a probe that forms a three-dimensional echo data capturing region, that is, a three-dimensional space. The 3D probe 50 has a 2D array transducer in this embodiment. The 2D array transducer is composed of vibration elements composed of hundreds or thousands of two-dimensional arrays. An ultrasonic beam is formed by the 2D array transducer, and a scanning plane is formed by scanning the ultrasonic beam in a predetermined direction. The above three-dimensional space is constructed by scanning the scanning plane in a direction orthogonal thereto. One ultrasonic beam corresponds to an echo data string arranged in time series. Volume data is configured as a whole three-dimensional space. The shape of the three-dimensional space is according to the electronic scanning method, and there are various shapes. Electronic linear scanning, electronic sector scanning, and the like are known as electronic scanning methods. The 3D probe 50 is used in contact with the body surface, but a probe of a type inserted into a body cavity may be used. Also, a probe that forms a three-dimensional space by combining electronic scanning and mechanical scanning may be used.

  The transmission / reception unit 52 functions as a transmission beam former and a reception beam former. That is, the transmission / reception unit 52 outputs a plurality of transmission signals to the 3D probe 50 in parallel with a predetermined delay relationship. As a result, a transmission beam is formed. On the other hand, the plurality of reception signals output from the 3D probe 50 are subjected to phasing addition processing in the transmission / reception unit 52. As a result, a reception beam is electronically formed, and a reception signal after phasing addition is output. The received signal is stored as echo data on the 3D memory 54 through necessary processes such as detection, A / D conversion, logarithmic conversion, and coordinate conversion.

  The 3D memory 54 has a three-dimensional storage space corresponding to the three-dimensional space in the living body. Each address on the 3D memory 54 corresponds to each address on the measurement coordinate. That is, volume data is stored on the 3D memory 54.

  The protrusion processing unit 56 executes a process of extracting protrusions in the tissue part based on the volume data. Further, as will be described in detail later, it has a function of measuring the volume and surface shape of the protrusion.

  The display processing unit 58 includes a tomographic image forming unit 60 and a three-dimensional image forming unit 62. The tomographic image forming unit 60 has a function of forming a tomographic image of a cutting plane arbitrarily set in the three-dimensional space based on the volume data. The three-dimensional image forming unit 62 forms a three-dimensional image of the organ based on the volume data based on, for example, a volume rendering method. In this case, the extracted protrusions are colored differently from other tissues. The display processing unit 58 executes a process for synthesizing the ultrasonic image and the graphic data.

  An arbitrary tomographic image or a three-dimensional image is displayed on the display unit 64. In addition, the volume and shape evaluation value calculated for the protrusion are displayed as numerical values.

  The control unit 66 is constituted by a processor that performs a program operation, and the operation of each component shown in FIG. 11 is controlled by the control unit 66. The operation panel 68 is connected to the control unit 66, and the operation panel 68 includes a keyboard, a trackball, and the like. Using the operation panel 68, the user can perform various settings and condition input. However, in this embodiment, there is an advantage that protrusions can be automatically extracted and automatically measured except for the setting of a three-dimensional region of interest (3D-ROI) described later.

  Next, a specific configuration example of the protrusion processing unit 56 illustrated in FIG. 11 will be described with reference to FIG.

  Each binarization processing unit 70 sequentially receives echo data constituting the volume data, that is, voxel data. The binarization processing unit 70 binarizes and converts each echo data with a predetermined threshold as a reference in order to discriminate the tissue part from the cavity part. In this case, the value 1 is given to the echo data corresponding to the tissue part, and the value 0 is given to the echo data corresponding to the cavity part. A threshold setting unit 74 and a 3D-ROI generation unit 76 are connected to the binarization processing unit 70. The threshold setting unit 74 is a module that variably sets a threshold serving as a reference for binarization processing. The 3D-ROI generation unit 76 is a module that generates a three-dimensional region of interest for limiting the target region for executing the protrusion extraction process. When such a three-dimensional region of interest is set, the projection extraction process is executed in that region. For example, a three-dimensional region of interest representing volume data may be displayed on a display, and the user may set a three-dimensional region of interest by designating a desired region on the three orthogonal surfaces. It is also possible to set a 3D region of interest on a 3D image.

  The memory 78 stores data after binarization processing. Note that each of the memories 78, 82, 86, 92, 94, 96, 106, 110, 116, and 122 shown in FIG. 12 corresponds to a three-dimensional memory having a three-dimensional storage space. However, they do not have to be physically separate, and each memory may be realized using one or a plurality of physical memories.

  The boundary surface extraction unit 80 extracts the boundary surface existing in the three-dimensional space by using the above-described edge detection method, specifically using a filter as shown in FIG. That is, voxels (data) corresponding to the boundary surface are extracted. The extracted data is stored in the memory 82. When each voxel on the boundary surface is specified, the center-of-gravity calculation unit 84 sets each voxel as a target voxel based on the binarized volume data stored in the memory 78, and sets a unit sphere for the voxel. The center of gravity of the tissue part element belonging to the unit sphere is calculated. In this case, the center of gravity of the cavity element may be calculated. The radius for the unit sphere can be variably set by the user or automatically. The calculation result of the center of gravity is stored on the memory 86. Accordingly, the memory 82 stores data representing each voxel on the boundary surface, and the memory 86 stores information specifying the coordinates of the center of gravity obtained for each voxel on the boundary surface. Therefore, the normal line calculation unit 88 calculates the direction of the normal line as a direction connecting the target voxel and the position of the center of gravity obtained for the target voxel. This is performed for each voxel on the interface.

  Then, the unit vector calculation unit 90 calculates a unit vector based on the normal for each voxel on the boundary surface. Specifically, as described with reference to FIG. 6, the x component, y component, and z component representing the unit vector are calculated. Each component is stored in three memories 92, 94, 96 provided at the subsequent stage of the unit vector calculation unit 90.

  The three standard deviation calculation units 98, 100, and 102 provided in the subsequent stage of the three memories 92, 94, and 96 each have a predetermined size centered on the target voxel, with each voxel on the boundary surface being the target voxel. Set the reference sphere. Then, the standard deviation is calculated for a plurality of unit vectors belonging to the reference sphere. Specifically, a standard deviation is calculated for each component. This standard deviation indicates the degree of variation for each component. The radius of the reference sphere can be changed by the user or automatically.

  The root determination unit 104 refers to the standard deviation for each component calculated for each voxel on the boundary surface, and determines a candidate voxel corresponding to the root of the protrusion from a plurality of voxels on the boundary surface. In this case, if any standard deviation of the x component, the y component, and the z component exceeds a predetermined threshold, the voxel is determined as a candidate voxel. That is, when a large variation is observed in any of the coordinate directions, that part is regarded as a projection root candidate. The memory 106 stores candidate voxels recognized as root candidates. The stored address of each candidate voxel specifies the rear position of the root part.

  The separation plane creating unit 108 sets a separation plane that divides it into two ring-shaped subgroups based on a plurality of candidate voxels stored in the memory 106, that is, a candidate voxel group. Although the setting method will be described later, in this embodiment, a condition for defining the size of the separation surface is given. For example, a circular shape having a radius twice or three times the diameter of the root is given. A separation plane is set as a region. Coordinate information representing the separation plane is stored on the memory 110. Alternatively, a parameter value specifying a function representing the separation plane is stored.

  The separation processing unit 114 executes a process of extracting a data portion corresponding to the protrusion from the binarized volume data using the separation surface obtained as described above. In this case, the center-of-gravity detector 112 calculates the coordinates corresponding to the center of gravity of the cavity based on the binarized volume data, and the portion existing on the center of gravity calculated when viewed from the separation plane is Recognized as a protrusion. Alternatively, by performing mask processing on the tissue part data on the side where the center of gravity does not exist when viewed from the separation surface, only the protrusions existing in the cavity part can remain as a result. Data corresponding to the protrusions is stored on the memory 116.

  The labeling unit 118 executes a three-dimensional labeling process in order to extract data corresponding to the protrusions stored in the memory 116, that is, a voxel set. This makes it possible to separate only the protrusions in the three-dimensional space for data processing.

  The volume calculation unit 124 calculates the volume of the extracted protrusion. In this case, the volume may be obtained by counting the number of voxels constituting the protrusion and multiplying it by a predetermined coefficient.

  The representative value calculation unit 120 extracts the largest standard deviation among the three standard deviations obtained for each voxel in the three standard deviation calculation units 98, 100, and 102 as a representative value, and extracts it as the memory 122. Is stored. The surface shape evaluation unit 126 receives voxels representing the protrusions extracted by the labeling unit 118, and the surface shape evaluation unit 126 applies the data representing the protrusions extracted from the labeling unit 118 to the surface of the protrusions. Only the corresponding representative values are extracted, and the surface shape evaluation value is calculated based on them. In the present embodiment, the surface shape evaluation value is obtained by further calculating the standard deviation (that is, the standard deviation of the standard deviation) with respect to the representative value for a plurality of voxels on the surface of the protrusion.

  The data representing the protrusion, the data representing the volume, and the data representing the surface shape evaluation value obtained as described above are respectively output to the display processing unit 58 shown in FIG. In the display processing unit 58, image formation is performed based on data representing protrusions. The calculated volume value and surface shape evaluation value are converted into numerical graphic data.

  In the calculation of the surface shape evaluation value, the representative value for the entire surface of the protrusion may be used as the basis of the calculation. However, the representative value for the voxel corresponding to the vicinity of the root is evaluated for the surface shape. It is desirable to exclude it from the calculation. Specifically, with respect to voxels belonging to a region having a certain thickness from the separation surface, an exclusion process is applied so that the representative value is not used in the calculation. In the above-described embodiment, the root of the protrusion is determined based on the degree of variation of the unit vector, and separately, the projection surface shape is evaluated based on the degree of variation of the unit vector. Variation is an object of determination, and in the latter surface shape evaluation, relatively small variation is an object of calculation, and since different objects are targeted, each calculation can be performed appropriately. In addition, it is desirable to appropriately set calculation conditions so that such calculation is properly performed. In particular, it is desirable to appropriately set a threshold value for determining the root.

  Next, the method for setting the separation surface will be described in more detail with reference to FIGS.

  FIG. 13 shows a candidate voxel group 132 corresponding to the root of the protrusion in the three-dimensional space (volume data space) 130. The candidate voxel population 132 has a ring shape. As described above, a separation plane is set as a plane that divides this candidate voxel group 132 into two pieces. There are several methods for setting the separation surface as follows.

  In the first method, a function ax + by + cz + d = 0 representing an arbitrary surface is set for the three-dimensional space 130. Here, a, b, c, and d are coefficients. A perpendicular is dropped from each candidate voxel with respect to this surface, and the distance from the surface to the candidate voxel on the perpendicular is obtained. And the total value of the distance about each voxel is calculated. The total value is a function of the coefficients a, b, c, d. By using the least square method and determining a, b, c, and d that minimize the total value, an optimal plane including the separation plane can be defined. However, this first method is complicated in processing and requires a long processing time. Therefore, the following second method, which is a simple method but can obtain the separation plane fairly accurately, will be described.

In FIG. 13, a direction in which the xy plane is viewed from the front is defined as a front direction, a direction in which the yz plane is viewed from the front is defined as a side direction, and a direction in which the xz plane is viewed from the front is defined as a top direction. In the second method, first, for the candidate voxel population 132 scattered in a ring shape, as shown in FIG. Further, the three-dimensional space 130 is divided into four by a plane passing through the entire center of gravity G and parallel to the upper surface and parallel to the side surface. In FIG. 14, each block (partial space) is represented by A, B, C, and D. Next, for each block A, B, C, D, the block centroid is calculated for a plurality of candidate voxels belonging to it. As shown in FIG. 15, the center of gravity of each block is G _A , G _B , G _C , and G _D. Among the block centroids G _A , G _B , G _C , and G _D , the first block centroid farthest from the overall centroid G and two blocks that are adjacent to the block having the first block centroid The second block centroid farther from the overall centroid G of the block centroids is obtained, and these are set as K1 and K2, respectively. Thereby, a plane passing through the three points G, K1, and K2 can be specified as a plane including the separation plane.

  In the above, the block centroid of the block including the first block centroid is referred to as the block centroid of the block including the first block centroid. This is because it is not specified as. That is, if the block centroid is determined as the second block centroid, the three points G, K1, and K2 are arranged in a straight line, and a plane cannot be determined. On the other hand, according to the above method, the three points of the two block centroids selected from the three blocks and the total centroid G are not arranged in a straight line but are separated from each other to some extent, so that the plane can be determined without error. .

  In the above second method, in most cases, the plane including the separation plane can be determined with high accuracy by the above method. However, the plane may not be determined depending on the distribution status of the candidate voxel population. As shown in FIG. 16, this is a case where the candidate voxel population 132 exists perpendicular to the lateral direction. FIG. 17 is a diagram in which the candidate voxel group 132 is projected from the front direction, but G, K1, and K2 obtained in the same manner as described above are linearly arranged and the plane cannot be determined. Therefore, in such a case, it is desirable to perform the four divisions described in FIG. 14 from the side surface side. That is, as shown in FIG. 18, the three-dimensional space 130 is divided into four by a plane passing through the center of gravity G and parallel to the front and a plane parallel to the upper surface. According to such division, the plane can be determined by obtaining the three points G, K1, and K2 by applying the same method as described above. The projection from the side is as shown in FIG.

  As described above, when the four points are divided from the front direction and three points are linearly arranged as a result of the series of processing, the above method is applied by dividing again into four blocks from the side direction. Good. On the other hand, from the beginning, the above method may be applied to both the front and side directions to obtain three points for each direction, and a non-linear direction may be employed. In addition, as a result, each plane formed from the three points obtained from both directions is substantially the same.

  The following several methods can be given as a method for determining the size of the separation surface. In FIG. 20, a circle 140 is set on the plane obtained as described above, with the overall center of gravity G as the center and a radius of about 2 to 3 times the distance between the overall center of gravity G and the first block center of gravity K1. The range of the circle 140 can be defined as the separation plane. FIG. 21 shows a state in which the protrusion 12B is separated from the root by the circular separation surface 140.

  As another method, as shown in FIG. 22, the size of the separation surface 142 may not be specified, and the separation surface 142 may be a plane corresponding to a cut surface that covers the entire three-dimensional space 130. In this case, as shown in FIG. 23, the barycentric coordinates of the cavity (for example, inside the gallbladder) in the volume data after binarization are obtained in advance, and within the two partial spaces divided into two by the separation plane, The partial space where the center of gravity of the cavity does not exist is specified, and the partial data is masked with the binarized data, and as a result, the protrusion 12B that exists in isolation is extracted.

  In the above-described embodiment, the gallbladder of the human body is a diagnosis target, but it is possible to apply the present invention to the other organs in the same manner as described above to process the protrusions. In the above embodiment, the unit vector is set using the unit sphere, but other methods may be used as long as the degree of change of the surface of each part on the boundary surface can be recognized. Further, it is desirable that the boundary surface be smoothed so as not to hinder the subsequent surface shape evaluation so that the change in shape of the boundary surface can be more clearly identified. The degree of smoothing may be varied depending on the shape of the protrusions, and the degree of smoothing may be reduced for polyps having enlarged tips and large constrictions. For a polyp or the like, the degree of smoothing may be increased so that the root can be discriminated more accurately.

It is a flowchart which shows an example of the protrusion extraction method which concerns on this invention. It is a conceptual diagram for demonstrating the protrusion extraction method shown in FIG. It is a figure for demonstrating the filter used in the case of extraction of a boundary surface. It is a figure for demonstrating the unit sphere for defining a unit vector. It is a figure for demonstrating the calculation of a normal vector. It is a figure for demonstrating each component in a unit vector. It is a figure for demonstrating the several unit sphere set to a boundary surface. It is a figure for demonstrating the difference in the dispersion | variation degree of the unit vector group for every site | part in a boundary surface. It is a figure for demonstrating the dispersion | variation degree of the predetermined component which changes with areas. It is a figure for demonstrating the separation process of the protrusion by a separation surface. 1 is a block diagram showing an overall configuration of an ultrasonic diagnostic apparatus according to the present embodiment. It is a block diagram which shows the specific structural example of the protrusion process part shown in FIG. It is a figure which shows the ring-shaped candidate voxel group which exists in a three-dimensional space. It is a figure for demonstrating the four division of a three-dimensional space. It is a figure which shows the projection surface of the three-dimensional space shown in FIG. It is a figure for demonstrating the four division of a three-dimensional space. It is a figure which shows the projection surface of the three-dimensional space shown in FIG. It is a figure which shows the other 4 division with respect to three-dimensional space. It is a figure which shows the projection surface of the three-dimensional space shown in FIG. It is a figure for demonstrating a circular separation surface. It is a figure for demonstrating the separation process of the processus | protrusion by a circular separation surface. It is a figure which shows the separation surface which crosses the whole three-dimensional space. It is a figure which shows the protrusion isolate | separated by the separation surface shown by FIG.

Explanation of symbols

  10 three-dimensional space, 12 tissue part, 12A body part, 12B protrusion, 14 cavity part, 16 boundary surface, 20 filter, 26 unit sphere, 54 3D memory, 56 protrusion processing part, 58 display processing part, 70 binarization process , 80 boundary plane extraction unit, 84 centroid calculation unit, 88 normal calculation unit, 90 unit vector calculation unit, 98, 100, 102 standard deviation calculation unit, 104 root determination unit, 108 separation plane creation unit, 114 separation processing unit , 118 Labeling unit, 120 Representative value calculation unit, 124 Volume calculation unit, 126 Surface shape evaluation unit.

Claims (12)

Wave transmitting / receiving means for acquiring volume data by transmitting / receiving ultrasonic waves to a three-dimensional space in a living body;
A boundary surface extraction means for extracting a boundary surface between a tissue part and a cavity in the target tissue in the three-dimensional space based on the volume data;
Based on the shape of the boundary surface, a protrusion extracting means for extracting a protrusion that protrudes from the main body portion of the tissue portion toward the cavity portion;
Only including,
The protrusion extracting means includes
A normal value calculation unit that calculates a normal at each position on the boundary surface, and a variance value that calculates a normal dispersion value at each position on the boundary surface based on the normal calculated at each position on the boundary surface A computing unit having a computing unit;
Separation surface setting means for setting a separation surface that crosses the base of the protrusion, based on the dispersion value of the normal calculated at each position on the boundary surface;
An ultrasonic diagnostic apparatus comprising: The apparatus of claim 1 .
The normal calculation unit is
A unit sphere setting unit that sets each position on the boundary surface as a target position and sets a unit sphere centered on the target position;
A centroid calculating unit that calculates the element centroid of the tissue element or cavity element included in the unit sphere;
A unit vector generation unit that generates a unit vector facing a direction connecting the target position and the element gravity center as the normal line;
An ultrasonic diagnostic apparatus comprising: The apparatus of claim 2 .
The separation surface setting means includes
A group extraction unit that extracts a ring-shaped group consisting of a plurality of candidate positions corresponding to the roots of the protrusions from a plurality of positions on the boundary surface, based on a variance of normals at each position on the boundary surface; ,
A function calculation unit that calculates a function representing the separation plane so that the ring-shaped group is divided into two ring-shaped subgroups by the separation plane;
An ultrasonic diagnostic apparatus comprising: Wave transmitting / receiving means for acquiring volume data by transmitting / receiving ultrasonic waves to a three-dimensional space in a living body;
A boundary surface extraction means for extracting a boundary surface between a tissue part and a cavity in the target tissue in the three-dimensional space based on the volume data;
A normal calculation unit that calculates a normal at each position on the boundary surface;
A dispersion value calculation unit for calculating a dispersion value of a normal at each position on the boundary surface based on a normal line calculated at each position on the boundary surface;
An evaluation value calculating means for calculating a shape evaluation value for the surface of the protrusion protruding from the main body portion of the tissue part toward the cavity , based on the dispersion value of the normal line at each position on the surface of the protrusion ;
An ultrasonic diagnostic apparatus comprising: The apparatus of claim 4.
Means for extracting a protrusion protruding from the main body portion of the tissue portion toward the cavity portion based on a dispersion value of a normal line at each position on the surface of the protrusion;
The ultrasonic diagnostic apparatus, wherein the evaluation value calculation means calculates a shape evaluation value for the surface of the extracted protrusion. The apparatus of claim 1.
The boundary surface extraction means includes
Means for discriminating between the tissue part and the cavity part based on the volume data;
Means for extracting the boundary surface based on the identification result of the tissue part and the cavity part;
An ultrasonic diagnostic apparatus comprising: The apparatus of claim 6 .
Means for setting a three-dimensional region of interest for the three-dimensional space;
The ultrasonic diagnostic apparatus, wherein the tissue part and the cavity part are identified in the three-dimensional region of interest. The apparatus of claim 1.
An ultrasonic diagnostic apparatus comprising volume calculating means for calculating the volume of the protrusion. The apparatus of claim 1.
An ultrasonic diagnostic apparatus comprising image processing means for identifying and displaying the protrusion on a screen. The apparatus of claim 1.
The ultrasonic diagnostic apparatus, wherein the protrusion is a polyp. The apparatus of claim 1.
The ultrasonic diagnostic apparatus, wherein the target tissue is a gallbladder. In a volume data processing method for processing volume data acquired by transmitting / receiving ultrasonic waves to / from a three-dimensional space in a living body,
Calculating boundary surface data representing a boundary surface between a tissue part and a cavity in the target tissue in the three-dimensional space based on the volume data;
Calculating shape variation data representing the degree of shape variation for each position on the boundary surface by calculating a normal variance value at each position on the boundary surface based on the boundary surface data;
Based on the shape change degree data calculated for each position on the boundary surface, calculating cutting surface data representing a cutting surface that crosses the root of the protrusion protruding from the main body portion of the target tissue toward the cavity, and
Identifying projection data representing the projection based on the cut surface data;
A volume data processing method comprising:

Images