Image damage detection support device, image damage detection support system, image damage detection support method and program

Abstract

Allow users to refer to images that can be used as a reference when reshooting. [Solution] According to the control unit of the console as a photo defect judgment support device, a photo defect judgment process is performed on the medical image using a trained model to generate a judgment result, and the medical image determined to be defective is input to the trained model to generate a target image, and the judgment result and the target image are output before reshooting is performed.

Description

Technical Field

[0001] The present invention relates to an imaging failure determination support device, an imaging failure determination support system, an imaging failure determination support method, and a program.

Background Art

[0002] Conventionally, an imaging failure determination technique for determining whether or not medical image imaging has failed (whether or not a medical image is an imaging failure) is known. For example, in Patent Document 1, it is determined whether a radiation image is an image taken with appropriate positioning, a predetermined feature amount is extracted from the image data of a specific region of the radiation image, and based on the learning result regarding the feature amount by a predetermined learning algorithm, a positioning determination device for determining whether a specific part is missing is described.

Prior Art Document

Patent Document

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] By the way, when it is determined that a medical image is an imaging failure, there may be a case where the user does not know how to change the positioning and imaging conditions at the time of re-imaging.

[0005] The present invention has been made in view of the above problems, and an object thereof is to enable a user to refer to an image that serves as a reference at the time of re-imaging.

Means for Solving the Problems

[0006] To solve the above problems, an imaging failure determination support device according to the present invention determination means for performing an imaging failure determination process on a medical image using a learned model and generating a determination result; A generation means generates a target image by inputting a medical image determined to be damaged by the aforementioned determination means into a trained model. Before reshooting is performed, an output means outputs the judgment result and the target image, It is equipped with.

[0007] Furthermore, the photo damage detection support system according to the present invention is A determination means that performs image defect detection processing on medical images using a pre-trained model and generates a determination result, A generation means generates a target image by inputting a medical image determined to be damaged by the aforementioned determination means into a trained model. Before reshooting is performed, an output control means controls the output of the determination result and the target image, It is equipped with.

[0008] Furthermore, the photo damage detection support method according to the present invention is A judgment step in which a pre-trained model is used to perform image defect detection on medical images and a judgment result is generated, The generation process involves inputting the medical image determined to be damaged in the above determination process into a trained model to generate a target image. Before reshooting is performed, an output step is performed which outputs the judgment result and the target image, It is equipped with.

[0009] Furthermore, the program according to the present invention is Computers A determination means that performs image defect detection processing on medical images using a pre-trained model and generates a determination result. A generation means generates a target image by inputting the medical image determined to be damaged by the aforementioned determination means into a trained model. Before reshooting is performed, an output means outputs the judgment result and the target image. To make it function as such. [Effects of the Invention]

[0010] According to the present invention, a user can refer to an image that serves as a reference during reshooting.

Brief Description of Drawings

[0011] [Figure 1] It is a block diagram showing an example of a radiation imaging system according to an embodiment of the present invention. [Figure 2] It is a block diagram showing the functional configuration of the console in FIG. 1. [Figure 3] It is a flowchart showing the flow of imaging control processing executed by the control unit in FIG. 2. [Figure 4] It is a diagram showing an example of display of the determination result and determination basis of defective image determination processing. [Figure 5] It is a diagram showing an example of a switching button for switching the ON / OFF of the marker for area display shown in FIG. 4. [Figure 6] It is a diagram showing an example of a switching button for switching the ON / OFF of the marker for area display shown in FIG. 4. [Figure 7] It is a diagram showing an example of a switching button for switching the ON / OFF of the marker for area display shown in FIG. 4. [Figure 8] It is a diagram showing an example of a switching button for switching the ON / OFF of the marker for area display shown in FIG. 4. [Figure 9] It is a diagram showing an example of a switching button for switching the ON / OFF of the marker for area display shown in FIG. 4. [Figure 10] It is a diagram showing an example of alert display. [Figure 11] It is a diagram showing a state where the area display shown in FIG. 4 follows and rotates when a rotation operation is performed on a radiation image. [Figure 12] It is a diagram showing a state where the area display is enlarged and displayed when the area display is small. [Figure 13] It is a diagram showing an example of display of a target image. [Figure 14] It is a diagram showing an example of a case where information indicating a determination basis is output as image information. [Modes for carrying out the invention]

[0012] Embodiments of the present invention will be described below with reference to the drawings. However, the technical scope of the present invention is not limited to the following embodiments and illustrated examples.

[0013] <1. Radiography System> First, the general configuration of the radiography system (hereinafter referred to as System 100) according to this embodiment will be described. Figure 1 is a block diagram of system 100.

[0014] As shown in Figure 1, system 100 includes a radiographic imaging device (hereinafter referred to as imaging device 1) and a console 2. Furthermore, the system 100 according to this embodiment further comprises a radiation generator (hereinafter referred to as generator 3) and an image management device 4. Each device 1 to 4 can communicate with each other via a communication network N (such as a LAN (Local Area Network), WAN (Wide Area Network), or the Internet).

[0015] The system 100 may be installed in the imaging room, or it may be configured to be mobile (for example, in a mobile medical unit). Furthermore, system 100 may be capable of communicating with a hospital information system (HIS), a radiology information system (RIS), and other systems not shown in the diagram.

[0016] [1-1. Radiation Generating Devices] The generator 3 comprises a generator 31, an irradiation instruction switch 32, and a radiation source 33.

[0017] The generator 31 applies a voltage to the radiation source 33 (tube) according to the preset imaging conditions based on the operation of the irradiation instruction switch 32.

[0018] The radiation source 33 is configured to generate radiation R (e.g., X-rays) in a dose corresponding to the applied voltage when a voltage is applied from the generator 31.

[0019] Furthermore, the generator 3 according to this embodiment generates radiation R in a manner corresponding to the form of the radiation image to be generated (still image, dynamic image having multiple frames). In the case of still images, radiation R is irradiated only once for each press of the irradiation instruction switch 32. In the case of dynamic images, pressing the irradiation instruction switch 32 once will cause pulsed radiation R to be irradiated multiple times per predetermined time (for example, 15 times per second), or the irradiation of radiation R will be continued for a predetermined time.

[0020] [1-2. Radiation imaging equipment] The imaging device 1 generates digital data of a radiographic image showing the part of the subject being photographed. The imaging device 1 according to this embodiment is a portable FPD (Flat Panel Detector). Specifically, the imaging device 1 according to this embodiment, although not shown in the figures, includes a sensor substrate in which an image sensor that generates an electric charge corresponding to the dose when it receives radiation R and a switch element that stores and releases charge are arranged in a two-dimensional (matrix) manner; a scanning unit that switches each switch element on and off; a reading unit that reads out the amount of charge emitted from each pixel as a signal value; a control unit that controls each unit and generates a radiation image from the multiple signal values ​​read out by the reading unit; and a communication unit that transmits the generated radiation image data and various signals to other devices (console 2, generator 3, image management device 4, etc.) and receives various information and various signals from other devices.

[0021] The imaging device 1 is configured to generate still image data (hereinafter referred to as still image data) or motion image data (hereinafter referred to as motion image data) by accumulating and releasing charge and reading out signal values ​​in synchronization with the timing of radiation R being irradiated from the generator 3. When generating still image data, the radiation image is generated only once for each press of the irradiation instruction switch 32. When generating dynamic image data, the generation of frames constituting the dynamic image is repeated multiple times per predetermined time (for example, 15 times per second) for each press of the irradiation instruction switch 32.

[0022] The imaging device 1 may be integrated with the generating device 3.

[0023] [1-3. Console] Console 2 is used to set various shooting conditions for at least one of the shooting device 1 and the generator 3. Console 2, on the other hand, consists of a PC, dedicated equipment, and other components. The imaging conditions include, for example, conditions related to the subject S (imaging area, imaging direction, body size, etc.), conditions related to radiation R irradiation (tube voltage, tube current, irradiation time, current-time product (mAs value), etc.), and conditions related to image reading by the imaging device 1. Console 2 may be configured to automatically set imaging conditions based on examination order information obtained from other systems (HIS, RIS, etc.), or it may be configured to (manually) set based on operations performed by a user (e.g., a technician) on the control unit 25.

[0024] Furthermore, the console 2 according to this embodiment also serves as a photo defect detection support device. In other words, Console 2 has a function to assist the user in deciding whether or not to consider a radiographic image as damaged (damage judgment). "Image loss" refers to marking (a flag in this embodiment) a radiographic image that has failed to be used for diagnosis when a reshoot is required due to a failed radiographic image. Details about this console 2 will be discussed later.

[0025] [1-4. Image Management Device] The image management device 4 manages the image data generated by the imaging device 1. Image management device 4 includes a Picture Archiving and Communication System (PACS), an image diagnostic workstation (IWS), etc.

[0026] <2. Console Details> Next, we will explain in detail the console 2, which has the function of a photo defect detection support device. Figure 2 is a block diagram showing the functional configuration of console 2, and Figure 3 is a flowchart showing the processing flow in console 2.

[0027] [2-1. Console Configuration] As shown in Figure 2, console 2 is composed of a control unit 21, a storage unit 22, a communication unit 23, a display unit 24, and an operation unit 25, and each of the units 21 to 25 is electrically connected by a bus or the like.

[0028] The control unit 21 is composed of a CPU (Central Processing Unit), RAM (Random Access Memory), ROM (Read Only Memory), and the like. ROM stores various programs that the CPU executes, as well as parameters necessary for program execution. The CPU then reads various programs stored in ROM, expands them into RAM, executes various processes according to the expanded programs, and centrally controls the operation of each part of Console 2. The control unit 21 functions as a determination means and a control means by executing the shooting control process described later.

[0029] The memory unit 22 is composed of non-volatile memory, a hard disk, or the like. Furthermore, the memory unit 22 is capable of storing image data of radiographic images acquired from other devices (such as the imaging device 1 and the image management device 4). For example, the memory unit 22 stores, in association with radiographic images, a flag indicating whether or not there is a photo defect, the result of the photo defect determination process that was performed, information indicating the basis for the determination, body part information, and information of the technician in charge.

[0030] Furthermore, the storage unit 22 according to this embodiment stores a plurality of trained models M. The multiple pre-trained models M correspond to each of the various types of image defect detection processes that can be executed on Console 2. They were generated by using machine learning (e.g., deep learning) with image data of radiation images and the correct image defect detection results (ground truth labels) corresponding to that image data. The trained model M then performs inference and outputs analysis results (details described later) when it receives image data of radiation images as input. Furthermore, the trained model M may be one that has been trained using not only image data and ground truth labels, but also information on the coordinates in which the subject S is located in the radiation image. Doing so can improve the accuracy of the analysis results.

[0031] Here, the type of image defect detection process (the algorithm for image defect detection) applied to the radiographic image differs depending on the area being photographed and the direction of photography. For example, if the area being photographed is a joint (knee joint, elbow joint, ankle joint, etc.), one or more image defect detection processes will be performed based on positioning (misalignment of the lateral and medial condyles), left-right reversal of the area, or incorrect area detection. Furthermore, if the image is of a limb (hand or foot), a processing step is performed to detect one or more defects, such as a left / right error in the image or an incorrect image location. Furthermore, if the imaging area is the torso (abdomen, spine, hip joint), one or more image degradation detection processes will be performed based on the following criteria: saturation due to excessive dose, body movement, or incorrect imaging location. Furthermore, if the imaging area is the chest, one or more image defect detection processes are performed, which include either lung field defects or incorrect location. The memory unit 22 stores information indicating the type of photo defects detection process to be performed on the radiation image for the area and direction of imaging (for example, the name of the photo defects detection algorithm) and a trained model M used in that photo defects detection process, associated with the imaging area and direction of imaging. In some cases, metal may be embedded in the imaging area, and in this case, separate photo defects detection processes are provided for cases with and without metal. The memory unit 22 stores information indicating the type of photo defects detection process to be performed on the radiation image that matches the conditions of the imaging area, direction of imaging, and presence or absence of metal, as well as a trained model M.

[0032] Furthermore, the memory unit 22 stores inspection order information transmitted from the RIS, etc.

[0033] The communication unit 23 consists of communication modules and the like. The communication unit 23 transmits and receives various signals and data to and from other devices (such as the imaging device 1, the generator 3, and the image management device 4) that are connected via the communication network N by wire or wireless.

[0034] The display unit 24 is composed of, for example, an LCD (Liquid Crystal Display), a CRT (Cathode Ray Tube), etc. The display unit 24 displays a radiation image or the like corresponding to the image signal received from the control unit 21.

[0035] The operation unit 25 includes a keyboard (cursor keys, number input keys, various function keys, etc.), a pointing device (mouse, etc.), and a touch panel laminated on the surface of the display unit 24. The operation unit 25 outputs control signals to the control unit 21 in response to operations performed by the user.

[0036] The console 2 may not include a display unit 24 and an operation unit 25, and may instead receive control signals from an input device provided separately from the console 2 via, for example, a communication unit 23, or output image signals to a display device (monitor) provided separately from the console 2. Furthermore, if other devices (such as the image management device 4) are equipped with a display unit and an operation unit, the device may receive control signals from the operation unit of the other device or output image signals to the display unit of the other device (the display unit and operation unit may be shared with other devices).

[0037] [2-2. Console Operation] Next, we will explain the operation of console 2 with reference to Figure 3. Console 2 executes the shooting control process shown in Figure 3. The shooting control process is executed, for example, when inspection order information is selected by the operation unit 25 from the inspection list screen displayed on the display unit 24, through the cooperation of the CPU of the control unit 21 and a program stored in the ROM.

[0038] First, the control unit 21 displays the inspection screen 241 for the selected inspection order information on the display unit 24 (step S1). The examination screen 241 (see, for example, Figure 4) includes a shooting selection button 241a that displays the details of each shooting included in the examination order information (shooting area, shooting direction, etc.), a setting area 241b for setting image reading conditions and image processing conditions for the selected shooting, an image display area 241c for displaying the captured radiographic image, a photo loss button 241d, an output button 241e, and so on. Note that at step S1, the radiographic image is not yet displayed in the image display area 241c.

[0039] When the shooting selection button 241a is pressed by the operation unit 25, the shooting to be performed (shooting area, shooting direction) is selected (step S2), and the control unit 21 sets the shooting conditions (image reading conditions, radiation irradiation conditions) for the shooting device 1 and the generator 3 (step S3). For example, the control unit 21 automatically sets the imaging conditions (image reading conditions, e.g., pixel size, image size, frame rate, etc.) on the imaging device 1 based on the imaging area, imaging direction, etc., selected by the pressed imaging selection button 241a, and also sets the imaging conditions (radiation irradiation conditions, e.g., tube voltage (kV), tube current (mA), irradiation time (ms), etc.) on the generator 3. Alternatively, the imaging conditions (image reading conditions) for the imaging to be performed may be set on the imaging device 1 in response to the user's operation on the operation unit 25 on the inspection screen 241. Furthermore, the radiation irradiation conditions may be set by the user from the control panel of the generator 3.

[0040] After pressing the shooting selection button 241a and setting the shooting conditions, the user (technician) positions the subject S between the radiation source 33 of the generator 3 and the imaging device 1. Then, when the user operates the irradiation instruction switch 32, the generator 3 irradiates the imaging area of ​​the subject S with radiation R. The imaging device 1 generates radiation images (still images, motion images) of the area being scanned at the moment it receives radiation R from the generator 3, and transmits the image data (still image data, motion image data) to the console 2.

[0041] When the image data of the radiation image is received (acquired) by the communication unit 23 (step S4), the control unit 21 previews the received radiation image in the image display area 241c of the inspection screen 241 (step S5).

[0042] Next, the control unit 21 determines the image defect determination process to be performed on the received radiation image based on the area of ​​the radiation image, the direction of the image, and / or the received radiation image (step S6). As described above, the memory unit 22 stores information indicating the type of image defect detection process to be performed on the radiation image for the area and direction of imaging, and the control unit 21 determines the image defect detection process to be performed on the received radiation image based on the area and direction of imaging. In some cases, for example, multiple types of image defect detection processes may be available for the same area and direction of imaging, depending on whether or not metal is embedded in that area. In such cases, the control unit 21 analyzes the received radiation image to determine the presence or absence of metal, and determines the image defect detection process to be performed on the radiation image based on the determination result. Any known method can be used to determine the presence or absence of metal from the radiation image. For example, metal regions can be recognized by binarization processing or advanced region extraction processing such as graph cut processing, and the presence or absence of metal can be determined based on the recognition result.

[0043] The image defect detection process to be performed on the received radiographic image may be determined when the shooting selection button 241a is pressed and the shooting area and shooting direction are set. Alternatively, the control unit 21 may not make the determination automatically, but the user may select it by operating the operation unit 25. In this case, it is necessary to prevent the user from making a selection error. Therefore, it is preferable for the control unit 21 to automatically change the selection of image defect detection processes based on the shooting area and shooting direction. For example, if the shooting area is the knee, the control unit 21 may prevent the selection of the image defect detection process for the side of the foot in the positioning-based image defect detection process. Also, for example, even if it is a radiographic image of the side of the ankle joint, if other conditions such as weight-bearing imaging that are not supported by the image defect detection processes provided in the console 2 are attached, the control unit may prevent the selection of any image defect detection process. In addition, the type of image defect detection process that does not correspond to the shooting area and shooting direction of the received radiographic image may be displayed in an identifiable manner. This makes it possible to prevent the user from making a selection error.

[0044] Next, the control unit 21 executes the determined image defect detection process (step S7). That is, the control unit 21 reads the trained model M corresponding to the image defect detection process determined in step S6 and executes the image defect detection process.

[0045] In step S7, the control unit 21 specifically first inputs the image data of the received radiation image into one of the multiple trained models M stored in the memory unit 22 that corresponds to the determined image defect determination process, and performs inference to output an analysis result corresponding to the image area. The trained model M according to this embodiment outputs numerical values ​​as analysis results, such as "the probability that subject S is photographed from the right (or left) is ○%" and "the displacement between the medial and lateral condyles is ○ mm". Furthermore, when performing inference, the trained model M performs segmentation processing to extract regions used to obtain analysis results from the input radiation image (for example, regions used to measure numerical values ​​of the analysis results, such as the region of the displacement between the medial and lateral condyles), and outputs an image of the extracted region.

[0046] Next, the control unit 21 generates information indicating the judgment result and the basis for the judgment of the image defect judgment process based on the analysis results output by the trained model M. The information indicating the basis for the judgment includes, for example, information indicating the type of image defect judgment process performed to generate the judgment result, numerical values ​​obtained by analyzing the radiographic image in the image defect judgment process and used to generate the judgment result, or superimposed display of markers for the region on the radiographic image used to generate the judgment result in the image defect judgment process.

[0047] For example, when generating a judgment result regarding left-right discrepancies in body parts, the control unit 21 compares the probability that the subject S is photographed from the right with the probability that it is photographed from the left, and determines that it is photographed from the direction with the higher probability. If the determined direction differs from the shooting direction of the received radiation image (shooting direction in the inspection order information), the text information "Caution: Left / Right" is generated as the result of the image defect detection process. In addition, text information indicating the type of image defect detection process performed (e.g., algorithm name) and text information indicating the probability of the image being taken from the right and the probability of the image being taken from the left are generated as information indicating the basis for the determination. Alternatively, the text information "Caution: Left / Right" may be generated and used as is to indicate the basis for the determination.

[0048] Furthermore, for example, when generating a judgment result regarding the misalignment between the medial and lateral condyles, the control unit 21 generates the judgment result of the image failure judgment process based on the analysis result (amount of misalignment) output in numerical form from the trained model M, in a form ranked according to the likelihood of the image being unsuccessful. Specifically, the control unit 21 refers to a pre-set judgment criterion (reference value) and assigns the generated analysis result (amount of displacement between the medial and lateral condyles) to one of several ranks. For example, the control unit 21 assigns the generated analysis result to "Rank A (Good)" if it is less than the first reference value (lowest probability of shooting failure), "Rank B (Acceptable)" if it is between the first and second reference values, and "Rank C (Reshoot)" if it is greater than or equal to the second reference value (highest probability of shooting failure). Then, it generates character information indicating the rank assigned to the amount of deviation in the analysis result as the result of the image failure judgment process. The ranking system may consist of two levels, "A / B rank" and "C rank" (with one criterion), or it may consist of four or more levels (with three or more criterion values). The control unit 21 then generates character information indicating the type of image defect detection process performed (e.g., algorithm name) and character information indicating the amount of displacement, as information indicating the basis for the determination. Furthermore, based on the image from which the displacement portions of the inner and outer condyles have been extracted, the control unit 21 superimposes colored markers onto the displacement portions of the inner and outer condyles in the radiographic image, and generates this as information indicating the basis for the determination.

[0049] When the image defect detection process is completed, the control unit 21 displays the detection result and the basis for the detection on the display unit 24 (step S8).

[0050] Figure 4 shows an example of an inspection screen 241 that displays a preview image of the radiation image acquired from the imaging device 1, as well as information indicating the judgment result and basis for the judgment of the image defect detection process. As shown in Figure 4, in step S8, the received radiation image is previewed in the image display area 241c of the inspection screen 241. Information indicating the judgment result and the basis for the judgment of the damage judgment process performed on the displayed radiation image is also displayed (indicated by symbols 241f to 241j in Figure 4). In addition, an alert 241k is displayed indicating that there is a judgment result indicating damage.

[0051] The code 241f, "Knee joint lateral view (without metal)," is the algorithm name for the positioning-related image defect detection process performed on the received radiographic image, and indicates the type of image defect detection process performed. The code 241g, "Positioning: B," is the result of the image defect detection process performed. The code 241h, "Amount of displacement: approximately 3.4 mm," is the displacement between the medial and lateral condyles, and is a numerical value (information indicating the basis for the judgment) obtained by analyzing the radiographic image and used to generate the result shown in code 241g. The code 241i, "Left / Right Caution," indicates the algorithm name and the result of the left / right error detection process performed on the received radiographic image. Note that, similar to positioning, the algorithm name for the left / right error detection process may also be a more detailed algorithm name with the imaging site added, such as "Knee joint lateral view, left / right." The display indicated by symbol 241j is a superimposed display of the region where the medial and lateral condyles of the knee joint (femur) are misaligned on the radiographic image (referred to as region display 241j), and it is information that indicates the basis for determining that the positioning rank is B.

[0052] Conventionally, while the determination of whether or not an image is damaged and the display of the determined image were performed, the basis for the determination—such as which damage detection process was used and the reason for the determination—was not shown. Therefore, there was a problem in that users were less likely to accept the determination results. In this embodiment, the basis for the determination result is output, allowing the user to confirm whether or not an appropriate damage detection process was performed, thereby improving user acceptance of the determination results. As a result, it is possible to help users quickly determine whether or not an image is damaged and whether or not reshooting is necessary. Furthermore, the judgment result is displayed during the preview stage before the shooting is completed, allowing users to decide whether or not reshooting is necessary before the shooting is finished.

[0053] The user reviews the radiographic image, the judgment result, the basis for the judgment, etc., and finally decides whether or not the image is damaged. If it is determined to be damaged (re-shooting required), the user presses the damaged image button 241d.

[0054] Here, the indication of the area used to determine the image defect (area indication 241j in Figure 4), which is shown on the radiographic image as the basis for the judgment, is cumbersome if it is always displayed, and since it is only a reference indication, it is desirable that it not interfere with image confirmation. Therefore, it is preferable to provide a switching means for switching the area display 241j ON / OFF (display / hide). For example, as shown in Figure 5, an ON / OFF button B1 for the region display 241j is provided near the judgment result and judgment basis (241f~241h) related to the region display 241j, and the control unit 21 switches the ON / OFF of the region display 241j in response to the pressing of button B1 by the operation unit 25. That is also acceptable. Alternatively, as shown in Figure 6, the display of the judgment result and judgment basis (241f~241h) related to the region display 241j may be used as the ON / OFF button for the region display 241j, and the control unit 21 may switch the ON / OFF of the region display 241j in response to the press of the display of the judgment result and judgment basis (241f~241h) by the operation unit 25. Alternatively, as shown in Figure 7, an ON / OFF button B2 for the area display 241j may be provided in the same row as the simple buttons 241l for image rotation and other operations, and the control unit 21 may switch the area display 241j ON / OFF in response to the press of button B2 by the operation unit 25. Alternatively, as shown in Figure 8, an ON / OFF button B3 for the area display 241j may be provided in the same row as the group of buttons for adjusting the advanced image processing conditions such as image density and frequency displayed in the setting area 241b, and the control unit 21 may switch the area display 241j ON / OFF in response to the press of button B3 by the operation unit 25. Alternatively, as shown in Figure 9, the radiation image displayed in the image display area 241c may be used as the ON / OFF button for the area display 241j, and the control unit 21 may switch the area display 241j ON or OFF in response to the operation unit 25 pressing the radiation image.

[0055] Furthermore, since the criteria for determining the positioning rank differ depending on the area being photographed, the facility, the shooting conditions, etc., if the area display 241j is always displayed in the same color, it becomes difficult to grasp the rank. Therefore, it is preferable for the control unit 21 to change the color of the area display 241j according to the determination result (rank) (for example, A is blue, B is yellow, C is red, etc.).

[0056] Furthermore, if the judgment result is unfavorable, there is a high possibility that a reshoot will be performed, so the user needs to recognize the result immediately. Therefore, it is preferable for the control unit 21 to display alerts in different ways depending on the rank of the judgment result. For example, if the rank of the judgment result is B, an alert A1 of a predetermined size (a size that does not hide the subject) is displayed as shown in Figure 10, and if the rank of the judgment result is C, a larger size alert A2 is displayed, for example, across the entire image as shown in Figure 10. Alternatively, if the rank of the judgment result is B, the alert may be displayed on the thumbnail image of the shooting selection button 241a (see 241k in Figure 4, etc.), and if the rank is C, the alert may be displayed on the text side of the shooting selection button 241a for better visibility, or the alert may be displayed on the radiographic image as shown in Figure 10.

[0057] Furthermore, the judgment result may be incorrect, for example, if the algorithm for the image defect detection process is selected incorrectly. In this case, it is necessary to redo (reprocess) the image defect detection process. Therefore, it is preferable that in step S8, the control unit 21 displays a reprocessing button (not shown) on the inspection screen 241, and reapplies the image defect detection process when the reprocessing button is pressed. At that time, it is also possible to automatically or in response to user operation to select and execute an image defect detection process different from the one already used.

[0058] Furthermore, if a photo loss detection process is applied to a radiographic image that does not correspond to the pre-prepared photo loss detection process in console 2, and some kind of detection result is produced, it is undesirable as an incorrect detection result will be displayed. Therefore, it is preferable that the control unit 21 does not execute the photo loss detection process if the received radiographic image is an unsupported image that does not correspond to the pre-prepared photo loss detection process, and instead displays a message such as "Analysis not possible" on the inspection screen 241 to notify that the image is unsupported.

[0059] Furthermore, if the region display 241j is always displayed, it interferes with image processing operations on the radiation image displayed in the image display region 241c (for example, adjustment operations such as image density and frequency, and zoom operations). Therefore, it is preferable that the control unit 21 automatically turns OFF (hides) the display of the region display 241j when the user starts a predetermined operation (for example, adjustment operations such as image density and frequency, and zoom operations) while the region display 241j is displayed. Furthermore, once the predetermined operation is completed, the control unit 21 may automatically switch the area display 241j to ON, or it may leave it OFF.

[0060] Furthermore, if operations are performed to change the shape, size, or position of an image, such as image rotation, flipping, enlargement, reduction, panning, or movement, the user may still view the area display 241j after the changes. Therefore, when the operation unit 25 performs an operation to change the shape, size, or position of the radiation image displayed in the image display area 241c, the control unit 21 controls the display of the area display 241j to follow the image changes. For example, as shown in Figure 11, if the radiation image is rotated, the area display 241j is rotated by the same angle to follow the rotation of the radiation image. Alternatively, the area display 241j may not be displayed while the image is being changed, and may only be displayed after the image has been changed to follow the changes. This can reduce the processing load. Also, if the processing load is small even when the area display 241j is displayed to follow the image changes while the image is being changed, such as during panning, the area display 241j may be displayed to follow the image changes even while the image is being changed.

[0061] Furthermore, if the area display 241j overlaps with the subject area, it may be difficult to see if there are foreign objects or other obstructions. Therefore, it is preferable for the control unit 21 to display the area display 241j with a transparency level above a predetermined threshold (for example, 30% or more).

[0062] Furthermore, the region display 241j may be difficult to see depending on the background density. For example, the color of the region display 241j is harder to see in high-signal areas (black) compared to low-signal areas (white). Therefore, it is preferable for the control unit 21 to change the color and transparency of the region display 241j according to the density of the background of the region display 241j. For example, if the signal value of the background of the region display 241j is in a high-signal area (the signal value is higher than a predetermined threshold), the control unit 21 will automatically lower the transparency of the region display 241j, add a border to the region display 241j, or increase the density.

[0063] Furthermore, when adding a border (frame) to the area display 241j, if the border thickness is constant, the area may become distorted and difficult to see depending on the thickness of the border when the display is enlarged or reduced. Therefore, it is preferable that the control unit 21 automatically adjusts the thickness of the border according to the enlargement or reduction ratio.

[0064] Furthermore, since the density and color that are easy to see vary from person to person, it is preferable that the density and color of the area display 241j can be changed. Therefore, the configuration may include setting means for the user to change the density, border thickness, and color of the area display 241j, and the control unit 21 may display the area display 241j with the set density, border thickness, and color. In this case, it is difficult to distinguish between multiple area displays if they are the same color. Therefore, if the user can change the color, it is preferable to control the system so that the color that has already been selected and a color similar to it are not displayed as options so that they cannot be selected.

[0065] Furthermore, if it is unclear where the measured displacement amount was taken when determining positioning, user satisfaction will be low. Therefore, it is preferable for the control unit 21 to display the measurement location of the displacement amount on the radiation image using markers ml such as arrows, as shown in Figure 12. Furthermore, if the region display 241j is small, the superimposed display of the region display 241j and marker ml may be difficult to see. Therefore, if the region display 241j is small (for example, if the area of ​​the region display 241j is below a predetermined threshold), the control unit 21 may superimpose the region display 241j and marker ml onto an enlarged image 241m, such as a pop-up, that is focused on the region where the region display 241j is displayed, as shown in Figure 12.

[0066] Returning to Figure 3, the control unit 21 determines whether or not the photo loss button 241d has been pressed by the operation unit 25 (step S9). If the control unit 25 determines that the image loss button 241d has been pressed (step S9; YES), the control unit 21 associates the image loss with a flag indicating that the image is lossy, the result and basis of the executed image loss determination process, body part information, and the information of the technician in charge, and saves this information to the storage unit 22 (step S10). By associating the image loss with the result and basis of the executed image loss determination process and the information of the technician in charge, etc., and storing this information, it can be used to educate future radiographers. After pressing the 241d button, the user readjusts the shooting conditions and positioning, and then takes the picture again. The control unit 21 returns to step S4(S3) and repeatedly performs steps S4(S3) to S8.

[0067] In this case, if there is no target image when rescanning, the user may not know how to change the positioning or shooting conditions. Therefore, the control unit 21 may use a GAN (Generative Adversarial Network, a type of machine learning method) or the like to generate a target image 241n of what it would look like if it were taken correctly, as shown in Figure 13, and display it on the inspection screen 241 as a reference when rescanning. For example, a trained model can be created by training the data patterns of an image with a positioning error and an image where the positioning has been corrected to zero using machine learning such as a GAN. The radiation image that was judged to be a failure is then input into this trained model to generate a target image 241n of what it would look like if it were taken correctly, and display it alongside the radiation image. This allows the user to rescan while referring to the target image.

[0068] In step S9, if the control unit 25 determines that the image loss button 241d has not been pressed (step S9; NO), the control unit 21 applies predetermined image processing to the radiation image and displays it as the final image in the image display area 241c (step S11). Furthermore, if the user manipulates the image processing conditions in the setting area 241b, the control unit 21 will perform image processing according to the operation.

[0069] The control unit 21 determines whether the output button 241e has been pressed by the operation unit 25. If it determines that the output button 241e has not been pressed (step S12; NO), the process returns to step S9. If the control unit 25 determines that the output button 241e has been pressed (step S12; YES), the control unit 21 associates the generated radiographic image as the final image with a flag indicating that there is no image defect, the result and basis of the executed image defect determination process, body part information, and information of the technician in charge, and saves it to the storage unit 22. In addition, the control unit 23 associates the generated radiographic image as the final image with patient information and examination information (examination ID, examination date, imaging area, imaging direction, etc.) and transmits it to the image management device 4 (step S13), and terminates the imaging control process.

[0070] As described above, the control unit 21 of console 2 executes one or more of the multiple types of image defect detection processes on the radiation image received from the imaging device 1, and generates a detection result and information indicating the basis for the detection result. The detection result and the information indicating the basis for the detection result are then output by the display unit 24. Therefore, users can verify whether or not appropriate image damage detection processing was performed, improving their satisfaction with the detection results. As a result, it is possible to help users quickly determine whether or not there is image damage and whether or not reshooting is necessary.

[0071] It goes without saying that the present invention is not limited to the embodiments described above, and can be modified as appropriate without departing from the spirit of the invention.

[0072] For example, in the above embodiment, we have described an example where the image damage detection process is performed by analyzing the radiation image using machine learning, but the image damage detection process may also be performed by analyzing the radiation image using image processing.

[0073] Furthermore, although the above embodiment described an example in which the functions of the image damage detection support device of the present invention are mounted on the console 2, the functions of the image damage detection support device may be mounted on a device other than the console 2, or on a dedicated device.

[0074] Furthermore, in the above embodiment, the control unit 21 of the console 2 was described as displaying the result of the photo defect determination and the basis for the determination on the display unit 24 as an output means, but it may also be displayed on a display device separate from the console 2. In other words, the determination means and output means of the present invention may be mounted on different devices and configured as a photo defect determination support system.

[0075] Furthermore, information indicating the results of the photo defect determination and the basis for that determination may be output not only on display but also as audio using an audio output device (not shown in the illustration).

[0076] Furthermore, the control unit 21 of console 2 may, as an output control means, control the output of the generated judgment result and information indicating the basis for the judgment to an external device (external system) in addition to the output means. Examples of external devices include an image management system 4 (PACS), a dose management system, a RIS, and a personal digital assistant (PDA). For example, the control unit 21 outputs character information and numerical information (for example, the length of the displacement and coordinate information of the displacement area) indicating the judgment result and the basis for the judgment to an external device via the communication unit 23. Alternatively, the judgment result and information indicating the basis for the judgment may be output as image information. For example, as shown in Figure 14(a), an image may be generated by superimposing the judgment result 231 and information indicating the basis for the judgment 232 and 233 onto the image used for the judgment, and output to an external device by the communication unit 23. Alternatively, as shown in Figure 14(b), an image may be generated by superimposing information indicating the basis for the judgment (information indicating the area that serves as the basis for the judgment) 233 onto the image used for the judgment, and output to an external device by the communication unit 23 along with the judgment result and text information indicating the basis for the judgment (for example, judgment result C, displacement amount 8 mm). Alternatively, as shown in Figure 14(c), a background-transparent image of the information indicating the basis for the judgment 233 may be generated and output to an external device by the communication unit 23 along with the judgment result, text information indicating the basis for the judgment (for example, judgment result C, displacement amount 8 mm), and the image used for the judgment. The file format of the output image information may be JPEG, BMP, PNG, TIFF, GIF, etc., but is not particularly limited.

[0077] Furthermore, settings may be provided to allow the output destination, whether or not to output, and the information to be output to be changed depending on the judgment result.The control unit 21 may then control the output of the judgment result and the information to be output according to the judgment result in the photo defect judgment process based on the settings. For example, the control unit 21 may be configured to output only the judgment result and the basis for the judgment of the image defect judgment process for the image that was ultimately deemed defective to the dose management system. Also, if there is an image that appears to have failed to be judged (for example, an image that is judged as equivalent to a defective image (e.g., rank B or rank C) in the image defect judgment process, but is not actually deemed defective), the control unit may, after obtaining permission from the operator, output the judgment result and the basis for the judgment to a new output destination such as customer support. In addition, the images used for analysis (used for judgment) may be output as information showing the basis for the judgment. Furthermore, if multiple images are taken in the same inspection, such as video recording or long-length recording by combining multiple images, the control unit may be configured to output only the image that served as the basis for determining the judgment result from among the images that were judged. If there are areas in the image that are unnecessary for judgment, such as when the subject is not visible, the image may be cropped and only a portion of it may be output. In addition, the program for image defect judgment and parameters such as the threshold used for judgment may also be output so that the image defect judgment process can be executed again at the output destination.

[0078] The timing for outputting the judgment result and information indicating the basis for the judgment to an external device may include, for example, the timing when the output button 241e is pressed or when the inspection end button (see Figure 4, etc.) is pressed, but is not limited to these.

[0079] Furthermore, in the above embodiment, information indicating the detection result and the basis for the detection is displayed along with the preview display, but it is also possible to display the detection result and the basis for the detection along with the final image after image processing. Note that when the image is redisplayed from the inspection list after shooting is complete, only the detection result is displayed, and the information is not superimposed on the image.

[0080] Furthermore, although the above embodiment described the case in which the present invention is applied to the detection of image defects in radiographic images, the present invention is not limited to radiographic images and may be applied to the detection of image defects in other medical images (for example, MRI images or ultrasound images).

[0081] Furthermore, in the case of follow-up imaging, for example, it is not possible to confirm the positioning (body position, etc.) or whether the body part being scanned was the left or right side during the previous scan. Therefore, there is a risk that the image taken may differ from the previous one in terms of positioning or left / right orientation. Therefore, in console 2, the control unit 21 may acquire past images of the same body part of the same patient from the image management device 4 or the like when examination order information is received, determine the left and right sides of the image using a trained model M that determines the left and right sides of the imaging area, and output an alert via display or sound if the left and right sides are reversed compared to the current examination order information. This makes it possible to prevent physician ordering errors in advance when taking follow-up images from past images. Furthermore, although it is rare, imaging may be performed in a position different from that specified in the examination order information. In some cases, even if imaging has been performed in a different position than specified in the examination order in the past, the technician may not notice. Therefore, it may be possible to determine the image defect by comparing the judgment result of the positioning defect detection process for past images with the judgment result of the positioning defect detection process for the newly acquired radiographic image. This makes it possible to reinforce the judgment of whether the positioning is good or bad (whether or not there is an image defect) based on past images.

[0082] Furthermore, while the above description discloses examples using hard disks, semiconductor non-volatile memory, etc., as computer-readable media for the program according to the present invention, the invention is not limited to these examples. Other computer-readable media include portable recording media such as CD-ROMs. Carrier waves can also be used as a medium for providing data for the program according to the present invention via a communication line. [Explanation of Symbols]

[0083] 100 radiography systems 1. Radiation imaging device 2. Console (Image Defect Detection Support Device) 21 Control Unit 22 Memory section 23 Communications Department 24 Display section 25 Control section 3. Radiation Generating Devices 31 Generator 32. Irradiation Indicator Switch 33 Radiation source 4. Image management device N Communication Network R radiation S Subject

Claims

1. A determination means that performs image defect detection processing on medical images using a pre-trained model and generates a determination result, A generation means generates a target image by inputting a medical image determined to be damaged by the aforementioned determination means into a trained model. Before reshooting is performed, an output means outputs the judgment result and the target image, A photo defect detection support device equipped with the following features.

2. The output means is a display means, The image damage determination support device according to claim 1, wherein the display means displays the medical image determined to be damaged and the target image side by side.

3. The determination means generates information indicating the basis for the determination result, The photo defect judgment support device according to claim 1, wherein the output means outputs information indicating the basis for the judgment.

4. The image defect determination support device according to claim 3, wherein the determination means performs one or more image defect determination processes from among a plurality of types of image defect determination processes on a medical image.

5. A determination means that performs image defect detection processing on medical images using a pre-trained model and generates a determination result, A generation means generates a target image by inputting a medical image determined to be damaged by the aforementioned determination means into a trained model. Before reshooting is performed, an output control means controls the output of the determination result and the target image, A photo damage detection support system equipped with the following features.

6. A judgment step in which a pre-trained model is used to perform image defect detection on medical images and a judgment result is generated, The generation process involves inputting the medical image determined to be damaged in the above determination process into a trained model to generate a target image. Before reshooting is performed, an output step is performed which outputs the judgment result and the target image, A method for supporting the detection of photo defects, comprising the following features.

7. Computers A determination means that performs image defect detection processing on medical images using a pre-trained model and generates a determination result. A generation means generates a target image by inputting the medical image determined to be damaged by the aforementioned determination means into a trained model. Before reshooting is performed, an output means outputs the judgment result and the target image. A program designed to function as such.

Images