Medical image processing apparatus, method, and storage medium

Abstract

The embodiments disclosed in the present specification and the accompanying drawings relate to a medical image processing apparatus, a method, and a storage medium. The medical image processing apparatus, the method, and the storage medium can appropriately perform analysis even in a case where medical image data in which a possibility of generation of an artifact caused by a gravity effect is high is analyzed. The medical image processing apparatus of the embodiments includes an acquisition unit, an identification unit, a first setting unit, and an analysis unit. The acquisition unit acquires medical image data that is an analysis target. The identification unit identifies a gravity direction with respect to the medical image data. The first setting unit sets a first weight for a pixel configuring the medical image data on the basis of the gravity direction. The analysis unit performs analysis of the medical image data on the basis of the first weight.

Description

[0001] Reference to relevant applications

[0002] This application enjoys the priority of Japanese Patent Application No. 2021-194186, filed on November 30, 2021, the entire contents of which are incorporated herein by reference. Technical Field

[0003] The embodiments disclosed in this specification and accompanying drawings relate to medical image processing apparatus, methods, and storage media. Background Technology

[0004] Previously, techniques were known for identifying medical image data that was highly likely to produce artifacts during the acquisition of medical image data. These artifacts were caused by gravity and were referred to as the gravity effect. For example, techniques were described that, based on clinical knowledge, determine the locations where artifacts caused by the gravity effect are likely to occur, and identify medical image data that is highly likely to produce artifacts based on the continuity of the high absorption region at that location. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a medical image processing apparatus, method and storage medium that can perform appropriate analysis even when analyzing medical image data that is highly likely to produce artifacts caused by gravity effects.

[0006] The medical image processing apparatus of this embodiment includes an acquisition unit, an identification unit, a first setting unit, and a parsing unit. The acquisition unit acquires medical image data that is the object of parsing. The identification unit identifies the direction of gravity relative to the medical image data. The first setting unit assigns a first weight to the pixels constituting the medical image data based on the direction of gravity. The parsing unit performs parsing of the medical image data based on the first weight.

[0007] Invention Effects

[0008] The medical image processing apparatus, method, and storage medium according to the embodiments can perform appropriate analysis even when analyzing medical image data that is highly likely to produce artifacts caused by gravity effects. Attached Figure Description

[0009] Figure 1 This is a diagram illustrating an example of the configuration of the medical image processing apparatus according to the first embodiment.

[0010] Figure 2 This is a flowchart illustrating the processing sequence performed by the processing functions of the processing circuit of the medical image processing apparatus of the first embodiment.

[0011] Figure 3This is a diagram illustrating an example of the setting process for the first weight in the first embodiment.

[0012] Figure 4 This is a diagram illustrating an example of the setting process for the first weight in Variation Example 2.

[0013] Figure 5 This is a diagram used to illustrate the analytical processing of variation example 3.

[0014] Figure 6 This is a diagram illustrating an example of the configuration of the medical image processing apparatus according to the second embodiment.

[0015] Figure 7 This is a flowchart illustrating the processing sequence of each processing function of the medical image processing apparatus of the second embodiment.

[0016] Figure 8 This is a diagram illustrating an example of the determination process in the second embodiment.

[0017] Figure 9 This is a diagram illustrating an example of the configuration of a medical image processing apparatus according to the third embodiment.

[0018] Figure 10 This is a flowchart illustrating the processing sequence of each processing function of the medical image processing apparatus of the third embodiment.

[0019] Figure 11A This is a diagram illustrating an example of the second weight setting process in the third embodiment.

[0020] Figure 11B This is a diagram illustrating an example of the second weight setting process in the third embodiment. Detailed Implementation

[0021] Hereinafter, embodiments of the medical image processing apparatus, method, and program will be described in detail with reference to the accompanying drawings. However, the medical image processing apparatus, method, and program of this application are not limited to the embodiments shown below. Furthermore, in the following description, common reference numerals are used to denote the same constituent elements, and repeated descriptions are omitted.

[0022] (First Implementation)

[0023] Figure 1 This is a diagram illustrating an example configuration of the medical image processing apparatus according to the first embodiment. For example, such as... Figure 1 As shown, the medical image processing apparatus 3 of this embodiment can be communicatively connected to the medical image diagnostic apparatus 1 and the medical image storage apparatus 2 via a network. Alternatively, it can also be... Figure 1The network shown is connected to various other devices and systems.

[0024] The medical image diagnostic device 1 captures images of the subject and generates medical image data. Furthermore, the medical image diagnostic device 1 transmits the generated medical image data to various devices on the network. For example, the medical image diagnostic device 1 can be an X-ray diagnostic device, an X-ray CT (Computed Tomography) device, an MRI (Magnetic Resonance Imaging) device, an ultrasound diagnostic device, a SPECT (Single Photon Emission Computed Tomography) device, or a PET (Positron Emission Computed Tomography) device.

[0025] The medical image storage device 2 stores various medical image data related to the patient being examined. Specifically, the medical image storage device 2 receives medical image data from the medical image diagnostic device 1 via a network and stores and stores the medical image data in its own storage circuitry. For example, the medical image storage device 2 is implemented using computer equipment such as a server or workstation. Alternatively, for example, the medical image storage device 2 is implemented using a PACS (Picture Archiving and Communication System) or similar system, storing the medical image data in a DICOM (Digital Imaging and Communications in Medicine) compliant format.

[0026] The medical image processing device 3 performs various processing related to medical image data. Specifically, the medical image processing device 3 receives medical image data from the medical image diagnostic device 1 or the medical image storage device 2 via a network, and uses the medical image data to perform various information processing. For example, the medical image processing device 3 is implemented by computer equipment such as servers or workstations.

[0027] For example, the medical image processing device 3 includes a communication interface 31, an input interface 32, a display 33, a storage circuit 34, and a processing circuit 35.

[0028] The communication interface 31 controls the transmission and communication of various data between the medical image processing device 3 and other devices connected via a network. Specifically, the communication interface 31 is connected to the processing circuit 35, sending data received from other devices to the processing circuit 35, or sending data sent from the processing circuit 35 to other devices. For example, the communication interface 31 is implemented by a network card, network adapter, NIC (Network Interface Controller), etc.

[0029] Input interface 32 receives various instructions and information input operations from the user. Specifically, input interface 32 is connected to processing circuit 35, converts the input operations received from the user into electrical signals, and sends them to processing circuit 35. For example, input interface 32 can be implemented as a trackball, switch button, mouse, keyboard, touchpad for input operations via contact surface, touchscreen integrating display screen and touchpad, contactless input interface using optical sensor, and voice input interface. Furthermore, in this specification, input interface 32 is not limited to physical operating components such as a mouse or keyboard. For example, an example where processing circuitry receives electrical signals corresponding to input operations from an external input device independent of the device and sends these electrical signals to control circuitry is also included in input interface 32.

[0030] The display 33 displays various information and data. Specifically, the display 33 is connected to the processing circuit 35 and displays various information and data received from the processing circuit 35. For example, the display 33 can be implemented using a liquid crystal display, a CRT (Cathode Ray Tube) display, a touch panel, etc.

[0031] The storage circuit 34 stores various data and programs. Specifically, the storage circuit 34 is connected to the processing circuit 35, storing data received from the processing circuit 35, or reading stored data and sending it to the processing circuit 35. For example, the storage circuit 34 is implemented using semiconductor memory elements such as RAM (Random Access Memory), flash memory, hard disks, optical disks, etc.

[0032] The processing circuit 35 controls the entire medical image processing device 3. For example, the processing circuit 35 performs various processes based on input operations received from the user via the input interface 32. For example, the processing circuit 35 receives data sent from other devices via the communication interface 31 and stores the received data in the storage circuit 34. Additionally, for example, the processing circuit 35 transmits the data received from the storage circuit 34 to other devices by sending the data to the communication interface 31. Furthermore, for example, the processing circuit 35 displays the data received from the storage circuit 34 on the display 33.

[0033] The above describes a configuration example of the medical image processing apparatus 3 according to this embodiment. For example, the medical image processing apparatus 3 according to this embodiment is installed in medical facilities such as hospitals and clinics to assist users such as physicians in various diagnoses and treatment plan formulations. For example, even when artifacts and abnormal shadows similar to artifacts appear simultaneously in the medical image data during the analysis of the medical image data, the medical image processing apparatus 3 performs various processing steps for appropriate analysis.

[0034] As mentioned above, techniques exist for identifying medical image data with a high probability of artifacts caused by gravity in medical image data, based on clinical knowledge. However, since it is difficult to completely distinguish such artifacts, as well as similar abnormal shadows (such as ground-glass opacities in the lungs (GGO),) even when the aforementioned techniques are used, all artifacts and abnormal shadows are treated as abnormal shadows when they occur simultaneously.

[0035] Therefore, the medical image processing apparatus 3 of this embodiment is configured to assign weights to the pixels constituting the medical image data based on the gravity direction in the medical image data, and perform analysis based on the assigned weights, thereby enabling appropriate analysis even when artifacts and abnormal shadows similar to artifacts simultaneously appear in the medical image data. The medical image processing apparatus 3 having this configuration will be described in detail below.

[0036] For example, such as Figure 1 As shown, in this embodiment, the processing circuit 35 of the medical image processing apparatus 3 executes a control function 351, an acquisition function 352, an extraction function 353, a recognition function 354, a first setting function 355, and a parsing function 356. Here, the acquisition function 352 is an example of an acquisition unit. The extraction function 353 is an example of an extraction unit. The recognition function 354 is an example of a recognition unit. The first setting function 355 is an example of a first setting unit. The parsing function 356 is an example of a parsing unit.

[0037] Control function 351 controls operations via input interface 32 to generate various GUIs (Graphical User Interfaces) and display information, which are then displayed on monitor 33. For example, control function 351 causes monitor 33 to display the GUI used for performing image processing on medical image data, the parsing results related to the medical image data, etc. Additionally, control function 351 generates various display images based on the medical image data acquired by acquisition function 352 and displays them on monitor 33.

[0038] The acquisition function 352 acquires medical image data of the subject to be analyzed from the medical image diagnostic device 1 or the medical image storage device 2 via the communication interface 31. Specifically, the acquisition function 352 acquires two-dimensional or three-dimensional medical image data (volume data) that is the subject of image analysis related to various diagnoses. Furthermore, the acquisition function 352 can also acquire multiple volume data obtained by taking multiple images in three dimensions over time. For example, the acquisition function 352 acquires CT image data, ultrasound image data, MRI image data, X-ray image data, Angio image data, PET image data, SPECT image data, etc., as the aforementioned medical image data. The processing circuit 35, by executing the acquisition function 352, receives the medical image data of the subject from the medical image diagnostic device 1 or the medical image storage device 2, and stores the received medical image data in the storage circuit 34.

[0039] Extraction function 353 extracts at least one structure of the subject from medical image data. Specifically, extraction function 353 extracts regions representing biological organs contained in the medical image data acquired by acquisition function 352. For example, extraction function 353 extracts regions corresponding to lungs or similar organs contained in the medical image data. The processing performed by extraction function 353 will be described in detail later.

[0040] The recognition function 354 identifies the gravity direction corresponding to the medical image data acquired by the acquisition function 352. The processing performed by the recognition function 354 will be described in detail later.

[0041] The first setting function 355 assigns a first weight to the pixels constituting the medical image data based on the direction of gravity. The processing performed by the first setting function 355 will be described in detail later.

[0042] The parsing function 356 parses the medical image data based on the first weight. The processing performed by the parsing function 356 will be described in detail later.

[0043] The aforementioned processing circuit 35 is implemented, for example, by a processor. In this case, the aforementioned processing functions are stored in the storage circuit 34 as programs executable by a computer. Furthermore, the processing circuit 35 implements the functions corresponding to each program by reading and executing the programs stored in the storage circuit 34. In other words, when the processing circuit 35 has read the programs, it has… Figure 1 The processing functions shown.

[0044] Alternatively, the processing circuit 35 can be configured by combining multiple independent processors, with each processor executing a program to implement its respective processing function. Furthermore, the processing functions of the processing circuit 35 can be appropriately distributed or unified across one or more processing circuits. Additionally, the processing functions of the processing circuit 35 can be implemented through a combination of hardware (such as circuits) and software. While an example has been described here where the program corresponding to each processing function is stored in a single storage circuit 34, the implementation is not limited to this. For example, a configuration could be adopted where the program corresponding to each processing function is distributed across multiple storage circuits, and the processing circuit 35 reads and executes each program from each storage circuit.

[0045] Next, regarding the order of processing performed by the medical image processing device 3, in the use of Figure 2 After the explanation, the details of each process will be explained. Figure 2 This is a flowchart showing the processing sequence of the processing functions performed by the processing circuit 35 of the medical image processing apparatus 3 of the first embodiment.

[0046] For example, such as Figure 2 As shown, in this embodiment, the acquisition function 352 acquires medical image data of the subject from the medical image diagnostic device 1 or the medical image storage device 2 (step S101). For example, the acquisition function 352 acquires specified medical image data according to the medical image data acquisition operation via the input interface 32. This process is implemented, for example, by the processing circuit 35 calling and executing the program corresponding to the acquisition function 352 from the storage circuit 34.

[0047] Next, the recognition function 354 identifies the direction of gravity in the acquired medical image data (step S102). For example, this can be achieved by the processing circuit 35 calling and executing the program corresponding to the recognition function 354 from the storage circuit 34.

[0048] Next, extraction function 353 extracts the structures contained in the acquired medical image data (step S103). This process is implemented, for example, by processing circuit 35 calling and executing the program corresponding to extraction function 353 from storage circuit 34. Additionally, in Figure 2The example shown is that the structure is extracted after the direction of gravity is identified, but the implementation is not limited to this. It is also possible to identify the direction of gravity after the structure is extracted, or to perform the identification of the direction of gravity and the extraction of the structure at the same time.

[0049] Next, the first setting function 355 sets a first weight for the pixels of the medical image data based on the direction of gravity (step S104). This process is implemented, for example, by the processing circuit 35 calling and executing the program corresponding to the first setting function 355 from the storage circuit 34.

[0050] Then, the parsing function 356 parses the medical image data using the first weight (step S105). This process is implemented, for example, by the processing circuit 35 calling and executing the program corresponding to the parsing function 356 from the storage circuit 34.

[0051] In addition, although Figure 2 Although not shown in the diagram, after parsing the medical image data, control function 351 enables the parsing results to be displayed on display 33. This process is implemented, for example, by processing circuit 35 calling and executing the program corresponding to control function 351 from storage circuit 34.

[0052] The details of each process performed by the medical image processing apparatus 3 will be described below. Furthermore, the following description will exemplify the process of acquiring three-dimensional chest X-ray CT image data as medical image data and performing lung-related analysis. However, the processing described in this embodiment is not limited to this; it can be applied to various other biological organs.

[0053] (Acquisition and processing of medical image data)

[0054] like Figure 2 As described in step S101, the acquisition function 352 acquires three-dimensional X-ray CT image data of the chest as the object by means of the medical image data acquisition operation via the input interface 32.

[0055] Furthermore, the acquisition and processing of medical image data in step S101 can be initiated either by user instruction via input interface 32 as described above, or it can be initiated automatically. In the latter case, for example, the acquisition function 352 monitors the medical image storage device 2 in advance and automatically acquires the data each time new three-dimensional chest X-ray CT image data is stored.

[0056] Alternatively, as described above, after acquiring three-dimensional chest X-ray CT image data by acquisition function 352, the acquired image data can be displayed on display 33.

[0057] (Identification and processing of gravity direction)

[0058] like Figure 2 As described in step S102, the identification function 354 identifies the direction of gravity in the three-dimensional chest X-ray CT image data acquired by the acquisition function 352. Specifically, the identification function 354 identifies the direction of gravity applied to the subject when the three-dimensional chest X-ray CT image data was captured (i.e., the direction of gravity of the subject depicted in the three-dimensional chest X-ray CT image data).

[0059] For example, recognition function 354 identifies the direction of gravity based on the patient's position as captured in three-dimensional chest X-ray CT image data. As an example, recognition function 354 infers the patient's position based on the image capture information from the three-dimensional chest X-ray CT image data and identifies the direction of gravity based on the inferred position. Here, recognition function 354 obtains the image capture information from the patient position, one of the header information of DICOM (Digital Imaging and Communications in Medicine), a standard for medical image data. For example, if the image capture information is HFS (Head First-Supine) or FFS (Feet First-Supine), recognition function 354 infers that the image was taken in a supine position. Additionally, if the shooting information is HFP (Head First-Prone), then the recognition function 354 infers that the shooting was taken in a prone position; if it is HFDL (Head First-Decubitus Left), then the recognition function 354 infers that the shooting was taken in a left lateral decubitus position.

[0060] Furthermore, if the body position is supine, the recognition function 354 identifies the direction of gravity as dorsal; if the body position is prone, it identifies the direction of gravity as ventral; and if the body position is left lateral decubitus, it identifies the direction of gravity as left of the body axis. Additionally, the following explanation uses 3D chest X-ray CT images taken in a supine position with the direction of gravity set to dorsal.

[0061] (Extraction and processing of structures)

[0062] like Figure 2As described in step S103, extraction function 353 extracts the structures contained in the three-dimensional chest X-ray CT image data. Specifically, extraction function 353 obtains the coordinate information of pixels representing the lungs, etc., in the three-dimensional chest X-ray CT image data. Here, extraction function 353 can extract the structures using various methods. For example, extraction function 353 can extract the region specified on the three-dimensional chest X-ray CT image data via input interface 32 as the lung. That is, extraction function 353 extracts the region manually specified by the user as the lung.

[0063] Additionally, for example, extraction function 353 can extract the lungs from 3D chest X-ray CT image data based on the anatomical structures depicted in the data using known region extraction techniques. For instance, extraction function 353 uses methods such as Otsu binarization based on CT values, region expansion, the Snake method, graph cut method, and mean shift method to extract the lungs from the 3D chest X-ray CT image data.

[0064] In addition, extraction function 353 can extract the lungs from three-dimensional chest X-ray CT image data using a learned model constructed based on pre-prepared learning data using machine learning techniques (including deep learning). Furthermore, by extracting the lungs using any of the methods described above, extraction function 353 can also extract the chest wall on the lateral (non-mediastinal) side of the lungs based on anatomical knowledge.

[0065] (Setting and processing the first weight)

[0066] like Figure 2 As described in step S104, the first setting function 355 assigns a first weight to the pixels of the three-dimensional chest X-ray CT image data. Specifically, the first setting function 355 sets the first weight based on the direction of gravity and at least one structure. For example, the first setting function 355 sets the first weight based on the direction of gravity and the distance from the edge of the extracted structure.

[0067] The aforementioned gravitational effect occurs because soft tissues, which are susceptible to gravity (e.g., the lungs), are supported by hard tissues, which are less susceptible to gravity (e.g., the chest wall), causing compression of the soft tissues. The closer the compression of the soft tissues is to the area where the load is applied, i.e., the closer it is to the hard tissues, the higher the probability of this occurring.

[0068] Therefore, the first setting function 355 assumes that there is a chest wall on the outside of the extracted lung, and determines an edge in the direction indicated by the gravity vector that is approximately orthogonal to the direction of gravity at the edge portion of the extracted lung. Moreover, the first setting function 355 sets a first weight based on the distance from the determined edge.

[0069] Figure 3 This is a diagram illustrating an example of the first weight setting process in the first embodiment. Here, Figure 3 This illustrates a cross-sectional image data 400 of a three-dimensional chest X-ray CT image, showing a body axial section. For example, cross-sectional image data 400 is provided to the user as image data for analysis. Recognition function 354 identifies [the image data] based on the image capture information of the three-dimensional chest X-ray CT image data. Figure 3 The direction of gravity is shown. Additionally, extraction function 353 extracts the lungs contained in the three-dimensional chest X-ray CT image data, and based on the extracted lungs, extracts the lung edges 410.

[0070] For example, the first setting function 355 determines, within the extracted lung edge 410, an edge 420 located in the direction indicated by the arrow pointing to the direction of gravity, which is approximately orthogonal to the direction of gravity (hereinafter, edge 420 will be referred to as reference edge 420). Furthermore, the first setting function 355 assigns a first weight to each pixel based on its distance from the determined reference edge 420. For example, the first setting function 355 assigns a first weight to each pixel based on the following equation (1).

[0071] w1(i)=0.5+0.1×min(l1(i),5)…(1)

[0072] Here, in equation (1), “w1(i)” represents the first weight of pixel i. In addition, “l1(i)” represents the distance (number of pixels) in the direction of gravity from the reference edge to the pixel. In addition, “min(l1(i), 5)” represents the function that takes the smaller value between “l1(i)” and “5”.

[0073] For example, the first setting function 355 inputs "l1=0" to the above formula (1) for pixel 430 which is connected to the reference edge 420, and sets the first weight "w1=0.5+0.1×0=0.5". In addition, the first setting function 355 inputs "l1=2" to the above formula (1) for pixel 440 which has "2" pixels in the gravity direction at a distance of "from" the reference edge 420, and sets the first weight "w1=0.5+0.1×2=0.7". In addition, the first setting function 355 sets the first weight "w1=0.5+0.1×5=1.0" to pixel 450 which has more than "5" pixels in the gravity direction at a distance of "from" the reference edge 420.

[0074] That is, in equation (1), it is shown that when the distance from the reference edge in the direction of gravity is less than a threshold (5 pixels in this example), the shorter the distance from the reference edge in the direction of gravity, the smaller the first weight. This becomes a formula that takes into account the above-mentioned mechanism of gravity effect. Of course, the setting method based on this formula is only one example, and the first weight can also be set by other methods.

[0075] In addition, Figure 3 In this case, the reference edge is in the same position in both the left and right lungs, but it is also possible that the reference edge is in different positions in the left and right lungs. Alternatively, the first weight can be set using different formulas for the right and left lungs.

[0076] In addition, Figure 3 The example shown is a two-dimensional example for simplicity, but it could also be a case where a reference edge (point / section) is set and a first weight is set in three-dimensional space. In this case, the first setting function 355 determines the reference edge in three-dimensional space based on the lung edge extracted from the in-volumetric data and the direction of gravity, and sets a first weight for each voxel according to the distance from the reference edge.

[0077] In addition, Figure 3 The example shown illustrates the case where the distance from the reference edge is determined by the number of pixels, but the implementation is not limited to this; it can also be determined by the distance in actual space. In this case, for example, similar to equation (1), a threshold is set for the actual distance in the direction of gravity from the reference edge, and for distances smaller than the set threshold, the first weight is set in such a way that the shorter the distance from the reference edge in the direction of gravity, the smaller the first weight.

[0078] Furthermore, in Equation (1) above, an example of setting the threshold to a constant (5 pixels) was given, but the implementation is not limited to this, and it can also be dynamically changed. For example, it can also be dynamically determined according to the size of the subject's body (e.g., the larger the body, the larger the threshold, etc.).

[0079] (Analysis and processing)

[0080] like Figure 2 As described in step S105, the parsing function 356 performs parsing processing on the three-dimensional chest X-ray CT image data using the first weight set for each pixel by the first setting function 355. For example, the parsing function 356 can analyze the distribution (mean, variance, etc.) of CT values ​​within the lungs. At this time, the parsing function 356 multiplies the CT value of each pixel by the first weight, and uses the sum of the CT values ​​multiplied by the first weight to calculate the weighted average and weighted variance. Here, the parsing function 356 can both... Figure 3 The entire cross-sectional image data 400 shown can be set as the parsing object, or a local area in the cross-sectional image data 400 can also be set as the parsing object.

[0081] Furthermore, the parsing function 356 parses the three-dimensional chest X-ray CT image data by inputting the calculated weighted average and weighted variance into a classifier for inferring disease names, or as input features for similar case retrieval. Additionally, the above example of parsing is only one instance; the parsing function 356 can also perform weighted parsing based on other image processing methods.

[0082] (Variation Example 1)

[0083] In the above embodiment, in step S102, the imaging information of the three-dimensional chest X-ray CT image data is obtained from the header information of the DICOM, the body position of the subject at the time of imaging is inferred, and the direction of gravity is identified based on the inferred body position. However, the embodiment is not limited to this, and the body position can be inferred and the direction of gravity identified by other methods. For example, the identification function 354 can infer the body position of the subject at the time of imaging based on RIS (Radiology Information Systems), electronic medical records, etc., and identify the direction of gravity based on the inferred body position.

[0084] Alternatively, the recognition function 354 may identify the direction of gravity through image analysis of medical image data. For example, the recognition function 354 infers the patient's position based on the results of image analysis of the medical image data, and identifies the direction of gravity based on the inferred position. As another example, the recognition function 354 measures the opening diameter between the spine and posterior mediastinum, and the contact area between the mediastinum and chest wall, etc., in three-dimensional chest X-ray CT image data through image processing, and infers the patient's position at the time the three-dimensional chest X-ray CT image data was taken based on the measured opening diameter and contact area. Then, the recognition function 354 identifies the direction of gravity based on the inferred position.

[0085] (Variation Example 2)

[0086] In the above embodiment, the case where the first weight is set based on the direction of gravity and the extracted structure in step S104 is described. However, the embodiment is not limited to this, and the first weight may be set by other methods. For example, the first setting function 355 calculates the angle at which the part corresponding to the pixel is supported by the structure based on the direction of gravity and the structure extracted from the medical image data, and sets the first weight for the pixel based on the calculated angle and the distance to the structure.

[0087] Figure 4 This is a diagram illustrating an example of the setting process for the first weight in Modification Example 2. Here, Figure 4 This image shows a magnified view of a cross-sectional image of an axial section from a three-dimensional chest X-ray CT image. (Identification function 354 pairs) Figure 4 The direction of gravity is identified as shown. In addition, the extraction function 353 extracts the lungs contained in the three-dimensional chest X-ray CT image data, and extracts the lung edges 510 based on the extracted lungs.

[0088] For example, the first setting function 355 extracts the approximate edge 512 obtained by approximating the curve of the lung edge 510. Then, the first setting function 355 determines the pixels j that are adjacent to the approximate edge 512, and calculates the angle "θ" between the tangent of the determined pixel j and the direction of gravity. j Here, let's set it to "0≤θ". j ≤π / 2.

[0089] For example, the first setting function 355 in Figure 4 In pixel 530 shown, the angle between the tangent 514 relative to the approximate edge 512 and the direction of gravity is calculated as "θ". 530 =0". Additionally, the first setting function 355 calculates the angle between the tangent 516 relative to the approximate edge 512 and the direction of gravity in pixel 535 as "θ". 535 ".

[0090] Then, the first setting function 355 will calculate the Euclidean distance from pixel i with the first weight to each pixel j that is in contact with the edge. 1j The minimum value calculated by (i) and “θj” is set as the first weight “w1(i)”. For example, the first setting function 355 sets the first weight for each pixel based on the following equation (2).

[0091]

[0092] Here, in equation (2), “min” j (·)” represents the value of the pixel j that minimizes the expression within the parentheses among all pixels j touched by the edge. In equation (2), for the Euclidean distance “l 1j (i)” plus |sinθ j |, if θ j >0 then |sinθ j|>0. Therefore, except for pixels with tangents perpendicular to the direction of gravity, the first weight “w1(i)” increases as the distance is assigned a positive value and the value increases. As mentioned above, the smaller the first weight, the more the influence of gravity is taken into account. That is, equation (2) becomes an equation that takes into account the following situation: gravity is dispersed about positions not perpendicular to the direction of gravity, so the first weight increases.

[0093] Furthermore, the first weight is not limited to the examples mentioned above, and various other weights can be set. For example, the first setting function 355 can also set a weight less than "1" for all pixels whose distance from the structure is less than a threshold.

[0094] (Variation Example 3)

[0095] In the above embodiment, the image processing in step S105 is described as a parsing process. However, the embodiment is not limited to this; for example, the parsing may be performed by a learner. In this case, for example, the parsing function 356 generates weighted image data based on a first weight, and performs parsing of medical image data based on this weighted image data.

[0096] Figure 5 This is a diagram illustrating an example of the analytical processing used in variation example 3. For example, as... Figure 5 As shown, the parsing function 356 is based on the medical image data "I k "Each pixel of 600 is assigned a first weight, and weighted image data W(I) is generated using the first weight as the pixel value." k )"610. Furthermore, the parsing function 356 will convert medical image data "I k "600 and weighted image data" W(I k The input 610 is fed into a deep learning-based learner and outputs the final parsing result through multiple convolutional layers. Here, Figure 5 The learner shown is, for example, a medical image data "I k "600 and weighted image data" W(I k A learner that takes "610" as input and outputs the presence or absence of a disease. Additionally, Figure 5 The learner shown is pre-generated and stored in storage circuit 34.

[0097] As described above, according to the first embodiment, the acquisition function 352 acquires medical image data that is the object of analysis. The identification function 354 identifies the direction of gravity relative to the medical image data. The first setting function 355 sets a first weight for the pixels constituting the medical image data based on the direction of gravity. The analysis function 356 analyzes the medical image data based on the first weight. Therefore, the medical image processing apparatus 3 of the first embodiment can perform analysis that takes into account the effect of gravity, and can appropriately analyze the medical image data.

[0098] Furthermore, according to the first embodiment, the extraction function 353 extracts at least one structure of the subject from the medical image data. The first setting function 355 sets a first weight based on the direction of gravity and the at least one structure. Therefore, the medical image processing apparatus 3 of the first embodiment can set a first weight for portions of the medical image data that are more likely to produce a gravitational effect, and can perform more appropriate parsing of the medical image data.

[0099] Furthermore, according to the first embodiment, the recognition function 354 infers the body position of the subject based on the image capture information of the medical image data, and identifies the direction of gravity based on the inferred body position. Therefore, the medical image processing device 3 of the first embodiment can easily identify the direction of gravity in the medical image data.

[0100] Furthermore, according to the first embodiment, the recognition function 354 infers the body position of the subject based on the image analysis results of the medical image data, and identifies the direction of gravity based on the inferred body position. Therefore, the medical image processing apparatus 3 of the first embodiment is able to identify the direction of gravity from the image.

[0101] Furthermore, according to the first embodiment, the first setting function 355 sets a first weight for each pixel based on the direction of gravity and the distance from the edge of the structure extracted from the medical image data. Therefore, the medical image processing apparatus 3 of the first embodiment can set a first weight corresponding to the ease of applying a load.

[0102] Furthermore, according to the first embodiment, the first setting function 355 calculates the angle at which the part corresponding to the pixel is supported by the structure based on the direction of gravity and the structure extracted from the medical image data, and sets a first weight for the pixel based on the calculated angle and the distance to the structure. Therefore, the medical image processing apparatus 3 of the first embodiment can set a first weight that better reflects the ease of applying load.

[0103] Furthermore, according to the first embodiment, the parsing function 356 generates weighted image data based on a first weight, and performs parsing of medical image data based on this weighted image data. Therefore, the medical image processing apparatus 3 of the first embodiment enables parsing to be performed easily.

[0104] (Second Implementation)

[0105] In the first embodiment described above, a first weight was set for the entire medical image data. In the second embodiment, the following situation will be described: It is determined whether a shadow indicating a gravitational effect exists in the medical image data; if it is determined to exist, the first weight is set only for the pixels corresponding to the region of the shadow indicating a gravitational effect. In other words, if it is determined whether a gravitational effect may occur in at least one pixel of the medical image data, and it is determined that it may occur, the first weight is set for the location (at least one pixel) where the gravitational effect may occur. Furthermore, as in the first embodiment, the image data of the object to be analyzed is three-dimensional chest X-ray CT image data, and the lung is extracted as a structure. Of course, the object is not limited to this; the above are merely examples used to illustrate the process of the medical image processing apparatus.

[0106] Figure 6 This is a diagram illustrating a configuration example of the medical image processing apparatus according to the second embodiment. Here, in Figure 6 In the middle, to and Figure 1 The medical image processing apparatus 3 of the first embodiment shown is labeled with the same reference numerals to perform the same operations. That is, compared with the medical image processing apparatus 3 of the first embodiment, the processing circuit 35a of the second embodiment newly executes the determination function 357, which is different from the processing content of the first setting function 355a executed by the processing circuit 35a. Hereinafter, the description will focus on these aspects.

[0107] Figure 7 This is a flowchart showing the processing sequence of the processing functions performed by the processing circuit 35a of the medical image processing apparatus 3a according to the second embodiment.

[0108] Figure 7 Steps S201 to S203 and Figure 2 Steps S101 to S103 are the same, and are implemented by the processing circuit 35a reading from the storage circuit 34 and executing the program corresponding to each processing function.

[0109] For example, in this embodiment, such as Figure 7 As shown, after extraction function 353 extracts the structure (step S203), determination function 357 determines whether a gravitational effect has occurred in the medical image data (step S204). This process is implemented, for example, by processing circuit 35a calling and executing the program corresponding to determination function 357 from storage circuit 34.

[0110] Next, the first setting function 355a sets a first weight for the pixels determined to have a gravity effect (step S205). This process is implemented, for example, by the processing circuit 35a calling and executing the program corresponding to the first setting function 355a from the storage circuit 34.

[0111] Figure 7 Step S206 and Figure 2 The same as step S105 is implemented by the processing circuit 35a calling and executing the program corresponding to the parsing function 356 from the storage circuit 34.

[0112] The following describes in detail the processes performed by the medical image processing device 3a.

[0113] (Determination and handling of gravitational effects)

[0114] like Figure 7 As described in step S204, the determination function 357 determines whether a gravitational effect has occurred in the three-dimensional chest X-ray CT image data acquired by the acquisition function 352. Specifically, the determination function 357 determines whether a gravitational effect is likely to occur in at least one pixel of the three-dimensional chest X-ray CT image data.

[0115] For example, the determination function 357 determines whether a shadow indicating a gravitational effect exists based on three-dimensional chest X-ray CT image data, the direction of gravity identified according to the patient's position, and the structure of the patient. Here, in this embodiment, the method described in the first embodiment is used to determine the edge located in the direction indicated by the gravity vector, and the presence of a shadow indicating a gravitational effect is determined based on the CT value of a pixel at a constant distance from the determined edge. That is, the determination function 357 determines whether a gravitational effect is likely to occur in at least one pixel of the three-dimensional chest X-ray CT image data based on the results of image analysis of the medical image data.

[0116] Figure 8 This is a diagram illustrating an example of the determination process in the second embodiment. Here, Figure 8 This image shows a cross-sectional image 700 of an axial section from a three-dimensional chest X-ray CT image. Recognition function 354 pairs. Figure 8 The direction of gravity is identified as shown. In addition, the extraction function 353 extracts the lungs contained in the three-dimensional chest X-ray CT image data, and extracts the lung edges 710 based on the extracted lungs.

[0117] For example, the determination function 357 first determines a reference edge 720 in the lung edge 710, which is located in the direction indicated by the arrow in the direction of gravity and is approximately orthogonal to the direction of gravity. Then, the determination function 357 determines the reference edge 720 at a threshold distance "l" along the direction of gravity.th At the position of "", set the threshold line to 760. Additionally, at... Figure 8 The diagram shows the threshold distance "l" th "An example of setting it to 4 pixels, but the threshold distance is "l th "It can be set arbitrarily. For example, the threshold distance can be changed based on the body shape of the subject, the biological organs being analyzed, etc." th In the case of "", it could also be due to the user manually changing the threshold distance "l" through input interface 32. th The situation of "".

[0118] If so Figure 8 If the threshold line 760 is set as shown, the determination function 357 will target the pixels inside the lung between the reference edge 720 and the threshold line 760 to determine whether there is a shadow that suggests a gravitational effect. For example, the determination function 357 will determine pixels inside the lung between the reference edge 720 and the threshold line 760 whose CT values ​​are in a range higher than a predetermined range as pixels that correspond to a shadow that suggests a gravitational effect.

[0119] Here, the range of CT values ​​used for determination can be arbitrarily set. For example, determination function 357 determines pixels whose CT values ​​are in a range (e.g., -800HU to -600HU) higher than a predetermined value (typically, the CT value inside the lung (around -1000HU)) as pixels corresponding to shadows that indicate gravitational effects. Alternatively, determination function 357 can also set the average CT values ​​of pixels inside the lung outside the reference edge 720 to the threshold line 760 in the cross-sectional image data 700 as a reference value, and determine pixels in a range of CT values ​​higher than the reference value as pixels corresponding to shadows that indicate gravitational effects.

[0120] The determination function 357 can also change the range of CT values ​​used depending on the situation. For example, the determination function 357 can change the range of CT values ​​used for determination based on the body shape of the subject and the respiratory phase (inhalation or exhalation) of the three-dimensional chest X-ray CT image data.

[0121] As described above, the determination function 357 sets the pixels inside the lung between the reference edge 720 and the threshold line 760 as the object, and determines whether there is a shadow that suggests a gravitational effect, thereby... Figure 8The region 770 shown is identified as the region corresponding to the shadow that evokes the gravitational effect. Furthermore, the determination process performed by the determination function 357 is not limited to the above-described process and can apply various other processes. For example, the determination function 357 can also determine, based on clinical knowledge, the location where artifacts caused by the gravitational effect are likely to occur, and, based on the continuity of the high-absorption region at that location, determine the region with a higher probability of artifact generation.

[0122] (Setting and processing the first weight)

[0123] like Figure 7 As described in step S205, the first setting function 355a sets the pixels contained in region 770, which is determined by the determination function 357 to correspond to the region of shadows that induce a gravitational effect, as objects and sets a first weight. For example, the first setting function 355a sets "0.7" as the first weight for all pixels constituting region 770. Furthermore, the above-described setting of the first weight is only one example, and the first weight can be set using various other methods. For example, the first setting function 355a can also set "0" as the first weight for the pixels in region 770 to disregard regions that induce a gravitational effect. Additionally, the first setting function 355a can also set the first weight for each pixel within region 770 using the method described in the first embodiment.

[0124] (Variation Example 1)

[0125] In the above embodiment, the case where the shadow indicating a gravitational effect is determined in step S204 based on the CT value of a pixel at a constant distance from the reference edge is described. However, the embodiment is not limited to this; it is also possible to use medical image data as input and perform the determination using a classifier that determines whether a gravitational effect is indicated. In this case, the determination function 357 determines, based on the classifier, whether a gravitational effect is likely to occur in at least one pixel of the medical image data.

[0126] Furthermore, the classifier used can determine whether a gravitational effect is indicated in each pixel, or it can extract regions that are considered to have a gravitational effect. For example, a classifier trained as follows can be used: labeled medical image data is manually input as the teacher image to identify regions (or pixels) that have a gravitational effect; if the medical image data is input, the region (pixel) that has a gravitational effect is identified.

[0127] As described above, according to the second embodiment, the determination function 357 determines whether a gravitational effect is likely to occur in at least one pixel of the medical image data. The first setting function 355a sets a first weight for the pixel based on the determination result. Therefore, since the medical image processing apparatus 3a of the second embodiment sets the first weight only for regions where the probability of a gravitational effect occurring is high, it is able to perform medical image data parsing more appropriately.

[0128] Furthermore, according to the second embodiment, the determination function 357 determines, based on the results of image analysis of the medical image data, whether a gravitational effect may occur in at least one pixel of the medical image data. Therefore, the medical image processing apparatus 3a of the second embodiment can easily perform the determination process.

[0129] Furthermore, according to the second embodiment, the determination function 357, based on a classifier, determines whether a gravitational effect may occur in at least one pixel of the medical image data. Therefore, the medical image processing apparatus 3a of the second embodiment can easily perform the determination process.

[0130] (Third Implementation)

[0131] In the first and second embodiments described above, the case of using a first weight based on the direction of gravity for analysis was explained. In the third embodiment, the following situation is described: for portions of medical image data deemed more important for analysis due to reasons different from the gravitational effect, a second weight is assigned, and analysis is performed using both the first and second weights. Furthermore, as in the first embodiment, the image data for analysis is three-dimensional chest X-ray CT image data, and the lungs are extracted as a construct. Additionally, the analysis object is used to infer lung diseases, particularly interstitial lung abnormalities.

[0132] Figure 9 This is a diagram illustrating a configuration example of the medical image processing apparatus according to the third embodiment. Here, in Figure 9 In the middle, to and Figure 1 The medical image processing apparatus 3 of the first embodiment shown has the same configuration for performing the same operations and is labeled with the same reference numerals. That is, compared to the medical image processing apparatus 3 of the first embodiment, the medical image processing apparatus 3b of the third embodiment differs in that the processing circuit 35b newly executes the second setting function 358, and the processing content of the parsing function 356b executed by the processing circuit 35b is different. Hereinafter, the description will focus on these aspects.

[0133] Figure 10 This is a flowchart showing the processing sequence of the processing functions performed by the processing circuit 35b of the medical image processing apparatus 3b in the third embodiment.

[0134] Figure 10 Steps S301 to S304 in the process Figure 2 Steps S101 to S104 are the same, and are implemented by the processing circuit 35b reading from the storage circuit 34 and executing the program corresponding to each processing function.

[0135] For example, in this embodiment, such as Figure 10 As shown, after the first setting function 355 sets the first weight (step S304), the second setting function 358 sets a second weight for the pixels of the medical image data based on the structure of the subject extracted from the medical image data (step S305). This process is implemented, for example, by the processing circuit 35b calling and executing the program corresponding to the second setting function 358 from the storage circuit 34.

[0136] Next, the parsing function 356b parses the medical image data based on the first weight and the second weight set for the medical image data (step S306). This process is implemented, for example, by the processing circuit 35b calling and executing the program corresponding to the parsing function 356b from the storage circuit 34.

[0137] The following describes in detail the processes performed by the medical image processing device 3b.

[0138] (Setting and processing the second weight)

[0139] like Figure 10 As described in step S305, the second setting function 358 sets a second weight for the pixels of the three-dimensional chest X-ray CT image data based on the lung extracted from the three-dimensional chest X-ray CT image data. Here, the second weight is set according to the purpose of the analysis. For example, in the case where the analysis object is interstitial lung disease as in this embodiment, based on the prior knowledge that information near the chest wall is more important, the second setting function 358 sets the second weight according to the distance from the chest wall.

[0140] Figure 11A as well as Figure 11B This is a diagram illustrating an example of the second weight setting process in the third embodiment. Here, Figure 11A This shows a cross-sectional image data 800 of an axial section from a three-dimensional chest X-ray CT image dataset. Additionally, Figure 11B Showing will Figure 11A A magnified portion of the image. Recognition function 354 pairs. Figure 11A The direction of gravity is identified as shown. Additionally, extraction function 353 extracts the lungs contained in the three-dimensional chest X-ray CT image data and extracts the lung edges 810 based on the extracted lungs. Furthermore, first setting function 355 determines a reference edge 820 based on the direction of gravity and the lung edges 810.

[0141] For example, the second setting function 358 assumes that there is a chest wall on the outside of the extracted lung, and sets a second weight for the pixels based on the distance from the edge of the extracted lung. As an example, the second setting function 358 sets a second weight for each pixel based on the following equation (3).

[0142] w2(i)=2-0.25×min(l2(i),4)…(3)

[0143] Here, in equation (3), “w2(i)” represents the second weight of pixel i. In addition, “l2(i)” represents the shortest distance from the edge of the lung to the pixel (converted to the number of pixels). In addition, “min(l2(i), 4)” represents a function that takes the smaller value between “l2(i)” and “4”.

[0144] For example, the second setting function 358 is for in Figure 11A For pixels 830 and 880 that are in contact with the lung edge 810, input "l2=0" into the above formula (3) and set the second weight "w2=2-0.25×0=2". In addition, for pixel 840, the second setting function 358 first, as shown in the figure, performs the second setting function 358. Figure 11B The shortest distance to the lung edge 810 is calculated as "l2 = √2". Then, the second setting function 358 inputs "l2 = √2" into the above formula (3) and sets the second weight "w2 = 2 - 0.25 × √2 = 1.7". In addition, for pixels 850 whose shortest distance to the lung edge 810 is greater than "4", the second setting function 358 sets the second weight "w2 = 2 - 0.25 × 4 = 1".

[0145] That is, in equation (3), it is shown that when the distance to the lung edge is less than a threshold (4 in this example), the shorter the distance to the lung edge, the greater the second weight. This becomes an equation that takes into account the prior knowledge mentioned above. Of course, the method of setting this equation is not limited to one example, and the second weight can also be set by other methods. For example, the prior knowledge that the edge of the mediastinal direction (e.g., the direction close to the heart and liver) is not very important can also be used, and the distance to the lung edge in the mediastinal direction can be ignored (set to be infinite).

[0146] (Analysis and processing)

[0147] like Figure 10As described in step S306, the parsing function 356b performs parsing processing on the three-dimensional chest X-ray CT image data using the first weight set for each pixel by the first setting function 355 and the second weight set by the second setting function 358. For example, the parsing function 356b performs the same parsing as in the first embodiment based on the weights synthesized by linearly summing the first weights and the second weights. Here, the parsing function 356b synthesizes the weights, for example, using the following equation (4).

[0148] w(i)=αw1(i)+(1-α)w2(i)…(4)

[0149] Here, in equation (4), “w(i)” represents the combined weight, and “α” represents the coefficient. For example, if “α” is increased, it becomes the weight that emphasizes the first weight “w1(i)”, and if “α” is decreased, it becomes the weight that emphasizes the second weight “w2(i)”. In this embodiment, it is set to “α = 0.5”. That is, the first weight and the second weight can be captured in the same way.

[0150] For example, if we take Figure 11A Taking the pixel in the example, for pixel 830, the parsing function 356b substitutes the first weight "w1 = 0.5" and the second weight "w2 = 2.0" into equation (4) to calculate the weight "w = 1.3". Similarly, for pixel 840, the parsing function 356b substitutes the first weight "w1 = 0.7" and the second weight "w2 = 1.7" into equation (4) to calculate the weight "w = 1.2".

[0151] Additionally, for pixel 850, the parsing function 356b substitutes the first weight "w1 = 1.0" and the second weight "w2 = 1.0" into equation (4) to calculate the weight "w = 1.0". Furthermore, for pixel 880, the parsing function 356b substitutes the first weight "w1 = 1.0" and the second weight "w2 = 2.0" into equation (4) to calculate the weight "w = 1.5".

[0152] Here, if we compare the calculated weights, then between pixel 830 (“w=1.3”) and pixel 880 (“w=1.5”), which are considered important by prior knowledge as the closest pixels to the chest wall, the weight of pixel 830, which is more likely to produce a gravitational effect, becomes smaller. That is, through Equation (4), even the same pixels closest to the chest wall are assigned weights that take into account the influence of gravitational effects.

[0153] Furthermore, the synthesis of the first and second weights is not limited to the method based on equation (4) described above, but can also be achieved through other methods. For example, weights can be synthesized using linear products, nonlinear sums, or nonlinear products. Alternatively, machine learning can be used to learn the appropriate composition of the coefficients.

[0154] As described above, if the weights are calculated, the parsing function 356b performs parsing processing of the medical image data based on the image processing described in the first embodiment.

[0155] (Variation Example 1)

[0156] In the above embodiment, the case where the second weight is set based on prior knowledge corresponding to the parsed object in step S305 is described. However, the embodiment is not limited to this; the second weight can also be set based on motion artifacts. In this case, for example, extraction function 353 extracts the heart, diaphragm, etc., as structures.

[0157] The second setting function 358 sets a second weight based on the distance from the extracted heart and diaphragm. That is, based on the prior knowledge that motion artifacts are generated due to the heartbeat and the movement of the diaphragm, the second setting function 358 sets a second weight such that pixels closer to the heart and diaphragm have a smaller second weight. In other words, the second setting function 358 sets a smaller second weight for pixels that are more likely to generate motion artifacts.

[0158] (Variation Example 2)

[0159] In the above embodiment, the case where image processing-based parsing is performed in step S306 is described. However, the embodiment is not limited to this; it may also be the same as the first embodiment, performing learner-based parsing. That is, the parsing function 356b generates weighted image data based on a first weight and a second weight, and performs parsing of medical image data based on this weighted image data. For example, the parsing function 356b generates synthetic weighted image data based on the synthesized weights, and sets the medical image data (three-dimensional chest X-ray CT image data) and the synthetic weighted image data as inputs to the learner, thereby obtaining the parsing result.

[0160] Furthermore, the weighted image data based on the first weight and the second weight is not limited to the synthetic weighted image data described above. For example, it can also be configured as follows: first weighted image data and second weighted image data are generated for the first weight and the second weight respectively, and deep learning is performed using medical image data, first weighted image data, and second weighted image data as input. In this case, it has the advantage that the coefficients used for synthesizing the weights do not need to be determined.

[0161] As described above, according to the third embodiment, the second setting function 358 sets a second weight for the pixels of the medical image data based on the structures extracted from the medical image data. The parsing function 356b performs parsing of the medical image data based on the first and second weights. Therefore, the medical image processing apparatus 3b of the third embodiment can perform parsing that takes into account both gravity effects and prior knowledge, and can perform parsing of medical image data more appropriately.

[0162] Furthermore, according to the third embodiment, the second setting function 358 sets a second weight for each pixel based on its distance from the edge of the structure. Therefore, the medical image processing apparatus 3b of the third embodiment can set a second weight corresponding to the purpose of the analysis.

[0163] Furthermore, according to the third embodiment, the parsing function 356b generates weighted image data based on the first weight and the second weight, and performs medical image data parsing based on this weighted image data. Therefore, the medical image processing apparatus 3b of the third embodiment can easily perform parsing using the first weight and the second weight.

[0164] (Other implementation methods)

[0165] In the above embodiments, the medical image data to be analyzed was described using three-dimensional chest X-ray CT image data. However, the embodiments are not limited to this, and may also include medical image data captured by other imaging devices (modality) and / or medical image data targeting other body parts as the analysis object.

[0166] Furthermore, the above-described embodiment describes the use of three-dimensional chest X-ray CT image data as the medical image data to be analyzed. However, the embodiment is not limited to this; for example, the medical image data to be analyzed could also be a sine curve.

[0167] Furthermore, the above-described embodiment focuses on CT image data captured by an X-ray CT scanner when the patient is in a supine position. However, the embodiment is not limited to this; for example, it may also focus on CT image data captured by an X-ray CT scanner when the patient is in an upright position. In this case, the direction affected by gravity is the direction below the Z-axis in either the sagittal or coronal section.

[0168] In the above embodiments, the display of analysis results has been described, but the display method of analysis results can be implemented in various ways. For example, analysis function 356 performs analysis considering gravity effects and analysis not considering gravity effects. Control function 351 can also display each analysis result separately. In addition, control function 351 can also control the display to notify the user when the analysis result based on analysis considering gravity effects differs from the analysis result based on analysis not considering gravity effects.

[0169] Furthermore, in the above embodiments, examples have been described in which the acquisition unit, extraction unit, identification unit, first setting unit, analysis unit, determination unit, and second setting unit in this specification are implemented by the acquisition function, extraction function, identification function, first setting function, analysis function, determination function, and second setting function of the processing circuit, respectively. However, the embodiments are not limited to this. For example, in addition to being implemented by the acquisition function, extraction function, identification function, first setting function, analysis function, determination function, and second setting function described in the embodiments, the acquisition unit, extraction unit, identification unit, first setting unit, analysis unit, determination unit, and second setting unit in this specification may also be implemented by hardware only, by software only, or by a combination of hardware and software.

[0170] Furthermore, the term "processor" used in the above description of the embodiments refers to circuits such as CPU (Central Processing Unit), GPU (Graphics Processing Unit), Application Specific Integrated Circuit (ASIC), and programmable logic device (e.g., Simple Programmable Logic Device (SPLD), Complex Programmable Logic Device (CPLD), and Field Programmable Gate Array (FPGA)). Here, instead of storing the program in a storage circuit, the program can be directly assembled within the processor's circuitry. In this case, the processor implements its function by reading and executing the program assembled within the circuitry. Additionally, the processors in this embodiment are not limited to being configured as a single circuit; multiple independent circuits can be combined to form a single processor to achieve its function.

[0171] Here, the medical image processing program, executed by the processor, is pre-assembled into ROM (Read Only Memory), storage circuitry, etc., and provided. Alternatively, this medical image processing program can also be provided using a non-transitory storage medium that can be read by a computer, such as a CD (Compact Disk)-ROM, FD (Flexible Disk), CD-R (Recordable), or DVD (Digital Versatile Disk), in a form that can be installed on or executed in these devices. Furthermore, the medical image processing program can also be stored in a computer connected to a network such as the Internet and provided or distributed by downloading it via the network. For example, the medical image processing program is composed of modules containing the aforementioned processing functions. As actual hardware, the CPU reads and executes the medical image processing program from a storage medium such as ROM, thereby loading each module onto the main storage device and generating it on the main storage device.

[0172] Furthermore, in the above-described embodiments and variations, the constituent elements of each device illustrated are functional concepts and do not necessarily need to be physically configured as shown in the illustrations. That is, the specific manner in which the devices are distributed or unified is not limited to that illustrated, and all or part of them can be functionally or physically distributed or unified in arbitrary units according to various loads, usage conditions, etc. Furthermore, all or any part of the processing functions performed in each device can be implemented by a CPU and the program parsed and executed in that CPU, or by hardware based on wiring logic.

[0173] Furthermore, all or part of the processes described in the above embodiments and variations, which are described as automatically executed processes, can also be performed manually, or all or part of the processes described as manually executed processes can be performed automatically using known methods. In addition, information including the processing order, control order, specific names, various data, and parameters shown in the above specification and drawings can be arbitrarily changed, unless specifically stated otherwise.

[0174] According to at least one embodiment described above, appropriate analysis can be performed even when analyzing medical image data that is highly likely to produce artifacts caused by gravity effects.

[0175] While several embodiments have been described, these embodiments are provided as examples and are not intended to limit the scope of the invention. These embodiments can be implemented in various other ways, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their variations are included within the scope and spirit of the invention, as well as within the scope of the invention as described in the claims and its equivalents.

Claims

1. A medical image processing device, comprising: The acquisition department acquires medical image data that will be used for analysis. The extraction unit extracts at least one structure of the subject from the medical image data; The recognition unit identifies the direction of gravity relative to the medical image data; The first setting unit determines the edge in the direction pointed to by the gravity direction in the edge of the structure, and sets a first weight for each pixel constituting the medical image data based on the distance to the edge in the direction pointed to by the gravity direction. as well as The parsing unit parses the medical image data based on the first weight. The first setting unit sets a first weight in such a way that the closer the weight is to the edge in the direction in which the gravity direction points, the smaller the weight value.

2. The medical image processing device as described in claim 1, The recognition unit infers the body position of the subject based on the captured information of the medical image data, and identifies the direction of gravity based on the inferred body position.

3. The medical image processing device as described in claim 1, The recognition unit infers the body position of the subject based on the image analysis results of the medical image data, and identifies the direction of gravity based on the inferred body position.

4. The medical image processing device as described in claim 1, It also includes a determination unit that determines whether a gravitational effect may occur in at least one pixel of the medical image data. The first setting unit sets the first weight for the pixels based on the result of the determination.

5. The medical image processing device as described in claim 4, The determination unit determines, based on the results of image analysis of the medical image data, whether a gravitational effect may occur in at least one pixel of the medical image data.

6. The medical image processing device as described in claim 4, The determination unit, based on a classifier, determines whether a gravitational effect may occur in at least one pixel of the medical image data.

7. The medical image processing device as described in claim 1, The first setting unit calculates the angle at which the part corresponding to the pixel is supported by the structure based on the direction of gravity and the structure extracted from the medical image data, and sets the first weight for the pixel based on the calculated angle and the distance to the structure.

8. The medical image processing device as described in claim 1, It also includes a second setting unit that sets a second weight for the pixels of the medical image data based on the structures extracted from the medical image data. The parsing unit parses the medical image data based on the first weight and the second weight.

9. The medical image processing apparatus as described in claim 8, The second setting unit sets the second weight for the pixel based on the distance from the edge of the structure.

10. The medical image processing apparatus as described in claim 1, The parsing unit generates weighted image data based on the first weight, and performs parsing of the medical image data based on the weighted image data.

11. The medical image processing apparatus as described in claim 8, The parsing unit generates weighted image data based on the first weight and the second weight, and performs parsing of the medical image data based on the weighted image data.

12. A medical image processing method, comprising the following steps: Obtain medical image data that will be analyzed; Extract at least one structure of the subject from the medical image data; Identify the direction of gravity relative to the medical image data; The edge in the direction of gravity in the edge of the structure is determined, and a first weight is set for each pixel constituting the medical image data such that the closer it is to the edge in the direction of gravity, the smaller the weight value. as well as Based on the first weight, the medical image data is parsed.

13. A storage medium that non-temporarily stores a program that causes a computer to perform the following processes: Obtain medical image data that will be analyzed; Extract at least one structure of the subject from the medical image data; Identify the direction of gravity relative to the medical image data; The edge in the direction of gravity in the edge of the structure is determined, and a first weight is set for each pixel constituting the medical image data such that the closer it is to the edge in the direction of gravity, the smaller the weight value. as well as Based on the first weight, the medical image data is parsed.

Images