JP5977596B2 - Image diagnosis support apparatus, method and program thereof - Google Patents

Description

  The present invention relates to an image diagnosis support apparatus, a method thereof, and a program.

  The result of the image diagnosis performed by the doctor observing the medical image is collected in an image diagnosis report. In the image diagnosis report, an image finding regarding an abnormal shadow discovered by a doctor from a medical image and an image diagnosis name for determining a medical classification name of the abnormal shadow are described. When the image findings are described in detail, the type of image findings (for example, long diameter, shape, etc.) and the evaluation value (for example, 15 mm, spherical shape, etc.) are described. In many cases, both the image findings and the name of the image diagnosis are determined based on the knowledge and experience of the doctor.

  In recent years, an image processing technique for detecting abnormal shadow candidates has been studied as an image diagnosis support technique using a computer. Patent Document 1 discloses a system that compares information on abnormal shadows detected by a computer with information on abnormal shadows discovered by a doctor and displays a comparison result. When this system is used, since there is a possibility that the computer can display the position of the abnormal shadow overlooked by the doctor, an effect of reducing the oversight of the abnormal shadow can be expected.

  In recent years, an inference technique for an image diagnosis name using SVM (Support Vector Machine), ANN (Artificial Neural Network), Bayesian network, and the like has been studied. An effect of improving the reliability of the image diagnosis name can be expected by referring to the image diagnosis name calculated using such an inference technique.

Japanese Patent No. 3085724

  As described above, since doctors determine image findings and diagnosis names based on their own knowledge and experience, there is a problem that they are not necessarily highly objective. In addition, since a doctor usually performs image diagnosis alone, even if an oversight or a mistake in image findings occurs, it is difficult for the doctor to notice it, so there is a problem that the reliability is not necessarily high. By utilizing the technique of Patent Document 1, an effect of reducing oversight of abnormal shadows can be expected. In addition, the use of inference techniques can be expected to improve the reliability of diagnostic imaging names. However, with any background technology, support information for making the image findings regarding the detected abnormal shadow objective and reliable cannot be obtained.

  The present invention has been made in view of the above problems, and an object thereof is to display support information for improving the objectivity and reliability of image findings determined by a doctor.

  In order to solve the above-described problem, an image diagnostic apparatus according to an aspect of the present invention includes an image finding acquisition unit that acquires a type and an evaluation value of an image finding regarding a predetermined three-dimensional region of a three-dimensional image to be diagnosed from a user; Characteristic value calculation means for calculating a characteristic value that can be compared with the evaluation value of the image findings based on an image feature from each of a plurality of cross-sectional images included in the predetermined three-dimensional region, and calculated from each of the plurality of cross-sectional images Representative characteristic value calculating means for calculating a representative characteristic value representing the predetermined three-dimensional region from the plurality of characteristic values obtained, and obtaining the cross-sectional image from which the representative characteristic value has been calculated from the plurality of cross-sectional images as a representative cross-sectional image Representative cross-sectional image acquisition means, comparison means for comparing the evaluation value of the image findings and the representative characteristic value, and the representative cross-section only when there is a significant difference in the comparison result of the comparison means And display means for displaying an image.

  According to the present invention, the representative cross-sectional image is displayed only when there is a significant difference between the evaluation value of the image finding acquired from the doctor and the representative characteristic value calculated by the image diagnosis support apparatus according to the present invention. Can do. As a result, the representative cross-sectional image can be displayed as support information that prompts the doctor to review the image findings only when the evaluation value of the image findings obtained from the doctor is questionable. Therefore, there is an effect that the objectivity and reliability of the image findings determined by the doctor can be improved.

1 is a diagram illustrating an example of an overall configuration of an image diagnosis support system including an image diagnosis support apparatus according to an embodiment of the present invention. 5 is a flowchart showing an example of a process flow of the image diagnosis support apparatus 10. An example of a diagnosis target image displayed on the display unit. The figure which shows an example of the relationship between the evaluation value of the image finding input by the user, and the characteristic value calculated by the apparatus. The figure which shows an example of the approximate cross-section image displayed on the display part 36, a representative cross-section image, and those incidental information.

  DESCRIPTION OF EMBODIMENTS Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 is a diagram showing an example of the overall configuration of an image diagnosis support system including an image diagnosis support apparatus according to an embodiment of the present invention.

  The image diagnosis support system includes an image diagnosis support apparatus 10 and a database 22, and these apparatuses are connected to each other via a communication unit 21 so as to communicate with each other. In the present embodiment, the communication means 21 is configured by a LAN (Local Area Network). The database 22 manages medical images and information associated therewith. The image diagnosis support apparatus 10 acquires a medical image (diagnosis target image) that is a diagnosis target managed in the database 22 and information associated therewith via the LAN 21.

  The diagnostic imaging support apparatus 10 includes a communication IF 31, a ROM 32, a RAM 33, a storage unit 34, an operation unit 35, a display unit 36, and a control unit 37 as functional configurations. The communication IF (Interface) 31 is realized by, for example, a LAN card or the like, and manages communication between the external apparatus (for example, the database 22) and the image diagnosis support apparatus 10 via the LAN 21. A ROM (Read Only Memory) 32 is realized by a nonvolatile memory or the like, and stores various programs. A RAM (Random Access Memory) 33 is realized by a volatile memory or the like, and temporarily stores various information. The storage unit 34 is realized by, for example, an HDD (Hard Disk Drive) or the like, and stores various types of information. The operation unit 35 is realized by, for example, a keyboard and a mouse, and inputs an instruction from the user into the apparatus. The display unit 36 is realized by, for example, a display or the like, and displays various types of information toward a user (for example, a doctor). The control unit 37 is realized by, for example, a CPU (Central Processing Unit) or the like, and performs overall control of processing in the diagnostic imaging support apparatus 10.

  The control unit 37 includes, as its functional configuration, an area acquisition unit 11, an image finding acquisition unit 12, a characteristic value calculation unit 13, an approximate characteristic value selection unit 14, a representative characteristic value calculation unit 15, and an approximate cross-sectional image acquisition unit 16. A representative cross-sectional image acquisition unit 17, a comparison unit 18, and a display processing unit 19 are provided. These configurations are realized, for example, when the CPU reads and executes a program stored in the ROM 32 or the storage unit 34 using the RAM 33 as a work area. A part or all of these configurations may be realized by a dedicated circuit or the like.

  The area acquisition unit 11 performs a predetermined three-dimensional area of a three-dimensional image to be diagnosed, a plurality of cross-sectional images included in the predetermined three-dimensional area, and an abnormal shadow existing in the predetermined three-dimensional area by a method described later. Get the area. The image finding acquisition unit 12 acquires from the user the type and evaluation value of the image finding related to the abnormal shadow. This acquisition is performed based on an operation by the user via the operation unit 35, for example.

  The characteristic value calculation unit 13 performs image processing according to the type of image findings acquired by the image finding acquisition unit 12 on each of the plurality of cross-sectional images acquired by the region acquisition unit 11. Thereby, the characteristic value calculation unit 13 calculates a characteristic value that can be compared with the evaluation value of the image finding acquired by the image finding acquisition unit 12 for each of the cross-sectional images.

  The approximate characteristic value selection unit 14 selects a value (approximate characteristic value) closest to the evaluation value of the image findings from the plurality of characteristic values calculated by the characteristic value calculation unit 13. The representative characteristic value calculation unit 15 calculates a characteristic value (representative characteristic value) representing the predetermined three-dimensional region from a plurality of characteristic values calculated from each of the plurality of cross-sectional images.

  The approximate cross-sectional image acquisition unit 16 acquires a cross-sectional image (approximate cross-sectional image) for which the approximate characteristic value is calculated from the plurality of cross-sectional images. The representative cross-sectional image acquisition unit 17 acquires a cross-sectional image (representative cross-sectional image) for which the representative characteristic value is calculated from the plurality of cross-sectional images.

  The comparison unit 18 compares the evaluation value of the image findings with the representative characteristic value, and determines whether there is a significant difference between these two values. Here, whether there is a significant difference is determined by, for example, whether the absolute value of the difference between the two values exceeds a predetermined threshold. The display processing unit 19 displays both the approximate cross-sectional image and the representative cross-sectional image only when receiving a determination result that the comparison result has a significant difference from the comparison unit 18. Further, the display processing unit 19 can display accompanying information such as a description of the cross-sectional image together with each of the approximate cross-sectional image and the representative cross-sectional image. These displays are performed by displaying a screen on the display unit 36, for example. Of course, the above display may be performed by printing with a printer (not shown).

  Of the functional configuration of the control unit 37 described above, the approximate characteristic value selection unit 14 and the approximate cross-sectional image acquisition unit 16 may be removed. In that case, the display processing unit 19 displays the representative cross-sectional image and its accompanying information only when the determination result that the comparison result has a significant difference is received from the comparison unit 18.

  Next, an example of the processing flow of the diagnostic imaging support apparatus 10 shown in FIG. 1 will be described with reference to FIG. When the user instructs acquisition of a three-dimensional image to be diagnosed managed on the database 22 via the operation unit 35, this process starts.

  When this processing starts, the diagnostic imaging support apparatus 10 acquires a three-dimensional image to be diagnosed from the database 22 via the LAN 21 according to the user instruction, and stores it in the RAM 33 (S101). The image diagnosis support apparatus 10 causes the display processing unit 19 to display an arbitrary cross-sectional image (at a position designated by the user) included in the diagnosis target three-dimensional image on the display unit 36. The diagnosis target three-dimensional image may be acquired from an external storage device connected to the image diagnosis support apparatus 10.

  Here, the user designates a predetermined region (a three-dimensional region where an abnormal shadow exists) of the diagnosis target three-dimensional image via the operation unit 35. Then, the image diagnosis support apparatus 10 acquires a predetermined area in accordance with the user operation in the area acquisition unit 11 and acquires a plurality of cross-sectional images included in the predetermined area (S102). At this time, the axial sectional image generated by the image capturing device may be acquired as it is, or a three-dimensional image may be reconstructed and then a sectional image orthogonal to an arbitrary normal direction may be generated and acquired. . For example, a coronal slice image or a sagittal slice image whose normal direction is orthogonal to the axial slice image may be acquired. The axial sectional image, the coronal sectional image, and the sagittal sectional image are collectively referred to as an orthogonal three sectional image.

  Further, the region acquisition unit 11 extracts and acquires an abnormal shadow region existing in a predetermined region using a known region extraction technique (such as a graph cut method or a level set method). An abnormal shadow region may be automatically acquired using a known abnormal shadow detection technique. In this case, after the abnormal shadow region is automatically acquired, the rectangular parallelepiped region surrounding the abnormal shadow region is automatically extracted and acquired as a predetermined three-dimensional region. Further, the processing procedure is to acquire a plurality of cross-sectional images included in the predetermined region.

Here, FIG. 3 shows an example of the diagnosis target image displayed on the display unit 36. Usually, since a three-dimensional image cannot be displayed as it is on the display unit 36, an axial sectional image, a coronal sectional image, and a sagittal sectional image are displayed in the three image display areas 41, 42, and 43, respectively. As described above, the user designates predetermined areas 51, 52, and 53 on each cross-sectional image via the operation unit 35. Here, the predetermined areas 51, 52, and 53 are rectangular parallelepiped projection figures (rectangles) drawn on each cross-sectional image in order to designate one rectangular parallelepiped on the three-dimensional image. Therefore, the following relationship holds.
X coordinate of region 51 = X coordinate of region 52 (1)
Y coordinate of area 51 = X coordinate of area 53 (2)
Y coordinate of region 52 = Y coordinate of region 53 (3)

  Therefore, the user can designate a predetermined area by drawing a rectangle on any two cross-sectional images of the three orthogonal cross-sectional images. The rectangular position on the remaining one cross-sectional image is automatically calculated using equations (1) to (3). Since there are a plurality of cross-sectional images in three orthogonal directions within the predetermined regions 51, 52, and 53, a plurality of cross-sectional images can be acquired in any direction by the processing of S102.

  Returning to the description of FIG. The user inputs an image finding via the operation unit 35. That is, the user who refers to the plurality of cross-sectional images displayed on the display unit 36 in the process of S101 inputs the type and evaluation value of the image findings regarding the abnormal shadow existing on the plurality of cross-sectional images. Then, the image diagnosis support apparatus 10 uses the image finding acquisition unit 12 to acquire the type and evaluation value of the image finding input according to the user operation (S103). Examples of types of image findings include the major axis (maximum diameter) of abnormal shadows, shape, spinous processes, irregular contours, calcification, air, vascular and bronchial entrapment, pleural insertion, bronchial clear image, etc. Information on presence and absence and degree. The process of S103 may be executed simultaneously with the process of S102, or may be executed before the process of S102.

  Subsequently, in the image diagnosis support apparatus 10, the characteristic value calculation unit 13 performs the processes of S <b> 104 and S <b> 105 for each of the plurality of cross-sectional images acquired by the region acquisition unit 11. First, in S104, an image feature of an abnormal shadow corresponding to the type of image finding acquired by the image finding acquisition unit 12 is calculated using a known image processing technique. Examples of image features include shape features of the abnormally shaded area (major axis, circularity of the borderline, flatness, irregularity, etc.) and density features (average and variance of concentration, specificities corresponding to calcification, air, etc.) For example, the existence ratio of concentration values within the concentration range. Further, in S105, for each of the plurality of cross-sectional images, a characteristic value that can be compared with the evaluation value of the image finding is calculated based on the image feature calculated in S104.

  Here, the calculation method of the characteristic value in the process of S105 will be described. For example, when the type of image finding is “major axis”, the characteristic value is obtained by measuring the major axis (maximum diameter) of the ellipse when the boundary line of the abnormal shadow is approximated to the ellipse. Further, for example, when the type of image finding is “rough outline”, the characteristic value is calculated according to the dispersion of the distance from the center of the abnormal shadow to each point on the boundary line. For example, when the type of image finding is “calcification”, a density histogram of an abnormal shadow is created, and a characteristic value is calculated according to a ratio of density values in a density range corresponding to calcification.

  The image diagnosis support apparatus 10 uses the approximate characteristic value selection unit 14 to select a value (approximate characteristic value) that is closest to the evaluation value of the image finding acquired by the image finding acquisition unit 12 from the plurality of characteristic values calculated by the characteristic value calculation unit 13. ) Is selected (S106). However, if any of the characteristic values is more than a predetermined distance from the evaluation value of the image findings, no approximate characteristic value may be set. Subsequently, the diagnostic imaging support apparatus 10 uses the approximate cross-sectional image acquisition unit 16 to obtain cross-section images (approximate cross-sectional images) having the characteristic values selected by the approximate characteristic value selection unit 14 by the region acquisition unit 11. Obtained from the image (S108).

  On the other hand, in the diagnostic imaging support apparatus 10, the representative characteristic value calculation unit 15 uses a characteristic value (representative characteristic value) representing a predetermined region acquired by the region acquisition unit 11 from a plurality of characteristic values calculated by the characteristic value calculation unit 13. ) Is calculated (S107). Subsequently, in the diagnostic imaging support apparatus 10, the representative cross-sectional image acquisition unit 17 has a plurality of cross-sections acquired by the region acquisition unit 11 with cross-sectional images (representative cross-sectional images) having the characteristic values calculated by the representative characteristic value calculation unit 15. Obtained from the image (S109).

  Here, the calculation method of the representative characteristic value in S107 will be described. The method of calculating the representative characteristic value varies depending on the type of image findings. For example, in many types of image findings such as long diameter, spinous process, calcification, air, vascular or bronchial entrainment, pleural insertion, bronchial clear image, the maximum value among the plurality of characteristic values is set. The representative characteristic value. For example, in some types of image findings such as contour irregularity, an average value of the plurality of characteristic values is used as a representative characteristic value. Further, for example, in some types of image findings such as a shape, a priority value is given in advance to all possible values of the characteristic value, so that the characteristic value with the highest priority is set as the representative characteristic value.

  Note that the processing of S106 and S108 and the processing of S107 and S109 may be executed in parallel, or any processing may be executed first. Alternatively, as described in the description of FIG. 1, the approximate characteristic value selection unit 14 and the approximate cross-sectional image acquisition unit 16 can be removed from the image diagnosis support apparatus 10, and when these are removed, S106 and S108 This process is not executed.

  Subsequently, in the comparison unit 18, the diagnostic imaging support apparatus 10 compares the evaluation value of the image finding acquired by the image finding acquisition unit 12 with the representative characteristic value calculated by the representative characteristic value calculation unit 15 (S110). Then, the comparison unit 18 determines whether there is a significant difference between the two values. For example, when two values have a magnitude relationship, it is determined that there is a significant difference when the absolute value of the difference between the two values exceeds a predetermined threshold. For example, when the two values do not have a magnitude relationship, it is determined that there is a significant difference when the two values are different. Further, for example, when the evaluation value and / or characteristic value of the image finding has a unique value, it is determined that there is a significant difference when only one of the values becomes a unique value.

  Finally, the image diagnosis support apparatus 10 has the following only in the display processing unit 19 when there is a significant difference between the evaluation value of the image findings and the representative characteristic value as a comparison result in the comparison unit 18. Information is displayed on the display unit 36 (S111). Of course, the above display may be performed by printing with a printer (not shown). The information displayed on the display unit 36 includes the approximate cross-sectional image acquired by the approximate cross-sectional image acquisition unit 16 and the representative cross-sectional image acquired by the representative cross-sectional image acquisition unit 17. Furthermore, you may display accompanying information, such as description of a cross-sectional image, with respect to each of these cross-sectional images. If S106 and S108 are not executed, the approximate cross-sectional image and its accompanying information are not displayed.

  FIG. 4 is a diagram illustrating an example of a relationship between an evaluation value of an image finding input by a user and a characteristic value calculated by the apparatus. In FIG. 4, the horizontal axis indicates the position of a plurality of cross-sectional images, and the vertical axis indicates the evaluation value or characteristic value of the image findings. In FIG. 4, an example of the evaluation value of the image finding acquired in S <b> 103 is indicated by a dotted line 61. An example of the characteristic value for each of the plurality of cross-sectional images calculated in S105 is indicated by a line graph 62. An example of the approximate characteristic value selected in S106 is indicated by a rectangle 63. An example of the representative characteristic value (64) calculated in S107 is indicated by a circle 64. As described above, the image diagnosis support apparatus 10 can display the approximate cross-sectional image, the representative cross-sectional image, and these on the display unit 36 only when there is a significant difference between the evaluation value 61 of the image finding and the representative characteristic value 64. Displays incidental information for.

  FIG. 5 is a diagram illustrating an example of the approximate cross-sectional image, the representative cross-sectional image, and their accompanying information displayed on the display unit 36. FIG. 5 shows an example 71 of an approximate cross-sectional image, an example 72 of a representative cross-sectional image, an example 73 of accompanying information of the approximate cross-sectional image, and an example 74 of auxiliary information of the representative cross-sectional image. The example of FIG. 5 shows a significant difference (for example, a value) between the evaluation value of the image finding (“user input value = 3”) and the representative characteristic value (“characteristic value calculated by the apparatus = 4.4”). Since there is a difference exceeding 1), it is displayed on the display unit 36. Since the approximate cross-sectional image 71 and the representative cross-sectional image 72 are displayed to the user, the user can confirm the difference between the cross-sectional images with his / her own eyes, and can review the image findings efficiently. . Therefore, there is an effect that the objectivity and reliability of the image findings determined by the doctor can be improved.

  The above is an example of a typical embodiment of the present invention, but the present invention is not limited to the embodiment described above and shown in the drawings, and can be appropriately modified and implemented without departing from the scope of the present invention. .

<Other embodiments>
The present invention is also realized by executing the following processing. That is, software (program) that realizes the functions of the above-described embodiments is supplied to a system or apparatus via a network or various storage media, and a computer (or CPU, MPU, etc.) of the system or apparatus reads the program. It is a process to be executed.

Claims (5)

Image finding acquisition means for acquiring from the user the type and evaluation value of an image finding related to a predetermined three-dimensional region of the three-dimensional image to be diagnosed;
A characteristic value calculating means for calculating a characteristic value that can be compared with the evaluation value of the image finding based on an image feature from each of a plurality of cross-sectional images included in the predetermined three-dimensional region;
Representative characteristic value calculating means for calculating a representative characteristic value representing the predetermined three-dimensional region from a plurality of characteristic values calculated from each of the plurality of cross-sectional images;
Representative cross-sectional image acquisition means for acquiring the cross-sectional image from which the representative characteristic value has been calculated as the representative cross-sectional image from the plurality of cross-sectional images;
Comparison means for comparing the evaluation value of the image findings and the representative characteristic value;
An image diagnosis support apparatus comprising: display means for displaying the representative cross-sectional image only when the comparison result of the comparison means has a significant difference. An approximate characteristic value selection means for selecting an approximate characteristic value closest to the evaluation value of the image findings from a plurality of characteristic values calculated from each of the plurality of cross-sectional images;
An approximate cross-sectional image acquisition means for acquiring the cross-sectional image from which the approximate characteristic value is calculated as the approximate cross-sectional image from the plurality of cross-sectional images;
The image diagnosis support apparatus according to claim 1, wherein the display unit displays both the approximate slice image and the representative slice image only when there is a significant difference in the comparison result of the comparison unit.   By providing a plurality of cross-sectional images for each of a plurality of predetermined directions as the plurality of cross-sectional images, the representative cross-sectional image acquisition unit and / or the approximate cross-sectional image acquisition unit can obtain a cross-sectional image in an arbitrary direction. The image diagnosis support apparatus according to claim 1, wherein the representative cross-sectional image and / or the approximate cross-sectional image can be acquired from the image. An image finding acquisition step of acquiring from the user the type and evaluation value of an image finding relating to a predetermined 3D region of the 3D image to be diagnosed;
A characteristic value calculating step of calculating a characteristic value that can be compared with the evaluation value of the image finding based on an image feature from each of a plurality of cross-sectional images included in the predetermined three-dimensional region;
A representative characteristic value calculating step of calculating a representative characteristic value representing the predetermined three-dimensional region from a plurality of characteristic values calculated from each of the plurality of cross-sectional images;
A representative cross-sectional image acquisition step of acquiring the cross-sectional image from which the representative characteristic value is calculated as a representative cross-sectional image from the plurality of cross-sectional images;
A comparison step of comparing the evaluation value of the image findings and the representative characteristic value;
And a display step of displaying the representative cross-sectional image only when there is a significant difference in the comparison result of the comparison step. An image finding acquisition step of acquiring from the user the type and evaluation value of an image finding relating to a predetermined 3D region of the 3D image to be diagnosed;
A characteristic value calculating step of calculating a characteristic value that can be compared with the evaluation value of the image finding based on an image feature from each of a plurality of cross-sectional images included in the predetermined three-dimensional region;
A representative characteristic value calculating step of calculating a representative characteristic value representing the predetermined three-dimensional region from a plurality of characteristic values calculated from each of the plurality of cross-sectional images;
A representative cross-sectional image acquisition step of acquiring the cross-sectional image from which the representative characteristic value is calculated as a representative cross-sectional image from the plurality of cross-sectional images;
A comparison step of comparing the evaluation value of the image findings and the representative characteristic value;
The program for making a computer implement | achieve the image diagnosis assistance method which has a display process which displays the said representative cross-sectional image only when the comparison result of the said comparison process has a significant difference.

Images