JP2015123306A - Blood vessel detection apparatus, magnetic resonance apparatus, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a blood vessel detection apparatus improved in accuracy in detecting a blood vessel such as an aorta.SOLUTION: A MR apparatus 100 includes image generation means 91 to generate images of m pieces of slices crossing an imaging portion including a lung, a liver, and an aorta, means to obtain an index for determining whether or not the slice crosses the liver on the basis of the image of the slice, selection means 95 to select a first slice crossing the liver from among the m pieces of slices on the basis of the index, and detection means 96 to detect a blood vessel in a second slice crossing the lung on the basis of a position of the aorta detected in the first slice.

Description

  The present invention relates to a blood vessel detection device that detects a blood vessel, a magnetic resonance device that acquires an image of a subject, detects the blood vessel from the acquired image, and a program for detecting the blood vessel.

  Conventionally, a magnetic resonance apparatus for imaging a subject using a contrast agent is known (Patent Document 1).

JP 2009-261904 A

  As an example of a method for imaging a subject using a contrast medium, there is a method in which an operator finds the position of the aorta from an image and sets a tracker region for detecting the contrast medium at the position of the aorta. In this method, imaging is performed when a predetermined amount of contrast agent reaches the tracker area. However, an inexperienced operator may take a long time to find the position of the aorta, or may set a tracker area at a location different from the aorta. Therefore, development of a method for automatically detecting the aorta in a short time is desired. An example of a technique for automatically detecting the aorta in a short time has been published in the paper “Med Imag Tech Vol.31 No.2 March 2013”.

  In the above paper, the blood signal in the aorta is reduced as much as possible by increasing the area of the Zdephaser of the gradient magnetic field. Therefore, an aorta can be detected for each slice by detecting a low signal area from each slice. However, in a slice crossing the lung, not only the aorta, but also the air region contained in the lung has a low signal, so the air region contained in the lung may be erroneously detected as the aorta. Therefore, a technique for improving the detection accuracy of blood vessels such as the aorta is desired.

According to a first aspect of the present invention, there is provided an image creating means for creating an image of m slices crossing an imaging region including a first region, a second region, and a blood vessel,
Means for determining an index for determining whether the slice crosses the first region based on the image of the slice;
Selection means for selecting a first slice across the first region from the m slices based on the index;
Detecting means for detecting the blood vessel in the first slice, the detecting means detecting a blood vessel in the second slice crossing the second part based on the position of the blood vessel in the first slice; ,
Is a blood vessel detection device having

  The 2nd viewpoint of this invention is a magnetic resonance apparatus which has said blood vessel detection apparatus.

According to a third aspect of the present invention, an image creation process for creating an image of m slices crossing an imaging region including the first region, the second region, and a blood vessel;
Processing for obtaining an index for determining whether or not the slice crosses the first region based on the image of the slice;
A selection process for selecting a first slice across the first region from the m slices based on the index;
A detection process for detecting the blood vessel in the first slice, the detection process for detecting a blood vessel in the second slice crossing the second part based on the position of the blood vessel in the first slice; ,
Is a program for causing a computer to execute.

  The possibility of erroneously detecting the second part as a blood vessel in the second slice can be reduced.

1 is a schematic view of a magnetic resonance apparatus according to a first embodiment of the present invention. It is a figure which shows the scan performed with this form. It is a figure which shows an imaging | photography site | part schematically. It is a figure which shows an example of the flow when performing localizer scan LS and this scan MS. It is explanatory drawing of the localizer scan LS. It is a figure which shows roughly the teacher data used in order to produce the discriminator C. It shows a search region R ₁ that is set in the axial image D _1. It is a diagram showing a window W set in the search area R _1. FIG. 10 is an explanatory diagram for obtaining the rotation angle θ, the number of pixels x, and the number of pixels y of the window W when the aorta A is inscribed in the distal end portion of the window W using the classifier C. Aorta A of the axial image _{D 1} is a diagram showing a window W of parameters (θ, x, y) = a _{(θ 1, x 1, y} 1) when the inscribed to the tip portion of the window W. Shows a search region R ₂ that is set in axial images D _2. The manner in which the search area R ₂ aorta A is detected is a diagram schematically showing. It shows a search region R ₃ that is set on the axial image D _3. It is a diagram showing a detected position of the aorta A from the search region R _3. The rotation angle θ = θ ₁ ~θ ₁₀ representing the position of the detected aorta A slice _SL 1 _{to SL 10} is a diagram schematically showing. It is a figure which shows the example of a misdetection. It shows a search region R ₂ that is set to the slice SL _2. It is a figure which shows the flow of step ST3 in this form. The axial image D ₁ of the selected slices SL ₁ is a diagram schematically showing. It is a figure which shows an example of seed area _| region R _seed . Binary image DA ₁ obtained by binarizing the image D ₁ of the slice SL ₁ in region growing method is a diagram schematically showing. Schematically illustrates the binary image DB ₁ obtained by binarizing the image D ₁ of the slice SL ₁ in the normal threshold processing. It is a figure for demonstrating the process performed by step ST340. It is an explanatory view when performing the step ST30 to the image of the slice SL _3. It is an explanatory view when performing the step ST3 to the image of the slice SL _7. It is a figure which shows the profile showing the relationship between each slice and area ratio R. From among the slices _SL 1 _{to SL 10,} it is an explanatory view of determining the slice which is positioned on the liver side. It shows a search region R ₅ set in the axial image D _5. The manner in which the search area R ₅ aorta A is detected is a diagram schematically showing. It shows a search region R ₆ set in the axial image D _6. It is a diagram showing a detected position of the aorta A from the search region R _6. It shows a search region R ₇ set in the axial image D _7. It is a diagram showing a detected position of the aorta A from the search region R _7. The rotation angle θ = θ ₆ ~θ ₁₀ representing the position of the detected aorta A slice _SL 6 _{to SL 10} schematically illustrates. It is a diagram showing a search area R ₄ set in the axial image D _4. It is a diagram showing a detected position of the aorta A from the search region R _4. It shows a search region R ₃ that is set on the axial image D _3. It is a diagram showing a detected position of the aorta A from the search region R _3. The rotation angle θ = θ ₄ ~θ ₁ representing the detected position of the aorta A slice _SL 4 to SL ₁ is a diagram schematically showing. The center of the aorta A slice _SL 1 _{to SL 10} is a diagram schematically showing. The tracker region R _t is a diagram schematically showing. It is a figure showing roughly MR apparatus 200 of the 2nd form. It is explanatory drawing of the flow when performing step ST3 in a 2nd form. The arm portion has been removed binary image DC ₁ is a diagram schematically showing. A binary image DB ₁ obtained by binarizing the axial image D ₁ of the slice SL _{1 by} normal threshold processing is schematically shown. The arm portion has been removed binary image DD ₁ is a diagram schematically showing. It is a figure for demonstrating the process performed by step ST340. It is a figure which shows the profile showing the relationship between each slice and area ratio R.

  Hereinafter, although the form for inventing is demonstrated, this invention is not limited to the following forms.

(1) First Embodiment FIG. 1 is a schematic view of a magnetic resonance apparatus according to a first embodiment of the present invention.
A magnetic resonance apparatus (hereinafter referred to as “MR apparatus”, MR: Magnetic Resonance) 100 includes a magnet 2, a table 3, a receiving coil 4, and the like.

  The magnet 2 has a bore 21 in which the subject 12 is accommodated. The magnet 2 includes a superconducting coil, a gradient coil, an RF coil, and the like.

  The table 3 has a cradle 3a. The cradle 3a is configured to be able to move into the bore 21. The subject 12 is transported to the bore 21 by the cradle 3a.

  The receiving coil 4 is attached to the subject 12. The receiving coil 4 receives a magnetic resonance signal from the subject 12.

  The MR apparatus 100 further includes a contrast medium injection device 5, a transmitter 6, a gradient magnetic field power source 7, a receiver 8, a control unit 9, an operation unit 10, a display unit 11, and the like.

The contrast agent injection device 5 injects a contrast agent into the subject 12.
The transmitter 6 supplies current to the RF coil, and the gradient magnetic field power source 7 supplies current to the gradient coil.
The receiver 8 performs signal processing such as detection on the signal received from the receiving coil 4.

  The control unit 9 transmits necessary information to the display unit 11 and reconstructs an image based on the data received from the receiver 8 so as to realize various operations of the MR apparatus 100. Control the operation of each part. The control unit 9 includes image creating means 91 to setting means 97 and the like.

The image creating unit 91 creates an axial image of the subject 12 (see, for example, FIG. 5).
The first binarization means 92 binarizes the image using the region expansion method.
The second binarization means 93 binarizes the image using threshold processing.
The calculating means 94 calculates an area ratio R (see, for example, FIG. 26) described later.
The selection means 95 selects a slice serving as a reference when detecting the aorta.
The detection means 96 detects the aorta.
The setting means 97 sets a tracker area for detecting a contrast agent.

  The control unit 9 is an example constituting the image creating means 91 to the setting means 97, and functions as these means by executing a predetermined program. The control unit 9 corresponds to a blood vessel detection device.

The operation unit 10 is operated by an operator and inputs various information to the control unit 9. The display unit 11 displays various information.
The MR apparatus 100 is configured as described above.

FIG. 2 is a diagram showing a scan executed in this embodiment, and FIG. 3 is a diagram schematically showing an imaging region.
In this embodiment, a localizer scan LS and a main scan MS are executed.

Localizer scan LS is a scan for acquiring an image to be used to set the slice position and the tracker region R _t. The tracker region _Rt is a region that is set for detecting a contrast agent flowing through the aorta A.
The main scan MS is executed after the localizer scan LS.

In the scan MS, the contrast agent is injected into the subject, a sequence for detecting the contrast medium from the tracker region R _t is repeatedly executed. Then, a predetermined amount of contrast medium to the tracker region R _t is when injected, imaging sequence for acquiring an image of the liver is performed. Hereinafter, an example of a flow when executing the localizer scan LS and the main scan MS will be described.

FIG. 4 is a diagram illustrating an example of a flow when executing the localizer scan LS and the main scan MS.
In step ST1, a localizer scan LS (see FIG. 2) is executed.

FIG. 5 is an explanatory diagram of the localizer scan LS.
The localizer scan LS is a scan for acquiring an image of a part including the lung and liver of the subject. FIG. 5 shows a slice when the localizer scan LS is executed. In the localizer scan LS, axial, sagittal, and coronal images are acquired. In FIG. 5, for convenience of explanation, m slices (in this case, m = 10 slices SL) for acquiring an axial image are acquired. _{1 to} SL ₁₀ ) are shown. Slice numbers n (n = 1 to 10) are associated with the slices SL _{1 to} SL ₁₀ . The slice number n is set to increase by “1” as the slice position moves from the S side to the I side. The image creating unit 91 (see FIG. 1) creates images D _{1 to} D ₁₀ of the slices SL _{1 to} SL ₁₀ based on the data collected by the localizer scan LS (hereinafter referred to as “axial image”). Called). Since the slices SL _{1 to} SL ₁₀ cross the aorta A, a cross section of the aorta A is depicted in the axial images D _{1 to} D ₁₀ . Comparing the position of the cross section of the aorta A in the axial images D _{1 to} D ₁₀ , it can be seen that the cross section of the aorta A rotates in the counterclockwise direction AN about the spine as it approaches the I side from the S side. .

The sequence used in the localizer scan LS is designed so that the echo signal of blood having a high flow rate is not converged as much as possible. Therefore, the blood signal having a high flow rate can be sufficiently reduced. For example, considering the flow rate of blood flowing through the aorta A, the flow rate of blood flowing through the aorta A is fast, so the blood signal of the aorta A can be sufficiently reduced. In FIG. 5, the aorta A is represented in black.
After creating an axial images _D 1 _{to D 10,} the process proceeds to step ST2.

In step ST2, cerebrospinal fluid CSF is detected from each of the axial images D _{1 to} D ₁₀ . As a method for detecting cerebrospinal fluid CSF, for example, a method described in “Med Imag Tech Vol.31 No.2 March 2013” can be used. After detecting cerebrospinal fluid CSF for each of the axial images D _{1 to} D ₁₀ , the process proceeds to step ST 3.

At step ST3, the detection means 96 (see FIG. 1) detects the aorta A from each of the axial images _D 1 _{to D 10.} In this embodiment, the aorta A is detected using a discriminator for identifying the aorta A. A method for creating a classifier will be described below.

  The discriminator is prepared in advance before imaging the subject. In this embodiment, the discriminator is created by machine learning. Specifically, a classifier C suitable for detecting the aorta A is created by preparing teacher data and learning the teacher data by machine learning (see FIG. 6).

FIG. 6 is a diagram schematically illustrating an example of teacher data used to create the classifier C.
The teacher data v _{1 to} v _p can be created using an abdominal axial image or a chest axial image obtained by photographing a plurality of actual humans.

The teacher data v _{1 to} v _p are defined so that the cross section of the aorta is inscribed on the right side of the rectangular region. The aorta of the teacher data v _{1 to} v _p is shown in black.

By discriminating these teacher data v _{1 to} v _p by machine learning, a discriminator C suitable for detecting an aorta from an axial image is created.

In the present embodiment, the detection unit 96 detects the aorta A from each of the slices SL _{1 to} SL ₁₀ using the discriminator C created as described above. Below, the detection method of the aorta A in this form is demonstrated. In the description of the method for detecting the aorta A in the present embodiment, in order to clarify the effect of the method for detecting the aorta A in the present embodiment, first, the detecting means 96 detects the aorta A by a method different from the present embodiment. A detection method will be described.

7 to 15 are diagrams showing a procedure when the aorta A is detected by a method different from the present embodiment.
First, the axial image D ₁ of the slice SL ₁ located on the high side, sets a search area for defining the search range of the aorta A (see FIG. 7).

FIG. 7 is a diagram showing the search area R ₁ set in the axial image D ₁ .
The detecting means 96 defines a line L0 extending horizontally from the cerebrospinal fluid CSF and a line L1 inclined by θ = θ _{0 with} respect to the line L0 with the cerebrospinal fluid CSF as the center. Region between the line L0 and the line L1 is set as the search region R ₁ that defines the search range of the aorta A. Search area R ₁ is set to include the aorta A. θ ₀ is, for example, θ ₀ = 45 °.

Next, the detection means 96, the search area R _1, sets a window to be used in detecting the aorta A (see FIG. 8).

Figure 8 is a diagram showing a window W set in the search area R _1.
The window W has a pixel size of x × y. The detection means 96 sets a window W on the line L0 with reference to the cerebrospinal fluid CSF. After setting the window W, the detection means 96, searches the region R _1, to change the θ rotation angle of the window W by rotating the window W around the cerebrospinal fluid CSF, further window W size (pixels The number x and the number of pixels y) are also changed. Then, using the discriminator C (see FIG. 6), the rotation angle θ of the window W, the number of pixels x, and the number of pixels y when the aorta A is inscribed in the tip of the window W are obtained (see FIG. 9).

  FIG. 9 is an explanatory diagram for obtaining the rotation angle θ, the number of pixels x, and the number of pixels y of the window W when the aorta A is inscribed in the distal end portion of the window W using the classifier C.

Detection means 96, each time you change one of the parameters of the three parameters of the window W set in the axial image _{D 1 (θ, x, y} ), extracts the pixel data in the window W. The discriminator C outputs a value for determining whether there is a high or low possibility that the aorta A of the axial image D ₁ is inscribed in the distal end portion of the window W based on the extracted data. Detection means 96, based on the output value Out of the classifier C, aorta A of axial images D ₁ to determine whether a high or low probability of being inscribed in the leading end portion of the window W. In this embodiment, the discriminator C is configured such that the output value Out increases as the possibility that the aorta A is inscribed at the tip of the window W increases. Therefore, when the output value Out of the classifier C is maximum, aorta A of axial images D ₁ it can be determined that are most likely to inscribed to the tip portion of the window W. Here, it is assumed that the output value Out of the discriminator C is maximized when the rotation angle θ = θ ₁ , the number of pixels x = x ₁ , and the number of pixels y = y ₁ . Accordingly, when the parameter (θ, x, y) = (θ ₁ , x ₁ , y ₁ ) of the window W, it is determined that the aorta A of the axial image D ₁ is inscribed in the distal end portion of the window W. . FIG. 10 shows the parameters (θ, x, y) = (θ ₁ , x ₁ , y ₁ ) of the window W when the aorta A of the axial image D ₁ is inscribed in the tip of the window W.

The rotation angle θ ₁ and the pixel number y ₁ of the window W represent the position of the aorta A in the slice SL ₁ , and the pixel number x ₁ of the window W represents the inner diameter of the aorta A in the slice SL ₁ . Therefore, the aorta A can be detected from the slice SL ₁ by obtaining the parameter (θ, x, y) = (θ ₁ , x ₁ , y ₁ ) of the window W.

After obtaining the parameter (θ, x, y) = (θ ₁ , x ₁ , y ₁ ) of the window W in the slice SL ₁ , the search range of the aorta A is defined in the axial image D ₂ in the adjacent slice SL ₂ . The search area to be set is set (see FIG. 11).

FIG. 11 is a diagram illustrating the search area R ₂ set in the axial image D ₂ .
Since the slice SL ₂ is located next to the slice SL _1, the position of the aorta A in the slice SL ₂ is believed to be close to the position of the aorta A in the slice SL _1. Therefore, the detection means 96, based on the rotational angle theta ₁ which represents the detected position of the aorta A Axial image D ₁ of the slice SL _1, defines the angular range θ _s2 ≦ θ ≦ θ _e2 including the rotation angle theta ₁ and sets this angle range θ _s2 ≦ θ ≦ θ _e2 as the search region _{R 2.}

Next, the detection means 96 detects the aorta A from the search region R ₂ using the window W. 12, how the shown schematically when it is detected aorta A from the search region R _2. As described above, the detection unit 96 changes the rotation angle θ of the window W and the size (number of pixels x, number y) of the window W, and the aorta A of the axial image D ₃ is located at the tip of the window W. The rotation angle θ of the window W when inscribed and the number of pixels x and y are obtained. From Figure 12, the aorta A, it can be seen that located on the theta = theta ₂ line.

After detecting the aorta A in the slice SL ₂ , the detection unit 96 sets a search region that defines the search range of the aorta A in the axial image D ₃ of the slice SL ₃ adjacent to the slice SL ₂ (see FIG. 13). .

FIG. 13 is a diagram showing the search area R ₃ set in the axial image D ₃ .
Since the slice SL ₃ is located next to the slice SL _2, the position of the aorta A in the slice SL ₃ is believed to be close to the position of the aorta A in the slice SL _2. Therefore, the detection means 96, and the rotation angle theta ₂ which represents the position of the aorta A detected from axial image D ₂ slices SL ₂ as a reference, defines the angular range θ _s3 ≦ θ ≦ θ _e3 including the rotation angle theta ₂ and, the angle range θ _s3 ≦ θ ≦ θ _e3, set as a search area _{R 3.}

Next, the detection means 96 detects the aorta A from the search region R ₃ using the window W. Figure 14 shows the position of the detected aorta A from the search region R _3. Aorta A is located on the theta = theta ₃ lines.

After detecting the aorta A of the slice SL ₃ , a search area is set in the axial image of the adjacent slice by the same method, and the aorta A is detected from the search area using the window W. Therefore, the aorta A in all slices SL _{1 to} SL ₁₀ can be detected. FIG. 15 schematically shows rotation angles θ = θ _{1 to} θ ₁₀ representing the positions of the aorta A detected from the slices SL _{1 to} SL ₁₀ .

As described above, in the method of FIGS. 7 to 15, the aorta A is detected in the order from the slice SL ₁ to the slice SL ₁₀ (in the order of the smaller slice number n). However, since the slice SL ₁ is set at a position crossing the lung, in addition to the aorta A, the lung region is also depicted with a low signal in the axial image D ₁ of the slice SL ₁ . Therefore, when detecting the aorta A Axial image D ₁ of the slice SL _1, the region of the lung that are visualized at low signal similar to the aorta A, it is erroneously detected as the aorta. FIG. 16 shows an example of erroneous detection. In FIG. 16, a partial region of the lung is erroneously detected as an aorta at the rotation angle θ = θ ₁ ′. Therefore, when setting the search area R ₂ to the slice SL ₂ next, based on the rotation angle θ = θ _{1 'of} the detected region erroneously as the aorta in the slice SL _1, setting the search area R ₂ slices SL ₂ is Is done. FIG. 17 shows the search region R ₂ set in the slice SL ₂ . In Figure 17, the aorta A in the slice SL ₂ is not included in the search region _{R 2,} can not be detected aorta A slice SL _2. Therefore, there is a problem that the aorta cannot be detected even in the remaining slices SL _{3 to} SL ₁₀ . As a method of dealing with this problem, it is conceivable to detect the aorta A in the order from the slice SL ₁₀ farthest from the lung toward the slice SL ₁ that crosses the lung (in descending order of the slice number n). However, the slice as in SL _10, the slice close to the lower end of the liver, because another cross-sectional also likely to appear in the arteries and aorta A, starts the detection of the aorta A slice SL _10, the aorta another artery A There is a possibility to detect as. Therefore, there is still a problem that the aorta A cannot be detected correctly. Therefore, in this embodiment, the process of step ST3 is devised so that the aorta of each slice can be detected correctly. Below, the process of step ST3 in this form is demonstrated.

FIG. 18 is a diagram showing a flow of step ST3 in the present embodiment.
In step ST30, an index for determining whether the slice crosses the liver is calculated. Step ST30 includes steps ST310 to ST340. Hereinafter, each step will be described.

In step ST 310, from the axial image _D 1 _{to D 10} of the slice _SL 1 _{to SL 10,} selects one of the axial image. Here, the axial image D ₁ of the slice SL ₁ across the lung has been selected. Figure 19 shows an axial image _{D 1} of the selected slices SL ₁ schematically. The axial image D _1, not only the torso of the subject, left arm and right arm have also been depicted. After selecting the axial image _{D 1} of the slice SL _1, the process proceeds to step ST 320.

In step ST 320, the first binarizing means 92, by using a region growing method, performing the binarization slice SL _1.

First binarization means 92 first sets the seed region in the background region in the axial image D ₁ (outside the region of the subject). FIG. 20 shows an example of the seed region R _seed . As a method of setting the seed region R _seed as the background region, there is a method of using position information of the cradle 3a. Since the cradle 3a is located outside the body of the subject, the seed region R _seed can be reliably set as the background region (external region) by setting the seed region R _seed at the position of the cradle 3a. The operator may manually set the seed region R _seed .

After setting the seed region R _seed , the first binarization unit 92 sets the signal value threshold v used when distinguishing between the outside and inside of the subject based on the signal value of the axial image D _1. Ask.

The space outside the body of the subject includes the cradle 3a and air in the bore. The signal value of the cradle 3a and the signal value of air are sufficiently smaller than the signal value of the body tissue of the subject. Value. Thus, since there is a large difference in signal value between the external and internal tissue, based on the signal value of the axial image D ₁ of the slice SL _1, signal for discriminating the outside and inside of the subject The threshold value v can be obtained. As an example of a method for obtaining the threshold value v, there is a method using a signal value in the seed region R _seed . Since the seed region R _seed is set outside the subject, the signal value of the seed region R _seed has a value (noise) that is sufficiently smaller than the signal value inside the subject. Therefore, by using the signal value of the seed region R _seed , it is possible to obtain the signal value threshold v for distinguishing between the outside and inside of the subject.

Next, the first binarization unit 92 expands the area of the pixel having a signal value smaller than the threshold value v while comparing the signal value of the pixel with the threshold value v, and the axial image D ₁ of the slice SL ₁ is expanded. Binarize. FIG. 21 schematically shows a binary image DA ₁ obtained by binarizing the image D ₁ of the slice SL _{1 by} the region expansion method. The signal values of most pixels outside the body of the subject are smaller than the threshold value v, but the signal values of most pixels on the body surface of the subject are larger than the threshold value v. Therefore, the pixel area extends over the entire external region of the subject, but does not extend inside the in-vivo region of the subject, so the background region (external to the subject) and the in-vivo region of the subject Can be distinguished. In FIG. 21, pixels in the expanded area are represented by a logical value 0 (black), and pixels in other areas are represented by a logical value 1 (white).

  When binarization is executed using the region expansion method, a logical value 0 is assigned to the background region (outside the subject), and a logical value 1 is assigned to the subject's body. Since the logical value 1 is assigned to the inside of the subject, the logical value 1 is also assigned to the lung.

After obtaining the binary image DA ₁ of the slice SL ₁ using the region expansion method, the process proceeds to step ST330.

In step ST330, the second binarizing means 93, without using the region growing method, using conventional thresholding, executes binarization axial image _{D 1} of the slice SL _1. Specifically, in the axial image D ₁ of the slice SL _1, assigned the logic value 0 the signal value of the pixel is the threshold value TH smaller area, whereas the logical value signal value of the pixel is the threshold value TH larger region 1 by assigning, binarizes the axial image _{D 1} of the slice SL _1. FIG. 22 schematically shows a binary image DB ₁ obtained by binarizing the axial image D ₁ of the slice SL ₁ by normal threshold processing. As the threshold value TH, for example, the threshold value v obtained when performing binarization using the region expansion method in step ST320 can be used.

As described above, there is a large difference in signal value between the body tissue and air, and the signal value of the body tissue increases, but the signal value of air decreases. Therefore, most of the pixels located in the body tissue have a signal value larger than the threshold value TH, but most of the pixels located in the space occupied by air have a signal value smaller than the threshold value TH. Since such a signal value difference occurs, the binary image DB ₁ obtained by the normal threshold processing is assigned a logical value 1 to most of the body tissue of the subject, but the background region (subject The logical value 0 is assigned to most of the inside of the lungs and the inside of the lungs (space containing air). After performing the binarization process, the process proceeds to step ST340.

FIG. 23 is a diagram for explaining the process executed in step ST340.
In step ST340, calculating means 94 (see FIG. 1), the area A of the region logical value 1 in the binary image DA ₁ is assigned the logical value 1 in the binary image DB ₁ is assigned The area B of the area is calculated. After obtaining the values of the areas A and B, the calculating means 94 calculates the area ratio R = B / A. The area ratio R is used as an index for determining whether the slice SL ₁ crosses the liver. A specific method for determining whether or not the slice crosses the liver using the area ratio R will be described later.

As shown in FIG. 23, when binarization using the region expansion method is executed, a logical value 1 can be assigned to the lung. On the other hand, when binarization using threshold processing is executed, a logical value 1 is assigned to the body tissue of the subject, but a logical value 0 is assigned to the inside of the lung. Therefore, when comparing the two binary images DA ₁ and DB _1, the area B of the area the logical value 1 in the binary image DB ₁ is assigned the logical value 1 in the binary image DA ₁ is allocated The value is smaller than the area A of the region. Therefore, the area ratio R in the slice SL ₁ becomes considerably smaller than 1.

In this way, step ST30 is executed. In FIG. 19 to 23 have been described the example in which the performance of step ST30 against axial image _{D 1} of the slice SL _1, against axial image of another slice _SL 2 _{to SL 10,} sliced Like the axial image _{D 1} of the SL _1, it executes step ST30. 24 and 25 show an example in which step ST30 is performed on images of other slices. Hereinafter, FIGS. 24 and 25 will be described in order.

Figure 24 is an explanatory view when performing the step ST30 to the image of the slice SL _3.
First, in step ST 310, it selects an axial image _{D 3} slices SL _3. Slice SL ₃ is located in the vicinity of the boundary between the lungs and liver, it traverses both organs the lungs and the liver. After selecting the axial image _{D 3} slices SL _3, the process proceeds to step ST 320.

In step ST 320, the first binarizing means 92 sets a seed region in the background region in the axial image D ₃ (extracorporeal area of the subject), to perform the binarization using a region growing method. Thus, the binary image DA ₃ slices SL ₃ is obtained. When binarization is executed using the region expansion method, a logical value 0 is assigned to the background region (outside the subject), and a logical value 1 is assigned to the subject's body. Since the inside of the subject is represented by a logical value 1, a logical value 1 is assigned to both the lung and the liver.
After obtaining the binary image DA ₃ slices SL ₃ using region growing method, the process proceeds to step ST330.

In step ST330, the second binarizing means 93, without using the region growing method, using conventional thresholding, executes binarization axial image _{D 3} slices SL _3. Thus, the binary image DB ₃ slices SL ₃ is obtained. In the binary image DB ₃ of the slice SL ₃ obtained by the normal threshold processing, a logical value 0 is assigned to most of the background region (external to the subject) and the inside of the lung (space containing air). However, a logical value of 1 is assigned to the majority of the liver. Thus, the binary image DB ₃ slices SL _3, compared with the binary image DB ₁ slice SL ₁ (see FIG. 23), the area of the region where the logical value 1 is allocated is widened. After performing the binarization process, the process proceeds to step ST340.

In step ST340, the calculation means 94 uses the area A of the binary image DA ₃ to which the logical value 1 is assigned and the area of the binary image DB ₃ to which the logical value 1 is assigned. B is calculated. After obtaining the values of the areas A and B, the calculating means 94 calculates the area ratio R = B / A.

When the binary images DA ₃ and DB ₃ are compared, the area B of the binary image DB _{3 to which} the logical value 1 is assigned is the area B of the region to which the logical value 1 of the binary image DA ₃ is assigned. Is a smaller value. Therefore, the area ratio R in the slice SL ₃ becomes a value smaller than 1.

In this manner, the step ST3 is performed on the axial image slices SL _3. Next, the case where performing step ST3 against axial image slices SL _7.

Figure 25 is an explanatory view when performing the step ST3 to the image of the slice SL _7.
First, in step ST 310, it selects an axial image _{D 7} of the slice SL _7. Slice SL ₇ is not across the lung, are across the liver. After selecting the axial image _{D 7} of the slice SL _7, the process proceeds to step ST 320.

In step ST 320, the first binarizing means 92 sets a seed region in the background region in the axial image D ₇ (extracorporeal area of the subject), to perform the binarization using a region growing method. Thus, the binary image DA ₇ slices SL ₇ is obtained. When binarization is executed using the region expansion method, a logical value 0 is assigned to the background region (outside the subject), and a logical value 1 is assigned to the subject's body. Since the inside of the subject is represented by a logical value 1, a logical value 1 is assigned to the liver.
After obtaining the binary image DA ₇ of the slice SL ₇ using the region expansion method, the process proceeds to step ST330.

In step ST330, the second binarizing means 93, without using the region growing method, using conventional thresholding, executes binarization axial image _{D 7} of the slice SL _7. Thus, the binary image DB ₇ of slices SL ₇ is obtained. Since the slice SL ₇ does not cross the lung but crosses the liver, the binary image DB ₇ is assigned a logical value 1 to most of the region in the body of the subject. After performing the binarization process, the process proceeds to step ST340.

In step ST340, the calculation means 94 uses the area A of the area assigned the logical value 1 in the binary image DA _{7 and} the area of the area assigned the logical value 1 in the binary image DB _7. B is calculated. After obtaining the values of the areas A and B, the calculating means 94 calculates the area ratio R = B / A.

Comparing the binary images DA ₇ and DB ₇ , the area B of the binary image DB _{7 to which} the logical value 1 is assigned is the area B of the region to which the logical value 1 of the binary image DA ₇ is assigned. Is almost the same. Therefore, the area ratio R in the slice SL ₇ becomes a value close to 1.

  As described above, the calculation unit 94 calculates the area ratio R for the image of each slice. Therefore, a profile representing the relationship between each slice and the area ratio R can be obtained. FIG. 26 shows a profile representing the relationship between each slice and the area ratio R.

Referring to the profile of FIG. 26, it can be seen that the area ratio R varies depending on whether the slice is located on the lung side or the liver side. The following, among slices _SL 1 _{to SL 10,} representatively, three slices _SL 1, SL _3, and taken up SL _7, three slices _SL 1, SL _3, and the difference between the area ratio R of SL ₇ explain. For convenience of explanation, the area ratio R of the slice SL ₁ that crosses the lung and the area ratio R of the slice SL ₇ that crosses the liver will be described first, and finally, the area of the slice SL ₃ that crosses the boundary between the liver and the lung. The ratio R will be described.

(1) In the area ratio R slice SL ₁ for the slice SL ₁ across the lungs, the binary lung region of the image DA ₁ logic value 1 is assigned, but the logic to the area of the binary image DB ₁ lung The value 0 is assigned. Accordingly, since the area B of the logical value 1 of the binary image DB ₁ is considerably smaller than the area A of the logical value 1 of the binary image DA ₁ , the area ratio R in the slice SL ₁ becomes a value considerably smaller than 1. .

(2) the slice SL ₇ for the area ratio R of the slice SL ₇ across the liver, in the region of the liver of the binary image DA ₇ logic value 1 is assigned the logical value in the region of the liver of the binary image DB ₇ 1 is assigned. Accordingly, since the area B of the logical value 1 of the binary image DB ₇ is substantially equal to the area A of the logical value 1 of the binary image DA ₇ , the area ratio R in the slice SL ₇ becomes a value close to 1.

(3) In the liver and the binary image DA area ratio R slice SL ₃ for slices SL ₃ crossing the boundary between the lung _3, the logical value 1 is assigned to the region of the liver, it has a logic value of 1 in the lung region Assigned. On the other hand, the binary image DB ₃ slices SL _3, but in the region of the liver are assigned the logic value 1, a logic value 0 is assigned to the lung region. Therefore, the area ratio R in the slice SL ₃ is larger than the area ratio R in the slice SL _1, it is smaller than the area ratio R in the slice SL _7.

Therefore, it can be seen that the closer the area ratio R is to 1, the higher the possibility that the slice will cross the liver, and the smaller the area ratio R, the lower the possibility that the slice will cross the liver.
After obtaining the area ratio R of the slices SL _{1 to} SL ₁₀ , the process proceeds to step ST31.

In step ST31, the selection unit 95 selects a slice SL _j (slice number n = j) serving as a reference when detecting the aorta A from the slices SL _{1 to} SL ₁₀ based on the area ratio R. A method for selecting the slice SL _j will be described below.

First, the selection unit 95 obtains a slice located on the liver side from the slices SL _{1 to} SL ₁₀ (see FIG. 27).

FIG. 27 is an explanatory diagram for obtaining a slice located on the liver side from the slices SL _{1 to} SL ₁₀ .

The selection means 95 obtains a difference ΔR between the maximum value max of the area ratio R and the minimum value min of the area ratio R. Then, based on ΔR, a threshold value TH for determining whether or not the area ratio R is close to 1 is set. The threshold value TH can be expressed by the following equation.
TH = min + k · ΔR (1)
Where min: minimum value of the area ratio R
k: coefficient

k is, for example, k = 0.9. In this embodiment, a slice having an area ratio R equal to or greater than the threshold value TH is obtained as a slice located on the liver side. In FIG. 27, slices SL _{5 to} SL ₁₀ are obtained as slices located on the liver side. As described above, it is considered that the closer the slice area ratio R is to 1, the higher the possibility that the slice is located on the liver side. Therefore, the slices SL _{5 to} SL having the area ratio R equal to or greater than the threshold TH. ₁₀ can be considered crossing the liver.

After obtaining the slices SL _{5 to} SL ₁₀ , the selection unit 95 selects one slice from the slices SL _{5 to} SL ₁₀ as a slice SL _j that serves as a reference when detecting the aorta. Here, j = 5, that is, the slice SL ₅ (slice number n = 5) located closest to the lung among the slices SL _{5 to} SL ₁₀ serves as a reference for detecting the aorta A SL _j Selected as. Since the slice SL ₅ has an area ratio R close to 1, it can be considered that the slice SL ₅ hardly crosses the lung. Therefore, around the aorta A in the slice SL _5, it believed lungs low signal area caused due does not exist. After you select a slice SL _5, the process proceeds to step ST32.

In step ST32, the detection unit 96 detects the aorta A slice SL ₅ selected in step ST31.

28 and 29 are explanatory view when detecting the aorta A slice SL _5.
First, the detection unit 96 sets a search area that defines a search range in the axial image D ₅ of the slice SL ₅ . Figure 28 shows a search region _{R 5} set in the axial image _{D 5.}

The detecting means 96 defines a line L0 extending horizontally from the cerebrospinal fluid CSF and a line L1 inclined by θ = θ _{0 with} respect to the line L0 with the cerebrospinal fluid CSF as the center. Region between the line L0 and the line L1 is set as the search area R ₅ defining the search range of the aorta A. Search area R ₅ is set to include the aorta A. θ ₀ is, for example, θ ₀ = 90 °.

Next, the detection means 96 detects the aorta A from the search region R ₅ using the window W. Figure 29 schematically illustrates a state in which the aorta A is detected from the search region R _5. As described above, the detection means 96 changes the rotation angle θ of the window W and the size (number of pixels x, number y) of the window W, and the aorta A of the axial image D ₅ is placed at the tip of the window W. The rotation angle θ of the window W when inscribed and the number of pixels x and y are obtained. From Figure 29, aorta A is found to be located on the line of θ = θ _5.
After detecting the aorta A slice SL _5, it proceeds to step ST33.

In step ST33, the detection unit 96 detects the aorta A slice _SL 6 _{to SL 10} arranged in a direction slice number n increases with respect to the slice SL _5.

FIGS. 30 34 is an explanatory diagram of when detecting the aorta A slice _SL 6 _{to SL 10.}
In step ST33, the detection unit 96, first, among the slice _SL 6 _{to SL 10,} detects the aorta A slice SL ₆ adjacent to the slice SL ₅ (see FIGS. 30 and 31).

30 and 31 are explanatory diagrams when the aorta A of the slice SL ₆ is detected.
Detection means 96, first, the axial image D ₆ slices SL _6, sets a search area for defining the search range. FIG. 30 shows the search region R ₆ set in the axial image D ₆ .

Since the slice SL ₆ is located next to the slice SL ₅ , the position of the aorta A in the slice SL ₆ is considered to be close to the position of the aorta A in the slice SL ₅ . Therefore, the detection means 96, based on the rotational angle theta ₅ representing the position of the aorta A detected from axial image D ₅ slices SL _5, defines the angular range θ _s6 ≦ θ ≦ θ _e6 comprising a rotation angle theta ₅ and sets this angle range θ _s6 ≦ θ ≦ θ _e6 as a search area _{R 6.}

As described above, the cross section of the aorta A rotates in the counterclockwise direction AN around the spine as it approaches the I side from the S side. Therefore, aorta A in the axial images D _6, so likely to be located in the counterclockwise direction AN respect _{θ = θ 5, θ s6 ≦} θ ≦ θ 5 angular range of the search area R _{6 May be} narrowed and the angle range of θ ₅ ≦ θ ≦ θ _e6 may be set wide. As described above, by setting the search region R ₆ based on the traveling direction of the aorta A, it is possible to exclude the region where the aorta A is unlikely to exist from the search region R ₆ .

After setting the search region R ₆ , the detection means 96 detects the aorta A from the search region R ₆ using the window W. Figure 31 shows the position of the detected aorta A from the search region R _6. Aorta A is located on the line of θ = θ _6.

After detecting the aorta A in the slice SL ₆ , the detection unit 96 searches the aorta A in the axial image D ₇ of the slice SL ₇ adjacent to the slice SL ₆ in the I direction (direction in which the slice number n increases). A search area for defining the range is set (see FIG. 32).

FIG. 32 is a diagram illustrating the search region R ₇ set in the axial image D ₇ .

Since the slice SL ₇ is located next to the slice SL ₆ , the position of the aorta A in the slice SL ₇ is considered to be close to the position of the aorta A in the slice SL ₆ . Therefore, the detection means 96, and the rotation angle theta ₆ representing the position of the aorta A detected from axial images D ₆ slices SL ₆ as a reference, defines the angular range θ _s7 ≦ θ ≦ θ _e7 including the rotation angle theta ₆ and sets this angle range θ _s7 ≦ θ ≦ θ _e7 as a search area _{R 7.} In consideration of the traveling direction of the aorta A, the angle range of θ _s7 ≦ θ ≦ θ _{6 in} the search region R ₇ may be narrowed and the angle range of θ ₆ ≦ θ ≦ θ _e7 may be set wide. .

After setting the search region R ₇ , the detection means 96 detects the aorta A from the search region R ₇ using the window W. Figure 33 shows the position of the detected aorta A from the search region R _7. Aorta A is located on the line of θ = θ _7.

After detecting the aorta A of the slice SL _7, the aorta A of the slices SL ₈ , SL ₉ , and SL ₁₀ is detected in the same manner. Therefore, it is possible to detect the aorta A slice _SL 6 _{to SL 10} to slice number n are arranged in the large direction relative slice SL _5. FIG. 34 schematically shows rotation angles θ = θ _{6 to} θ ₁₀ representing the position of the aorta A detected from the slices SL _{6 to} SL ₁₀ .

When detecting the aorta A of the slice SL ₈ , the detection unit 96 detects the search region R in the axial image D ₈ of the slice SL ₈ based on the rotation angle θ = θ ₇ representing the position of the aorta detected from the slice SL _7. set _8, it detects the aorta a from the search region R _8. In Figure 34, the aorta A slice SL ₈ is located on the line of θ = θ _8. After detecting the aorta A of the slice SL ₈ , the search area R ₉ is set in the axial image D ₉ of the adjacent slice SL ₉ based on θ = θ ₈ , and the aorta A is detected from the search area R _9. . In FIG. 34, the aorta A of the slice SL ₉ is located on the line θ = θ ₉ . After detecting the aorta A of the slice SL ₉ , the search area R ₁₀ is set in the axial image D ₁₀ of the adjacent slice SL ₁₀ based on θ = θ ₉ , and the aorta A is detected from the search area R _10. . In this way, the aorta A of the slices SL _{6 to} SL ₁₀ can be detected.
After detecting the aorta A slice _SL 6 _{to SL 10,} the process proceeds to step ST34.

In step ST34, the detection unit 96 detects the aorta A slice _SL 4 to SL ₁ arranged in a direction slice number n is smaller than the slice SL _5.

35 to 39 are explanatory diagrams when the aorta A of the slices SL _{4 to} SL ₁ is detected.
In step ST34, the detection unit 96, _first, among the SL 4 to SL _1, detects the aorta A slice SL4 adjacent to the slice SL ₅ (see FIGS. 35 and 36).

Figures 35 and 36 are explanatory view when detecting the aorta A slice SL _4.

First, the detection unit 96 sets a search area that defines a search range in the axial image D ₄ of the slice SL ₄ . FIG. 35 shows the search region R ₄ set in the axial image D ₄ .

Since the slice SL ₄ is located next to the slice SL ₅ , the position of the aorta A in the slice SL ₄ is considered to be close to the position of the aorta A in the slice SL ₅ . Therefore, the detection means 96, based on the rotational angle theta ₅ representing the position of the aorta A detected from axial image D ₅ slices SL _5, defines the angular range θ _s4 ≧ θ ≧ θ _e4 including a rotation angle theta ₅ and sets this angle range θ _s4 ≧ θ ≧ θ _e4 as the search region _{R 4.}

From FIG. 35, the cross section of the aorta A rotates clockwise about the spine as it approaches the S side from the I side. Therefore, since the aorta A in the axial image D ₄ is likely to be positioned in the clockwise direction CR with respect to θ = θ ₅ , the angle range of θ _s4 ≧ θ ≧ θ _{5 in} the search region R ₄ is The angle range of θ ₅ ≧ θ ≧ θ _e4 may be set wide. As described above, by setting the search region R ₄ based on the traveling direction of the aorta A, it is possible to exclude the region where the aorta A is unlikely to exist from the search region R ₄ .

After setting the search region R ₄ , the detection means 96 detects the aorta A from the search region R ₄ using the window W. Figure 36 shows the position of the detected aorta A from the search region R _4. Aorta A is located on the line of θ = θ _4.

After detecting the aorta A in the slice SL ₄ , the detecting unit 96 searches the aorta A in the axial image D ₃ of the slice SL ₃ adjacent to the slice SL ₄ in the S direction (direction in which the slice number n decreases). A search area that defines the range is set (see FIG. 37).

FIG. 37 is a diagram showing the search area R ₃ set in the axial image D ₃ .
Since the slice SL ₃ is located next to the slice SL _4, the position of the aorta A in the slice SL ₃ is believed to be close to the position of the aorta A in the slice SL _4. Therefore, the detection means 96, based on the rotational angle theta ₄ representing the position of the aorta A detected from axial image D ₄ slices SL _4, defines the angular range θ _s3 ≧ θ ≧ θ _e3 including the rotation angle theta ₄ and sets this angle range θ _s3 ≧ θ ≧ θ _e3 as the search region _{R 3.} In consideration of the traveling direction of the aorta A, the angle range of θ _s3 ≧ θ ≧ θ _{4 in} the search region R ₃ may be narrowed and the angle range of θ ₄ ≧ θ ≧ θ _e3 may be set wide. .

After setting the search area R ₃ , the detection means 96 detects the aorta A from the search area R ₇ using the window W. Figure 38 shows the position of the detected aorta A from the search region R _3. Aorta A is located on the theta = theta ₃ lines.

After detecting the aorta A slice SL _3, in the same manner, it detects the aorta A slice SL ₂ and SL _1. Therefore, it is possible to detect the aorta A of the slices SL _{4 to} SL ₁ arranged in the direction in which the slice number n is smaller than the slice SL ₅ . FIG. 39 schematically shows rotation angles θ = θ _{4 to} θ ₁ representing the position of the aorta A detected from the slices SL _{4 to} SL ₁ .

When detecting the aorta A of the slice SL ₂ , the detecting unit 96 detects the search region R in the axial image D ₂ of the slice SL ₂ based on the rotation angle θ = θ ₃ representing the position of the aorta detected from the slice SL _{3. 2} set to detect the aorta a from the search region R _2. In Figure 39, the aorta A slice SL ₂ is located on the theta = theta ₂ line. After detecting the aorta A of the slice SL ₂ , the search area R ₁ is set in the axial image D ₁ of the adjacent slice SL ₁ based on θ = θ ₂ , and the aorta A is detected from the search area R _1. . In this way it is possible to detect the aorta A slice _SL 4 to SL _1.

Therefore, the aorta A of all the slices SL _{1 to} SL ₁₀ can be detected by executing steps ST32 to ST34. After detecting the aorta A of all the slices SL _{1 to} SL ₁₀ , the process proceeds to step ST4 (see FIG. 4).

In step ST4, the detection unit 96 determines the center of the detected aorta A slice _SL 1 _{to SL 10.} FIG. 40 schematically shows the center of the aorta A of the slices SL _{1 to} SL ₁₀ . FIG. 40 shows only the center of the aorta A of the slices SL ₁ , SL ₂ , SL ₃ , and SL ₁₀ due to space limitations. Since the cross-sectional shape of the aorta A can be regarded as a substantially circular shape, the center of the aorta A can be obtained by considering a circle inscribed in the window W and detecting the center of the circle. After obtaining the center of the aorta A, the process proceeds to step ST5.

In step ST5, the setting means 97 (see FIG. 1) sets a tracker region for detecting a contrast agent based on the detected blood vessel center. Figure 41 shows a set tracker region R _t schematically. The tracker region R _t is set so as to be located inside the aorta A. After setting the tracker region _{R t,} the process proceeds to step ST6.

In step ST6, the main scan MS (see FIG. 2) is executed. In the scan MS, the contrast agent is injected into the subject, a sequence for detecting the contrast medium from the tracker region R _t is repeatedly executed. Then, when a predetermined amount of contrast medium was injected into the tracker region R _t, scan to acquire images of the liver is performed, the flow is terminated.

In this embodiment, a profile representing the relationship between the slices SL _{1 to} SL ₁₀ and the area ratio R is obtained (see FIG. 27), and a slice SL _{5 serving} as a reference when detecting the aorta A is selected based on this profile. . Of the slices _SL 1 _{to SL 10,} it is first detected aortic slice SL ₅ selected. Since the area ratio R of the slice SL ₅ is a value close to 1, on the periphery of the aorta A in the slice SL _5, believed lungs low signal area caused due does not exist. Therefore, when detecting the aortic slice SL _5, lungs a low signal area caused due to not erroneously detected as the aorta, it is possible to correctly detect the aorta. In this embodiment, after detecting the aortic slice SL _5, with reference to the position of the aortic slice SL _5, sets the search area and the search area of the slice _SL 4 to SL ₁ slice _SL 6 _{to SL 10.} Therefore, as the slice SL _1, even slices across the lungs, a low signal area by pulmonary can be prevented from being erroneously detected as the aorta.

In this embodiment, after detecting the aortic slice _SL 6 _{to SL 10} (step ST33), and detects the aortic slice _SL 4 to SL ₁ (step ST34). However, a step ST33 for detecting the aortic slice _SL 6 _{to SL 10,} and a step ST34 for detecting aortic slice _SL 4 to SL ₁ may be executed in parallel. By executing steps ST33 and ST34 in parallel, the time required to detect the aorta in the slices SL _{1 to} SL ₁₀ can be shortened.

In this embodiment, the slice SL ₅ of the slice _SL 5 _{to SL 10,} is selected as a slice SL _j as a reference in detecting the aorta. However, from the slice _SL 5 _{to SL 10,} another slice from the slice SL ₅ (e.g., a slice SL _4), and may be selected as the reference and the slice SL _j in detecting the aorta. However, the slice as in SL _10, the slice close to the lower end of the liver, because another cross-sectional also likely to appear in the arteries and aorta A, starts the detection of the aorta A slice SL _10, the aorta another artery A There is a possibility to detect as. Therefore, when selecting a slice SL _{j serving} as a reference for detecting the aorta from the slices SL _{5 to} SL ₁₀ , it is desirable to select a slice at a position away from the lower end of the liver (slice close to the lung). .

  In this embodiment, the window W is set based on the cerebrospinal fluid CSF. However, the window W may be set based on a part other than the cerebrospinal fluid CSF. Moreover, although the aorta A is detected in this embodiment, the present invention can also be applied to the case of detecting a blood vessel different from the aorta A.

  In this embodiment, the aorta is detected based on the axial image, but the aorta is determined based on an image of a surface different from the axial surface (for example, an oblique surface that obliquely intersects the axial surface). Also good.

  In this embodiment, the window W has a rectangular shape, but may have another shape (for example, an elliptical shape).

(2) Second Mode In the first mode, step ST3 is executed according to the flow shown in FIG. 18, but in the second mode, step ST3 is executed according to a flow different from the flow shown in FIG. An example will be described.

FIG. 42 is a diagram schematically showing an MR apparatus 200 according to the second embodiment.
The MR apparatus 200 according to the second embodiment is different from the MR apparatus 100 according to the first embodiment in that the MR apparatus 200 according to the second embodiment includes a deleting unit 98 for deleting the arm from the binary image. Is the same as the MR apparatus 100 of the first embodiment. Therefore, in the description of the MR apparatus 200 of the second embodiment, the deleting unit 98 will be mainly described.
The control unit 9 is an example constituting the image creating unit 91 to the deleting unit 98, and functions as these units by executing a predetermined program.

FIG. 43 is an explanatory diagram of a flow when step ST3 is executed in the second embodiment.
Steps ST310 and ST320 are the same as those in the first embodiment, and a description thereof will be omitted.

By step ST320, a binary image DA ₁ (see FIG. 21) is obtained. Note that the binary image DA _1, not only the torso of the subject, left arm and right arm have also been depicted. In the second mode, a process of deleting the left arm part and the right arm part is performed. In order to execute this process, the process proceeds to step ST321.

In step ST321, deletion means 98, from the binary image DA _1, executes a process of deleting the arms (see Fig. 44).

Figure 44 is a binary image DC ₁ that arm portion has been deleted is a view schematically showing.
In step ST321, from the binary image DA _1, executes image processing for deleting the left arm and the right arm. Thus, it is possible to obtain the arm portion is deleted binary image DC _1. As a method of deleting the arm, for example, as described in “Med Imag Tech Vol.31 No.2 March 2013”, a method of executing reduction processing, region expansion processing, and expansion processing is used. Can do. After the arm portion to obtain a binary image DC ₁ that have been removed, the process proceeds to step ST330.

In step ST330, the second binarizing means 93, without using the region growing method, using conventional thresholding, executes binarization axial image _{D 1} of the slice SL _1. This binarization is the same as the binarization executed in step ST330 of the first form. FIG. 45 schematically shows a binary image DB ₁ obtained by binarizing the axial image D ₁ of the slice SL ₁ by normal threshold processing. After obtaining the binary image DB ₁ , the process proceeds to step ST331.

In step ST331, the deletion unit 98 performs image processing for deleting the left arm portion and the right arm portion on the binary image DB ₁ . The image processing executed in step ST331 can use a method of executing reduction processing, region expansion processing, and enlargement processing, as in step ST321. Since the logical value 0 is assigned to the lung in the binary image DB ₁ , when the reduction process, the area expansion process, and the enlargement process are executed, not only the arm part is deleted but also the body part area becomes narrow. FIG. 46 schematically shows a binary image DD ₁ from which the arm portion is deleted. After deleting the arm, the process proceeds to step ST340 (see FIG. 47).

FIG. 47 is a diagram for explaining the process executed in step ST340.
In step ST340, the calculation means 94 uses the area A of the binary image DC ₁ to which the logical value 1 is assigned and the area of the binary image DD ₁ to which the logical value 1 is assigned. B is calculated. After obtaining the values of the areas A and B, the calculating means 94 calculates the area ratio R = B / A.

  As shown in FIG. 47, since the area B is considerably smaller than the area A, it can be seen that the area ratio R is considerably smaller than 1.

In this way, step ST30 is executed. Note that the description will be given of an example of executing in FIGS. 44 to 47, the step ST30 against axial image _{D 1} of the slice SL _1. However, for each of the image of another slice _SL 2 _{to SL 10,} similarly to the axial image _{D 1} of the slice SL _1, and executes step ST30, determining the area ratio R. Therefore, a profile representing the relationship between each slice and the area ratio R can be obtained. FIG. 48 shows a profile representing the relationship between each slice and the area ratio R. After executing step ST30 on the axial image of each slice, the process proceeds to step ST31.
Since steps ST31 to ST34 are the same as those in the first embodiment, the description thereof is omitted.

  In the second embodiment, the area ratio R is obtained using a binary image from which the arm portion is deleted. The area ratio R is close to 1 on the liver side and close to 0 on the lung side. Therefore, since the difference between the area ratio on the liver side and the area ratio on the lung side is larger than that in the first embodiment (see FIG. 26), the detection accuracy of the boundary between the lung and the liver can be further improved. It becomes possible.

In the first and second embodiments, the area ratio R = A / B is used as an index for determining whether or not the slice crosses the liver. However, an index different from the area ratio R may be used as long as the slice SL _{j serving} as a reference when detecting the aorta A can be obtained. For example, instead of the area ratio R, the area B may be used as an index for determining whether the slice crosses the liver.

2 Magnet 3 Table 3a Cradle 4 Receiving coil 5 Contrast medium injection device 6 Transmitter 7 Gradient magnetic field power supply 8 Receiver 9 Control unit 10 Operation unit 11 Display unit 12 Subject 21 Bore 91 Image creation means 92 First binarization means 93 Second binarization means 94 Calculation means 95 Selection means 96 Detection means 97 Setting means 98 Deletion means

Claims (18)

Image creating means for creating an image of m slices crossing the imaging region including the first region, the second region, and the blood vessel;
Means for determining an index for determining whether the slice crosses the first region based on the image of the slice;
Selection means for selecting a first slice across the first region from the m slices based on the index;
Detecting means for detecting the blood vessel in the first slice, the detecting means detecting a blood vessel in the second slice crossing the second part based on the position of the blood vessel in the first slice; ,
A blood vessel detection device comprising: The selection means includes
Obtaining two or more slices crossing the first region from the m slices based on the index, and selecting the first slice from the two or more slices; Item 4. The blood vessel detection device according to Item 1. The selection means includes
The blood vessel detection device according to claim 2, wherein a slice closest to the second part is selected as the first slice among the two or more slices. The detection means includes
The blood vessel detection device according to any one of claims 1 to 3, wherein a search region for the blood vessel is set for each slice, and the blood vessel is detected from the search region. Each of the m slices is associated with slice numbers n = 1, 2, 3,.
The slice number n of the first slice is n = j (1 <j <m),
The detection means includes
A first step of detecting the blood vessel in the first slice;
A second step of detecting the blood vessels in slices with slice numbers n = j + 1 to m arranged in a direction in which the slice number increases with respect to the first slice;
A third step of detecting the blood vessels in the slices of slice numbers n = j−1 to 1 arranged in a direction in which the slice number decreases with respect to the first slice;
The blood vessel detection device according to claim 4, wherein: The detection means includes
In the second step,
Based on the position of the blood vessel in the slice of slice number n (j ≦ n ≦ m−1), a first search area is set in the slice of slice number n + 1 (j ≦ n ≦ m−1),
In the third step,
6. The second search region is set in a slice having a slice number n−1 (j ≧ n ≧ 2) based on a position of the blood vessel in a slice having a slice number n (j ≧ n ≧ 2). Blood vessel detection device. The detection means includes
In the second step,
Based on the first angle representing the position of the blood vessel in the slice with slice number n (j ≦ n ≦ m−1), the first search region is set,
In the third step,
The blood vessel detection device according to claim 6, wherein the second search region is set based on a second angle representing the position of the blood vessel in a slice having a slice number n (j ≧ n ≧ 2). The detection means includes
The blood vessel detection device according to any one of claims 5 to 7, wherein the second step and the third step are executed in parallel. The means for obtaining the index is:
A first binarizing image of each of the m slices such that the first and second portions have a first logical value and the background region has a second logical value Binarization means of
The images of each of the m slices are binarized so that the first part has the first logical value and the second part and the background region have the second logical value. Second binarizing means for
Based on the feature quantity of the first binary image obtained by the first binarization means and the feature quantity of the second binary image obtained by the second binarization means, the A calculation means for calculating an index;
The blood vessel detection device according to claim 1, comprising: The feature amount of the first binary image is an area of the first logical value in the first binary image,
The blood vessel detection device according to claim 9, wherein the feature amount of the second binary image is an area of the first logical value in the second binary image. The calculating means includes
An area of the first logical value in the first binary image and an area of the first logical value in the second binary image are obtained, and the index is calculated based on a ratio of these areas. The blood vessel detection device according to claim 10 for calculation. The means for obtaining the index is:
A first binarizing image of each of the m slices such that the first and second portions have a first logical value and the background region has a second logical value Binarization means of
The images of each of the m slices are binarized so that the first part has the first logical value and the second part and the background region have the second logical value. Second binarizing means for
Deletion means for executing a first deletion process for deleting a predetermined part from the first binary image and a second deletion process for deleting the predetermined part from the second binary image;
Calculating means for calculating the index based on a feature amount of the first binary image after the first deletion process and a feature amount of the second binary image after the second deletion process;
The blood vessel detection device according to claim 1, comprising: The feature amount of the first binary image after the first deletion process is an area of the first logical value in the first binary image after the first deletion process,
The feature quantity of the second binary image after the second deletion process is an area of the first logical value in the second binary image after the second deletion process. Blood vessel detection device. The calculating means includes
The area of the first logical value in the first binary image after the first deletion process and the area of the first logical value in the second binary image after the second deletion process The blood vessel detection device according to claim 13, wherein the index is calculated based on a ratio of these areas. The first binarization means is binarization using a region expansion method,
The blood vessel detection device according to any one of claims 9 to 14, wherein the second binarization means is binarization using threshold processing.   The blood vessel detection device according to any one of claims 1 to 15, wherein the first part includes a liver and the second part includes a lung.   A medical device comprising the blood vessel detection device according to any one of claims 1 to 16. An image creation process for creating an image of m slices crossing the imaging region including the first region, the second region, and the blood vessel;
Processing for obtaining an index for determining whether or not the slice crosses the first region based on the image of the slice;
A selection process for selecting a first slice across the first region from the m slices based on the index;
A detection process for detecting the blood vessel in the first slice, the detection process for detecting a blood vessel in the second slice crossing the second part based on the position of the blood vessel in the first slice; ,
A program to make a computer execute.