Disease determination assistance device, disease determination assistance method, and program

Abstract

The purpose of the present invention is to facilitate verification of the validity of an inference. This disease determination assistance device processes image data obtained by imaging the maxillofacial region by at least one of X-ray imaging and MRI imaging, and provides information for determining a disease, the disease determination assistance device comprising a storage device for storing image data, a processing device, and a display device for displaying information generated by the processing device. The processing device identifies an abnormality determination evaluation relating to an abnormal region on the basis of the image data, estimates a disease in a region of interest including the abnormal region on the basis of the abnormality determination evaluation, and simultaneously or selectively displays, on the display device, the abnormality determination evaluation and the name of the disease.

Description

Disease determination support apparatus, disease determination support method and program

[0001] The present disclosure relates to technology for supporting disease determination.

[0002] Patent Document 1 discloses a technique of inputting an image into a trained model to infer the position of a lesion area or the stage of a lesion.

[0003] Non-Patent Document 1 discloses a tool for inferring lesions by AI analysis.

[0004] Japanese Unexamined Patent Application Publication No. 2021-174394

[0005] DGNCT LLC "CBCT images", Internet, [retrieved on February 19, 2024] (https: / / diagnocat.com / us / products / radiology-report / )

[0006] Even if only the inference result is presented, it may be difficult to examine the validity of the inference.

[0007] Even if the position of a lesion area is indicated along with an inference result, if it is unclear what characteristics of the lesion area the lesion inference is based on, it may still be difficult to examine the validity of the inference.

[0008] Accordingly, an object of the present disclosure is to facilitate verification of the validity of inference.

[0009] The disease determination support apparatus according to the present disclosure is a disease determination support apparatus that provides information for determining a disease by processing image data obtained by imaging the maxillofacial region by at least one of X-ray imaging and MRI imaging, comprising: a storage device that stores the image data; a processing device; and a display device that displays information generated by the processing device, wherein the processing device specifies an abnormality determination evaluation related to an abnormal region based on the image data, estimates the disease in a region of interest including the abnormal region based on the abnormality determination evaluation, and displays the abnormality determination evaluation and the name of the disease simultaneously or selectively on the display device.

[0010] The disease determination support method relating to this disclosure is a disease determination support method that provides information for determining a disease by processing image data obtained by taking X-ray images and MRI images of the maxillofacial region, and includes identifying an abnormality determination evaluation regarding an abnormal area based on the image data, estimating the disease in the area of ​​interest including the abnormal area based on the abnormality determination evaluation, and outputting the abnormality determination evaluation and the name of the disease in a manner that can be recognized by the user.

[0011] The program relating to this disclosure is a program that provides information for determining a disease by processing image data obtained by imaging the maxillofacial region using at least one of X-ray and MRI, and causes a computer to perform the following processes: identify an abnormality judgment evaluation regarding an abnormal area based on the image data; estimate the disease in the area of ​​interest including the abnormal area based on the abnormality judgment evaluation; and output the abnormality judgment evaluation and the name of the disease in a manner that can be recognized by the user.

[0012] According to this disclosure, the validity of the inference can be easily verified.

[0013] Figure 1 is a schematic diagram showing a disease diagnosis support device connected to a CT scanner according to the first embodiment. Figure 2 is a block diagram showing the electrical configuration of the disease diagnosis support device and the CT scanner. Figure 3 is a functional block diagram of the calculation circuit. Figure 4 is a block diagram showing an example of application of a trained model. Figure 5 is a block diagram showing a training example. Figure 6 is a flowchart showing an example of processing by the disease diagnosis support device. Figure 7 is a diagram showing an example of display on a display device. Figure 8 is a functional block diagram of the calculation circuit according to the second embodiment. Figure 9 is a diagram showing the position of the dentition constituent tissues in the maxillofacial region. Figure 10 is a partial enlarged view of the teeth and jawbone. Figure 11 is a block diagram showing an example of application of a trained model. Figure 12 is an explanatory diagram showing a dentition unfolded image. Figure 13 is an explanatory diagram showing the dentition constituent tissue region. Figure 14 is an explanatory diagram showing an example of generating a dentition unfolded image. Figure 15 is a flowchart showing an example of processing by the disease diagnosis support device. Figure 16 is a functional block diagram showing the data flow. Figure 17 is a diagram showing an example of display on a display device. Figure 18 is a diagram showing an example of display on a display device. Figure 19 is a diagram showing an example of display on a display device. Figure 20 is an explanatory diagram showing the state in which the head is positioned by the head holding part. Figure 21 is an explanatory diagram showing the positional relationship between the head retaining part and the dental arch constituent tissues. Figure 22 is a schematic diagram showing an image processing device according to a modified example. Figure 23 is a block diagram showing the electrical configuration of the same image processing device.

[0014] {First Embodiment} The disease determination support device, disease determination support method, and program according to the first embodiment will be described below.

[0015] <About the disease diagnosis support device and CT scanning device> Figure 1 is a schematic diagram showing the disease diagnosis support device 20 connected to the CT scanning device 10.

[0016] The CT scanner 10 is an example of an imaging device that obtains image data by imaging the maxillofacial region. The CT scanner 10 is an example of an imaging device that obtains three-dimensional image data by X-ray imaging the maxillofacial region. In this embodiment, the CT scanner 10 is an imaging device that obtains image data by CT scanning (Computed Tomography) of the dentition tissue. The obtained image data includes data related to the dentition tissue. The disease determination support device 20 processes the image data and provides information for determining a disease.

[0017] In the illustrated example, a CT scanner 10 is shown for explanation, but the scanning device may be a panoramic X-ray scanner that takes panoramic X-ray images of the maxillofacial region. Alternatively, the scanning device may be an intraoral X-ray scanner that takes intraoral images by placing film, an imaging plate, a two-dimensional X-ray sensor, etc., inside the mouth. Image data obtained from CT scanning, image data obtained from panoramic X-ray scanning, and image data obtained from intraoral X-ray scanning are all image data obtained from X-ray scanning and can be considered "X-ray image data."

[0018] Furthermore, the imaging device may be an MRI imaging device 500 that performs MRI imaging on the maxillofacial region. The image data to be processed in this disclosure may be image data obtained by performing X-ray imaging and MRI imaging on the maxillofacial region.

[0019] For example, the CT scanner 10 comprises an X-ray generator 11, an X-ray detector 12, a swivel arm 13, a support column 14, and an imaging processing unit 15. In the space where the CT scanner 10 exists, a three-dimensional coordinate system is calculated. The three-dimensional coordinate system is set, for example, with the orientation of the subject positioned for imaging as the reference, and includes a Z-direction along the body axis, an X-direction perpendicular to the Z-direction and along the left-right direction of the subject, and a Y-direction perpendicular to the Z-direction and the X-direction and along the front-back direction of the subject. In this embodiment, since the subject is positioned in a standing position, the Z-direction is the direction perpendicular to the floor surface, where the support column 14 extends.

[0020] The X-ray generator 11 includes an X-ray tube and is configured to emit an X-ray beam toward the object. The X-ray detector 12 includes an X-ray detection sensor. The X-rays emitted from the X-ray generator 11 pass through the object and are detected by the X-ray detector 12. The X-ray generator 11 may also be called an X-ray generator. The X-ray detector may also be called an X-ray detector.

[0021] The swivel arm 13 is, for example, a U-shaped member that opens downward. An X-ray generator 11 and an X-ray detector 12 are supported at each end of the swivel arm 13 in an opposing position. A subject can be placed between the X-ray generator 11 and the X-ray detector 12. The subject is a part of the human body including the maxillofacial region, i.e., the head including the jaw. The part of the human body including the maxillofacial region includes the tissues that make up the dentition. The swivel arm may also be called a supporter or a rotation arm.

[0022] The support column 14 is erected so as to extend along the direction of gravity (vertical direction). A cantilever arm 14a is supported on this support column 14 so as to be able to move up and down. A swivel arm 13 is supported on the cantilever arm 14a so as to be able to swivel. The height position of the swivel arm 13 is adjusted along the support column 14 to match the height position of the head. The cantilever arm may also be called a support bracket or a cantilever arm.

[0023] With the subject's head positioned between both ends of the swivel arm 13, the swivel arm 13 is rotated. This causes the X-ray generator 11 and X-ray detector 12 to rotate around the head. As a result, an X-ray CT scan of the head is performed, and image data is obtained.

[0024] For example, when the rotating arm 13 rotates, X-ray imaging is performed at small rotation angles. This yields X-ray projection image data (frame data) for each small rotation angle. Based on the set of X-ray projection image data (frame data) taken at different rotation angles, three-dimensional volume data of the subject is generated. This three-dimensional volume data is three-dimensional image data that shows the distribution of the X-ray absorption rate of the subject in a three-dimensional coordinate system.

[0025] The CT scanner 10 may include a head support section 9. The head support section 9 is the part that holds the head to be scanned. The head support section 9 may include a chin rest 9a that supports the chin. The chin rest 9a can support the front and lower part of the chin of the head. This positions the head in a fixed position in the front-back and up-down directions. The head support section 9 may also include side support sections 9b that position the head from both sides. The side support sections 9b may be, for example, ear rods that contact both ears of the head. The side support sections 9b position the head in a fixed position in the left-right direction. The head support section 9 may also be called a head holder.

[0026] The disease diagnosis support device 20 provides information for diagnosing a disease by processing three-dimensional image data. The three-dimensional image data may be three-dimensional volume data obtained by CT scanning of the maxillofacial region. The three-dimensional image data may also be data generated based on a group of X-ray projection image data for each minute rotation angle obtained by CT scanning, or a group of tomographic images. In other words, the three-dimensional image data is data based on data obtained by CT scanning. In this embodiment, an example is described in which three-dimensional image data is generated based on a group of X-ray projection image data for each rotation angle. In other words, here the three-dimensional image data is three-dimensional CT image data obtained by processing projection images obtained by scanning the maxillofacial region.

[0027] The disease diagnosis support device 20 comprises a processing unit 30, a display device 22, and user interfaces 24 and 26.

[0028] The processing unit 30 is comprised of a computer or workstation. The processing unit 30 is connected to the imaging processing unit 15 of the CT imaging device 10 by wired or wireless connection, and can send and receive various types of data with the imaging processing unit 15. In this embodiment, the processing unit 30 can receive data obtained from CT imaging from the imaging processing unit 15. Some or all of the functions of the processing unit 30 may be implemented by a cloud server.

[0029] The display device 22 is, for example, a liquid crystal display or an organic electroluminescent (EL) display, and is connected to the processing unit 30 by wire or wireless connection. The display device 22 displays information generated by the processing unit 30 based on the output for display from the processing unit 30. The display device 22 may also be referred to as a display.

[0030] User interfaces 24 and 26 are devices that receive instructions from the user (operator) of the disease diagnosis support device 20. User interface 24 may be, for example, a switch device such as a keyboard. User interface 26 may be, for example, a pointer device such as a mouse. If the user interface is a switch device, the user's instructions can be input through the switch device alone. If the user interface is a pointer device, the user's instructions can be input through operations on characters or images displayed on the display device 22. The user interface may also be a touch panel.

[0031] <Regarding the electrical configuration of the disease diagnosis support device and CT imaging device> Figure 2 is a block diagram showing the electrical configuration of the disease diagnosis support device 20 and the CT imaging device 10.

[0032] The CT imaging apparatus 10 comprises an imaging processing unit 15, an imaging unit drive mechanism 18, and a user interface 19.

[0033] The image processing unit 15 is comprised of a computer that includes an arithmetic circuit 16 and a storage device 17.

[0034] The arithmetic circuit 16 includes a processor 16a. The processor 16a may be a CPU (Central Processing Unit). The processor 16a may also include a GPU (Graphics Processing Unit).

[0035] The storage device 17 is composed of a non-volatile storage device such as flash memory or a hard disk drive. The storage device 17 may also be a memory circuit. The storage device 17 may also be called memory. The storage device 17 stores a program. The program describes the procedure for the CT scanning device 10 to perform a CT scan.

[0036] The imaging unit drive mechanism 18 is connected to the imaging processing unit 15. The imaging unit drive mechanism 18 includes a rotation drive mechanism for rotating the rotation arm 13. The rotation drive mechanism includes an actuator and a transmission mechanism. The actuator is an electric motor or the like that generates rotational driving force. The transmission mechanism is a gear and pulley or the like that transmits the rotational driving force of the actuator. The rotational driving force of the actuator is transmitted to the rotation arm 13 via the transmission mechanism, and the rotation arm 13 rotates at a timing and rotational speed corresponding to the command from the imaging processing unit 15.

[0037] The user interface 19 is an interface for giving instructions to the CT scanning device 10, and is a switch device or pointer device, etc. The user interface 19 is connected to the imaging processing unit 15. The user can give various instructions to the CT scanning device 10 through the user interface 19.

[0038] The X-ray generator 11 and X-ray detector 12 are also connected to the imaging processing unit 15. The X-ray generator 11 irradiates X-rays at timings and outputs corresponding to instructions from the imaging processing unit 15, and the X-ray detector 12 outputs the detection results to the imaging processing unit 15.

[0039] The processor 16a executes processing according to the program in the storage device 17, thereby controlling the rotational movement of the swivel arm 13 and the X-ray irradiation operation of the X-ray generator 11. For each minute rotation angle of the swivel arm 13, the detection result of the X-ray detector 12 is input to the imaging processing unit 15, and the processor 16a generates X-ray projection image data for each minute rotation angle based on the detection result.

[0040] The processor 16a may generate three-dimensional image data based on the X-ray projection image data group. Data 17a including the generated three-dimensional image data is stored in the storage device 17.

[0041] Note that the process of generating three-dimensional image data based on the data group obtained by X-ray CT imaging may be performed by another computer. For example, a data group obtained by X-ray CT imaging may be transmitted to the disease determination support apparatus 20, and three-dimensional image data may be generated in the disease determination support apparatus 20 based on the data group.

[0042] The disease determination support apparatus 20 includes a processing unit 30, a display device 22, and user interfaces 24 and 26.

[0043] The processing unit 30 is connected to the display device 22 and the user interfaces 24 and 26. The processing unit 30 can receive instructions from a user via the user interfaces 24 and 26. The processing unit 30 can control display on the display device 22.

[0044] The processing unit 30 is configured by a computer including an arithmetic circuit 32 serving as a processing device and a storage device 34.

[0045] The arithmetic circuit 32 includes a processor 32a. The processor 32a may be a CPU (Central Processing Unit). The processor 32a may include a GPU (Graphics Processing Unit) or a processor for AI (Artificial Intelligence).

[0046] The storage device 34 is configured by a non-volatile storage device such as a flash memory or a hard disk drive. The storage device 34 may be a storage circuit. The storage device 34 stores a program 34a and data 34b.

[0047] The program 34a describes procedures for the disease determination support apparatus 20 to provide information for determining a disease by processing three-dimensional image data.

[0048] The data 34b includes the three-dimensional image data transmitted from the CT imaging apparatus 10 and data generated by processing for providing the above-mentioned information, etc. After processing, the data 34b may be retained as historical data or may be deleted.

[0049] The processing unit 30 includes a connection port 36. The processing unit 30 is connected to the CT imaging apparatus 10 via the connection port 36. The connection port 36 may include a terminal connected to a signal line of a wired cable extending from the CT imaging apparatus 10 and a circuit for communication processing. The processing unit 30 and the CT imaging apparatus 10 may be connected wirelessly; in this case, the connection port 36 may include a circuit for wireless processing. The three-dimensional image data 17a of the CT imaging apparatus 10 is transmitted to the processing unit 30 via the connection port 36 and stored in the storage device 34.

[0050] The arithmetic circuit 32 reads the program 34a stored in the storage device 34 and executes the processing described in the program 34a, whereby the arithmetic circuit 32 can execute various processes for providing information for determining a disease, as will be described later. The processing unit 30 may control CT imaging performed by the CT imaging apparatus 10.

[0051] As shown in FIG. 3, the processing function implemented by the arithmetic circuit 32 reading the program 34a includes a determination support processing unit 33 that supports disease determination. The determination support processing unit 33 may include an abnormality determination evaluation unit 33a, a disease estimation unit 33b, and a display control unit 33c. It may be considered that the arithmetic circuit 32 has a function as the determination support processing unit 33, and the determination support processing unit 33 has respective functions as the abnormality determination evaluation unit 33a, the disease estimation unit 33b, and the display control unit 33c.

[0052] The abnormality determination evaluation unit 33a specifies an abnormality determination evaluation related to an abnormal region based on the three-dimensional image data. The abnormality determination evaluation may be an evaluation performed through image feature determination processing for at least one of the reason, basis, or estimation material leading to the result of disease estimation.

[0053] An abnormal area is, for example, an abnormal area in the dentition tissue. The dentition tissue is, for example, the tissue including the dentition and the alveolar bone. The dentition tissue may also include the dentition and the jawbone. The jawbone in the dentition tissue may be the jawbone in the area that supports the teeth, and may include not only the alveolar bone but also the surrounding area. The alveolar bone is a partial area of ​​the jawbone and may be understood as the part of the jawbone that constitutes the alveolar bone and supports the teeth. This understanding is in line with the explanation in the Dental Medicine Dictionary. The dentition tissue may have a horseshoe shape in plan view. Note that plan view may be considered a plan view in which the line of sight is along the direction from the head to the legs in the axial direction of the body axis. The dentition tissue may extend so as to have thickness in the buccolingual direction. The dentition tissue may be, for example, tissue that has at least the thickness of the dentition in the buccolingual direction. The dentition tissue may be tissue that has the thickness of the area of ​​the jawbone that supports the teeth. The dentition may include upper and lower dental arches and upper and lower jawbones that support them. Each of the upper and lower dental arches contains several teeth arranged in an arch. The upper jawbone has alveolar bone that supports the teeth of the upper dental arch. The lower jawbone has alveolar bone that supports the teeth of the lower dental arch.

[0054] An abnormal area in the dentition is a visually distinctive region in three-dimensional image data, and it is a region that can be identified as a distinctive region through computer processing using these visual features. The computer processing for determining and evaluating the abnormal area may be artificial intelligence-based processing or logic-based computer processing.

[0055] More specifically, an abnormal area is, for example, a region that shows an abnormal distribution of X-ray absorption rates, different from that of healthy dentition tissue. An abnormal area is, for example, a region of dentition tissue where a lesion has occurred, i.e., a lesional area. A lesion is a change caused by disease. A lesional area shows a distribution of X-ray absorption rates different from that of normal dentition tissue. A lesional area does not have to be a region directly caused by a lesion; it may be a region indirectly caused by a lesion. For example, if a tooth is filled with a filling as a result of caries, the area of ​​the filling may also be identified as an abnormal area.

[0056] Diseases include, for example, chronic suppurative apical periodontitis and periapical granuloma. In the case of apical periodontitis, a lesion area with lower X-ray absorption than the normal state and its surroundings occurs at the periphery of the tooth root. Therefore, if a lesion area with low X-ray absorption is widespread at the periphery of the tooth root, that lesion area can be identified as an abnormal area. For other lesion areas as well, the lesion area in the 3D image data can be identified as an abnormal area based on its location in the dentition tissue (e.g., tooth region or jawbone region), the spread pattern or distribution pattern of X-ray absorption, etc.

[0057] Abnormal areas can be distinguished by the names of disease-related features, such as lesion names. These disease-related feature names may include, for example, oval-shaped areas, proximal caries, fillings, impacted teeth, periodontal ligament space enlargement, or bone resorption. Oval-shaped areas indicate shadows that are nearly circular; proximal caries indicates caries near the pulp; fillings indicate fillings in teeth; and impacted teeth indicate teeth that are partially or completely embedded in the jawbone or gums. Periodontal ligament space enlargement is a condition where the periodontal ligament space is larger than normal, and bone resorption is a condition where the alveolar bone supporting the tooth is dissolved and absorbed. These lesion names can also be identified based on their location within the dentition (e.g., tooth region or jawbone region), the spread pattern or distribution pattern of X-ray absorption, etc. Note that when multiple abnormal areas are identified, these areas may partially or completely overlap.

[0058] An abnormality assessment may include an assessment of the presence or absence of an abnormal area corresponding to at least one lesion. In other words, an abnormality assessment may be understood as including an assessment of the presence or absence of one or more lesions. For example, if there is a cavity in the alveolar cavity that is not normally seen, it is considered abnormal because it is different from the norm. An abnormality assessment may also include an assessment of the degree of the abnormal area. The degree of the abnormal area may be, for example, the size, length, shape, or X-ray absorption rate of the abnormal area. In other words, the degree of the abnormal area indicates the size, length, shape, or degree of the lesion area, such as a cavity, inflammation area, or border abnormal area. An abnormality assessment may also include an assessment of the condition of the abnormal area. The condition of the abnormal area may be, for example, the clarity of the boundary of the abnormal area. The degree or condition of the abnormal area may be evaluated according to the nature of each lesion. For example, for a lesion that is roughly oval-shaped, the degree of its longest diameter may be evaluated as the degree of the abnormal area. For a lesion that is roughly oval-shaped, the clarity of its boundary may be evaluated as the condition of the abnormal area.

[0059] Anomaly detection evaluations may include cases where the result is unknown. For example, if there is a possibility of an anomaly area existing, but the degree of that possibility is unknown, it may be evaluated as unknown. If it is difficult to classify the degree or circumstances of the anomaly area into a categorized classification, it may be evaluated as unknown which classification it belongs to.

[0060] Anomaly detection evaluations may be digitized so that they can be objectively recognized by humans. Such digitization may be called human-based digitization. By providing users with human-based digitized anomaly detection evaluations, it becomes possible to judge the validity of the anomaly detection evaluations. Furthermore, the anomaly detection evaluations can be modified as needed.

[0061] Anomaly detection may be performed, for example, by applying a pre-trained model 40 that has been machine-learned to perform anomaly detection.

[0062] The trained model may be composed of, for example, a model suitable for image recognition, such as a multilayer neural network. The trained model may also be trained using, for example, a training dataset that includes three-dimensional data of abnormal areas based on actual CT images and the name, degree, and condition of the lesion associated with the abnormal area.

[0063] By applying three-dimensional image data to a trained model, for example, the presence, extent, and circumstances of an abnormal area where a lesion has occurred can be evaluated. If the trained model outputs a score regarding the reliability of the fit for the presence or absence of an abnormal area, a probability corresponding to that score may be displayed on the display device 22. Depending on the score, in addition to the presence or absence of an abnormal area, an unknown status may also be evaluated.

[0064] The disease estimation unit 33b performs disease estimation processing, which is the process of estimating diseases in the area of ​​interest that includes abnormal areas based on the abnormality judgment evaluation. If no abnormal areas exist, disease estimation processing does not need to be performed.

[0065] In this embodiment, the region of interest is a region that includes an abnormal region. That is, the region of interest is a region that is positive with respect to an abnormality. The region of interest may be the abnormal region itself, or it may be a region that is wider than the abnormal region. The region of interest may include multiple abnormal regions.

[0066] More specifically, the area of ​​interest may be a region distinguished on a tooth-by-tooth basis, for example. If a tooth has multiple root canals, the area of ​​interest may be a region further subdivided on a root canal-by-root-canal basis. The tooth and the alveolar bone may be considered the same region or separate regions.

[0067] The region of each tooth in the 3D image data can be identified, for example, by referring to a standard tooth position, a tooth position set by the user, or by extracting the tooth region based on the 3D image data. By referring to the identified region of each tooth, it can be determined which tooth's region of interest the identified region of interest belongs to. In other words, the abnormality detection evaluation described above is associated with a region of interest corresponding to any of the teeth.

[0068] The area of ​​focus corresponding to each tooth may include one abnormality assessment or multiple abnormalities.

[0069] The disease estimation process performed by the above-mentioned calculation circuit 32 as a disease estimation unit 33b may be carried out, for example, by applying a machine learning-trained model 40, as shown in Figure 4.

[0070] The trained model 40 is, for example, a model suitable for solving classification problems, such as a multilayer neural network or an SVM (Support Vector Machine), and is stored in the memory device 34. At least one anomaly judgment evaluation associated with any of the areas of interest is input to the disease estimation unit 33b. The arithmetic circuit 32 then reads the program and parameters written in the trained model 40 and performs identification processing to identify the name of the disease in the area of ​​interest based on the input at least one anomaly judgment evaluation.

[0071] The trained model 40 is generated by a machine learning device 50, which is comprised of a computer equipped with a storage device that stores the training model 40B and a processor as a circuit that functions as a model generation unit 52, as shown in Figure 5. The model generation unit 52 learns a process to infer the name of a disease from at least one anomaly judgment evaluation based on a training dataset. The training dataset can be a plurality of datasets in which disease names are associated with anomaly judgment evaluations. Then, a trained model 40 that infers the name of a disease based on anomaly judgment evaluations is generated from the training dataset.

[0072] Here, the abnormality judgment evaluation in the data used as the training dataset may be an evaluation by the abnormality judgment evaluation unit 33a or an evaluation by the trainer. In the data used as the training dataset, the disease name associated with the abnormality judgment evaluation may be a disease name identified by the trainer after looking at the abnormality judgment evaluation, or a disease name identified separately by a computer for disease name estimation.

[0073] The probability of the estimated disease name may be calculated. For example, if the trained model 40 outputs a score regarding the suitability of the estimated disease, the probability of the disease may be output according to that score.

[0074] The display control unit 33c performs the process of simultaneously or selectively displaying the abnormality judgment evaluation and the name of the disease on the display device 22. If there are multiple candidate disease names, multiple names may be displayed. In this case, the names may be displayed in order of probability from highest to lowest.

[0075] <Examples of processing by the disease diagnosis support device> Examples of processing by the disease diagnosis support device 20 will be explained with reference to the flowchart in Figure 6.

[0076] For example, an instruction to identify the disease name is given to the processing unit 30 via the user interface 24, 26, etc. Then, in step S1, the arithmetic circuit 32 reads three-dimensional image data from the storage device 34.

[0077] In the next step S2, the calculation circuit 32 performs an abnormality determination evaluation process based on the three-dimensional image data. This identifies the abnormal area in the three-dimensional image data. The abnormal area is identified along with the name and location of the lesion. As described above, the abnormal area may also include the degree or state of the abnormality.

[0078] The identified anomaly detection evaluations are associated with the region of interest. In other words, if an anomaly exists in the 3D image data, at least one anomaly detection evaluation is associated with at least one region of interest.

[0079] In the next step S3, for the region of interest which includes at least one abnormal region, the name of the disease is estimated based on at least one abnormality judgment evaluation corresponding to that at least one abnormal region.

[0080] In the next step S4, control is performed to display the abnormality judgment evaluation and disease name on the display device 22.

[0081] Figure 7 shows an example of a display on the display device 22.

[0082] In the diagram, the disease name is displayed at the top of the screen. The abnormality assessment items may include multiple items, such as "circular shape" and "tooth structure," as shown in the diagram. Also, for the item indicating the presence or absence of a circular shape, which is an example of an abnormal area, "present," "absent," or "unknown" may be displayed. As an example of the degree of abnormality assessment, the dimension item for the major axis direction of the circular shape may be displayed. For example, for whether the "major axis is within 8 mm," "present," "absent," or "unknown" may be displayed. Also, as an example of the state of abnormality assessment, the item indicating whether the boundary of the circular shape is unclear may be displayed. For example, for whether the "boundary is unclear," "present," "absent," or "unknown" may be displayed.

[0083] Furthermore, as examples of assessments for abnormalities related to tooth structure, the following items are displayed: "cariform caries near the pulp," "large fillings in the crown," "presence of a tooth," and "impacted tooth," with the options being "present," "absent," or "unknown."

[0084] In Figure 7, "Yes," "No," and "Unknown" are displayed for each item of the abnormality judgment evaluation. The determined evaluation for each item is displayed in a visually identifiable manner. For example, one of "Yes," "No," and "Unknown" may be distinguished from the others by increasing or decreasing the density of its background color. Alternatively, the text color of each item may be increased or decreased to distinguish it from the others. A mark may be displayed to distinguish the selectively determined evaluation from the others, the background or text of the selectively determined evaluation may flash, or only the determined evaluation may be displayed.

[0085] In the example shown in Figure 7, the disease name is displayed as "periapical cyst." In addition, a halftone pattern is added to the background of the selectively determined evaluation items. Therefore, "approximately oval" is displayed as "present," "longest diameter within 8 mm" is displayed as "absent," "unclear boundary" is displayed as "absent," "cariform caries near the pulp" is displayed as "present," "large filling in the crown" is displayed as "present," "is there a tooth?" is displayed as "present," and "impacted tooth" is displayed as "absent."

[0086] In the example above, the anomaly detection evaluation displays not only whether or not an anomaly exists, but also the degree and circumstances of the anomaly. However, the anomaly detection evaluation may also only show whether or not an anomaly exists.

[0087] Furthermore, in the above example, assuming the existence of an "oval-shaped" abnormality, "carcinosis near the pulp," "large filling in the crown," "presence or absence of a tooth," and "impacted tooth" are displayed regardless of whether or not they are abnormal. These abnormalities are referenced when estimating a disease based on a specific abnormality, and may be called reference abnormalities.

[0088] Users who view the display device 22 can understand that, based on the evaluation of each item in the above abnormality judgment evaluation, the disease name is estimated to be a periapical cyst.

[0089] In Figure 7, the disease name and abnormality assessment are displayed simultaneously on the display device 22. The disease name and abnormality assessment may be selectively displayed on the display device 22. For example, when a tomographic image or panoramic image based on CT image data is displayed on the display device 22, the disease name and abnormality assessment may be selectively displayed in the remaining area of ​​the display device 22. Furthermore, the disease name may be editable by the user via the user interface 24, 26, etc.

[0090] The result of the abnormality assessment may be referred to as a "positive abnormality assessment" when the element of abnormality in the target area is large (high probability or degree), and as a "negative abnormality assessment" when the element of abnormality in the target area is small (low probability or degree). The case of a large element of abnormality includes cases where the area is abnormal or has an abnormality. The case of a small element of abnormality includes cases where the area is not abnormal or does not have an abnormality.

[0091] <Effects> With the disease determination support device 20, disease determination support method, and disease determination support program 34a configured as described above, the name of the estimated disease and the abnormality judgment evaluation used as the basis for estimating the disease are displayed on the display device 22 simultaneously or selectively. Therefore, users can recognize the abnormality judgment evaluation and verify the validity of the estimated disease name, making it easier to verify the validity of the inference.

[0092] For example, users can observe tomographic or panoramic images based on separately displayed CT image data to verify the validity of the abnormality assessment. If the abnormality assessment is deemed valid, they can then make a final diagnosis, recognizing that the inferred disease is also likely to be valid.

[0093] Furthermore, for example, if the abnormality assessment is deemed invalid based on the image observation results, the user can verify the validity of the disease name while being aware that the abnormality assessment is invalid. The user can consider the possibility of other diseases, taking into account the invalid abnormality assessment while referring to the estimated disease name.

[0094] Furthermore, if the 3D image data is 3D CT image data obtained by CT scanning, the disease can be easily estimated by scanning the CT scan.

[0095] {Second Embodiment} A disease determination support device, disease determination support method, and program according to the second embodiment will be described. In this description of the second embodiment, the differences from the first embodiment will be mainly explained. Components similar to those described in the first embodiment may be denoted by the same reference numerals and their descriptions may be omitted.

[0096] As a CT imaging apparatus for obtaining three-dimensional CT image data, an apparatus with the same configuration as the CT imaging apparatus 10 described in the first embodiment with reference to Figures 1 and 2 may be used.

[0097] The disease determination support device 20 described in the first embodiment has a program 34a implemented for the implementation of the second embodiment.

[0098] In the second embodiment, Figure 8 shows the processing function realized by the arithmetic circuit 132, which corresponds to the arithmetic circuit 32, reading the program 34a. The processing function of the arithmetic circuit 132 includes a judgment support processing unit 133 that assists in the determination of a disease. In this embodiment, the judgment support processing unit 133 includes a focus area extraction unit 133a, an abnormality determination evaluation unit 133b, a disease estimation unit 133c, a panoramic image generation unit 133d, and a display control unit 133e. It may also be considered that the arithmetic circuit 132 has the function of the judgment support processing unit 133, and the judgment support processing unit 133 has the functions of the focus area extraction unit 133a, an abnormality determination evaluation unit 133b, a disease estimation unit 133c, a panoramic image generation unit 133d, and a display control unit 133e.

[0099] The area of ​​interest extraction unit 133a extracts areas of interest from three-dimensional image data. An area of ​​interest is a unit area that is subject to abnormality judgment evaluation. For example, the three-dimensional image data is data obtained by CT imaging of the maxillofacial region and includes the dentition tissue. The dentition tissue includes at least one area of ​​interest and may include multiple areas of interest. It is preferable that abnormality judgment evaluation be performed on multiple areas of interest units.

[0100] Figure 9 shows the location of the dentition tissues E in the maxillofacial region F. Figure 10 is a magnified view of a tooth T and jawbone Hj.

[0101] The cylindrical region is the CT imaging region Ect. The dentition tissues E are located within the CT imaging region Ect. The region Ee, where the dentition tissues E extend, includes the dental arch Ht, where multiple teeth T are arranged, and the upper and lower jawbones Hj. The dentition tissues E may also include the temporomandibular joint Hjo. In the three-dimensional image data, the region Ee where the dentition tissues E extend may be called the dentition tissue region Ee. The dentition tissue region Ee is a horseshoe-shaped region set in calculations, and it is sufficient if it is a region that matches the shape and position of the dentition tissues. The shape of the dentition tissue region Ee may match the shape of the dentition tissues E of a standard skeleton, and if the shape of the dentition tissues E of an individual is known, it may be set to a shape that matches that individual.

[0102] The area of ​​interest R is, for example, a unit divided by tooth T. The alveolar bone Hja may be divided by the supporting tooth T. For example, the upper and lower alveolar bone Hja may be divided by the adjacent region above or below each tooth T. Alternatively, for example, the upper and lower alveolar bone Hja may be divided into the upper extension region unit for each upper tooth T and the lower extension region unit for each lower tooth. For example, suppose there is a tooth T1 in the upper jaw, and there is an alveolar AL1 that supports this tooth T1. The portion of the alveolar bone that forms this alveolar AL1 is the region unit alveolar bone Hja1. If there is a tooth T2 next to tooth T1, the portion of the alveolar bone that forms the alveolar AL2 that supports this tooth T2 is the region unit alveolar bone Hja2 adjacent to region unit alveolar bone Hja1. The area of ​​interest may also be divided by tooth root Tr.

[0103] The tooth T and the alveolar bone Hja supporting it may be considered the same region of focus R, or they may be considered different regions of focus R. Note that the alveolar bone Hja is part of the jawbone. The region of focus R may also include the portion of the jawbone Hj surrounding the alveolar bone Hja.

[0104] In this embodiment, an example is described in which the tooth T and the alveolar bone Hja supporting the tooth T are considered to be the same region of interest R.

[0105] Extraction of the region of interest R may be performed, for example, by applying a trained machine learning model. For example, a large amount of data is prepared as training data, in which tooth regions and alveolar bone regions in the upper and lower jawbones are mapped onto 3D image data. Using this training data, a trained model is prepared that is trained to segment tooth regions and alveolar bone regions in the upper and lower jawbones in the 3D image data. The trained model may be, for example, a model trained based on a semantic segmentation algorithm. By applying the above trained model to the 3D image data, each region of multiple teeth T and each region of multiple alveolar bones Hja are identified in the 3D coordinate system. This makes it possible to distinguish between the regions of multiple teeth T and the regions of multiple alveolar bones Hja corresponding to each tooth T based on the 3D image. Then, the region of interest R is extracted based on the distinguished regions of multiple teeth T and the regions of multiple alveolar bones Hja corresponding to each tooth T.

[0106] Furthermore, if each tooth T and jawbone Hj are extracted, the above-mentioned dentition constituent tissue E will also be extracted, making it easy to generate the panoramic image described later.

[0107] The identification of each region of multiple teeth T and each region of multiple alveolar bone Hja in three-dimensional image data may be performed by a pre-defined region extraction process, such as a pattern matching process.

[0108] Furthermore, the dentition tissues E can also be distinguished by tissues such as the root canal, apex, or mandibular canal. Therefore, the area of ​​interest may be defined as a region distinguished by tissues such as the root canal, apex, or mandibular canal.

[0109] The anomaly determination and evaluation unit 133b identifies an anomaly determination and evaluation related to the anomaly area based on the three-dimensional image data.

[0110] The abnormality detection evaluation is the evaluation of the image characteristics of living organisms in image data such as 3D image data. The abnormality detection evaluation unit 133b does not estimate diseases.

[0111] Furthermore, the abnormality determination evaluation may be performed by determining the characteristics of image features related to at least one of the reasons, grounds, or estimation materials that lead to the disease estimation result.

[0112] As previously mentioned, the abnormal area is, for example, an abnormal area in the tissues that make up the dentition.

[0113] The anomaly determination and evaluation unit 133b, like the anomaly determination and evaluation unit 33a, may identify an anomaly determination and evaluation for the entire 3D image data. In this case, a process may be performed to determine which of the multiple areas of interest R the anomaly area corresponding to the identified anomaly determination and evaluation belongs to. This process can be performed, for example, by referring to the location of the extracted area of ​​interest R and the anomaly area in a 3D coordinate system. Alternatively, for example, the anomaly area may be associated with the area of ​​interest corresponding to a standard tooth by referring to the location of a pre-set standard tooth and the anomaly area in a 3D coordinate system.

[0114] As in the first embodiment, when the result of the abnormality judgment evaluation indicates that the element of abnormality in the target area is large (high probability or degree), it may be called a "positive abnormality judgment evaluation," and when the element of abnormality in the target area is small (low probability or degree), it may be called a "negative abnormality judgment evaluation." An evaluation in which it is unclear whether it is on the negative or positive side may be called a "general abnormality judgment evaluation."

[0115] Anomaly detection evaluations may be digitized so that they can be objectively recognized by humans. Such digitization may be called human-based digitization. By providing users with human-based digitized anomaly detection evaluations, it becomes possible to judge the validity of the anomaly detection evaluations. Furthermore, as will be discussed later, anomaly detection evaluations may be modified as needed.

[0116] The X-ray projection image data that is the subject of a reconstruction may be called the "original X-ray projection image data," and the X-ray CT scan from which the original X-ray projection image data was obtained may be called the "original X-ray CT scan." Alternatively, the reconstruction may be called the "current reconstruction," and the three-dimensional image data generated as a result of the current reconstruction may be called the "current three-dimensional image data." The abnormality detection evaluation may be performed using the current three-dimensional image data as the processing target.

[0117] If we change the order of the explanation, it will be as follows:

[0118] First, an X-ray CT scan (original X-ray CT scan) is performed. Next, X-ray projection data (original X-ray projection image data) is obtained. Then, a reconstruction (current reconstruction) process (RC1) is performed to generate 3D image data (current 3D image data). Finally, the 3D image data (current 3D image data) is subjected to an abnormality detection evaluation process.

[0119] The anomaly determination evaluation unit 133b may identify an anomaly determination evaluation based on the 3D image data in units of the extracted area of ​​interest R. An anomaly determination evaluation can be identified in association with the area of ​​interest R.

[0120] Anomaly detection and evaluation can be achieved using artificial intelligence or logic-based computer processing, as described in the first embodiment.

[0121] The disease estimation unit 133c estimates a disease based on at least one abnormality judgment evaluation for the area of ​​interest R.

[0122] The disease estimation process may be based solely on abnormality assessment, or it may be based on abnormality assessment and other information.

[0123] Other information may include data contained in the 3D image data, data based on the analysis of the 3D image data, or information including the results of human judgment.

[0124] The estimation of the disease is a process that is carried out using, at least, the abnormality judgment evaluation performed by the abnormality judgment evaluation unit 133b with respect to the image characteristics as material.

[0125] In this embodiment, an example of estimating a disease based on abnormality detection evaluation and reference region image data that includes the abnormal region among three-dimensional image data is described.

[0126] In other words, the disease estimation unit 133c estimates the disease in the area of ​​interest R, which includes the abnormal area, based on the results of the abnormality judgment evaluation. At this time, the abnormality judgment evaluation for at least one abnormal area associated with the area of ​​interest R is referenced. Reference area image data included in the area of ​​interest R is also referenced.

[0127] The reference region image data includes the abnormal region that is subject to abnormality judgment evaluation. More specifically, the reference region image data includes the abnormal region that has been judged as positive abnormality. The reference region image data may be the abnormal region itself, or it may be a wider area than the abnormal region, for example, an area including the area surrounding the abnormal region. If the region of interest R includes multiple abnormal regions, all of the abnormal regions may be referenced, or some of the abnormal regions may be referenced. The reference region image data may be the region of interest R itself.

[0128] The 3D image data used for the abnormality detection evaluation and the reference area image data do not need to be the same data; they just need to be image data obtained from the same maxillofacial region. For example, if the abnormality detection evaluation is performed on 3D image data 3D1 (current 3D image data) generated by reconstructing X-ray projection data PD1 (original X-ray projection image data) obtained from X-ray CT scanning CT1 (original X-ray CT scanning) using a certain reconstruction RC1 (current reconstruction) process, then the reference area image data may be image data obtained from the same X-ray projection data PD1 (original X-ray projection image data). Furthermore, 3D image data obtained from CT scanning may be used for the abnormality detection evaluation, while panoramic images or simple transmission X-ray images may be used as the reference area images.

[0129] The 3D image data used for the anomaly detection evaluation and the reference area image data may be the same image data. For example, if the anomaly detection evaluation is performed on 3D image data 3D1 (current 3D image data) generated by reconstructing X-ray projection data PD1 (original X-ray projection image data) obtained by performing an X-ray CT scan CT1 (original X-ray CT scan) using a certain reconstruction RC1 (current reconstruction) process, then the 3D image data 3D1 (current 3D image data) generated by reconstructing using the reconstruction RC1 (current reconstruction) process may be used as the reference image data.

[0130] The abnormality detection of image data by the abnormality detection evaluation unit 133b and the abnormality detection of image data by the disease estimation unit 133c may use different algorithms. The accuracy of the disease estimation process can be improved by the disease estimation unit 133c referencing the reference range image data using its own algorithm.

[0131] The image data used for abnormality detection and the reference region image data may be image data obtained by photographing the same maxillofacial region.

[0132] For example, if a certain X-ray imaging performed on a certain maxillofacial region is referred to as "original X-ray imaging," and the X-ray image data obtained from the original X-ray imaging XE1 is referred to as "original X-ray image data," then image data generated from the original X-ray image data Xi1 may be used for abnormality detection evaluation, and image data generated from the same original X-ray image data Xi1 may be used as reference area image data.

[0133] Furthermore, for example, if a certain X-ray CT scan performed on a certain maxillofacial region is referred to as the "original X-ray CT scan," and the X-ray projection data obtained from the original X-ray CT scan CT1 is referred to as the "original X-ray projection image data," then the image data generated by reconstructing the original X-ray projection image data PD1 may be used for abnormality detection evaluation, and the same image data generated by reconstructing the original X-ray projection image data PD1 may be used as the reference area image data.

[0134] For example, if a certain reconstruction that processes original X-ray projection image data is called "current reconstruction," the image data generated by current reconstruction RC1 in the abnormality detection evaluation may be used, and the image data generated by another current reconstruction RC2 may be used as reference domain image data.

[0135] Alternatively, the image data generated by the current reconstruction RC1 used in the anomaly detection evaluation may be used, and the same image data generated by the current reconstruction RC1 may be used as the reference domain image data.

[0136] Furthermore, when the 3D image data generated by reconstructing with a certain current reconstruction is referred to as "current 3D image data," the current 3D image data generated by current reconstruction RC1 in the anomaly detection evaluation may be used, and the current 3D image data generated by another current reconstruction RC2 may be used as the reference domain image data.

[0137] Alternatively, the current 3D image data 3D1 generated by the current reconstruction RC1 in the anomaly detection evaluation may be used, and the same current 3D image data 3D1 may be used as the reference area image data.

[0138] Let's assume that the information from the abnormality judgment evaluation is information ABa, and the information from the reference image area data at that time is information IMA. For example, if the disease estimation unit 133c has learned that when information ABa and information IMA are in an AND (logical conjunction) state, then when information ABa and information IMA for a given subject are in an AND (logical conjunction) state, the disease estimation unit 133c can determine that the disease name for that subject is disease name IMA.

[0139] The disease estimation process performed by the above-mentioned calculation circuit 132 as a disease estimation unit 133c may be carried out, for example, by applying a trained model 140, as shown in Figure 11.

[0140] As with the first embodiment, a model suitable for solving the classification problem can be used as the pre-trained model 140. In the example in Figure 11, the pre-trained model 140 differs from that of the first embodiment in that it is trained to infer disease names based on abnormality judgment evaluation and reference range image data.

[0141] In other words, the trained model 140 is trained on multiple training datasets in which disease names are associated with abnormality judgment evaluations and reference range image data.

[0142] Here, the reference domain image data used as the training dataset may be region data extracted by a human, or region data extracted by the same process used to extract the reference domain image data.

[0143] The panoramic image generation unit 133d generates an image including a dental arch unfolded image Ep, which shows the dental arch constituent tissue E unfolded based on the three-dimensional image data (see Figure 12).

[0144] The dental arch unfolded image Ep is an image in which the dental arch constituent tissues E, which form an arch shape or horseshoe shape in a plan view, are unfolded to form a straight line in a plan view.

[0145] For example, the dental arch unfolded image Ep is generated as follows: In other words, the dental arch constituent tissue E exists in the 3D image data as a dental arch constituent tissue region Ee in which the horseshoe-shaped region in plan view extends vertically (see Figure 13).

[0146] At each coordinate position in the vertical direction, a curve forming an arch or horseshoe shape that passes through the buccal-lingual center of the dentition tissue E is calculated. At each position along this curve, the presence or absence or transparency of the image of the dentition tissue E in the direction perpendicular to the tangent to the curve is calculated. The presence or absence or transparency of the image at each coordinate on the curve is expressed as the presence or absence or transparency of the image at each coordinate on the straight line. By performing the above process at each coordinate position in the vertical direction and superimposing them vertically, a dental arch development image Ep is generated (see Figure 14).

[0147] Furthermore, if the 3D image data of the dentition tissue E is represented by the translucent data of the surface of the dentition tissue E, then in the unfolded dentition image Ep, the areas that form the boundaries of the dentition tissue will be represented by low transparency and a dark color.

[0148] A dental arch unfolding image Ep is a type of panoramic image that shows the dental arch unfolded in a planar manner. The dental arch unfolding image Ep may also extend to the alveolar bone. Furthermore, the dental arch unfolding image Ep may also extend to the jawbone. The dental arch unfolding image Ep may be an image of the dental arch constituent tissues E transmitted in the thickness direction, or it may be a tomographic image of an arbitrary position in the thickness direction.

[0149] The dental arch unfolding image Ep should be an image that shows the entire area of ​​the dental arch constituent tissue E in a normal view. An image that shows the entire area of ​​the dental arch constituent tissue E in a normal view is, as shown in Figure 14, an image of the entire dental arch constituent tissue E viewed from the front of the midline of the head H (from the normal viewing direction Dv that shows the anterior tooth region) when the dental arch unfolding image Ep is unfolded in a direction perpendicular to the midline Md of the head H.

[0150] The area set in the calculation when the dentition tissue region Ee is unfolded can be considered as the dentition unfolded region Ex. The unfolding of the dentition tissue region Ee in a flat manner can also be called a flat unfolding, and in particular, the dentition unfolded region Ex that has been flattened can be considered as the dentition flat unfolded region Exp.

[0151] The unfolding of the dentition tissue E may be performed by an unfolding process that unfolds the image data of the dentition tissue E in the dentition tissue region Ee to conform to the shape of the dentition unfolding region Ex. Specifically, this process may be performed by extracting the 3D volume data of the dentition tissue E in the dentition tissue region Ee and generating unfolded 3D volume data that is unfolded and deformed to conform to the shape of the dentition unfolding region Ex. Transparency may be given to the dentition unfolding image Ep, which consists of the unfolded 3D volume data, so that the teeth or dental arch in the dentition unfolding region Ex can be seen through. In the following description, the region occupied by the dentition tissue E and the dentition tissue region Ee are assumed to coincide, and the region occupied by the dentition unfolding image Ep is assumed to coincide with the dentition unfolding region Ex.

[0152] The state of the dentition tissue region Ee before unfolding may be called the pre-unfolding curved state, and the state of the unfolded dentition region Ex after unfolding may be called the post-unfolding list state. The unfolding process is preferably performed such that the buccal-lingual direction BLD of the pre-unfolding curved state corresponds to the normal viewing direction NVD of the post-unfolding list state.

[0153] The direction of normal viewing may be perpendicular to the normal viewing plane of the dental arch development area Ex, or it may be the normal direction, but a viewing direction based on a different concept may also be set. For example, a virtual viewpoint Ey1 may be set on the normal viewing direction of the center of the dental arch development area Ex, and image processing may be performed along the viewing direction EVD from this viewpoint.

[0154] Since the dentition tissue region Ee has thickness in the buccolingual direction, the dentition exposure region Ex also has thickness in the emmetropic direction corresponding to the buccolingual thickness. Furthermore, since the dentition tissue E has thickness in the buccolingual direction as described above, the dentition exposure image Ep may also have thickness in the emmetropic direction corresponding to the buccolingual direction. As already mentioned, the dentition exposure image Ep may be an image transmitted through the thickness direction of the dentition tissue E, or it may be a tomographic image at an arbitrary position in the thickness direction.

[0155] In this way, a panoramic image is generated in which the dentition tissues E contained in the 3D image are unfolded.

[0156] The display control unit 133e performs the process of simultaneously or selectively displaying the abnormality judgment evaluation and the name of the disease on the display device 22. In this embodiment, the display control unit 133e further displays image information of at least a portion of the maxillofacial region on the display device 22. The abnormality judgment evaluation displayed on the display device 22, or the display of the abnormality judgment evaluation on the display device 22, may be referred to as "abnormality judgment evaluation display," and the name of the disease displayed on the display device 22, or the display of the name of the disease on the display device 22, may be referred to as "disease name display."

[0157] <Example of processing by the disease diagnosis support device> An example of processing by the calculation circuit 132 in the disease diagnosis support device 20 will be explained with reference to the flowchart in Figure 15.

[0158] For example, an instruction to identify the disease name is given to the processing unit 30 via the user interface 24, 26, etc. Then, in step S11, the arithmetic circuit 132 reads three-dimensional image data from the storage device 34.

[0159] In the next step S12, the calculation circuit 132 extracts the region of interest R based on the three-dimensional image data. Here, the region of interest R includes the tooth T and the alveolar bone Hja, and is extracted on a tooth T-by-tooth basis.

[0160] In the next step S13, a panoramic image is generated based on the three-dimensional image data. The generation of the panoramic image may be performed at any time before it is displayed on the display device 22.

[0161] In the next step S14, an anomaly judgment evaluation is identified for the area of ​​interest R. For example, at least the presence or absence of an anomaly is determined for each area of ​​interest R, and an anomaly judgment evaluation is associated with the area of ​​interest R that contains an anomaly. Once the determination of the presence or absence of an anomaly has been completed for all areas of interest R, the process proceeds to the next step S15.

[0162] In the next step S15, a disease is estimated for the region of interest R, which includes at least one abnormal region. The disease is estimated based on one abnormality judgment evaluation associated with the region of interest R and reference region image data.

[0163] In step S14, if the area of ​​interest R does not have an abnormal area, i.e., if the judgment evaluation is negative, the process may be terminated. However, if the disease estimation unit 133c refers to the reference image data, the process may proceed to step S15. The disease estimation unit 133c may verify the specific data obtained from the judgment evaluation by the abnormality judgment evaluation unit 133b and, if it finds an abnormal area, modify the judgment evaluation to estimate the disease.

[0164] In step S15, the estimation that a disease is present may be called a "positive disease estimation," and the estimation that a disease is absent may be called a "negative disease estimation."

[0165] In the next step S16, the abnormality assessment and disease name are displayed on the display device 22 (display of abnormality assessment and display of disease name). The display device 22 may also display image information of at least a portion of the maxillofacial region. This allows the user of the disease assessment support device 20 to recognize the abnormality assessment and the disease name based on the abnormality assessment by looking at the display device 22.

[0166] In the next step S17, it is determined whether or not there is an instruction to correct the evaluation. The user can check the abnormal judgment evaluation displayed on the display device 22 and correct the abnormal judgment evaluation by looking at the user interfaces 24 and 26. In other words, the user interfaces 24 and 26 are operation user interfaces (operation interfaces) that accept operations from the operator and accept the corrected abnormal judgment evaluation after the abnormal judgment evaluation has been corrected. The operation to correct the abnormal judgment evaluation may also be called an "abnormal judgment evaluation correction operation". If it is determined that a correction instruction has been made through the user interfaces 24 and 26, the process returns to step S15.

[0167] The processes in steps S15 and S16 are repeated, which leads to the estimation of diseases based on the corrected abnormality assessment and subsequent display processing. In other words, circuit 132 estimates diseases in the area of ​​interest R based on the corrected abnormality assessment.

[0168] If it is determined that there are no instructions for evaluation correction in the next step S17, the process ends.

[0169] In this embodiment, the disease name is estimated after the initial abnormality assessment, but such processing is not necessarily required. Before disease estimation based on the initial abnormality assessment, the initial abnormality assessment may be modified by the user, and then the initial disease estimation process may be performed based on the modified abnormality assessment.

[0170] Figure 16 is a functional block diagram showing the data flow.

[0171] As shown in the figure, the area of ​​interest R is extracted when the 3D image data is input to the area of ​​interest extraction unit 133a.

[0172] When the area of ​​interest R is input to the abnormality determination evaluation unit 133b, if the area of ​​interest R has an abnormality area, an abnormality determination evaluation (positive abnormality determination evaluation) is identified for that area of ​​interest R.

[0173] The abnormality judgment evaluation is input to the disease estimation unit 133c. Also, for example, image data of the area of ​​interest R is input to the disease estimation unit 133c as reference area image data. The disease estimation unit 133c estimates the disease for the area of ​​interest R associated with the abnormality judgment evaluation, based on the abnormality judgment evaluation and the reference area image data. Estimating a disease can be thought of as estimating the name of the disease.

[0174] The abnormality assessment and the name of the estimated disease are displayed on the display device 22.

[0175] Three-dimensional image data is provided to the panoramic image generation unit 133d. This generates a panoramic image.

[0176] Three cross-sectional images of the region of interest R can be generated based on separate 3D image data. The three cross-sectional images are three images aligned along three mutually orthogonal planes.

[0177] These panoramic images and three cross-sectional images can also be displayed on the display device 22.

[0178] By viewing the display device 22, the user of the disease diagnosis support device 20 can understand the abnormality judgment evaluation, the name of the disease, the panoramic image, and the three cross-sectional images. The user can examine the validity of the abnormality judgment evaluation by viewing the panoramic image and the three cross-sectional images. If, as a result of the examination, it is determined that the abnormality judgment evaluation should be revised, the abnormality judgment evaluation is revised. Based on the revised abnormality judgment evaluation and the reference range image data, the disease estimation unit 133c estimates the disease again. The name of the estimated disease is displayed on the display device 22. The abnormality judgment evaluation displayed on the display device 22 may also be revised to reflect the revised evaluation.

[0179] If the anomaly detection evaluation is modified, the trained model used for the anomaly detection evaluation may be further trained or modified based on the data including the modified anomaly detection evaluation. This is expected to make the determination of the anomaly detection evaluation unit 133b more accurate as the anomaly detection evaluation is modified.

[0180] Furthermore, modifications to disease names may be accepted via user interfaces 24 and 26. The operation of modifying disease names may be called a "disease name modification operation." The trained model 140 for disease estimation may undergo additional training or modification training based on the data including the accepted modified disease names. As a result, it is expected that the more disease names are modified, the more accurate the estimation by the disease estimation unit 133c will become.

[0181] <Display Example> Figure 17 shows an example of the display on the display device 22.

[0182] In the figure, a CT image including the abnormal area is displayed on the display screen. In this embodiment, the CT image displays three cross-sectional images D. More specifically, the three cross-sectional images D include three mutually orthogonal cross-sectional images Da, Db, and Dc.

[0183] Each of the three cross-sectional images Da, Db, and Dc may include the abnormal region W. The magnification of the three cross-sectional images Da, Db, and Dc may be set so that the entire abnormal region W is displayed. The centers of the three cross-sectional images Da, Db, and Dc may be set to coincide with the geometric center of the abnormal region W that appears in each cross-section.

[0184] If the area of ​​interest R includes multiple abnormal areas W, the magnification may be set so that the entirety of all abnormal areas W can be displayed.

[0185] Each of the three cross-sectional images Da, Db, and Dc may be an image showing the distribution of X-ray transmittance along a single plane, or an image showing the distribution of X-ray transmittance in a thick slice layer.

[0186] The display device 22 displays the abnormality judgment evaluation Ev. The abnormality judgment evaluation Ev may be the same as in the first embodiment. In this embodiment, in addition to the same items as in the first embodiment, an item called "Possible cause of this root" is displayed, and "Yes," "No," or "Unknown" is displayed for that item.

[0187] The positional and relative sizes of the three cross-sectional images D and the anomaly detection evaluation Ev are arbitrary. For example, the three cross-sectional images D and the anomaly detection evaluation Ev may be displayed side by side.

[0188] The default cross-sectional display conditions, such as the cross-sectional position and direction of the three cross-sectional images, may be configured to change depending on the estimated disease name. For example, if the estimated disease name is a radicular cyst, the three cross-sections will pass through the three-dimensional centroid of the cyst, and one of the three cross-sectional images will have a direction parallel to the tangent to the dental arch closest to the site and an extension in the axial direction. If the estimated disease name is caries confined to the enamel of the tooth crown, the three cross-sections will pass through the three-dimensional centroid of the caries, and one of the three cross-sectional images will have a direction normal to the dental arch closest to the site and an extension in the axial direction.

[0189] By viewing the display device 22, the user can understand the image of the abnormal region W, the abnormality assessment evaluation, and the name of the disease. The user can correct the abnormality assessment evaluation as needed. Based on the corrected abnormality assessment evaluation, the disease is re-estimated, and the name of the re-estimated disease is displayed on the display device 22.

[0190] For example, initially, as shown in Figure 17, the system displays "None" for the "Longest diameter is within 8 mm" aspect of the abnormality assessment, and "None" for the "Boundary is unclear" aspect of the above situation.

[0191] Suppose a user views the three cross-sectional images D, identifies the abnormal area W, and determines that the major axis of the circular area is within 8 mm and the boundary is unclear. Then, as shown in Figure 18, the user can correct whether the "major axis is within 8 mm" to "yes" and whether the "boundary is unclear" to "yes". The correction may be made, for example, by clicking the relevant area on the screen with the pointer. For example, it may be corrected by clicking the "yes" area on the screen with the pointer. The correction may also be made by input via the keyboard or other means. In other words, user interfaces 24 and 26 accept corrections regarding the presence or absence of abnormalities and the degree of abnormalities.

[0192] Then, based on the revised abnormality assessment, the disease is estimated again, and the name of the newly estimated disease, for example, where it was previously estimated and displayed as "periapical cyst," is displayed as "periapical granuloma" on the display device 22. This makes it possible to estimate the disease in a way that reflects human evaluation.

[0193] Individual items related to the degree of abnormality, such as whether the "longest diameter is within 8 mm," can be called "abnormality assessment items." Similarly, assessment elements such as "yes," "no," or "don't know" can be called "abnormality assessment elements." Abnormality assessment elements may include both negative and positive assessment elements. For example, for the item "whether the longest diameter is within 8 mm," there may be a negative assessment element like "yes" and a positive assessment element like "no." Negative assessment elements may be called "negative abnormality assessment elements," and positive assessment elements may be called "positive abnormality assessment elements."

[0194] Evaluation elements that are unclear whether they are negative or positive, such as "I don't know," can be called "general abnormality assessment elements."

[0195] If user modifications are accepted, the system may include at least one abnormality assessment item in which either a positive abnormality assessment element or a negative abnormality assessment element can be selected.

[0196] As already mentioned, the presence or absence of a disease, its severity, or its condition may be assessed as "unknown," that is, unclear. For example, an abnormality assessment may include not only the presence or absence of an abnormality, but also an assessment of the "abnormal range" (based on positive abnormality assessment), the "non-abnormal range" (based on negative abnormality assessment), and the "abnormal range boundary" (i.e., an area where abnormality is either not abnormal or unclear) which is the boundary between the abnormal and non-abnormal ranges (based on general abnormality assessment).

[0197] In this case, the disease may be inferred from other abnormality assessments or from reference range image data.

[0198] Furthermore, the above-mentioned "abnormal region boundary" may be modified by the user to either an "abnormal region" or a "non-abnormal region."

[0199] The items for anomaly detection evaluation are not limited to the examples above and should be set according to the nature of the anomaly area. Examples of evaluation items include size (length, width, area, volume, etc.), boundary conditions, and shape. Furthermore, the evaluation of these items does not need to be forcibly associated with any particular evaluation level or evaluation category. The evaluation of each item may be associated with an evaluation level or evaluation category corresponding to possibility, or with an evaluation of "unknown."

[0200] As shown in Figure 19, the display device 22 may display a panoramic image P. The display position of the panoramic image P is arbitrary. In Figure 19, three cross-sectional images D, the disease name, and the abnormality judgment evaluation are displayed simultaneously above the panoramic image P.

[0201] The panoramic image P may display a mark Q indicating the area of ​​interest R. This allows for the assessment of the validity of the disease while understanding the observation position within the entire dentition. If there is a mix of areas of interest R that include abnormalities and areas of interest R that do not, only the areas of interest R that include abnormalities may be indicated with mark Q, or the display may be differentiated by changing the color of mark Q depending on whether or not it includes abnormalities. If there are multiple areas of interest R that include abnormalities, multiple mark Qs may be displayed simultaneously, or if there are multiple areas of interest R that include abnormalities, the mark Qs may be sequentially changed to indicate the areas of interest R through operation on the user interfaces 24 and 26, and the three cross-sectional images D, disease name, and abnormality judgment evaluation may be switched to content corresponding to the changed areas of interest R.

[0202] Operations on user interfaces 24 and 26 may include, for example, click or touch operations on the panoramic image, or operations that specify the transition direction using keyboard keys.

[0203] The panoramic image and the three cross-sectional images D may be selectively displayed by operations on the user interfaces 24 and 26.

[0204] One or both of the panoramic image and the three-section image D may be switched to display the disease name and abnormality assessment.

[0205] When the disease estimation unit 133c refers to reference range image data, it performs an abnormality judgment evaluation on the reference range image data, similar to the abnormality judgment evaluation display by the abnormality judgment evaluation unit 133b, and may enable the display of this abnormality judgment evaluation with abnormality judgment evaluation items. Furthermore, in this abnormality judgment evaluation display, modification operations of the abnormality judgment evaluation elements may be accepted via the user interface 24, 26, etc., and a new disease estimation may be performed.

[0206] The criteria for determining abnormalities may differ depending on the type of area of ​​interest. For example, if the area of ​​interest is the root apex and its surrounding region, criteria may include the presence or absence of a roughly oval shape, whether the longest diameter is within 8 mm, whether the boundary is unclear, and the presence or absence of proximal caries. If the area of ​​interest is the crown region, criteria may include the presence or absence of caries, and whether the caries is confined to the enamel or has reached the dentin.

[0207] If an abnormality assessment and disease name identification have been performed for a subject, this abnormality assessment and the identified disease name may be stored in the memory device 17. For example, suppose that an abnormality assessment performed by the abnormality assessment unit 133b at a certain point in time was abnormality assessment Ev1, and the disease name estimated by the disease estimation unit 133c was disease name DN1. Later, when another abnormality assessment and disease name estimation are performed, the similarity of the abnormality assessments may be searchable. For example, suppose that a later abnormality assessment is abnormality assessment Ev2. If the similarity between abnormality assessment Ev1 and abnormality assessment Ev2 is high or they match, disease name DN1 may be output. This disease name DN1 is displayed on the display device 22. Not only disease name DN1, but also abnormality assessment Ev1 may be output and displayed.

[0208] The presence or absence of similarity may be determined based on the degree of agreement of each evaluation element for multiple anomaly detection evaluation items, or it may be determined by referring to the similarity (of the image) of the reference domain image data in addition to the degree of agreement of each evaluation element for multiple anomaly detection evaluation items.

[0209] Furthermore, if a user modifies the abnormality judgment evaluation made by the abnormality judgment evaluation unit 133b, the combination of the abnormality judgment evaluation before modification and the abnormality judgment evaluation after modification may be stored. For example, suppose the abnormality judgment evaluation made by the abnormality judgment evaluation unit 133b at a certain point in time was abnormality judgment evaluation Ev1, and the disease name estimated by the disease estimation unit 133c was disease name DN1. Suppose the user modifies this abnormality judgment evaluation Ev1 to abnormality judgment evaluation Ev2. Suppose that as a result of this modification, the disease name estimated by the disease estimation unit 133c becomes disease name DN2. These abnormality judgment evaluations Ev1 and Ev2 are stored as a pair. Later, when another abnormality judgment evaluation is made, the similarity of the abnormality judgment evaluations may be searchable. For example, suppose another abnormality judgment evaluation is made later, and this becomes abnormality judgment evaluation Ev3. If the similarity between this abnormality judgment evaluation Ev1 and abnormality judgment evaluation Ev3 is high or they match, abnormality judgment evaluation Ev2 may be output. This abnormality detection evaluation Ev2 is displayed on the display device 22. In addition to the abnormality detection evaluation Ev2, the abnormality detection evaluation Ev1 may also be output and displayed.

[0210] Furthermore, the disease name DN2 may be output and displayed. The process for responding to such corrections in the abnormality judgment evaluation may be called the "abnormality judgment evaluation correction response process." The users do not need to be the same; for example, if the user who made the correction was user U1, then when another user, user U2, uses the system and performs abnormality judgment evaluation Ev3, abnormality judgment evaluation Ev1 may be output.

[0211] Furthermore, if the user modifies the disease name estimated by the disease estimation unit 133c, the combination of this abnormality judgment evaluation and the modified disease name may be stored. For example, suppose the abnormality judgment evaluation by the abnormality judgment evaluation unit 133b at a certain point in time was abnormality judgment evaluation Ev1, and the disease name estimated by the disease estimation unit 33b was disease name DN1. Suppose the user modifies this disease name DN1 to disease name DN2. This abnormality judgment evaluation Ev1 and disease name DN2 are stored as a pair. Later, when estimating a disease name with another abnormality judgment evaluation, the similarity of the abnormality judgment evaluations may be searchable. For example, suppose another abnormality judgment evaluation is performed later, and it becomes abnormality judgment evaluation Ev2. If the similarity between this abnormality judgment evaluation Ev1 and abnormality judgment evaluation Ev2 is high or they match, disease name DN2 may be output. This disease name DN2 is displayed on the display device 22. Not only disease name DN2, but also abnormality judgment evaluation Ev1 may be output and displayed. The process of handling such corrections to disease names may be called "disease name correction response processing." The users do not need to be the same; for example, if the user who made the correction was user U1, then when another user, user U2, uses the system and performs an abnormality assessment Ev2, the disease name DN2 may be output. This produces an effect similar to obtaining a second opinion regarding the abnormality assessment and disease name estimation.

[0212] <Effects> The disease determination support device 20, disease determination support method, and disease determination support program 34a configured as described above can achieve the same effects as those of the first embodiment.

[0213] In addition, diseases in the area of ​​interest R are estimated based on the revised abnormality assessment.

[0214] Furthermore, since it is possible to correct for the presence or absence of abnormalities and the degree of abnormalities, it is easier to make a reasonable estimation of the disease.

[0215] Furthermore, since the disease is estimated by also referring to reference region image data that includes the abnormal region W within the 3D image data, the disease is estimated by taking into account image features that are difficult to express in abnormality judgment evaluation.

[0216] For example, relying solely on categorized abnormality assessments may make it difficult to accurately predict subtle diseases. In such cases, incorporating image-related elements can help infer the disease.

[0217] Furthermore, if the area of ​​interest R is divided into areas at the tooth or root level, then disease can be estimated at the tooth or root level.

[0218] Furthermore, a region of interest R is extracted based on the three-dimensional image data, and the disease is estimated based on at least one abnormality determination evaluation for the extracted region of interest R.

[0219] Furthermore, if at least a portion of the maxillofacial region image information is displayed on the display device 22, the user can view at least a portion of the maxillofacial region image information and verify the validity of the estimated disease name. This makes it easier to verify the validity of the estimation.

[0220] Furthermore, if the above image information includes a CT image with abnormal areas, users can view the CT image containing the abnormal areas to verify the validity of the estimated disease name.

[0221] Furthermore, if the CT image includes three mutually orthogonal cross-sectional images Da, Db, and Dc, users can easily verify the validity of the estimated disease name by looking at the three mutually orthogonal cross-sectional images Da, Db, and Dc.

[0222] If the image information includes a dental arch development image Ep, the entire structure of the dental arch tissue E can be visualized, and the validity of the estimated disease name can be verified.

[0223] Furthermore, by displaying image information and anomaly detection evaluations simultaneously on the display device 22, users can view both at the same time, making it easier to examine the validity of the anomaly detection evaluation. Users can also easily modify the anomaly detection evaluation as needed.

[0224] {Modification} In the second embodiment, the acceptance of corrections to the abnormality judgment evaluation and the re-estimation of the disease based on the corrected abnormality judgment evaluation can also be applied to the process of estimating the disease without relying on reference range image data in the first embodiment.

[0225] In each of the above embodiments, it was explained that abnormal areas may be associated with standard tooth positions.

[0226] In this case, performing a CT scan as follows makes it easier to accurately correlate the position in the 3D image data with the standard tooth position.

[0227] For example, as shown in Figure 20, when performing a CT scan with the CT scanner 10, the head holder 9 positions the head H in a fixed position. As a result, as shown in Figure 21, the dental arch tissues E are also positioned in a fixed position relative to the head holder 9.

[0228] The positional relationship between the head holder 9 and the X-ray generator 11 and X-ray detector 12 can be a known positional relationship. Furthermore, the area in the head H where the dentition tissue E is located can be considered to be somewhat constant. For this reason, in the three-dimensional image data based on CT scan data, the area of ​​the standard dentition tissue E can be treated as the area of ​​the actual dentition tissue E.

[0229] In each of the above embodiments, a case in which the disease determination support device 20 is used in conjunction with the CT imaging device 10 was described.

[0230] The disease diagnosis support device 20 can also be configured as a separate device from the CT scanning device 10.

[0231] In this case, as shown in Figures 22 and 23, the disease determination support device 220 may include a data access device 230 connected by rotating the connection port 36. The data access device 230 is, for example, a device that can read data recorded on the recording medium 228. The recording medium 228 may be an optical recording medium such as an optical disc, or a flash memory such as a USB memory or SD card. The data access device 230 may be an optical disc reader or a card reader.

[0232] The recording medium 228 records image data obtained from CT scanning. The image data may be a group of X-ray projection image data, a group of three-dimensional image data, or a group of multiple tomographic images.

[0233] The disease diagnosis support device 220 reads image data obtained from CT scanning via the data access device 230 and generates three-dimensional image data from the read data as necessary. The disease diagnosis support device 220 can perform the same processing as in each of the above embodiments based on the read three-dimensional image data or the three-dimensional image data generated based on the read image data.

[0234] Image data obtained from CT scans is stored in a server device, and the disease diagnosis support device 220 may access the server device via a communication network to obtain the image data obtained from CT scans. In this case, the server device may be located within the facility where the disease diagnosis support device 220 is located, or it may be located outside the facility. The server device may also be a cloud server. The disease diagnosis support device 220 can access the server device via a dedicated line or a public network to obtain the image data obtained from CT scans.

[0235] The fundamental idea behind this disclosure can also be expressed as follows:

[0236] A disease diagnosis support device that provides information for determining a disease by processing three-dimensional image data obtained by imaging the maxillofacial region using at least one of X-ray and MRI, comprising: a storage device for storing the three-dimensional image data; a processing device; and a display device for displaying information generated by the processing device, wherein the processing device identifies an abnormality judgment evaluation, which is a judgment evaluation performed by a judgment processing on at least one of the reasons, grounds, or estimation materials that lead to the estimation result of the disease based on the three-dimensional image data; estimates the disease in the area of ​​interest including the abnormal area based on the abnormality judgment evaluation; and simultaneously or selectively displays the abnormality judgment evaluation and the name of the disease on the display device.

[0237] Furthermore, the configurations described in each of the above embodiments and modifications can be combined as appropriate, as long as they do not contradict each other.

[0238] This disclosure discloses the following aspects:

[0239] The first embodiment is a disease diagnosis support device that provides information for determining a disease by processing image data obtained by imaging the maxillofacial region using at least one of X-ray and MRI, comprising a storage device for storing the image data, a processing device, and a display device for displaying information generated by the processing device, wherein the processing device identifies an abnormality judgment evaluation regarding an abnormal area based on the image data, estimates the disease in the area of ​​interest including the abnormal area based on the abnormality judgment evaluation, and simultaneously or selectively displays the abnormality judgment evaluation and the name of the disease on the display device.

[0240] In this case, the name of the suspected disease and the abnormality assessment used as the basis for the suspected disease are displayed simultaneously or selectively on the display device. Users can recognize the abnormality assessment and verify the validity of the suspected disease name, making it easier to verify the validity of the estimation.

[0241] The second embodiment is a disease determination support device according to the first embodiment, wherein the abnormality determination evaluation includes a plurality of abnormality determination evaluation items.

[0242] This allows for the estimation of a disease based on multiple abnormality assessment criteria.

[0243] A third embodiment is a disease determination support device according to the first or second embodiment, wherein the image data includes three-dimensional image data obtained by X-ray imaging of the maxillofacial region.

[0244] This allows for disease diagnosis based on three-dimensional information.

[0245] A fourth embodiment is a disease determination support device according to any one of the first to third embodiments, comprising an operation interface for receiving operations from an operator, the operation interface receiving a modified abnormality determination evaluation obtained by modifying the abnormality determination evaluation, and the processing device estimating the disease in the area of ​​interest based on the modified abnormality determination evaluation.

[0246] This allows for the correction of abnormality assessments and the estimation of diseases.

[0247] The fifth aspect is a disease determination support device according to the fourth aspect, wherein the abnormality determination evaluation includes an evaluation of the presence or absence of an abnormality and the degree of the abnormality, and the operation interface accepts corrections to the presence or absence of an abnormality and the degree of the abnormality.

[0248] This allows for the estimation of a disease by correcting for the presence or absence of abnormalities and their severity.

[0249] An abnormality assessment may include multiple abnormality assessment items. Each abnormality assessment item may be evaluated using evaluation elements related to the applicability of an abnormality. Evaluation elements that indicate the absence of an abnormality may be designated as negative abnormality assessment elements, and evaluation elements that indicate the presence of an abnormality may be designated as positive abnormality assessment elements. Evaluation elements where the presence or absence of an abnormality is unclear may be designated as no-abnormality assessment elements.

[0250] The modifiable evaluation elements should include at least two: a negative abnormality judgment evaluation element and a positive abnormality judgment evaluation element. The modifiable evaluation elements may further include a no-abnormality judgment evaluation element.

[0251] A sixth embodiment is a disease determination support device according to any one of the first to fifth embodiments, comprising an operation interface that accepts operations from an operator, the operation interface accepting a disease name modification operation for modifying the name of the disease displayed on the display device.

[0252] This may allow for a correction of the estimated disease name.

[0253] The seventh embodiment is a disease determination support device according to any one of the first to sixth embodiments, wherein the processing device estimates the disease in the area of ​​interest including the abnormal area based on the abnormality determination evaluation and reference area image data including the abnormal area among the image data.

[0254] This allows for disease estimation by referencing reference region image data, including abnormal areas, within the 3D image data. As a result, the disease can be estimated by taking into account image features that are difficult to represent in abnormality detection evaluations.

[0255] The image data used for anomaly detection evaluation and the reference domain image data may be the same data or different data, as long as they correspond to each other's regions. If they are different data, the image data used for anomaly detection evaluation and the reference domain image data may be different data before and after image processing. The image processing here is, for example, the process of reconstructing a set of X-ray projection image data into three-dimensional image data.

[0256] The image data used for abnormality detection and the reference region image data may be image data obtained by photographing the same maxillofacial region.

[0257] For example, if a certain X-ray image taken on a certain maxillofacial region is referred to as the "original X-ray image," and the X-ray image data obtained from the original X-ray image is referred to as the "original X-ray image data," then the image data generated from the original X-ray image data may be used for abnormality detection evaluation, and the same image data generated from the original X-ray image data may be used as the reference area image data.

[0258] Furthermore, if a certain X-ray CT scan performed on a certain maxillofacial region is referred to as the "original X-ray CT scan," and the X-ray projection data obtained from the original X-ray CT scan is referred to as the "original X-ray projection image data," then the image data generated by reconstructing the original X-ray projection image data may be used for abnormality detection evaluation, and the same image data generated by reconstructing the original X-ray projection image data may be used as the reference area image data.

[0259] For example, if a certain reconstruction that processes original X-ray projection image data is called the "current reconstruction," the image data generated by the current reconstruction may be used in the anomaly detection evaluation, and image data generated by another current reconstruction may be used as the reference domain image data.

[0260] Alternatively, the image data generated by the current reconstruction used in the anomaly detection evaluation may be used, and the same image data generated by the current reconstruction may be used as the reference domain image data.

[0261] Furthermore, when three-dimensional image data generated by reconstruction using one current reconstruction method is referred to as "current three-dimensional image data," the current three-dimensional image data generated by the current reconstruction method may be used in the anomaly detection evaluation, and the current three-dimensional image data generated by another current reconstruction method may be used as the reference domain image data.

[0262] Alternatively, the current 3D image data generated by the current reconstruction in the anomaly detection evaluation may be used, and the same current 3D image data may be used as the reference domain image data.

[0263] Furthermore, the image data used for evaluating the anomaly detection and the reference range image data may be images obtained through different imaging processes.

[0264] The eighth aspect is a disease determination support device according to any one of the first to seventh aspects, wherein the area of ​​interest is an area divided into tooth or tooth root units, and the processing device estimates the disease based on at least one abnormality determination evaluation for the area of ​​interest.

[0265] This allows for the estimation of disease at the level of the area of ​​interest.

[0266] The ninth embodiment is a disease determination support device according to any one of the first to eighth embodiments, wherein the processing device extracts the area of ​​interest based on the three-dimensional image data and estimates the disease based on at least one abnormality determination evaluation for the extracted area of ​​interest.

[0267] This allows for disease estimation on a region-by-region basis.

[0268] The tenth embodiment is a disease determination support device according to any one of the first to ninth embodiments, wherein the three-dimensional image data is three-dimensional CT image data obtained by processing a projection image obtained by photographing the maxillofacial region.

[0269] This makes it easier to estimate diseases based on three-dimensional CT image data obtained from CT scans.

[0270] The eleventh embodiment is a disease determination support device according to any one of the first to tenth embodiments, wherein the processing device further displays image information of at least a portion of the maxillofacial region on the display device.

[0271] This allows users to view image information of at least a portion of the maxillofacial region and verify the validity of the estimated disease name, making it easier to verify the validity of the estimation.

[0272] The twelfth embodiment is a disease determination support device according to the eleventh embodiment, wherein the image information is a CT image including the abnormal area.

[0273] This allows us to examine CT images that include abnormal areas and verify the validity of the estimated disease name.

[0274] The thirteenth aspect is a disease determination support device according to the twelfth aspect, wherein the CT image includes three mutually orthogonal cross-sectional images.

[0275] This allows us to verify the validity of the estimated disease name by looking at three mutually orthogonal cross-sectional images.

[0276] The 14th embodiment is a disease determination support device according to any one of the 11th to 13th embodiments, wherein the image information includes a dental arch unfolded image obtained by unfolding the dental arch constituent tissue based on three-dimensional image data obtained by X-ray imaging of the maxillofacial region.

[0277] This allows for visualizing the entire structure of the dental arch while verifying the validity of the estimated disease name.

[0278] The 15th embodiment is a disease determination support device according to any one embodiment of the 11th to 14th embodiments, wherein the processing device simultaneously displays the image information and the abnormality determination evaluation on the display device.

[0279] This allows users to view image information and anomaly detection evaluations simultaneously, making it easier to assess the validity of the anomaly detection evaluation.

[0280] The sixteenth aspect is a disease determination support method that provides information for determining a disease by processing image data obtained by imaging the maxillofacial region using at least one of X-ray and MRI, wherein the method identifies an abnormality determination evaluation regarding an abnormal area based on the image data, estimates the disease in the area of ​​interest including the abnormal area based on the abnormality determination evaluation, and outputs the abnormality determination evaluation and the name of the disease in a manner that can be recognized by the user.

[0281] In this case, the name of the estimated disease and the abnormality assessment used as the basis for estimating the disease are output in a way that is recognizable to the user. The user can recognize the abnormality assessment and verify the validity of the estimated disease name, making it easier to verify the validity of the estimation.

[0282] The seventeenth aspect is a program that provides information for determining a disease by processing image data obtained by imaging the maxillofacial region using at least one of X-ray and MRI, and causes a computer to perform the following processes: identify an abnormality judgment evaluation regarding an abnormal area based on the image data; estimate the disease in the area of ​​interest including the abnormal area based on the abnormality judgment evaluation; and output the abnormality judgment evaluation and the name of the disease in a manner that can be recognized by the user.

[0283] In this case, the name of the estimated disease and the abnormality assessment used as the basis for estimating the disease are output in a way that is recognizable to the user. The user can recognize the abnormality assessment and verify the validity of the estimated disease name, making it easier to verify the validity of the estimation.

[0284] From a mechanical standpoint, the first to the seventeenth aspects may be modified as follows, resulting in the first modified aspect to the seventeenth modified aspect.

[0285] The first modified embodiment is a disease diagnosis support device that provides information for determining a disease by processing three-dimensional image data obtained by imaging the maxillofacial region using at least one of X-ray and MRI, comprising a memory for storing the three-dimensional image data, a processor, and a display for displaying information generated by the processor, wherein the processor identifies an abnormality judgment evaluation, which is a judgment evaluation regarding an abnormal area, performed by a judgment processing on at least one of the reasons, grounds, and estimation materials that lead to the estimation result of the disease based on the three-dimensional image data, estimates the disease in the area of ​​interest including the abnormal area based on the abnormality judgment evaluation, and simultaneously or selectively displays the abnormality judgment evaluation and the name of the disease on the display.

[0286] In this case, the name of the suspected disease and the abnormality assessment used as the basis for the suspected disease are displayed on the screen simultaneously or selectively. Users can recognize the abnormality assessment and verify the validity of the suspected disease name, making it easier to verify the validity of the estimation.

[0287] The second modified embodiment is a disease determination support device according to the first modified embodiment, wherein the abnormality determination evaluation includes a plurality of abnormality determination evaluation items.

[0288] This allows for the estimation of a disease based on multiple abnormality assessment criteria.

[0289] The third modified embodiment is a disease determination support device according to the first modified embodiment, wherein the image data includes three-dimensional image data obtained by X-ray imaging of the maxillofacial region.

[0290] This allows for disease diagnosis based on three-dimensional information.

[0291] A fourth modified embodiment is a disease determination support device according to the first modified embodiment, comprising an operation interface that accepts operator input, the operation interface accepts a modified abnormality determination evaluation obtained by correcting the abnormality determination evaluation, and the processor estimates the disease in the area of ​​interest based on the modified abnormality determination evaluation.

[0292] This allows for the correction of abnormality assessments and the estimation of diseases.

[0293] The fifth modification is a disease determination support device according to the fourth modification, wherein the abnormality determination evaluation includes an evaluation of the presence or absence of an abnormality and the degree of the abnormality, and the operation interface accepts corrections to the presence or absence of an abnormality and the degree of the abnormality.

[0294] This allows for the estimation of a disease by correcting for the presence or absence of abnormalities and their severity.

[0295] An abnormality assessment may include multiple abnormality assessment items. Each abnormality assessment item may be evaluated using evaluation elements related to the applicability of an abnormality. Evaluation elements that indicate the absence of an abnormality may be designated as negative abnormality assessment elements, and evaluation elements that indicate the presence of an abnormality may be designated as positive abnormality assessment elements. Evaluation elements where the presence or absence of an abnormality is unclear may be designated as no-abnormality assessment elements.

[0296] The modifiable evaluation elements should include at least two: a negative abnormality judgment evaluation element and a positive abnormality judgment evaluation element. The modifiable evaluation elements may further include a no-abnormality judgment evaluation element.

[0297] The sixth modification is a disease determination support device according to the first modification, comprising an operation interface that accepts operations from an operator, the operation interface accepting a disease name modification operation for modifying the name of the disease displayed on the display.

[0298] The operating interface may be a physical interface.

[0299] This may allow for a correction of the estimated disease name.

[0300] The seventh modified embodiment is a disease determination support device according to the third modified embodiment, wherein the processor estimates the disease in the area of ​​interest including the abnormal area based on the abnormality determination evaluation and the reference area image data including the abnormal area among the image data.

[0301] This allows for disease estimation by referencing reference region image data, including abnormal areas, within the 3D image data. As a result, the disease can be estimated by taking into account image features that are difficult to represent in abnormality detection evaluations.

[0302] The image data used for anomaly detection evaluation and the reference domain image data may be the same data or different data, as long as they correspond to each other's regions. If they are different data, the image data used for anomaly detection evaluation and the reference domain image data may be different data before and after image processing. The image processing here is, for example, the process of reconstructing a set of X-ray projection image data into three-dimensional image data.

[0303] The image data used for abnormality detection and the reference region image data may be image data obtained by photographing the same maxillofacial region.

[0304] For example, if a certain X-ray image taken on a certain maxillofacial region is referred to as the "original X-ray image," and the X-ray image data obtained from the original X-ray image is referred to as the "original X-ray image data," then the image data generated from the original X-ray image data may be used for abnormality detection evaluation, and the same image data generated from the original X-ray image data may be used as the reference area image data.

[0305] Furthermore, if a certain X-ray CT scan performed on a certain maxillofacial region is referred to as the "original X-ray CT scan," and the X-ray projection data obtained from the original X-ray CT scan is referred to as the "original X-ray projection image data," then the image data generated by reconstructing the original X-ray projection image data may be used for abnormality detection evaluation, and the same image data generated by reconstructing the original X-ray projection image data may be used as the reference area image data.

[0306] For example, if a certain reconstruction that processes original X-ray projection image data is called the "current reconstruction," the image data generated by the current reconstruction may be used in the anomaly detection evaluation, and image data generated by another current reconstruction may be used as the reference domain image data.

[0307] Alternatively, the image data generated by the current reconstruction used in the anomaly detection evaluation may be used, and the same image data generated by the current reconstruction may be used as the reference domain image data.

[0308] Furthermore, when three-dimensional image data generated by reconstruction using one current reconstruction method is referred to as "current three-dimensional image data," the current three-dimensional image data generated by the current reconstruction method may be used in the anomaly detection evaluation, and the current three-dimensional image data generated by another current reconstruction method may be used as the reference domain image data.

[0309] Alternatively, the current 3D image data generated by the current reconstruction in the anomaly detection evaluation may be used, and the same current 3D image data may be used as the reference domain image data.

[0310] Furthermore, the image data used for evaluating the anomaly detection and the reference range image data may be images obtained through different imaging processes.

[0311] The eighth modified embodiment is a disease determination support device according to the first modified embodiment, wherein the area of ​​interest is an area divided into tooth or tooth root units, and the processor estimates the disease based on at least one abnormality determination evaluation for the area of ​​interest.

[0312] This allows for the estimation of disease at the level of the area of ​​interest.

[0313] The ninth modified embodiment is a disease determination support device according to the third modified embodiment, wherein the processor extracts the area of ​​interest based on the three-dimensional image data and estimates the disease based on at least one abnormality determination evaluation for the extracted area of ​​interest.

[0314] This allows for disease estimation on a region-by-region basis.

[0315] The tenth modified embodiment is a disease determination support device according to the third modified embodiment, wherein the three-dimensional image data is three-dimensional CT image data obtained by processing projection images obtained by photographing the maxillofacial region.

[0316] This makes it easier to estimate diseases based on three-dimensional CT image data obtained from CT scans.

[0317] The eleventh modified embodiment is a disease determination support device according to the first modified embodiment, wherein the processor further displays image information of at least a portion of the maxillofacial region on the display.

[0318] This allows users to view image information of at least a portion of the maxillofacial region and verify the validity of the estimated disease name, making it easier to verify the validity of the estimation.

[0319] The twelfth modification is a disease determination support device relating to the eleventh modification, wherein the image information is a CT image including the abnormal area.

[0320] This allows us to examine CT images that include abnormal areas and verify the validity of the estimated disease name.

[0321] The thirteenth modified embodiment is a disease determination support device according to the twelfth modified embodiment, wherein the CT image includes three mutually orthogonal cross-sectional images.

[0322] This allows us to verify the validity of the estimated disease name by looking at three mutually orthogonal cross-sectional images.

[0323] The 14th modified embodiment is a disease determination support device according to the 11th modified embodiment, wherein the image information includes a dental arch unfolded image obtained by unfolding the dental arch constituent tissue based on three-dimensional image data obtained by X-ray imaging of the maxillofacial region.

[0324] This allows for visualizing the entire structure of the dental arch while verifying the validity of the estimated disease name.

[0325] The 15th modification is a disease determination support device according to the 11th modification, wherein the processor simultaneously displays the image information and the abnormality determination evaluation on the display.

[0326] This allows users to view image information and anomaly detection evaluations simultaneously, making it easier to assess the validity of the anomaly detection evaluation.

[0327] The sixteenth modified aspect is a disease determination support method that provides information for determining a disease by processing image data obtained by imaging the maxillofacial region using at least one of X-ray and MRI, wherein the method identifies an abnormality determination evaluation regarding an abnormal area based on the image data, estimates the disease in the area of ​​interest including the abnormal area based on the abnormality determination evaluation, and outputs the abnormality determination evaluation and the name of the disease in a manner recognizable to the user.

[0328] In this case, the name of the estimated disease and the abnormality assessment used as the basis for estimating the disease are output in a way that is recognizable to the user. The user can recognize the abnormality assessment and verify the validity of the estimated disease name, making it easier to verify the validity of the estimation.

[0329] The 17th modified form is a program that provides information for determining a disease by processing image data obtained by imaging the maxillofacial region using at least one of X-ray and MRI, and causes a computer to perform the following processes: identify an abnormality judgment evaluation regarding an abnormal area based on the image data; estimate the disease in the area of ​​interest including the abnormal area based on the abnormality judgment evaluation; and output the abnormality judgment evaluation and the name of the disease in a manner that can be recognized by the user.

[0330] In this case, the name of the estimated disease and the abnormality assessment used as the basis for estimating the disease are output in a way that is recognizable to the user. The user can recognize the abnormality assessment and verify the validity of the estimated disease name, making it easier to verify the validity of the estimation.

[0331] The above description is illustrative in all respects, and the invention is not limited thereto. It is understood that countless variations not illustrated can be conceivable without falling outside the scope of this invention.

[0332] 10 CT scanning device 17a 3D image data 20, 220 Disease determination support device 22 Display device 24, 26 User interface 30 Processing unit 32, 132 Arithmetic circuit 32a Processor 33, 133 Determination support processing unit 33a, 133b Abnormality determination evaluation unit 33b, 133c Disease estimation unit 33c Display control unit 34 Storage device 34a Program 34b Data 40, 140 Trained model 133a Area of ​​focus extraction unit 133d Panoramic image generation unit 133e Display control unit 228 Recording medium D 3 cross-sectional image E Dental arch constituent tissue Ep Dental arch unfolded image Ev Abnormality determination evaluation Hja Alveolar bone P Panoramic image Q Mark R Area of ​​focus T Tooth Tr Tooth root F Maxillofacial region

Claims

1. A disease diagnosis support device that provides information for determining a disease by processing image data obtained by imaging the maxillofacial region using at least one of X-ray and MRI, comprising: a storage device for storing the image data; a processing device; and a display device for displaying information generated by the processing device, wherein the processing device identifies an abnormality judgment evaluation regarding an abnormal area based on the image data, estimates the disease in the area of ​​interest including the abnormal area based on the abnormality judgment evaluation, and simultaneously or selectively displays the abnormality judgment evaluation and the name of the disease on the display device.

2. A disease determination support device according to claim 1, wherein the abnormality determination evaluation includes a plurality of abnormality determination evaluation items.

3. A disease diagnosis support device according to claim 1, wherein the image data includes three-dimensional image data obtained by X-ray imaging of the maxillofacial region.

4. A disease determination support device according to claim 1, comprising an operation interface for receiving operations from an operator, the operation interface receiving a modified abnormality determination evaluation obtained by modifying the abnormality determination evaluation, and the processing device estimating the disease in the area of ​​interest based on the modified abnormality determination evaluation.

5. A disease determination support device according to claim 4, wherein the abnormality determination evaluation includes an evaluation of the presence or absence of an abnormality and the degree of the abnormality, and the operation interface accepts corrections of the presence or absence of an abnormality and the degree of the abnormality.

6. A disease determination support device according to claim 1, comprising an operation interface for receiving operations from an operator, wherein the operation interface receives a disease name modification operation for modifying the name of the disease displayed on the display device.

7. A disease determination support device according to claim 3, wherein the processing device estimates the disease in the area of ​​interest including the abnormal area based on the abnormality determination evaluation and reference area image data including the abnormal area among the image data.

8. A disease determination support device according to any one of claims 1 to 7, wherein the area of ​​interest is an area divided into units of teeth or tooth roots, and the processing device estimates the disease based on at least one abnormality determination evaluation for the area of ​​interest.

9. A disease determination support device according to claim 3, wherein the processing device extracts the area of ​​interest based on the three-dimensional image data, and estimates the disease based on at least one abnormality determination evaluation for the extracted area of ​​interest.

10. A disease diagnosis support device according to claim 3, wherein the three-dimensional image data is three-dimensional CT image data obtained by processing a projection image obtained by photographing the maxillofacial region.

11. A disease determination support device according to claim 1, wherein the processing device further displays image information of at least a portion of the maxillofacial region on the display device.

12. A disease determination support device according to claim 11, wherein the image information is a CT image including the abnormal area.

13. A disease determination support device according to claim 12, wherein the CT image includes three cross-sectional images that are orthogonal to each other.

14. A disease determination support device according to claim 11, wherein the image information includes a dental arch unfolded image obtained by unfolding the dental arch constituent tissue based on three-dimensional image data obtained by X-ray imaging of the maxillofacial region.

15. A disease determination support device according to claim 11, wherein the processing device simultaneously displays the image information and the abnormality determination evaluation on the display device.

16. A disease diagnosis support method that provides information for determining a disease by processing image data obtained by imaging the maxillofacial region using at least one of X-ray and MRI, the method comprising: identifying an abnormality judgment evaluation regarding an abnormal area based on the image data; estimating the disease in the area of ​​interest including the abnormal area based on the abnormality judgment evaluation; and outputting the abnormality judgment evaluation and the name of the disease in a manner recognizable to the user.

17. A program that provides information for determining a disease by processing image data obtained by imaging the maxillofacial region using at least one of X-ray and MRI, the program causing a computer to perform the following processes: identify an abnormality judgment evaluation regarding an abnormal area based on the image data; estimate the disease in the area of ​​interest including the abnormal area based on the abnormality judgment evaluation; and output the abnormality judgment evaluation and the name of the disease in a manner recognizable to the user.

Images