JP2018000374A - Image processor and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a technique capable of improving image interpretation accuracy regardless of an image reader. An image processing apparatus has an alignment means for aligning 3D images I1, I2, and I3. Further, the image processing device obtains the voxel value corresponding to the part Q in the imaging site IP for each of the aligned 3D images IM1, IM2, and IM3, and obtains each of the 3D images IM1, IM2, and IM3. It has a specific means for identifying the type of disease at the imaging site based on the voxel values S1, S2, and S3 corresponding to the portion Q. [Selection diagram] Fig. 5

Description

  The present invention relates to an image processing apparatus for processing an image and a program applied to the image processing apparatus.

  Due to the fact that high-speed imaging is possible in CT / MR inspection, inspection using multi-phase dynamic scan has become widespread (for example, see Patent Document 1).

JP 2014-012133 A

  There is a clinical advantage that differential diagnosis of a disease becomes possible by executing a multi-phase dynamic scan. However, there are cases in which the change in dark staining between time phases is difficult to understand, and there is a problem in that different diagnoses may make different diagnoses.

  Under such circumstances, a technique capable of improving the interpretation accuracy regardless of the interpreter is desired.

A first aspect of the present invention is an image processing apparatus that processes a plurality of images of imaging regions obtained in different time phases,
For each image, a voxel value corresponding to a first portion in the imaging region is obtained, and the first of the imaging region is determined based on the voxel value corresponding to the first portion obtained for each image. It is an image processing apparatus which has the specific means which specifies the kind of the disease in this part.

A second aspect of the present invention is an image processing apparatus that processes a plurality of images of imaging regions obtained in different time phases,
For each image, a pixel value corresponding to a first portion in the imaging region is determined, and the first value of the imaging region is determined based on the pixel value corresponding to the first portion determined for each image. It is an image processing apparatus which has the specific means which specifies the kind of the disease in this part.

A third aspect of the present invention is a program applied to an image processing apparatus that processes a plurality of images of imaging regions obtained in different time phases,
For each image, a voxel value corresponding to a first portion in the imaging region is obtained, and the first of the imaging region is determined based on the voxel value corresponding to the first portion obtained for each image. It is a program for making a computer perform the specific process which specifies the kind of disease in this part.

A fourth aspect of the present invention is a program applied to an image processing apparatus that processes a plurality of images of imaging regions obtained at different time phases,
For each image, a pixel value corresponding to a first portion in the imaging region is determined, and the first value of the imaging region is determined based on the pixel value corresponding to the first portion determined for each image. It is a program for making a computer perform the specific process which specifies the kind of disease in this part.

  For each of a plurality of images in different time phases, a voxel value or pixel value corresponding to the first portion in the imaging region is obtained, and the voxel value or pixel corresponding to the first portion obtained for each image is obtained. Based on the value, the type of disease in the first portion of the imaging region is specified. Therefore, it is possible to improve the interpretation accuracy regardless of the interpreter.

It is a figure which shows roughly the hardware-like structure of the image processing apparatus 1 which concerns on a 1st form. 2 is a functional block diagram of the image processing apparatus 1. FIG. It is a figure which shows the flow for performing the diagnostic method in a 1st form. 2 is a diagram schematically showing a plurality of images input to the image processing apparatus 1. FIG. It is a figure which shows roughly the image obtained by pre-processing 3D image I1, I2, and I3. It is explanatory drawing of the contrast pattern which appears in 3D image IM1, IM2, and IM3. It is explanatory drawing of a table. It is a figure which shows schematically the diagnostic assistance images DS1, DS2, and DS3. It is a functional block diagram in the image processing apparatus 1 of a 2nd form. It is a figure which shows the flow performed in a 2nd form. It is a figure which shows roughly the image obtained by assigning R gradation, G gradation, and B gradation to 3D image IM1, IM2, and IM3, respectively. It is explanatory drawing of a table. It is a functional block diagram in the image processing apparatus 1 of a 3rd form. It is a figure which shows the flow performed in a 3rd form. It is a figure which shows the synthesized image DF schematically. It is a figure which shows roughly the synthetic | combination image DF and the diagnostic assistance image DS which were displayed on the display apparatus. It is a figure which shows an example of the synthetic | combination image DF and the diagnostic assistance image DS.

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

(1) First Embodiment FIG. 1 is a diagram schematically showing a hardware configuration of an image processing apparatus 1 according to a first embodiment. The image processing apparatus 1 can be configured by a computer, for example. The image processing apparatus 1 is connected to the storage unit M. The image processing apparatus 1 can read a program stored in a computer-readable storage unit M and execute each process. In the storage unit M, the computer C can also download and execute a program from a network N.

  FIG. 2 is a functional block diagram of the image processing apparatus 1. As shown in the figure, the image processing apparatus 1 includes a preprocessing unit 2, an alignment unit 3, an ROI setting unit 4, and a specifying unit 5 as functional blocks. In this example, by causing a computer to execute a predetermined program, the computer is caused to function as these units.

The preprocessing unit 2 performs preprocessing on the image read into the image processing apparatus 1. As preprocessing, smoothing processing or the like can be performed.
The alignment unit 3 aligns a plurality of images read by the image processing apparatus 1.
The ROI setting unit 4 sets a region of interest (ROI). The ROI setting unit 4 can set an ROI based on information input by a user (for example, an interpreting doctor). Further, the ROI setting unit 4 can set the ROI using a segmentation method or the like without using input information by the user.
The specifying unit 5 specifies the type of disease at the imaging site based on a plurality of images at different time phases. A method for identifying the type of disease will be described in detail later.
Below, the example which performs the differential diagnosis of a disease using the image processing apparatus 1 is demonstrated.

FIG. 3 is a diagram illustrating a flow for executing the diagnosis method according to the first embodiment.
In step ST1, a plurality of images to be processed are input to the image processing apparatus 1. In this embodiment, it is assumed that a plurality of images in different time phases obtained by contrast imaging of a subject are input as a plurality of images to be processed. A plurality of images are input to the image processing apparatus 1 via the network N, for example. FIG. 4 schematically shows a plurality of images input to the image processing apparatus 1. FIG. 4 shows an example in which 3D images I1, I2, and I3 are input to the image processing apparatus 1, but a 2D image may be used instead of the 3D image. The 3D images I1, I2, and I3 are images obtained by contrast imaging of the imaging region IP including the liver of the subject. The 3D images I1, I2, and I3 represent 3D images in the first time phase P1, the second time phase P2, and the third time phase P3, respectively. Here, it is assumed that the 3D images IM1, IM2, and IM3 are grayscale images. After inputting the 3D images I1, I2, and I3, the process proceeds to step ST2.

  In step ST2, the preprocessing means 2 (see FIG. 2) performs preprocessing on the 3D images I1, I2, and I3. FIG. 5 is a diagram schematically showing images obtained by preprocessing the 3D images I1, I2, and I3. In FIG. 5, the 3D image after the pre-processing is indicated by reference signs “IM1”, “IM2”, and “IM3”. Voxels Q1, Q2, and Q3 shown in the 3D images IM1, IM2, and IM3 represent points corresponding to the portion Q in the imaging region IP. As the preprocessing, for example, one or more of the following three processes can be executed.

-Smoothing processing-Contrast enhancement processing by smoothing the histogram-Processing for making the dynamic range (range from the minimum value to the maximum value) the same between different time phases

FIG. 6 is an explanatory diagram of contrast patterns appearing in the 3D images IM1, IM2, and IM3.
FIG. 6 schematically shows the contrast pattern of the disease X, the contrast pattern of the disease Y, and the contrast pattern of the disease Z as the contrast pattern of the liver disease.

FIG. 6A is an explanatory diagram of a contrast pattern when disease X is found in the liver.
FIG. 6A schematically shows images D1, D2, and D3 of a cross section CS across the liver for convenience of explanation.
In the case of the disease X, the whole tumor is strongly emphasized in the cross-sectional image D1 of the first time phase P1, the whole tumor is moderately emphasized in the cross-sectional image D2 of the second time phase P2, and the third time phase P3 In the cross-sectional image D3, the entire tumor has a lower luminance value than the liver parenchyma.

FIG. 6B is an explanatory diagram of a contrast pattern when a disease Y is observed in the liver.
FIG. 6B also schematically shows images D1, D2, and D3 of the cross section CS across the liver, as in FIG. 6A.
In the case of the disease Y, in the cross-sectional image D1 of the first time phase P1, the tumor has a lower luminance value than the liver parenchyma, and becomes a lower luminance value as it goes from the edge to the center. Further, in the cross-sectional image D2 of the second time phase P2, the tumor margin has a slightly higher luminance value than the liver parenchyma, and becomes a lower luminance value from the margin toward the center. In the cross-sectional image D3 of the third time phase P3, the tumor has a lower luminance value than the liver parenchyma, and becomes a lower luminance value as it goes from the edge to the center.

FIG. 6C is an explanatory diagram of a contrast pattern when a disease Z is observed in the liver.
FIG. 6C also schematically shows images D1, D2, and D3 of the cross section CS across the liver, as in FIGS. 6A and 6B.
In the case of the disease Z, in the cross-sectional image D1 of the first time phase P1, the tumor has a higher luminance value than the liver parenchyma, and becomes a higher luminance value as it goes from the center to the edge. Furthermore, in the cross-sectional image D2 of the second time phase P2, the entire tumor is emphasized moderately. In the cross-sectional image D3 of the third time phase P3, the entire tumor has a lower luminance value than the liver parenchyma.

Thus, in general, the contrast pattern in a liver tumor differs depending on the type of disease. The inventor of the present application pays attention to this point, and has found that the type of disease can be specified if the contrast pattern of the subject can be specified to which contrast pattern. Hereinafter, a method for identifying a contrast pattern of a disease will be specifically described.
After pre-processing in step ST2, the process proceeds to step ST3.

  In step ST3, the alignment means 3 (see FIG. 2) aligns the 3D image on which preprocessing has been executed.

  Ideally, the imaging region of the subject does not move during the scan, but in practice, the imaging region cannot be prevented from moving at all. Therefore, in order to minimize the positional deviation of the 3D image between the time phases, the alignment of the 3D image on which preprocessing has been performed is performed in step ST3. As an alignment algorithm, known algorithms such as a mutual information amount, a standard gradient field, a cross-correlation coefficient, and a point set method can be used. After alignment, the process proceeds to step ST4.

  In step ST4, the radiogram interpreter determines whether to set an ROI. When the interpretation target is the entire 3D image, the ROI is not necessary. In this case, the process proceeds to step ST6. On the other hand, when the target of interpretation is a specific part in the image (for example, a predetermined organ or a specific part in the predetermined organ), it is easier to observe the lesion when the ROI is set. Become. Therefore, when the interpretation target is a specific part in the image, the process proceeds to step ST5 in order to set the ROI.

  Below, the case where it determines with not setting ROI first is demonstrated. Accordingly, the process proceeds to step ST6.

  In step ST6, the specifying means 5 (see FIG. 2) specifies the type of disease. In the first embodiment, processing for specifying the type of disease is performed using a table representing the correspondence between each of the plurality of diseases and the voxel values of the image. This table is prepared in advance before performing differential diagnosis. The table will be described below.

FIG. 7 is an explanatory diagram of a table.
The table has two columns 1 and 2.

  Column 1 represents n types of diseases. In the first embodiment, for convenience of explanation, n = 3, that is, three diseases X, Y, and Z are shown as disease types.

  Column 2 shows the range of voxel values of the image corresponding to each disease for each time phase. Hereinafter, the relationship between each disease and the range of voxel values will be described.

(1) Relationship between Disease X and Voxel Value Range First, the relationship between the disease X and the voxel value range in the time phase P1 is considered. When the voxel value of the 3D image in the time phase P1 is expressed by “S1”, the range X1 of the voxel value S1 corresponding to the disease X is expressed as follows with respect to the time phase P1.
Range X1: SX ₁₁ ≦ S1 ≦ SX ₁₂
Here, SX ₁₁ : When it is considered that there is a risk of disease X
Lower limit of voxel value in time phase P1
SX _12: when is considered that there is a risk of disease X
Upper limit of voxel value in time phase P1

Next, consider the relationship between the disease X and the range of voxel values in the time phase P2. When the voxel value of the 3D image in the time phase P2 is expressed by “S2”, the range X2 of the voxel value S2 corresponding to the disease X is expressed as follows with respect to the time phase P2.
Range X2: SX ₂₁ ≦ S2 ≦ SX ₂₂
Where SX ₂₁ : when there is a risk of disease X
Lower limit of voxel value in time phase P2
SX ₂₂ : When there is a possibility of disease X
Upper limit of voxel value in time phase P2

Next, consider the relationship between the disease X and the range of voxel values in the time phase P3. When the voxel value of the 3D image in the time phase P3 is expressed by “S3”, the range X3 of the voxel value S3 corresponding to the disease X is expressed as follows with respect to the time phase P3.
Range X3: SX ₃₁ ≦ S3 ≦ SX ₃₂
Here, SX ₃₁ : When it is considered that there is a risk of disease X
Lower limit of voxel value in time phase P3
SX ₃₂ : When there is a risk of disease X
Upper limit of voxel value in time phase P3

  For example, paying attention to the voxels Q1, Q2, and Q3 corresponding to the portion Q in the imaging region IP of the 3D images IM1, IM2, and IM3 (see FIG. 5), the voxel values S1 in the voxels Q1, Q2, and Q3, Consider S2 and S3. When the voxel values S1, S2, and S3 in the voxels Q1, Q2, and Q3 are included in the ranges X1, X2, and X3, respectively, the portion Q in the imaging region IP is caused by the disease X It is determined that there is a high risk of being. On the other hand, if any one of the voxel values S1, S2, and S3 is out of the range of voxel values corresponding to the disease X, it is determined that the risk of the disease X is low.

(2) Relationship between Disease Y and Voxel Value Range In the case of disease Y, the range Y1 of the voxel value S1 for the time phase P1, the range Y2 of the voxel value S2 for the time phase P2, and the voxel value S3 for the time phase P3. The range Y3 is expressed as follows.
Range Y1: SY ₁₁ ≦ S1 ≦ SY ₁₂
Range Y2: SY ₂₁ ≦ S2 ≦ SY ₂₂
Range Y3: SY ₃₁ ≦ S3 ≦ SY ₃₂

  When the voxel values S1, S2, and S3 in the voxels Q1, Q2, and Q3 are included in the ranges Y1, Y2, and Y3, respectively, the portion Q in the imaging region IP has a disease Y. It is determined that there is a high risk of being. On the other hand, if any one of the voxel values S1, S2, and S3 is out of the range of the voxel values corresponding to the disease Y, it is determined that the risk of the disease Y is low.

(3) Regarding the relationship between the disease Z and the range of voxel values In the case of disease Z, the range Z1 of the voxel value S1 for the time phase P1, the range Z2 of the voxel value S2 for the time phase P2, and the voxel value S3 for the time phase P3. The range Z3 is expressed as follows.
Range Z1: SZ ₁₁ ≦ S1 ≦ SZ ₁₂
Range Z2: SZ ₂₁ ≦ S2 ≦ SZ ₂₂
Range Z3: SZ ₃₁ ≦ S3 ≦ SZ ₃₂

  When the voxel values S1, S2, and S3 in the voxels Q1, Q2, and Q3 are included in the ranges Z1, Z2, and Z3, respectively, the portion Q in the imaging region IP is caused by the disease Z. It is determined that there is a high risk of being. On the other hand, if any one of the voxel values S1, S2, and S3 is out of the range of the voxel values corresponding to the disease Z, the risk of the disease Z is determined to be low.

Therefore, by determining which of the diseases X, Y, and Z the voxel value (S1, S2, S3) corresponding to the portion Q in the imaging region IP corresponds to the imaging region The type of disease occurring in the part Q in the IP can be specified. In the above description, the portion Q in the imaging region IP has been described. However, in step ST6, the voxel values (S1, S2, S3) of the respective regions in the imaging region IP are the diseases X, Y, and Z. It is determined which of these diseases corresponds to the voxel value. Therefore, it is possible to determine whether or not diseases X, Y, and Z have occurred in the imaging region IP.
After making this determination, the process proceeds to step ST7.

  In step ST7, a diagnosis support image reflecting the determination result in step ST6 is displayed on the display device. FIG. 8 schematically shows the diagnosis support images DS1, DS2, and DS3. Diagnosis support images DS1, DS2, and DS3 represent cross-sectional images across the liver in time phases P1, P2, and P3, respectively. Each diagnosis support image shows a diseased part (disease part) r0 and a disease name ("disease X" in FIG. 8). Therefore, the radiogram interpreter can recognize the type of disease of the subject by referring to the diagnosis support image.

  In the above description, an example is described in which it is determined in step ST4 (see FIG. 3) that ROI is not set. Next, an example in which it is determined in step ST4 that ROI is set will be described.

  When it is determined in step ST4 that the ROI is set, the process proceeds to step ST5.

  In step ST5, ROI is set. When setting the ROI, the user inputs information for setting the ROI using, for example, a drawing tool. The ROI setting unit 4 (see FIG. 2) sets the ROI range based on the input information. Instead of using the drawing tool, the ROI setting unit 4 may set the ROI by using a segmentation method such as a region expansion method, a level set method, or a method based on edge detection. Further, a threshold value may be set for each time phase image, and the corresponding voxel group may be set as the ROI. As a threshold, when a general CT value is known for each organ as in a CT image, the known value may be set, or a threshold may be set from histogram analysis. One ROI may be set for each time phase, or a common ROI may be used for all time phases. Since the optimal segmentation method varies depending on the type of organ or lesion, it is desirable that the system be appropriately used depending on the examination target. For example, the segmentation technique used to create the ROI may be individually set according to the type of organ, lesion, and imaging sequence. After setting the ROI, the process proceeds to step ST6.

  In step ST6, the specifying unit 5 performs a process for specifying the type of disease within the ROI range set in step ST5. After specifying the type of disease, the process proceeds to step ST7, a diagnosis support image is displayed, and the flow ends.

  In the first embodiment, based on the table (see FIG. 7), for each time phase, it is determined whether or not the voxel value of the image is included in the range of voxel values corresponding to each disease. Specific types. Therefore, since it is possible to identify a disease without depending on the experience and intuition of the interpreter, the objectivity and reproducibility of diagnosis can be improved, and the interpretation accuracy can be greatly improved. In the first mode, a diagnosis support image is displayed (see FIG. 8). Since the diseased part (lesion) r0 is shown in the diagnosis support image, the image interpreter can easily visually recognize the position of the disease in the imaging region.

(2) Second Embodiment FIG. 9 is a functional block diagram of the image processing apparatus 1 according to the second embodiment.
In the second embodiment, the image processing apparatus 1 includes preprocessing means 2 to specifying means 5. Of the preprocessing means 2 to the specifying means 5, the preprocessing means 2, the alignment means 3, and the ROI setting means 4 are the same as those in the first embodiment, so that the description thereof is omitted, and the assigning means 41 and the specifying means 5 are omitted. Will be described.

The assigning means 41 assigns R gradation to the image in the first time phase P1, and assigns the second time phase P.
G gradation is assigned to the image in 2, and B gradation is assigned to the image in the third time phase P3.

  The specifying unit 5 specifies the type of the disease based on the image after the RGB gradation is assigned by the assigning unit 41. A method for identifying the type of disease will be described in detail later.

FIG. 10 is a diagram illustrating a flow executed in the second mode.
Steps ST1 to ST4 are the same as those in the first embodiment, and a detailed description thereof will be omitted. In step ST4, it is determined whether or not an ROI is set. If the ROI is not set, the process proceeds to step ST51. On the other hand, when setting ROI, it progresses to step ST5 and ROI is set. As in the first embodiment, the ROI setting method may be set arbitrarily by the user or automatically. After setting the ROI, the process proceeds to step ST51.

In step ST51, the assigning means 41 (see FIG. 9) assigns a display color of red, green, or blue (R, G, B) to each time phase image. Usually, medical images such as CT and MR are displayed in black and white gradation (gray scale) of about 8-16 bits. In the second embodiment, any gradation of R, G, or B is used for the black and white gradation. assign. For example, the R gradation is assigned to the 3D image IM1 in the first time phase P1, the G gradation is assigned to the 3D image IM2 in the second time phase P2, and the B gradation is assigned to the 3D image IM3 in the third time phase P3. FIG. 11 is a diagram schematically showing images obtained by assigning R gradation, G gradation, and B gradation to the 3D images IM1, IM2, and IM3, respectively. In FIG. 11, the 3D image after each gradation is assigned is indicated by symbols “IMR”, “IMG”, and “IMB”. Voxels Q1, Q2, and Q3 shown in the 3D red image IMR, the 3D green image IMG, and the 3D blue image IMB represent points corresponding to the portion Q in the imaging region IP.
After assigning RGB gradations, the process proceeds to step ST6.

  In step ST6, the specifying means 5 (see FIG. 9) specifies the type of disease. In the second embodiment, a process for specifying the type of disease is performed using a table (see FIG. 12) representing the correspondence between each of a plurality of diseases and the voxel value of the image. This table is prepared in advance before performing differential diagnosis. The table used in the second form will be described below.

FIG. 12 is an explanatory diagram of a table.
The table has columns 1 and 2.

Column 1 represents n types of diseases. Also in the second form, similarly to the first form, n = 3, that is, three kinds of diseases X, Y, and Z are shown as types of diseases.
Column 2 shows the range of voxel values corresponding to each disease for each time phase. Hereinafter, the relationship between each disease and the range of voxel values will be described.

(1) Relationship between Disease X and Voxel Value Range First, the relationship between the disease X and the voxel value range in the time phase P1 is considered. When the voxel value of the 3D red image in the time phase P1 is expressed by “S1”, the range X1 of the voxel value S1 corresponding to the disease X is expressed as follows with respect to the time phase P1.
Range X1 (R gradation): SX _R1 ≦ S1 ≦ SX _R2
Where SX _R1 : if there is a risk of disease X
Lower limit of voxel value in time phase P1
SX _R2 : When there is a risk of disease X
Upper limit of voxel value in time phase P1

Next, consider the relationship between the disease X and the range of voxel values in the time phase P2. When the voxel value of the 3D green image in the time phase P2 is represented by “S2”, the range X2 of the voxel value S2 corresponding to the disease X is expressed as follows with respect to the time phase P2.
Range X2 (G gradation): SX _G1 ≦ S2 ≦ SX _G2
Where SX _G1 : When there is a risk of disease X
Lower limit of voxel value in time phase P2
SX _G2 : When there is a risk of disease X
Upper limit of voxel value in time phase P2

Next, consider the relationship between the disease X and the range of voxel values in the time phase P3. When the voxel value of the 3D blue image in the time phase P3 is expressed by “S3”, the range X3 of the voxel value S3 corresponding to the disease X is expressed as follows with respect to the time phase P3.
Range X3 (B gradation): SX _B1 ≦ S3 ≦ SX _B2
Here, SX _B1 : When it is considered that there is a risk of disease X
Lower limit of voxel value in time phase P3
SX _B2 : When there is a possibility of disease X
Upper limit of voxel value in time phase P3

  For example, paying attention to the voxels Q1, Q2, and Q3 corresponding to the portion Q in the imaging region IP of the 3D red image IMR, 3D green image IMG, and 3D blue image IMB (see FIG. 11), the voxels Q1, Q2, Consider the voxel values S1, S2, and S3 at Q3 and Q3. When the voxel values S1, S2, and S3 in the voxels Q1, Q2, and Q3 are included in the ranges X1, X2, and X3, respectively, the portion Q in the imaging region IP is caused by the disease X It is determined that there is a high risk of being. On the other hand, if any one of the voxel values S1, S2, and S3 is out of the range of voxel values corresponding to the disease X, it is determined that the risk of the disease X is low.

(2) Relationship between Disease Y and Voxel Value Range In the case of disease Y, the range Y1 of the voxel value S1 for the time phase P1, the range Y2 of the voxel value S2 for the time phase P2, and the voxel value S3 for the time phase P3. The range Y3 is expressed as follows.
Range Y1 (R gradation): SY _R1 ≦ S1 ≦ SY _R2
Range Y2 (G gradation): SY _G1 ≦ S2 ≦ SY _G2
Range Y3 (B gradation): SY _B1 ≤ S3 ≤ SY _B2

  When the voxel values S1, S2, and S3 in the voxels Q1, Q2, and Q3 are included in the ranges Y1, Y2, and Y3, respectively, the portion Q in the imaging region IP has a disease Y. It is determined that there is a high risk of being. On the other hand, if any one of the voxel values S1, S2, and S3 is out of the range of the voxel values corresponding to the disease Y, it is determined that the risk of the disease Y is low.

(3) Regarding the relationship between the disease Z and the range of voxel values In the case of disease Z, the range Z1 of the voxel value S1 for the time phase P1, the range Z2 of the voxel value S2 for the time phase P2, and the voxel value S3 for the time phase P3. The range Z3 is expressed as follows.
Range Z1 (R gradation): SZ _R1 ≦ S1 ≦ SZ _R2
Range Z2 (G gradation): SZ _G1 ≦ S2 ≦ SZ _G2
Range Z3 (B gradation): SZ _B1 ≦ S3 ≦ SZ _B2

  When the voxel values S1, S2, and S3 in the voxels Q1, Q2, and Q3 are included in the ranges Z1, Z2, and Z3, respectively, the portion Q in the imaging region IP is caused by the disease Z. It is determined that there is a high risk of being. On the other hand, if any one of the voxel values S1, S2, and S3 is out of the range of the voxel values corresponding to the disease Z, the risk of the disease Z is determined to be low.

  Therefore, by determining which of the diseases X, Y, and Z the voxel value (S1, S2, S3) corresponding to the portion Q in the imaging region IP corresponds to the imaging region The type of disease occurring in the part Q in the IP can be specified. In the above description, the portion Q in the imaging region IP has been described. However, in step ST6, the voxel values (S1, S2, S3) of the respective regions in the imaging region IP are the diseases X, Y, and Z. It is determined which of these diseases corresponds to the voxel value. Therefore, it is possible to determine whether or not diseases X, Y, and Z have occurred in the imaging region IP. If the ROI is set in step ST5, the voxel value (S1, S2, S3) corresponds to the voxel value of any of the diseases X, Y, and Z within the ROI range. Is determined. After making this determination, the process proceeds to step ST7.

  In step ST7, diagnosis support images DS1, DS2, and DS3 (see FIG. 8) reflecting the determination result of step ST6 are output. In the second form, the diagnosis support images DS1, DS2, and DS3 are displayed as images to which R gradation, G gradation, and B gradation are assigned, respectively. When the diagnosis support image is displayed, the flow ends.

  In the second embodiment, based on the table (see FIG. 12), for each time phase, it is determined whether or not the voxel value of the image is included in the range of voxel values corresponding to each disease. Specific types. Therefore, in the second embodiment, as in the first embodiment, the disease can be identified without depending on the experience and intuition of the image interpreter, so that the objectivity and reproducibility of diagnosis can be improved, and the image interpretation can be improved. The accuracy can be greatly improved. In the second mode, the diagnosis support images DS1, DS2, and DS3 are assigned different colors. Therefore, the radiogram interpreter can confirm which part of the three time phases is stained with the contrast agent at which time phase among the three time phases by comparing the diagnosis support images.

  In the second mode, RGB gradations are assigned to the monochrome gradations (grayscale) of the 3D images IM1, IM2, and IM3 (step ST51), and the type of disease is specified using a table (see FIG. 12). (Step ST6). In general, the dynamic range of voxel values on a grayscale image varies depending on the medical device and imaging conditions, so when identifying the type of disease based on the voxel value of a grayscale image, It is necessary to prepare a table (see FIG. 7). On the other hand, in the second embodiment, as described above, RGB gradations are assigned, so that voxel values can be standardized. Therefore, the second embodiment has an advantage that the number of tables used for specifying the type of disease can be reduced. Furthermore, since the voxel values are standardized, improvement in diagnostic accuracy and reproducibility can be expected.

(3) Third Embodiment FIG. 13 is a functional block diagram of the image processing apparatus 1 according to the third embodiment.
In the third embodiment, the image processing apparatus 1 has preprocessing means 2 to determination means 53. Of the preprocessing means 2 to the determination means 53, the preprocessing means 2, the alignment means 3, the ROI setting means 4, the assigning means 41, and the specifying means 5 are the same as those in the second embodiment, and thus description thereof is omitted. The opacity setting means 51, the synthesis means 52, and the determination means 53 will be described.

  The opacity setting means 51 sets opacity for each time-phase image displayed in one of R, G, and B gradations. The synthesizing unit 52 synthesizes the images of the respective time phases displayed with R, G, and B gradations to generate a synthesized image. The determination unit 53 determines a background image used for generating the composite image and the diagnosis support image.

  Note that more detailed functions of the units 51, 52, and 53 will be described together with the description of the processing flow in the image processing apparatus 1.

FIG. 14 is a diagram illustrating a flow executed in the third mode.
Steps ST1 to ST4 are the same as those in the first embodiment, and a detailed description thereof will be omitted. In step ST4, it is determined whether or not an ROI is set. If the ROI is not set, the process proceeds to step ST51. On the other hand, when setting ROI, it progresses to step ST5. Here, first, the case where the ROI is not set will be described. Accordingly, the process proceeds to step ST51.

  In step ST51, the assigning means 41 (see FIG. 13) assigns a display color of red, green, or blue (R, G, B) to each time phase image. The display color assignment method is the same as in the second embodiment. Therefore, as shown in FIG. 11, a 3D red image IMR, a 3D green image IMG, and a 3D blue image IMB are generated. After assigning RGB gradations, the process proceeds to step ST6.

  In step ST6, the specifying means 5 (see FIG. 13) specifies the type of disease. Since the method for identifying the type of disease is the same as in the second embodiment, description thereof is omitted. After identifying the disease, the process proceeds to step ST61.

  In step ST61, the opacity setting means 51 (see FIG. 13) sets opacity (sets a weight) for each time phase image displayed in one of R, G, and B gradations. . The opacity may be set equal to 1/3 of each time phase image displayed in R, G, B, and the opacity in a predetermined time phase may be set to be higher than the opacity in other time phases. A high value may be set. For example, when a lesion is characteristically depicted in the second time phase image, the opacity of the second time phase image can be set high.

In each time phase image, the contrast noise ratio (CNR) and the signal noise ratio (SNR) are not always constant. In such a case, it is desirable to change the opacity according to the CNR or SNR. For example, a substantially equal weight is set by setting the integrated value of CNR and opacity to be a constant value at each time phase based on the following equation.
CNR ₁ xT ₁ = CNR ₂ xT ₂ = CNR ₃ xT ₃
T ₁ + T ₂ + T ₃ = 1
Where CNR _k : contrast noise ratio of the kth time phase
T _k : opacity of the kth time phase

  After setting the opacity, the process proceeds to step ST62.

In step ST62, the combining unit 52 (see FIG. 13) performs the 3D red image IMR (first time phase P1), the 3D green image IMG (second time phase P2), and the 3D blue image IMB (third time). Phase P3) is synthesized to create a composite image. The RGB value in the composite image is determined by the following equation.
R _F = T ₁ xV ₁
G _F = T ₂ xV ₂
B _F = T ₃ xV ₃
Here, R _F , G _F , B _F : RGB components in the composite image
Vk: voxel value of the image of the kth time phase

  FIG. 15 schematically shows the composite image DF. After creating the composite image DF, the process proceeds to step ST7.

  In step ST7, the composite image DF and the diagnosis support image DS are displayed on the display device. FIG. 16 schematically shows the composite image DF and the diagnosis support image DS displayed on the display device. The diagnosis support image DS represents an image obtained by superimposing characters “Disease X” on the composite image DF. When the composite image DF and the diagnosis support image DS are displayed, the flow ends.

  In the third mode, as in the second mode, based on the table (see FIG. 12), whether the voxel value of the image is included in the range of voxel values corresponding to each disease for each time phase. By determining whether or not, the type of disease is specified. Therefore, since it is possible to identify a disease without depending on the experience and intuition of the interpreter, the objectivity and reproducibility of diagnosis can be improved, and the interpretation accuracy can be greatly improved. In the third embodiment, the composite image DF and the diagnosis support image DS are displayed as RGB images. Therefore, the radiogram interpreter can interpret images without comparing the images of the three time phases, so that the time required for image interpretation can be shortened and the image interpretation efficiency can be improved.

  In the above example, the case where it is determined not to set the ROI in step ST4 (see FIG. 14) is described. Next, the case where it is determined to set the ROI in step ST4 will be described.

  If it is determined in step ST4 that ROI is set, the process proceeds to step ST5. In step ST5, the ROI setting means 4 (see FIG. 13) sets the ROI. The ROI may be set using a drawing tool, or may be set using a segmentation technique or the like without using a drawing tool. After setting the ROI, the process proceeds to step ST50.

  In step ST50, the determination means 53 (refer FIG. 13) determines background image DB used when producing | generating the synthetic | combination image DF and the diagnostic assistance image DS (refer FIG. 17) mentioned later. The determination unit 53 determines, for example, an arbitrary calculation image such as any one of the 3D images read out in step ST1 or an addition average image of all time phases as the background image DB. Can do. After determining the background image DB, the process proceeds to step ST51. In step ST51, RGB is assigned (see FIG. 11). After assigning RGB, the process proceeds to step ST6. In step ST6, the specifying unit 5 performs processing for specifying the type of disease only within the ROI range set in step ST5. After specifying the type of disease, opacity is set in step ST61, and the process proceeds to step ST62. In step ST62, the synthesis unit 52 creates a synthesized image using the background image determined in step ST50. After creating the composite image, the process proceeds to step ST7. In step ST7, the composite image DF and the diagnosis support image DS are displayed. FIG. 17 shows an example of the composite image DF and the diagnosis support image DS. The composite image DF is an image generated by superimposing the image part of the diseased part r0 on the background image DB selected in step ST50. The diagnosis support image DS is an image generated by adding a disease name (disease X) to the composite image DF. In the composite image DF and the diagnosis support image DS, since the diseased part r0 is superimposed on the background image DB, the image interpreter easily visually distinguishes the region where the disease appears from the region where the disease does not appear. be able to.

  Although the opacity Tk is set in the third embodiment, it is desirable that the value of the opacity Tk can be manually changed by a radiographer. By making it possible to change the value of the opacity Tk, the RGB brightness of the composite image DF and the diagnosis support image DS can be individually changed for each color. Therefore, the radiogram interpreter can adjust the brightness of the composite image DF and the diagnosis support image DS to a brightness suitable for visually recognizing a disease by adjusting the value of the opacity Tk. Become.

  In the first to third embodiments, examples in which a disease is specified based on an image of three time phases are described. However, the present invention is not limited to images of three time phases, and can be applied to a case where a disease is specified based on images of m (≧ 2) time phases.

  In the first to third embodiments, examples of specifying a disease based on a plurality of images obtained by contrast imaging are described. However, in the present invention, an image used for specifying a disease may be an image obtained in a different time phase, and is not limited to an image obtained by contrast imaging.

  Note that the image processing apparatus 1 according to the first to third embodiments may be provided with an updating unit for updating the information in the table used in the first embodiment. By updating the information in the table, it is possible to specify the type of disease with higher accuracy. The table information may be updated manually by the user or automatically using a machine learning method.

  Moreover, the 1st-3rd form demonstrates the method of performing a differential diagnosis based on 3D image. However, differential diagnosis may be performed based on 2D images instead of 3D images. When a 2D image is used, the type of disease can be specified based on the pixel value of the 2D image at each time phase. In the case of using a 2D image, it is possible to specify the type of the disease at the imaging site by using a table representing the correspondence between each disease and the range of pixel values.

DESCRIPTION OF SYMBOLS 1 Image processing apparatus 2 Preprocessing means 3 Positioning means 4 ROI setting means 5 Specification means 41 Assignment means 51 Opacity setting means 52 Composition means 53 Determination means

Claims (13)

An image processing apparatus that processes a plurality of images of imaging regions obtained at different time phases,
For each image, a voxel value corresponding to a first portion in the imaging region is obtained, and the first of the imaging region is determined based on the voxel value corresponding to the first portion obtained for each image. An image processing apparatus having a specifying means for specifying the type of disease in the portion. The specifying means is:
Using a table representing a correspondence relationship between each of a plurality of diseases and a voxel value for each time phase, a voxel value corresponding to the first portion obtained for each image is calculated from among the plurality of diseases. The image processing apparatus according to claim 1, wherein a disease voxel value is determined and a disease type in the first portion of the imaging region is specified based on the determination result. Setting means for setting the region of interest of the imaging region;
The specifying means is:
For each image, a voxel value corresponding to a first portion in the region of interest is determined, and based on the voxel value corresponding to the first portion determined for each image, the first in the region of interest. The image processing apparatus according to claim 1, wherein a type of a disease in one portion is specified. An image processing apparatus that processes a plurality of images of imaging regions obtained at different time phases,
For each image, a pixel value corresponding to a first portion in the imaging region is determined, and the first value of the imaging region is determined based on the pixel value corresponding to the first portion determined for each image. An image processing apparatus having a specifying means for specifying the type of disease in the portion. The specifying means is:
A pixel value corresponding to the first portion determined for each image is obtained from the plurality of diseases by using a table representing a correspondence relationship between each of the plurality of diseases and the pixel value for each time phase. The image processing apparatus according to claim 4, wherein a disease value corresponding to a pixel value is determined, and a disease type in the first portion of the imaging region is specified based on the determination result. Setting means for setting the region of interest of the imaging region;
The specifying means is:
For each image, a pixel value corresponding to a first portion in the region of interest is determined, and a first value of the region of interest is determined based on the pixel value corresponding to the first portion determined for each image. The image processing apparatus according to claim 4, wherein the type of disease in the part is specified.   The image processing apparatus according to claim 1, wherein each of the plurality of images is a grayscale image.   The image processing apparatus according to claim 1, wherein each of the plurality of images is an image obtained by assigning color information to a grayscale image.   The plurality of images include a red image to which red color information is assigned, a green image to which green color information is assigned, and a blue image to which blue color information is assigned. Image processing device. Alignment means for aligning the plurality of images;
The specifying means is:
The image processing apparatus according to claim 1, wherein a type of a disease in the first portion of the imaging region is specified based on the plurality of images that have been aligned.   The image processing apparatus according to claim 1, further comprising an updating unit that updates information included in the table. A program applied to an image processing apparatus that processes a plurality of images of imaging regions obtained in different time phases,
For each image, a voxel value corresponding to a first portion in the imaging region is obtained, and the first of the imaging region is determined based on the voxel value corresponding to the first portion obtained for each image. Specific processing to identify the type of disease in the part of
A program that causes a computer to execute. A program applied to an image processing apparatus that processes a plurality of images of imaging regions obtained in different time phases,
For each image, a pixel value corresponding to a first portion in the imaging region is determined, and the first value of the imaging region is determined based on the pixel value corresponding to the first portion determined for each image. Specific processing to identify the type of disease in the part of
A program that causes a computer to execute.