Image processing device, image processing system, image processing method, and program

Abstract

The present invention provides an image processing device that enables the early notification of image findings of a patient to an external physician, even in a mobile health checkup vehicle or screening vehicle. [Solution] The console 20 is an image processing device mounted on a vehicle 1 having a radiography device 10, which processes radiographic images taken by the radiography device 10, and comprises an acquisition unit that acquires radiographic images of a subject M, an analysis unit 26 that performs analysis on the radiographic images of the subject M acquired by the acquisition unit using a trained model 27, and an output unit that outputs information regarding the analysis results by the analysis unit 26 to the PACS. The control unit 21 includes the functions of the acquisition unit and the output unit.

Description

Technical Field

[0001] The present disclosure relates to an image processing apparatus, an image processing system, an image processing method, and a program.

Background Art

[0002] Conventionally, in mobile health examination vehicles and screening vehicles, after the radiation imaging of all examinees is completed, the image data obtained by imaging is stored on a medium such as a DVD, and the medium is transferred to a medical facility or the like. In this case, examinees with abnormal findings are generally notified of diagnostic results or the like prompting them to undergo a re-examination several days later.

[0003] Also, as an alternative means to media transfer, for example, a system that utilizes a cloud environment and transmits image data from a health examination vehicle or the like to a PACS via a network is known. Even in this case, on the day of the health examination, the number of imaging cases is large, and generally, after all the imaging on that day is completed in terms of rotation rate, the image data and the like are collectively transmitted to the cloud. Therefore, even if an abnormal finding is discovered during the screening, the examinee cannot receive the diagnostic result unless a certain period of time has passed.

[0004] On the other hand, in recent years, guidelines for STAT image reports that require early reporting of results regarding abnormal findings from a radiographer to a doctor when an abnormal finding is discovered have been presented. In the future, a process for early reporting of image findings from a radiographer to a doctor will also be required in a health examination environment such as a health examination vehicle. As a means to support the diagnosis of a radiographer, for example, it is assumed to utilize AI technology. Patent Document 1 proposes a diagnostic method in which symptoms of parts other than those indicated can also be discovered by the judgment of AI in the imaging of the parts indicated by a doctor during the examination or health examination of animals.

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

[0006] Conventional technologies can use AI to display abnormalities in images of animals and other subjects. However, no means have been disclosed for notifying radiologists, clinicians, etc., who are outside of mobile health checkup units or similar locations, of the image findings once abnormalities are detected in the images.

[0007] Therefore, the purpose of this disclosure is to provide an image processing device, an image processing system, an image processing method, and a program that can promptly notify external physicians of the image findings of a patient, even in a mobile health checkup vehicle or screening vehicle, in order to solve the above problems. [Means for solving the problem]

[0008] The image processing device relating to this disclosure is An image processing device mounted on a vehicle equipped with a radiography device, which processes radiographic images captured by the radiography device, multiple An acquisition unit that acquires the aforementioned radiographic image of the subject, The acquisition unit acquired the multiple An analysis unit that performs analysis on the aforementioned radiographic images of the subject using machine learning, An output unit that outputs information regarding the analysis results from the aforementioned analysis unit to an external device, Equipped with 、 The analysis unit performs an analysis of the radiation image before outputting the radiation image to the outside of the vehicle. If the output unit finds an abnormal finding as a result of the analysis by the analysis unit, it outputs information regarding the analysis results of the subject in which the abnormal finding was found to the external device before outputting the radiographic images of other subjects to the outside of the vehicle. .

[0009] The image processing system related to this disclosure is An image processing device mounted on a vehicle equipped with a radiography device, which processes radiographic images taken by the radiography device, and an image management device that stores the radiographic images, Includes an information terminal that acquires the radiographic image from the image management device. An image processing system, The aforementioned image processing device is An acquisition unit that acquires the radiation image of the subject; For the radiation image of the subject acquired by the acquisition unit, Before outputting the aforementioned radiation image to the image management device, An analysis unit that performs analysis using machine learning; Information regarding the analysis result by the analysis unit The aforementioned image management device An output unit that outputs to; and Prepare, The information terminal acquires information from the image management device regarding the analysis results of subjects in whom abnormal findings were found as a result of the analysis by the analysis unit. .

[0010] The image processing method according to the present disclosure is Mounted on a vehicle having a radiation imaging device, and an image processing device that processes a radiation image taken by the radiation imaging device, multiple An acquisition step of acquiring the radiation image of the subject; For the acquired multiple radiation image of the subject, an analysis step of performing analysis using machine learning; An output step of outputting information regarding the analysis result to an external device; and has death, The analysis step involves analyzing the radiation image before outputting the radiation image to the outside of the vehicle. If an abnormal finding is found as a result of the analysis performed in the analysis step, the output step outputs information regarding the analysis results of the subject in which the abnormal finding was found to the external device before outputting the radiographic images of other subjects to the outside of the vehicle. .

[0011] The program according to the present disclosure is A computer included in an image processing device mounted on a vehicle having a radiation imaging device and processing a radiation image taken by the radiation imaging device, multiple An acquisition unit that acquires the radiation image of the subject, For the radiation image of the subject acquired by the acquisition unit, multiple An analysis unit that performs analysis using machine learning, An output unit that outputs information regarding the analysis result by the analysis unit to an external device, functioning as height, The analysis unit performs an analysis of the radiation image before outputting the radiation image to the outside of the vehicle. If the output unit finds an abnormal finding as a result of the analysis by the analysis unit, it outputs information regarding the analysis results of the subject in which the abnormal finding was found to the external device before outputting the radiographic images of other subjects to the outside of the vehicle. .

Advantages of the Invention

[0012] According to this disclosure, abnormal findings in radiographic images taken in a vehicle can be communicated to the patient and physician early on without burdening the radiographer. [Brief explanation of the drawing]

[0013] [Figure 1] This figure shows an example of a schematic configuration of the image processing system according to this embodiment. [Figure 2] This figure shows an example of a schematic configuration of a vehicle according to this embodiment. [Figure 3] This flowchart shows an example of the main console operation when taking a radiographic image of the subject's chest in the vehicle according to this embodiment. [Figure 4] This flowchart shows an example of the flow of events on the physician's side when patient information of a patient in whom abnormal findings were found during a health checkup according to this embodiment is obtained from a radiologist. [Modes for carrying out the invention]

[0014] The following describes in detail, with reference to the attached drawings, an image processing apparatus, an image processing system, an image processing method, and a program according to a preferred embodiment of the present disclosure.

[0015] [Example of a schematic configuration of the image processing system 100] Figure 1 shows an example of the schematic configuration of the image processing system 100 according to this embodiment. The image processing system 100 comprises a vehicle 1, a PACS 2, and an information terminal 3 which is an example of an external device. PACS is an abbreviation for Picture Archiving and Communication System. The vehicle 1, PACS 2, and information terminal 3 are connected to each other so as to be able to communicate with each other via a network N. Examples of the network N include the Internet, WAN, LAN, etc. The communication method of the network N may be wired communication or wireless communication. LAN is an abbreviation for Local Area Network. WAN is an abbreviation for Wide Area Network.

[0016] Vehicle 1 is a mobile health checkup vehicle, such as a mobile medical examination vehicle, capable of performing chest and stomach radiography, electrocardiogram examinations, etc. A mobile health checkup vehicle is a vehicle used for health examinations. A mobile medical examination vehicle is a vehicle that enables screening for specific diseases to be conducted outside of hospital facilities. Vehicle 1 is equipped with a radiography device 10 and a console 20, which is an example of an image processing device. The radiography device 10 takes radiographic images of a designated area of ​​the subject using radiation such as X-rays and acquires radiographic images. The console 20 analyzes the radiographic images of the subject using AI technology and transmits information Ia of the analysis results to PACS2 via the network N. Details of the radiography device 10 and console 20 will be described later.

[0017] PACS2 receives, stores, and manages information Ia regarding the patient's analysis results transmitted from vehicle 1 via network N. PACS2 is sometimes also referred to as a medical image management system.

[0018] Information terminal 3 is, for example, a tablet, a notebook personal computer, or a smartphone. Information terminal 3 is, for example, carried by a physician who is outside vehicle 1 and interpreting and diagnosing radiographic images, or is installed near the physician's waiting area. If an abnormality is found in a patient during an examination in vehicle 1, the radiologist inside vehicle 1 notifies the physician of patient information Ib, such as the patient's registration number, via information terminal 3. The physician uses information terminal 3 to access PACS2, downloads diagnostic information Ic, such as the patient's radiographic images linked to patient information Ib, and displays it on the screen. This allows the physician to quickly review and diagnose the patient's radiographic images. Information terminal 3 is pre-installed with an application that allows viewing radiographic images from PACS2.

[0019] [Example configuration of Vehicle 1] Figure 2 shows an example of the schematic configuration of vehicle 1 according to this embodiment. As described above, vehicle 1 is equipped with a radiography device 10 and a console 20, etc. The radiography device 10 and the console 20 are connected to each other so as to be able to communicate with each other by wiring 30, such as an Ethernet cable. The radiography device 10 and the console 20, etc. conform to the DICOM standard, and communication between each device is performed in accordance with the DICOM standard. DICOM is an abbreviation for Digital Image and Communications in Medicine.

[0020] The radiography apparatus 10 comprises a radiation source 11, a radiation irradiation control device 12, a radiation detection unit 13, and a reading control device 14. The radiation irradiation control device 12 and the reading control device 14 are connected via a communication cable or the like, and synchronize the radiation irradiation operation and the image reading operation by exchanging synchronization signals with each other. In Figure 2, the radiation detection unit 13 and the reading control device 14 are configured as separate units, but the radiation detection unit 13 and the reading control device 14 may be configured as an integrated unit.

[0021] The radiation source 11 is positioned opposite the radiation detection unit 13, with the subject M in between. The radiation source 11 irradiates the subject M with radiation such as X-rays according to the control of the radiation irradiation control device 12. The radiation irradiation control device 12 controls the radiation source 11 to perform radiography based on the radiation irradiation conditions input from the console 20. Examples of radiation irradiation conditions include pulse rate, pulse width, pulse interval, number of imaging frames per imaging, X-ray tube current value, X-ray tube voltage value, and additional filter type. The pulse rate is the number of radiation irradiations per second and is the same as the frame rate, which will be described later. The pulse width is the radiation irradiation time per radiation irradiation. The pulse interval is the time from the start of one radiation irradiation to the start of the next radiation irradiation and is the same as the frame interval, which will be described later.

[0022] The radiation detection unit 13 is composed of, for example, a semiconductor image sensor such as an FPD. FPD is an abbreviation for Flat Panel Detector. The FPD has a substrate made of glass or the like. Multiple detection elements, including pixels, are arranged in a matrix at predetermined positions on the substrate. The multiple detection elements detect radiation irradiated from the radiation source 11 that has passed through at least the subject M according to its intensity, and convert the detected radiation into an electrical signal and store it. Each pixel has a switching unit such as a TFT. TFT is an abbreviation for Thin Film Transistor. FPDs can be of the indirect conversion type or the direct conversion type, and either type may be used. The indirect conversion type is a method in which radiation is converted into an electrical signal by a photoelectric conversion element via a scintillator. The direct conversion type is a method in which radiation is directly converted into an electrical signal.

[0023] The reading control device 14 controls the switching unit of each pixel of the radiation detection unit 13 based on image reading conditions input from the console 20 or the like. The reading control device 14 switches the reading of the electrical signals accumulated in each pixel of the radiation detection unit 13 and acquires image data by reading the electrical signals accumulated in the radiation detection unit 13. The image data consists of each frame image of a still image or a moving image. In this embodiment, still images and moving images are referred to as radiation images. If a structure exists between the radiation source 11 and the radiation detection unit 13, the amount of radiation reaching the radiation detection unit 13 decreases due to the structure. In this case, the signal value of each pixel of the image data changes according to the structure of the subject M. The signal value includes pixel value, density value, etc. The reading control device 14 outputs the acquired radiation image data to the console 20. Examples of image reading conditions include frame rate, frame interval, pixel size, image size, etc. The frame rate is the number of frames acquired per second and is the same as the pulse rate. The frame interval is the time from the start of one frame image acquisition operation to the start of the next frame image acquisition operation, and it coincides with the pulse interval.

[0024] The console 20 is comprised of a computer, such as a personal computer or workstation. The console 20 includes a control unit 21, a storage unit 22, an operation unit 23, a display unit 24, a communication unit 25, and an analysis unit 26. The control unit 21, storage unit 22, operation unit 23, display unit 24, communication unit 25, and analysis unit 26 are connected to each other by wiring such as a bus 28.

[0025] The control unit 21 has at least one processor such as a CPU, RAM, etc. CPU is an abbreviation for Central Processing Unit. RAM is an abbreviation for Random Access Memory. The CPU 21 reads programs 22a, such as system programs and various processing programs, stored in the memory unit 22, expands them into RAM, and executes processing related to radiography of each subject M according to the expanded programs 22a.

[0026] In this embodiment, the control unit 21 functions as an acquisition unit, a determination unit, and an output unit, etc. The acquisition unit acquires radiographic images of the subject M taken by the radiography device 10. The determination unit determines whether or not to output the information Ia of the analysis results from the analysis unit 26, or the timing of the output. Specifically, the determination unit determines that if a radiologist finds an abnormality based on the analysis results, it is necessary to output the information Ia of the analysis results to an external device. The determination unit can also determine whether to output to PACS2 immediately or after a certain period of time has elapsed, depending on the severity of the lesion based on the analysis results. The output unit outputs the information Ia of the analysis results from the analysis unit 26 to an external device such as PACS2, based on the determination result of the determination unit. Examples of outputting to an external device include transmitting the information Ia of the analysis results to PACS2 via the network N, or outputting the information Ia of the analysis results to a recording device that writes data to media such as a DVD.

[0027] Regarding the timing of outputting the analysis result information Ia, if abnormal findings are found in the analysis results, the output unit will, from the perspective of speed, basically output only the analysis result information Ia for the subject M to the external device, but it is not limited to this. Specifically, if 50 examinations are scheduled to be performed on a day, for example, abnormal findings may be found in the analysis results of the 22nd examination. If the output destination is PACS2, a communication line such as network N is used, so outputting the analysis result information Ia for the patient of the 22nd examination to PACS2 in advance is not particularly inefficient. On the other hand, if the output destination is media such as DVD, depending on the remaining number and capacity of the media, it may be inefficient to output only the analysis result information Ia for one patient to the media. In such cases, it is desirable to output the analysis result information Ia for each patient from the 1st to the 22nd examination to the media all at once. Another case is when the imaging system and the health checkup reception system are not linked. In an environment where the health check registration number and the order of imaging are the same, even if information Ia regarding the analysis results for one patient is output in advance, it may not be possible to determine which examination the abnormal image belongs to. In such cases, it is easier to identify the abnormal image if information Ia regarding the analysis results for each patient, including the 22nd examination in which abnormal findings were observed, and examinations 1 through 21 prior to that examination, is output to the media all at once. Therefore, it may be possible to specify whether to output only information Ia regarding the analysis results for the patient in which abnormal findings were observed, or to output information Ia regarding the analysis results for each of the multiple patients up to the patient in which abnormal findings were observed. These specifications may be set in advance on the console 20 screen, for example, or they may be selectable when outputting information Ia regarding the analysis results.

[0028] The storage unit 22 is a non-volatile semiconductor memory, hard disk, etc. The storage unit 22 stores data such as various programs 22a executed by the control unit 21, parameters necessary for executing processing by programs 22a, and processing results. The various programs 22a are stored in the form of readable program code. Specifically, program 22a causes the console 20, which functions as a computer, to function as an acquisition unit for acquiring radiation images of the subject M, an analysis unit for performing machine learning analysis on the radiation images of the subject M, and an output unit for outputting information Ia regarding the analysis results to an external device.

[0029] The operation unit 23 includes at least one of the following: a keyboard, mouse, switch, button, and sensor. The operation unit 23 may also be a touch panel combined with the display screen of the display unit 24. The operation unit 23 receives various instructions from the user's input and outputs instruction signals corresponding to the received instructions to the control unit 21.

[0030] The display unit 24 is a display device such as a liquid crystal display or an organic EL display. EL is an abbreviation for Electro Luminescence. The display unit 24 displays a GUI, etc., for accepting various input operations from users such as radiologists, in accordance with the instructions of the display signals input from the control unit 21. GUI is an abbreviation for Graphical User Interface. The display unit 24 displays radiographic images acquired from the radiography device 10, analysis result information showing the results of the analysis of lesions in the radiographic images by the analysis unit 26, etc.

[0031] The communication unit 25 includes a LAN adapter, modem, etc. The communication unit 25 transmits and receives various data such as radiographic images and imaging conditions, as well as various signals, etc., to and from the radiography device 10, PACS 2, information terminal 3, etc., which are connected to the network N.

[0032] The analysis unit 26 has at least one processor such as a CPU or GPU, and memory such as RAM. GPU is an abbreviation for Graphics Processing Unit. The analysis unit 26 uses radiographic images acquired from the radiography device 10 to analyze and infer lesions contained in the radiographic images. Specifically, the analysis unit 26 inputs the acquired radiographic images into a trained model 27 that has been trained to output analysis results regarding lesions, and obtains analysis result information regarding lesions output from the trained model 27. The trained model 27 is stored, for example, in a memory or storage unit 22 of the analysis unit 26 (not shown). The functions of the analysis unit 26 may be provided within the control unit 21. In this case, the analysis unit 26 and the control unit 21 described above may be configured as a single processing unit.

[0033] Here, we will explain the process for creating the pre-trained model described above. A training device, including, for example, a computer, can be used to create the pre-trained model. The training device comprises one or more processors such as CPUs and GPUs, memory for storing programs, an operation unit, a display unit, and a communication unit. The training device uses pairs of radiographic images and ground truth labels as training data to generate a pre-trained model that is trained to output analysis result information about lesions in response to radiographic image input. Specifically, the training device compares the output data output by the pre-trained model to be trained with the ground truth labels in response to the training radiographic image input, and updates the parameters of the pre-trained model based on the error resulting from the comparison. For example, if the pre-trained model is implemented using a convolutional neural network, the training device may adjust the parameters of the pre-trained model according to the error between the output result and the ground truth data according to backpropagation until a predetermined completion condition is met.

[0034] [Example configuration of pre-trained model 27] As shown in Figure 2, the trained model 27 includes a first trained model 27a, a second trained model 27b, and a third trained model 27c. The first trained model 27a is a classifier trained using, for example, chest images, which are one of the diagnostic targets of health checkups, and case images that are easily found in health checkups. The second trained model 27b is a classifier trained using, for example, case images of pre-specified cases. For example, the case images may consist of a case collection held by each medical facility that undertakes health checkups. The third trained model 27c is a classifier trained using, for example, normal images. According to the third trained model 27c, by comparing the radiographic image obtained by imaging with a normal image and extracting the difference, radiographic images that are different from normal images can be screened as abnormal. In this embodiment, an example using three trained models 27 is described, but the number of trained models 27 is not limited to the number shown in Figure 2. Furthermore, the first trained model 27a, etc., may be stored in the storage unit 22 instead of the memory of the analysis unit 26.

[0035] It is known that the types of lesions discovered during health checkups differ depending on gender and age. For example, women over 40 are more susceptible to certain specific cases. Therefore, when creating a trained model, it is possible to increase the proportion of gender- and age-specific case images during training. Furthermore, if the time slots for health checkups are divided by gender and age, it is preferable to switch to the appropriate trained model 27 for each time slot and perform radiographic image analysis using AI technology. Also, the first trained model 27a, the second trained model 27b, and the third trained model 27c shown in Figure 2 may be used in series or in parallel. When the first trained model 27a, etc. are arranged in series, the analysis unit 26 uses each first trained model 27a, etc. in sequence. If the analysis unit 26 detects an abnormality in a particular trained model 27 while using each first trained model 27a, etc. in sequence, it stops using the remaining trained model 27 at that point. Furthermore, if the first trained models 27a, etc. are arranged in parallel, the analysis unit 26 may use each first trained model 27a, etc., for example once, and compare the analysis results of each trained model.

[0036] Furthermore, the radiographic images used for analysis by the analysis unit 26 may be dynamic images obtained by dynamic imaging. In this case, a trained model 27 capable of receiving multiple frame images for time-dependent analysis may be prepared separately from the first trained model 27a, etc., which targets a single frame image as described above. Before inputting the dynamic images into the trained model 27, the analysis unit 26 can select which frame images to input to the trained model 27. For example, if the dynamic image is of the chest, the radiographic image to be input to the trained model 27 may be a frame image taken during maximum respiration, where the lung area is large or the diaphragm is at its lowest point. Alternatively, in dynamic imaging, the breath-holding interval may be determined, and frame images from the breath-holding interval and frame images from outside the breath-holding interval may be selected.

[0037] In this embodiment, dynamic imaging refers to acquiring a series of images of a subject M by repeatedly irradiating the subject M with pulsed radiation, such as X-rays, at predetermined time intervals in response to a single imaging operation. Repeatedly irradiating with pulsed radiation at predetermined time intervals is called pulsed irradiation. Alternatively, dynamic imaging refers to acquiring a series of images of a subject M by continuously irradiating the subject M at a low dose rate without interruption in response to a single imaging operation. Continuously irradiating with radiation without interruption is called continuous irradiation. A series of images obtained by dynamic imaging is called a dynamic image. In addition, each of the images that make up a dynamic image may be called a frame image. Here, dynamic imaging includes video recording, but does not include still images taken while displaying video. Also, dynamic images include video, but do not include images obtained by taking still images while displaying video.

[0038] [Flow of a health checkup] Figure 3 is a flowchart showing an example of the operation of the console 20, mainly when taking a radiographic image of the chest of subject M in vehicle 1 according to this embodiment. The control unit 21, analysis unit 26, etc. of the console 20 execute the program 22a stored in the storage unit 22 to realize various processes including acquisition steps, analysis steps, and output steps.

[0039] First, subject M registers at the health check venue and obtains a registration number. Subject information Ib, including subject M's name, gender, date of birth, and registration number, is entered into the console 20 via the operation unit 23, etc. (step S10). Subject information Ib may also be transmitted from RIS, HIS, etc., as order information.

[0040] The control unit 21 performs a radiographic imaging of the subject M's chest (step S11). Specifically, the control unit 21 transmits radiation irradiation conditions corresponding to the imaging area to the radiation irradiation control device 12 and transmits image reading conditions to the reading control device 14. Radiation corresponding to the radiation irradiation conditions is irradiated onto the subject M from the radiation source 11. The radiation detection unit 13 detects the radiation that has passed through the subject M and outputs image data of the radiation image corresponding to the detected radiation to the console 20. The control unit 21 and analysis unit 26 of the console 20 acquire the image data of the subject M output from the radiation detection unit 13 of the radiographic imaging device 10.

[0041] The analysis unit 26 analyzes the acquired radiographic images using a trained model 27, which is a machine learning model (step S12). In other words, the analysis unit 26 performs the analysis process before outputting the radiographic images, etc., to the outside of the vehicle 1. Specifically, the analysis unit 26 inputs the radiographic images acquired from the radiography device 10 into the trained model 27 and obtains analysis result information, including the detection results of lesions, output from the trained model 27. The analysis result information is information to support diagnosis, and may be information that highlights lesion areas in the radiographic images by coloring or surrounding them with a frame, or it may be radiographic images with this information embedded. The analysis result information may also be information such as text or figures indicating "abnormality detected" to indicate the detection of a lesion, "no abnormality detected" to indicate the absence of a lesion, the number of detected lesions, and details of the lesions.

[0042] The control unit 21 displays the acquired radiographic image of subject M and its analysis result information on the screen of the display unit 24 (step S13). For example, if the analysis result information indicates "abnormality," the screen of the display unit 24 displays a radiographic image in which the lesion area has been highlighted, such as by coloring or surrounding it with a frame. In this case, the radiologist inside the vehicle 1 can check the radiographic image and its analysis result information displayed on the screen of the display unit 24 and make a preliminary determination as to whether it is necessary to notify the doctor immediately. If the radiologist determines that it is necessary to notify the doctor, they operate the operation unit 23, etc., to input an output instruction to output the radiographic image, etc., to PACS2. For example, the screen of the display unit 24 may be provided with an output button corresponding to the output instruction to output the radiographic image, etc., to PACS2. When the control unit 21 receives an output instruction, it adds output instruction information indicating the output instruction by the radiologist to the analysis result information. The above-described display control may also be performed by the analysis unit 26.

[0043] Furthermore, there are cases where the analysis results for a given subject M may not be available between the completion of the examination for that subject M and the start of the examination for the next subject M. In other words, there may be a time lag until the analysis results for the previous subject M are output, and the system transitions to imaging for the next subject M, displaying the imaging screen for the next subject M on the console 20's display unit 24. In this case, even if the analysis results, including any abnormalities in the previous subject M, are obtained later, the radiographer may not be able to recognize the abnormality in the previous subject M because the system has already moved on to the imaging screen for the next subject M. Therefore, even when the examination for the next subject M has begun, it may be possible to allow the radiographer to check the analysis results of examinations performed before the current examination on the console 20 during the intervals between imaging or examinations. For example, the control unit 21 may display the analysis results of the previous subject M, which may have a time lag, on the imaging screen of the console 20 each time an examination is completed or during the intervals between imaging for an examination. In this case, since the target examination performed on the previous subject M may not be the same as the examination being performed on the current subject M, it is preferable to link the target examination information to the analysis results of the previous subject M and display it on the console 20. This allows the radiographer to accurately confirm the analysis results of the examination performed on the previous subject M, which may have been delayed, before proceeding to the next examination. In addition, the doctor's diagnosis can be notified to the patient before they return home from the health checkup site.

[0044] The control unit 21 determines whether or not to transmit the analysis result information Ia to PACS2 based on at least one piece of information: output instruction information and analysis result information (step S14). In other words, the control unit 21 performs the determination process before outputting the radiographic image, etc., to the outside of vehicle 1. The analysis result information Ia includes at least one of the following: the original radiographic image of subject M, analysis result information, and subject information Ib, which includes the reception number of subject M, etc. The analysis result information Ia may also include, as analysis result information, information regarding abnormalities in the radiographic image of subject M, information regarding the presence or absence of abnormalities in the radiographic image of subject M, etc. For example, if output instruction information is input, the control unit 21 determines that it is necessary to transmit the analysis result information Ia to PACS2 and proceeds to step S15. On the other hand, if output instruction information is not input, the control unit 21 determines that it is not necessary to transmit the analysis result information Ia to PACS2 and performs radiography on other subjects M.

[0045] Furthermore, the control unit 21 may be configured to determine whether or not to output the analysis result information Ia to PACS2 based solely on the analysis result information from the analysis unit 26, without the need for a radiologic technologist's judgment. For example, if the analysis result information indicates an "abnormality," the control unit 21 may decide to output the analysis result information Ia to PACS2. Alternatively, the control unit 21 may use both the output instruction information and the analysis result information to determine whether or not to output the analysis result information Ia to PACS2. In addition, the setting for the control unit 21 to automatically output the analysis result information Ia without the need for a radiologic technologist's judgment can be switched for each facility or mobile health checkup unit. This is because there are two possible scenarios: one where it is better for the control unit 21 to automatically output the analysis result information when the analysis result is urgent, and another where it is better for a radiologic technologist to confirm the analysis result information precisely because the analysis result is urgent. In another case, the setting for automatically outputting the analysis result information Ia described above may be switched depending on the level of urgency and the congestion of examination orders. Specifically, the urgency level threshold for displaying analysis result information on the console 20's imaging screen can be varied according to the number of pending health checkup examination orders. For example, when there is a backlog of health checkup examination orders, i.e., a long waiting list, radiologists often do not have the time to continuously check the analysis result information on the console 20's imaging screen. In this case, the analysis result information Ia is automatically sent to the PACS2 in advance, where the PACS2 makes its own judgment. Furthermore, in another case, the setting for automatically outputting the analysis result information Ia described above may be switched depending on the accuracy of the judgment algorithm of the AI-trained model 27. For example, if the version of the trained model 27 is old, the analysis accuracy is low, and judgment by a radiologist may be required. On the other hand, if the version of the trained model 27 is the latest, the analysis accuracy is high, and judgment by a radiologist may not be necessary.

[0046] Furthermore, in addition to the output determination described above, the control unit 21 may select at least one of the following output targets, output methods, and output destinations depending on the analysis result information, etc. For example, if the analysis result from the analysis unit 26 is "abnormal", the control unit 21 may control the transmission of the subject M's radiographic images, etc. to PACS2 until additional radiography is performed. Also, if the analysis result from the analysis unit 26 concerns findings related to pneumoconiosis, the control unit 21 may control the transmission of two types of images to PACS2: the original radiographic image and the image processed for pneumoconiosis. Also, if the radiographic image is a dynamic image and the analysis result is "abnormal", the control unit 21 may control the transmission of only one representative frame image from the dynamic image to PACS2 in order to convey the urgency. Furthermore, the control unit 21 may determine whether or not to transmit the analysis result information Ia to PACS2 based on the urgency of the analysis result from the analysis unit 26, the severity of the lesion, etc. In this case, settings such as the correspondence between the urgency of the analysis results, the severity of the lesion, etc., and whether or not transmission is necessary can be made on the screen of the display unit 24.

[0047] Furthermore, the control unit 21 may output information Ia regarding the analysis results to PACS2 according to the analysis result information, etc., based on the output target, output timing, and output destination shown in (1) to (3) below. (1) In cases where the analysis results are urgent and require immediate rescanning of the patient, the analysis results will first be displayed on the console 20's imaging screen, and then information Ia regarding the analysis results for subject M will be output to PACS2. (2) If the analysis results show abnormalities but are not urgent, additional tests will be necessary. However, to avoid unnecessarily causing anxiety to the subject M, the display on the console 20's imaging screen will be kept to a minimum, while information Ia regarding the analysis results for subject M will be prioritized and output to PACS2. (3) If no abnormal findings are found in the analysis results, the information Ia regarding the analysis results for each subject M is stored in a primary storage device such as RAM, and for example, when all examinations for the day are completed, the information Ia regarding the analysis results for each subject M is moved to the media. On the other hand, if abnormal findings are found in the analysis results, the information Ia regarding the analysis results for all subjects M up to that point, or only the information Ia regarding the analysis results for the subject M in which abnormal findings were found, is moved to the media at the time the abnormality is confirmed.

[0048] The control unit 21 transmits information Ia regarding the analysis results of subject M, in which abnormal findings were found, to PACS2 via the network N (step S15). In this case, the control unit 21 may adjust the timing of transmitting information Ia regarding the analysis results to PACS2 based on the urgency of the analysis results by the analysis unit 26, the severity of the lesion, etc. For example, if the severity level of the lesion is high, the control unit 21 immediately transmits information Ia regarding the analysis results to PACS2. When PACS2 receives information Ia regarding the analysis results transmitted from the console 20, it saves and manages the received information Ia regarding the analysis results. For example, PACS2 stores the reception number of subject M, the radiographic image of subject M, and the analysis result information of this radiographic image linked together. The information output and stored in PACS2 may also include information indicating whether or not the analysis results by the analysis unit 26 were explained to the patient, subject M. In this case, the information indicating whether or not the analysis results were explained to the patient is stored in PACS2 linked to the radiographic image and analysis result information mentioned above.

[0049] The radiologic technologist notifies the physician's information terminal 3 of the patient information Ib of the patient M in whom abnormal findings were found (step S16). For example, the radiologic technologist contacts the physician's information terminal 3 using functions such as calling, emailing, or SNS on an information terminal such as a mobile phone or smartphone. At this time, the radiologic technologist provides the reception number of the patient M in whom abnormal findings were found. The patient information Ib of the patient M in whom abnormal findings were found may also be transmitted from the console 20 to the physician's information terminal 3 via the network N. The timing of the transmission of the patient information Ib may be approximately the same as the timing of the transmission of the analysis result information Ia to the PACS2 as described above, or it may be immediately before or after that. In other words, it is preferable to notify the physician of the patient information Ib of the patient M in whom abnormal findings were found before the patient M heads to the next examination location or before the diagnosis of the patient M is completed and the patient leaves the health check venue. The patient M in whom abnormal findings were found waits, for example, in the vehicle 1 or in a room prepared at the health check venue.

[0050] Figure 4 is a flowchart illustrating an example of the flow on the physician's side when the physician obtains patient information Ib from a radiologic technologist for a patient M in whom abnormal findings were found during a health checkup according to this embodiment. The physician obtains patient information Ib, such as the reception number, for patient M in whom abnormal findings were found in the radiographic image, upon contact from the radiologic technologist (step S20). For example, the physician can use functions such as telephone, email, or SNS on the physician's information terminal 3 as a means of communication with the radiologic technologist. The physician may be at the health checkup venue or at a different medical facility.

[0051] The physician uses the acquired patient information Ib for patient M to access PACS2 from information terminal 3 and downloads diagnostic information Ic, which is linked to patient information Ib for patient M, from PACS2 (step S21). Diagnostic information Ic includes the radiographic images and analysis results for patient M in which abnormal findings were found. The radiographic images and analysis results for patient M downloaded from PACS2 are displayed on the screen of the display unit of information terminal 3.

[0052] The physician checks the radiographic images of subject M displayed on the information terminal 3 and makes a diagnosis of subject M. If the diagnosis indicates that a re-examination is necessary, the physician informs subject M that a detailed examination is required (step S22). If subject M is in vehicle 1, the physician uses information terminal 3 to notify the radiologist in vehicle 1 of the diagnosis of subject M (step S22). The radiologist receives the diagnosis from the physician and, if the diagnosis indicates that a re-examination is necessary, informs subject M that a detailed examination is required. Subject M, having received notification that a re-examination is necessary, may download their radiographic images showing abnormal findings from the cloud or obtain them on media such as a DVD so that they can undergo a re-examination at a medical facility.

[0053] This embodiment provides the following advantages and benefits. In conventional environments, such as those found in mobile health checkup vehicles, media such as DVDs containing the radiographic images of all patients M are transported to the medical facility after all radiographic imaging for the day has been completed. However, when sending all of the patients M's radiographic images for the day to the medical facility at once, there is a problem that even if abnormal findings are discovered, there will be a delay in contacting the patients M for further examination. Therefore, it is conceivable to send the radiographic images to the medical facility using the cloud or the like after each radiographic imaging, but this would lead to problems such as a decrease in the turnaround time of radiographic imaging work and an increase in the communication load on the image processing system.

[0054] In contrast, according to this embodiment, even in the environment of a vehicle 1 such as a mobile health checkup unit, a radiologic technologist can make a preliminary judgment based on the results of the analysis of radiologic images by the analysis unit 26 before sending a day's worth of radiologic images from the vehicle 1 to a medical facility. This allows information Ia regarding the analysis results of patients M to be sent to PACS2 in advance, limited to patients M who require early diagnosis and examination. In other words, output processing can be limited to those who need output by triage using AI technology. Therefore, it is possible to prevent a significant interruption in the radiologic technologist's radiography work, and to quickly notify the doctor of patient information Ib of patients M in whom abnormal findings have been found. Furthermore, for patients M in whom abnormal findings have been found, the doctor's diagnosis can be communicated early, for example, before they go home, so that they can receive appropriate examinations and treatment before their symptoms worsen. In addition, according to this embodiment, the communication load on the image processing system 100 can be reduced by limiting the output targets to PACS2.

[0055] Although preferred embodiments of this disclosure have been described in detail above with reference to the attached drawings, the technical scope of this disclosure is not limited to these examples. Furthermore, various modifications and improvements naturally fall within the technical scope of this disclosure, within the scope of the technical ideas described in the claims for those skilled in the art. [Explanation of Symbols]

[0056] 1 vehicle 2 PACS 3. Information terminal (external device) 10. Radiography equipment 20 Console (Image Processing Unit) 21 Control Unit (Acquisition Unit, Output Unit) 26 Analysis Department 100 Image Processing Systems

Claims

1. An image processing device mounted on a vehicle equipped with a radiography device, which processes radiographic images captured by the radiography device, An acquisition unit that acquires the aforementioned radiographic images of multiple subjects, An analysis unit performs analysis using machine learning on the radiographic images of the multiple subjects acquired by the acquisition unit, An output unit that outputs information regarding the analysis results from the aforementioned analysis unit to an external device, Equipped with, The analysis unit performs an analysis of the radiation image before outputting the radiation image to the outside of the vehicle. The output unit is an image processing device that, if an abnormal finding is found as a result of the analysis by the analysis unit, outputs information regarding the analysis results of the subject in which the abnormal finding was found to the external device before outputting the radiographic images of other subjects to the outside of the vehicle.

2. The system includes a determination unit that determines whether or not to output the information or the timing of outputting the information, based on the information obtained from the analysis unit. The determination unit performs a determination process before outputting the radiation image to the outside of the vehicle. The image processing apparatus according to claim 1.

3. The output unit determines at least one of the output target, output method, and output destination according to the analysis results of the analysis unit. The image processing apparatus according to claim 1.

4. The information includes at least one of the following: the radiographic image, information regarding abnormalities in the radiographic image, information regarding the presence or absence of abnormalities in the radiographic image, and the subject's information. The image processing apparatus according to claim 1.

5. An image processing system comprising: an image processing device mounted on a vehicle equipped with a radiography device, which processes radiographic images captured by the radiography device; an image management device for storing the radiographic images; and an information terminal for acquiring the radiographic images from the image management device, The aforementioned image processing device is An acquisition unit that acquires the aforementioned radiographic image of the subject, An analysis unit performs analysis using machine learning on the radiographic image of the subject acquired by the acquisition unit before outputting the radiographic image to the image management device. An output unit that outputs information regarding the analysis results from the analysis unit to the image management device, Equipped with, The aforementioned information terminal is an image processing system that acquires information from the image management device regarding the analysis results of subjects in whom abnormal findings have been found as a result of the analysis by the analysis unit.

6. An image processing device, mounted on a vehicle equipped with a radiography device, processes radiographic images captured by the radiography device, An acquisition step of acquiring the aforementioned radiographic images of multiple subjects, An analysis step in which machine learning is used to analyze the acquired radiographic images of the multiple subjects, An output step that outputs information regarding the analysis results to an external device, It has, The analysis step involves analyzing the radiation image before outputting the radiation image to the outside of the vehicle. The output step is an image processing method in which, if an abnormal finding is found as a result of the analysis performed in the analysis step, information regarding the analysis results of the subject in which the abnormal finding was found is output to the external device before outputting the radiographic images of other subjects to the outside of the vehicle.

7. A computer included in an image processing device that is mounted on a vehicle equipped with a radiography device and processes radiographic images taken by the radiography device, Acquisition unit that acquires the aforementioned radiographic images of multiple subjects, An analysis unit performs analysis using machine learning on the radiographic images of the multiple subjects acquired by the acquisition unit. An output unit that outputs information regarding the analysis results from the aforementioned analysis unit to an external device. To make it function as, The analysis unit performs an analysis of the radiation image before outputting the radiation image to the outside of the vehicle. The output unit is a program that, if an abnormal finding is found as a result of the analysis by the analysis unit, outputs information regarding the analysis results of the subject in which the abnormal finding was found to the external device before outputting the radiographic images of other subjects to the outside of the vehicle.

Images