Dynamic image analysis device and program

Abstract

This method accurately calculates characteristic features related to pulmonary blood flow, excluding major vessels, from dynamic images of the chest. [Solution] The control unit 31 of the diagnostic console 3 acquires dynamic images of the chest obtained by dynamic radiography, extracts the lung field region from the acquired dynamic images, and calculates feature quantities related to blood flow (inter-frame difference values) from the extracted lung field region. Then, it determines an upper limit for the calculated blood flow feature quantity value and imposes a restriction based on the upper limit for the blood flow feature quantity value.

Description

Technical Field

[0001] The present invention relates to a dynamic image analysis device and a program.

Background Art

[0002] Conventionally, techniques for analyzing pulmonary blood flow based on dynamic images of the chest have been proposed. For example, in Patent Document 1, after performing warping processing (alignment processing) on a plurality of frame images of a dynamic image of the chest to match the shapes (including blood vessels) of the lung field regions of the plurality of frame images, the blood vessel region is extracted, and the blood flow feature amount is calculated by calculating the amount of density change between the plurality of frame images from the blood vessel region.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, since blood vessels move three-dimensionally, it is difficult to accurately perform alignment in a two-dimensional dynamic image of the chest, and it is difficult to accurately calculate the blood flow feature amount.

[0005] In general, pulmonary blood flow scintigraphy (pulmonary blood flow scintigraphy) is known as a modality for evaluating the left-right ratio of pulmonary blood flow (the blood flow ratio of the left and right lungs). In pulmonary blood flow scintigraphy, the blood flow ratio of the peripheral blood vessels of the left and right lungs is evaluated. When performing an evaluation equivalent to that of pulmonary blood flow scintigraphy in a dynamic image of the chest, for example, a method of calculating the left-right ratio of pulmonary blood flow based on the feature amount related to the pulmonary blood flow in a region excluding the region of a specific blood vessel (main blood vessel) part is used. However, since the peripheral blood vessels also overlap behind the main blood vessel, excluding the region of the main blood vessel part from the evaluation region makes it impossible to perform an accurate evaluation.

[0006] The objective of this invention is to accurately calculate characteristic quantities related to pulmonary blood flow, excluding the main blood vessels, from dynamic images of the chest. [Means for solving the problem]

[0007] To solve the above problems, the motion image analysis device of the present invention is An acquisition unit that acquires dynamic images of the chest obtained by dynamic imaging using radiation, An extraction unit for extracting lung field regions from the aforementioned dynamic image, A blood flow feature calculation unit calculates feature quantities related to blood flow from the lung field region, A limiting unit that imposes restrictions on the value of the calculated characteristic quantity related to blood flow, It is equipped with.

[0008] Furthermore, the program of the present invention, Computers Acquisition unit that acquires dynamic images of the chest obtained by dynamic imaging using radiation, An extraction unit for extracting the lung field region from the aforementioned dynamic image, A blood flow feature calculation unit calculates feature quantities related to blood flow from the lung field region. A limiting unit that imposes restrictions on the value of the calculated characteristic quantity related to blood flow, To make it function as such. [Effects of the Invention]

[0009] According to the present invention, it is possible to accurately calculate characteristic quantities related to pulmonary blood flow, excluding the main blood vessels, from dynamic images of the chest. [Brief explanation of the drawing]

[0010] [Figure 1] This figure shows the overall configuration of the dynamic image analysis system in an embodiment of the present invention. [Figure 2] This flowchart shows the shooting control process performed by the control unit of the shooting console shown in Figure 1. [Figure 3]It is a flowchart showing the pulmonary blood flow analysis process executed by the control unit of the diagnostic console in FIG. 1. [Figure 4] It is a diagram for explaining the prerequisite conditions of the simulation of the signal change accompanying the cardiac ejection in the dynamic image of the chest. [Figure 5] It is a diagram for explaining an example of setting a lower upper limit value as the distance from the center of the lung field increases. [Figure 6] It is a diagram showing the brachiocephalic artery and the subclavian artery in the radiographic image of the chest. [Figure 7] It is a diagram for explaining an example of a method for generating a lung field mask for blood flow ratio using deep learning. [Figure 8] It is a diagram for explaining an example of a method for generating a lung field mask for blood flow ratio using deep learning. [Figure 9] It is a diagram for explaining an example of a method for generating a lung field mask for blood flow ratio using deep learning. [Figure 10] It is a diagram for explaining an example of a method for generating a lung field mask for blood flow ratio using deep learning. [Figure 11] It is a diagram for explaining an example of a method for generating a lung field mask for blood flow ratio using deep learning. [Figure 12] It is a diagram for explaining the setting of the analysis region based on the density information of the lung field region. [Figure 13] It is a diagram for explaining the setting of the analysis region based on the inter-frame difference value of the lung field region. [Figure 14] It is a diagram for explaining a method for calculating the volume of the lung field region overlapping with other organs using CT images. [Figure 15] It is a diagram for explaining the region used for calculating the organ background correction coefficient. [Figure 16] It is a diagram showing an example of an analysis result screen displaying the left-right ratio of pulmonary blood flow. [Figure 17] It is a diagram showing an example of displaying only the lung field mask for blood flow ratio without distinguishing the region of the organ background. [Figure 18]This figure shows how the analysis results screen changes when the upper limit is changed by user input. [Figure 19] This figure shows the changes in the analysis results screen when the generation of lung masks for blood flow ratios is changed from automatic to manual by user input. [Figure 20] This figure shows how the analysis results screen changes when the application of the organ correction coefficient is changed from enabled to disabled by the user. [Figure 21] This figure shows an example of an analysis results screen that includes the assumed error in the left-right ratio of pulmonary blood flow. [Modes for carrying out the invention]

[0011] Embodiments of the present invention will be described in detail below with reference to the drawings. However, the scope of the invention is not limited to the illustrated examples.

[0012] [Configuration of the motion image analysis system 100] First, the configuration of this embodiment will be described. Figure 1 shows an example of the overall configuration of the dynamic image analysis system 100 in this embodiment. As shown in Figure 1, the dynamic image analysis system 100 is configured such that the imaging device 1 and the imaging console 2 are connected by a communication cable, and the imaging console 2 and the diagnostic console 3 are connected via a communication network NT such as a LAN (Local Area Network). Each device constituting the dynamic image analysis system 100 conforms to the DICOM (Digital Image and Communications in Medicine) standard, and communication between each device is performed in accordance with DICOM.

[0013] [Configuration of imaging device 1] The imaging device 1 is an imaging means for capturing periodic (cycle-like) movements of the chest, such as the morphological changes of lung expansion and contraction associated with respiratory movement, and the beating of the heart. Motion imaging refers to acquiring multiple images that show the movement of a subject by repeatedly irradiating it with pulsed radiation such as X-rays at predetermined time intervals (pulsed irradiation), or by continuously irradiating it at a low dose rate without interruption (continuous irradiation). A series of images obtained by motion imaging is called a motion image. In addition, each of the multiple images that make up a motion image is called a frame image. Here, "dynamic photography" includes video recording, but does not include taking still images while displaying video. Similarly, "dynamic images" include video, but does not include images obtained by taking still images while displaying video.

[0014] As shown in Figure 1, the imaging device 1 is configured to include a radiation source 11, a radiation irradiation control device 12, a radiation detection unit 13, a reading control device 14, and the like.

[0015] The radiation source 11 is positioned opposite the radiation detection unit 13 with the subject M in between, and irradiates the subject M with radiation (X-rays) according to the control of the radiation irradiation control device 12. The radiation irradiation control device 12 is connected to the imaging console 2 and controls the radiation source 11 to perform radiography based on the radiation irradiation conditions input from the imaging console 2. The radiation irradiation conditions input from the imaging console 2 include, for example, the pulse rate, pulse width, pulse interval during continuous irradiation, the number of imaging frames per exposure, the value of the X-ray tube current, the value of the X-ray tube voltage, and the 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 in continuous imaging and is the same as the frame interval, which will be described later.

[0016] The radiation detection unit 13 is composed of a semiconductor image sensor such as an FPD. The FPD has, for example, a glass substrate, and multiple detection elements (pixels) are arranged in a matrix at predetermined positions on the substrate. These 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 for storage. Each pixel is configured with a switching unit such as a TFT (Thin Film Transistor). FPDs can be of the indirect conversion type, which converts X-rays into electrical signals via a scintillator using a photoelectric conversion element, or the direct conversion type, which directly converts X-rays into electrical signals. Either type may be used. The radiation detection unit 13 is positioned to face the radiation source 11 with the subject M in between.

[0017] The reading control device 14 is connected to the imaging console 2. Based on the image reading conditions input from the imaging console 2, the reading control device 14 controls the switching unit of each pixel of the radiation detection unit 13, switching the reading of the electrical signals accumulated in each pixel, and acquires image data by reading the electrical signals accumulated in the radiation detection unit 13. This image data is a frame image. The reading control device 14 then outputs the acquired frame image to the imaging console 2. The image reading conditions are, for example, the frame rate, frame interval, pixel size, image size (matrix size), etc. The frame rate is the number of frame images acquired per second and is the same as the pulse rate. The frame interval is the time from the start of the acquisition operation of one frame image to the start of the acquisition operation of the next frame image and is the same as the pulse interval.

[0018] Here, the radiation irradiation control device 12 and the reading control device 14 are connected to each other and exchange synchronization signals to synchronize the radiation irradiation operation and the image reading operation.

[0019] [Configuration of shooting console 2] The imaging console 2 outputs radiation irradiation conditions and image reading conditions to the imaging device 1 to control the radiography and radiographic image reading operations of the imaging device 1, and also displays the dynamic images acquired by the imaging device 1 for the radiographer to confirm positioning and whether the images are suitable for diagnosis. As shown in Figure 1, the shooting console 2 is configured to include a control unit 21, a storage unit 22, an operation unit 23, a display unit 24, and a communication unit 25, and each unit is connected by a bus 26.

[0020] The control unit 21 is composed of a CPU (Central Processing Unit), RAM (Random Access Memory), etc. The CPU of the control unit 21 reads system programs and various processing programs stored in the memory unit 22 in response to operations on the operation unit 23, expands them into RAM, and executes various processes, including the imaging control process described later, according to the expanded programs, thereby centrally controlling the operation of each part of the imaging console 2, as well as the radiation irradiation and reading operations of the imaging device 1.

[0021] The storage unit 22 is composed of non-volatile semiconductor memory, a hard disk, or the like. The storage unit 22 stores data such as various programs executed by the control unit 21, parameters necessary for processing by the programs, or processing results. For example, the storage unit 22 stores a program for executing the imaging control processing shown in Figure 2. The storage unit 22 also stores radiation irradiation conditions and image reading conditions associated with the area to be examined. Various programs are stored in the form of readable program code, and the control unit 21 sequentially executes operations according to the program code.

[0022] The operation unit 23 is configured with a keyboard equipped with cursor keys, number input keys, and various function keys, and a pointing device such as a mouse, and outputs instruction signals input by key operations on the keyboard or mouse to the control unit 21. The operation unit 23 may also be equipped with a touch panel on the display screen of the display unit 24, in which case it outputs instruction signals input via the touch panel to the control unit 21.

[0023] The display unit 24 is composed of monitors such as LCDs (Liquid Crystal Displays) and CRTs (Cathode Ray Tubes), and displays input instructions and data from the operation unit 23 according to the instructions of the display signals input from the control unit 21.

[0024] The communication unit 25 is equipped with a LAN adapter, modem, TA (Terminal Adapter), etc., and controls the transmission and reception of data between it and each device connected to the communication network NT.

[0025] [Configuration of diagnostic console 3] The diagnostic console 3 is a dynamic image analysis device that acquires dynamic images from the imaging console 2, performs analysis on the acquired dynamic images, and displays the dynamic images and analysis results for the physician's interpretation. As shown in Figure 1, the diagnostic console 3 is comprised of a control unit 31, a storage unit 32, an operation unit 33, a display unit 34, and a communication unit 35, with each unit connected by a bus 36.

[0026] The control unit 31 is composed of a CPU, RAM, etc. The CPU of the control unit 31 reads system programs and various processing programs stored in the memory unit 32 in response to operations on the operation unit 33, expands them into RAM, and executes various processes, including the pulmonary blood flow analysis process described later, according to the expanded programs, and centrally controls the operation of each part of the diagnostic console 3. The control unit 31 functions as an acquisition unit, extraction unit, blood flow feature calculation unit, limiting unit, analysis area setting unit, and analysis unit of the present invention.

[0027] The storage unit 32 is composed of non-volatile semiconductor memory, a hard disk, or the like. The storage unit 32 stores various programs, including the program for executing pulmonary blood flow analysis processing in the control unit 31, as well as data such as parameters necessary for executing the processing by the programs, and processing results. These various programs are stored in the form of readable program code, and the control unit 31 sequentially executes operations according to the program code.

[0028] The operation unit 33 is configured with a keyboard equipped with cursor keys, number input keys, and various function keys, and a pointing device such as a mouse, and outputs instruction signals input by key operations on the keyboard or mouse operations to the control unit 31. The operation unit 33 may also be equipped with a touch panel on the display screen of the display unit 34, in which case it outputs instruction signals input via the touch panel to the control unit 31.

[0029] The display unit 34 is composed of a monitor such as an LCD or CRT, and displays input instructions and data from the operation unit 33 according to the instructions of the display signal input from the control unit 31.

[0030] The communication unit 35 is equipped with a LAN adapter, modem, TA, etc., and controls the transmission and reception of data between it and each device connected to the communication network NT.

[0031] [Operation of the motion image analysis system 100] Next, the operation of the dynamic image analysis system 100 described above will be explained.

[0032] (Operation of imaging device 1 and imaging console 2) First, we will explain the shooting operation using the shooting device 1 and the shooting console 2. Figure 2 shows the shooting control process performed in the control unit 21 of the shooting console 2. The shooting control process is performed in cooperation with the control unit 21 and the program stored in the storage unit 22.

[0033] First, the radiographer operates the control panel 23 of the imaging console 2 to input patient information (patient's name, height, weight, age, gender, etc.) of the subject (subject M) (step S1).

[0034] Next, the radiation irradiation conditions are read from the storage unit 22 and set in the radiation irradiation control device 12, and the image reading conditions are read from the storage unit 22 and set in the reading control device 14 (step S2). Here, the frame rate (pulse rate) is preferably 7.5 frames / second or more, taking into account the human heart rate cycle. Also, the number of frames to be captured is preferably one heart rate cycle or more.

[0035] Next, the control unit 23 is set up to receive an instruction for radiation irradiation (step S3). At this point, the person performing the imaging, such as the radiographer, positions the subject (subject M) for frontal chest imaging and instructs the subject to hold their breath. Once the imaging preparations are complete, the control unit 23 is operated to input the instruction for radiation irradiation.

[0036] When a radiation irradiation instruction is input via the operation unit 23 (step S3; YES), a start-up instruction is output to the radiation irradiation control device 12 and the reading control device 14, and dynamic imaging begins (step S4). That is, radiation is irradiated by the radiation source 11 at pulse intervals set in the radiation irradiation control device 12, and frame images are acquired by the radiation detection unit 13. When the acquisition of a predetermined number of frames is completed, the control unit 21 outputs an instruction to the radiation irradiation control device 12 and the reading control device 14 to end the acquisition, and the acquisition operation is stopped. The number of frames acquired is the number that can capture at least one heartbeat cycle.

[0037] The frame images acquired through imaging are sequentially input to the imaging console 2, stored in the memory unit 22 in association with a number indicating the imaging order (step S5), and displayed on the display unit 24 (step S6). The imaging technician checks the positioning, etc., from the displayed dynamic images and determines whether an image suitable for diagnosis has been acquired through imaging (imaging OK) or whether re-imaging is necessary (imaging NG). Then, the technician operates the operation unit 23 to input the result of the determination.

[0038] When a judgment result indicating OK for shooting is input through a predetermined operation of the operation unit 23 (step S7; YES), information such as an identification ID for identifying the dynamic image, patient information, examination target area, radiation irradiation conditions, image reading conditions, and a number indicating the shooting order (frame number) is attached to each of the series of frame images acquired by dynamic imaging (for example, written in the header area of ​​the image data in DICOM format), and transmitted to the diagnostic console 3 via the communication unit 25 (step S8). Then, this process ends. On the other hand, when a judgment result indicating NG for shooting is input through a predetermined operation of the operation unit 23 (step S7; NO), the series of frame images stored in the storage unit 22 are deleted (step S9), and this process ends.

[0039] (Operation of diagnostic console 3) Next, we will explain the operation of the diagnostic console 3. In the diagnostic console 3, when a series of dynamic images of the chest are received from the imaging console 2 via the communication unit 35, the control unit 31 and the program stored in the memory unit 32 work together to execute the pulmonary blood flow analysis process shown in Figure 3. In the pulmonary blood flow analysis process, the left-right ratio of pulmonary blood flow is calculated.

[0040] The following explanation will describe the flow of pulmonary blood flow analysis, referring to Figure 3. In the following explanation, it will be assumed that the dynamic images of the chest are images taken from the front of the chest.

[0041] First, a series of frame images of the chest in motion are acquired (step S11). Next, a pulmonary blood flow summary image is generated based on the acquired dynamic images (step S12).

[0042] A pulmonary blood flow summary image is an image that shows characteristic quantities related to blood flow in each block region (small area) of the lung field in a dynamic image of the chest. A pulmonary blood flow summary image is generated, for example, by the following processes (1) to (8).

[0043] (1) First, logarithmic transformation is applied to each frame image of the acquired motion image. While logarithmic transformation is desirable, it can be omitted.

[0044] (2) Extract the lung field region from each frame image after logarithmic transformation. Any method can be used to extract the lung region. For example, a threshold can be determined by discriminant analysis from the histogram of the pixel values ​​(intensity values) of each pixel in the frame image, and regions with higher signals than this threshold can be extracted as candidate lung regions. Next, edge detection can be performed near the boundaries of the extracted candidate lung regions, and the points where the edge is maximized in the small blocks near the boundaries can be extracted along the boundaries to extract the boundaries of the lung regions. Alternatively, the frame image with the outlines of the lung field regions automatically extracted using the above method may be displayed on the display unit 34, allowing the user to manually adjust the lung field regions using the operation unit 33.

[0045] (3) Blocking is performed on the extracted lung field region of each frame image. In the blocking process, the lung field area is divided into rectangular block regions, for example, 10mm x 10mm in size. The signal value (density value) of each pixel within the block region is replaced with a representative value (average value, etc.) of the signal value within the block region, and then smoothed. Alternatively, a 10mm x 10mm smoothing process is performed on each pixel, and each pixel is set as a block region.

[0046] (4) Set an ROI (Region of Interest) within the cardiac region and determine the parameters of a bandpass filter to remove noise at frequencies other than the blood flow signal. First, an ROI is set in the cardiac region of each frame image. The cardiac region can be extracted using known methods, such as template matching or deep learning techniques. Next, a waveform showing the time evolution of the average signal value of the ROI (a waveform representing the heart rate cycle) is generated, and the parameters of the bandpass filter are determined based on the frequency of the generated waveform.

[0047] (5) For each block region of the lung area, the waveform of the time change in signal value is acquired, and a bandpass filter is applied in the time direction using the parameters determined in (4). This eliminates the influence of noise at frequencies other than blood flow signals.

[0048] (6) A reference frame image is set for each block region after bandpass filtering. First, an initial reference frame image is selected based on a waveform that shows the time evolution of the average signal value of the ROI. For example, the frame image with the minimum average signal value of the ROI is set as the initial reference frame image. Here, the frame image with the smallest average signal value (average density value) of the ROI means that the blood volume in the region designated as the ROI is maximum. When blood flows into an organ, the transmission of radiation is obstructed by the blood, so the amount of radiation transmitted in the radiographic image decreases, the signal value becomes smaller, and the image appears whitish (i.e., low density). When viewed in relation to cardiac blood flow flowing into and out of the heart and pulmonary blood flow flowing into and out of the lungs, when a lot of blood is flowing into the heart, the transmission of radiation is obstructed in the cardiac area, so the signal value on the radiographic image is small and it appears relatively whitish (low density). In contrast, at this timing, there is little blood flowing into the lungs, so the amount of radiation transmitted in the lungs increases, the signal value on the radiographic image becomes larger and it appears relatively dark (high density). In other words, the frame image with the least amount of blood flowing into the lungs is set as the initial reference frame image. Next, a reference frame image is set for each block region. For example, the frame image showing the highest signal value within a predetermined search range (for example, ±0.1333 seconds (2 frames before and after) in the case of 15fps) from the initial reference frame image is set as the reference frame image for that block region.

[0049] (7) For each frame image, the difference in signal value (inter-frame difference) is calculated for each block region of the extracted lung area compared to the corresponding block region (block region in the same position) in the reference frame image. The inter-frame difference represents the change in density from the reference frame image and is a characteristic quantity related to blood flow in each block region for each frame image. In this embodiment, the inter-frame difference value is calculated by subtracting the signal value of the reference frame image from the signal value of each frame image. Therefore, a negative value in the inter-frame difference value indicates blood flow rate.

[0050] (8) Next, a pulmonary blood flow summary image is generated by aggregating the blood flow features of each block region for each frame image. For example, a representative value in the time direction of the inter-frame difference value (either the cumulative value, mean, minimum, maximum, or median) is extracted for each block region to generate a pulmonary blood flow summary image.

[0051] Once the generation of the pulmonary blood flow summary image is complete, an upper limit setting process is performed on the pulmonary blood flow summary image (step S13).

[0052] In dynamic chest images, factors that cause density changes in the lung field region with the same period as the heartbeat include vascular displacement, changes in blood thickness (changes in blood vessel diameter due to pulsation, etc.), and changes in blood density (changes in blood concentration (blood pressure)). Of these, density changes due to vascular displacement are noise and not density changes that indicate blood flow, while changes in blood thickness and blood density are the density changes that indicate blood flow. Since changes in blood density are estimated to be at most 0.0001%, the density changes that indicate blood flow can be considered to be changes in blood thickness.

[0053] Furthermore, since pulmonary perfusion scintigraphy evaluates blood flow in peripheral vessels, in order to perform blood flow evaluation equivalent to that of pulmonary perfusion scintigraphy, it is necessary to remove the concentration changes caused by the main vessels (changes in blood thickness and density in the main vessels) and extract the changes in blood thickness (and density) in the peripheral vessels.

[0054] In other words, the inter-frame difference values, which are the signal values ​​for each block region of a pulmonary perfusion summary image, include noise such as density changes due to the movement of blood vessels and density changes due to the main blood vessels. To obtain density changes due to blood flow with high accuracy, similar to those obtained from pulmonary perfusion scintigraphy, these noises must be removed. Noise such as density changes due to the movement of blood vessels and density changes due to the main blood vessels are greater than density changes due to blood flow in peripheral blood vessels.

[0055] Therefore, in the upper limit setting process of step S13, an upper limit (absolute upper limit) is set for the signal value (inter-frame difference value) of each block region of the pulmonary blood flow summary image. If the value exceeds the upper limit, it is replaced with the upper limit, thereby limiting the inter-frame difference value. This removes the concentration changes due to the movement of blood vessels and the concentration changes due to the main blood vessels from the inter-frame difference value, and accurately calculates the features related to pulmonary blood flow. Furthermore, as described above, in this embodiment, blood flow rate corresponds to a negative value of the inter-frame difference. Therefore, if positive values ​​are used directly when calculating the blood flow ratio relative to blood flow rate, they will become a source of error. For this reason, positive values ​​may be replaced with 0 at this point.

[0056] Here, the upper limit of the interframe difference can be determined based on the results of simulating the signal changes in peripheral blood vessels associated with cardiac output in dynamic images of the chest. Alternatively, it can be determined based on the results of simulating the signal changes in peripheral blood vessels associated with cardiac output in dynamic images of the chest and the results of determining the signal changes of the main blood vessels from multiple dynamic images of the chest.

[0057] The following describes a simulation of signal changes in peripheral blood vessels associated with cardiac output in dynamic images of the chest. As a prerequisite for the simulation, we assume that the blood equivalent to the stroke volume is uniformly diffused throughout the lung field, as shown in Figure 4. We also assume that the change in the thickness of the diffused blood corresponds to a change in the signal in the radiographic image.

[0058] It is known that the X-ray dose I after passing through the subject M can be expressed by the following (Equation 1). I = I0 × exp[-μ × ρ × X] ···(Equation 1) Here, I0 is the amount of X-rays irradiated, and μ is the mass absorption coefficient [cm²]. 2 [g / g], ρ is density [g / cm³]. 3 ], where X is the subject transmission distance [cm].

[0059] From the above (Equation 1), the rate of signal change caused by changes in blood thickness can be calculated using the following (Equation 2). Signal change rate = (I0 - I) / I0 = 1 - exp[-μ × ρ × X] ... (Equation 2)

[0060] In this context, the subject distance X is the thickness of the blood, and if we let SV be the stroke volume and SL be the size (area) of the lung field, then, based on the above assumptions, X = SV / SL [cm] In other words, the rate of signal change caused by changes in blood thickness can be simulated using the values ​​of parameters such as stroke volume SV, lung field size SL, mass absorption coefficient μ, and density ρ.

[0061] Stroke volume SV is the blood flow rate to the lung field in one heartbeat. A standard value may be used as the stroke volume SV, but if stroke volume information for each subject is available (for example, stored in the memory unit 32 or included in the patient information), it is preferable to use that value. Furthermore, stroke volume SV tends to be higher with height, for example. Therefore, a table associating at least one physical characteristic (height, weight, age, sex) with the corresponding stroke volume SV (for example, a value obtained statistically) may be stored in the memory unit 32, and the stroke volume SV corresponding to the physical characteristic obtained from the patient's patient information may be determined by referring to the table.

[0062] The lung field size SL here refers to the area of ​​the lung field. A standard value may be used for lung field size SL, but it is preferable to use the value if information on the lung field size for each subject is available (for example, if it is stored in the memory unit 32 or included in the patient information). Also, if data exists that can measure the lung field volume of the subject M, such as from a CT scan, the volume may be determined using that measurement data, and the area may be calculated from the determined volume. Furthermore, lung field size SL tends to vary depending on physical characteristics, for example, it tends to be larger with height. Therefore, a table that associates at least one physical characteristic, such as height, weight, age, and sex, with the lung field size SL corresponding to that physical characteristic (for example, a value obtained by statistics, etc.) may be stored in the memory unit 32, and the lung field size corresponding to the physical characteristic obtained from the patient information may be determined by referring to the table.

[0063] The mass absorption coefficient μ can be determined from literature values ​​based on the imaging conditions (tube voltage, average energy, etc.) and the area for which the blood flow characteristics are calculated (in this case, peripheral blood vessels of the lungs). That is, the mass absorption coefficient μ for each combination of imaging conditions and area is stored in the memory unit 32, and the mass absorption coefficient μ corresponding to the imaging conditions and measurement area (in this case, peripheral blood vessels of the lungs) of the chest dynamic image to be processed can be read out from the memory unit 32 and obtained. The density ρ can be determined from literature values ​​from the site for which the blood flow characteristics are calculated (in this case, the peripheral blood vessels of the lungs). That is, the density ρ of the peripheral blood vessels of the lungs can be stored in the memory unit 32, and the density ρ of the peripheral blood vessels of the lungs can be read from the memory unit 32 to be obtained.

[0064] In other words, in the upper limit setting process, the signal change rate due to peripheral pulmonary blood vessels can be calculated based on at least the acquisition conditions of the dynamic chest image and information on the blood vessels (peripheral pulmonary blood vessels) that are subject to the calculation of blood flow features, and the upper limit of the blood flow features can be determined based on the signal change rate. Furthermore, the signal change rate due to peripheral pulmonary blood vessels can be calculated based on one or more pieces of information from the subject's physical information in the dynamic chest image, the area of ​​the lung field, or the volume, and the upper limit of the blood flow features can be determined based on the signal change rate.

[0065] For example, a normal stroke volume is 60-130 ml, and a standard lung field size is 400 cm². 2 The mass absorption coefficient, determined from the literature based on the imaging conditions and target area, is 0.216 cm². 2 / g, density 1.056 g / cm³ 3 In that case, from the calculation formula (Equation 2) above, the rate of signal change due to changes in blood thickness in peripheral blood vessels can be estimated to be 3.4 to 7.1%. Therefore, the signal change due to changes in blood thickness The upper limit of the rate can be determined, for example, between 3.4% and 7.1%, depending on the stroke volume.

[0066] Furthermore, visual evaluation of multiple cases of actual chest dynamic images revealed that the signal change rate due to changes in the thickness of the main blood vessels in the hilar region and the signal change rate due to the movement of the main blood vessels were both 7.5% or higher in most cases, and the signal change rate due to the movement of the main blood vessels was 5% or higher. Therefore, the upper limit for the signal change rate may be set at 5%. Verification has shown that this upper limit has a high correlation with pulmonary perfusion scintigraphy.

[0067] When the imaging position is supine, as per the above preconditions, it is possible to assume that blood flow is uniformly distributed throughout the lung field and to set the upper limit of the signal change rate uniformly regardless of the position in the lung field. However, when imaging is performed in an upright position, it is known that blood flow is greater in the lower part of the lung field due to the effect of gravity. For example, the blood flow ratio between the lung apex and the lung base is apex:lung base = 1:10 (see "Normal Structure and Function of the Human Body," edited by Takeo Sakai and Katsumasa Kawahara, Nippon Iji Shinpo Co., Ltd., January 2021). Therefore, when the imaging position is upright, the upper limit of the signal change rate may be varied according to the position (vertical position) in the lung field region based on this blood flow distribution. For example, if the upper limit of the signal change rate at the lung apex is set to 1, the upper limit of the signal change rate at the lung base may be set to 10, and the upper limit of the signal change rate may be determined according to the position in the lung field region.

[0068] Furthermore, blood flow decreases as it spreads outward from the hilum of the lung. Therefore, the upper limit of the signal change rate may be varied according to the location in the lung field, reflecting this distribution of blood flow. For example, as shown in Figure 5, the upper limit may be set lower the further away from the center of the lung field.

[0069] In step S13, once the upper limit of the signal rate of change is determined, the upper limit of the inter-frame difference value for each block region is calculated using a predetermined calculation formula that utilizes the determined upper limit of the signal rate of change. If the inter-frame difference value is greater than or equal to the upper limit, it is replaced with the upper limit.

[0070] Next, the analysis region setting process is performed, and the region obtained by removing blood vessels that carry blood flow to areas other than the lungs from the lung field region is set as the analysis region for the pulmonary blood flow ratio (step S14).

[0071] Here, the brachiocephalic artery and subclavian artery overlap with the lung field region in the two-dimensional frontal chest image, but they are blood vessels that carry blood to areas other than the lungs. Although it is difficult to visualize the shape of these arteries in dynamic chest images, as shown by the dashed line in Figure 6, they are visible as areas of decreased density (white areas) in the upper inner part of the generally recognized lung field region, and it was found that these blood flows have a large signal change due to blood flow, and thus have a significant impact as noise when evaluating pulmonary blood flow. Therefore, in step S14, the region obtained by excluding the area affected by the concentration changes due to either the brachiocephalic artery or the subclavian artery from the already extracted lung field region in the pulmonary blood flow summary image is set as the analysis region for the pulmonary blood flow ratio.

[0072] The process of defining the analysis region as the area excluding the region affected by the concentration changes due to either the brachiocephalic artery or the subclavian artery from the lung field region may be performed using, for example, a trained model obtained by machine learning such as deep learning (DL processing), or the analysis region may be defined by image processing without using machine learning. The following describes examples of processing in step S14. Examples (A) to (E) below use deep learning, and examples (F) to (G) define the analysis area using image processing without machine learning. In the following explanation, the lung field region that is generally extracted as a lung field region, including the brachiocephalic artery and subclavian artery (for example, the lung field region extracted in step S12), will be referred to as the lung field mask, and the lung field region excluding the brachiocephalic artery and subclavian artery will be referred to as the lung field mask for blood flow ratio.

[0073] (A) As shown in Figure 7, a trained model M1 that outputs a lung field mask for blood flow ratio from an input original image of the chest is stored in the memory unit 32. This trained model is generated by training a deep learning model using sets (multiple sets) of chest radiographs (original images) and images in which a physician or other person has annotated a lung field mask for blood flow ratio in the original image (ground truth images of the lung field mask for blood flow ratio) as training data. Then, in step S14, for example, the original image of one frame from among multiple frames of dynamic images of the chest (called the representative frame image) is input to the trained model M1 to obtain a lung field mask for blood flow ratio, and the region of the obtained lung field mask for blood flow ratio is set as the analysis region.

[0074] (B) As shown in Figure 8, a trained model M2 is stored in the memory unit 32 by training the model using deep learning with sets (multiple sets) of training data, which output a lung field mask for blood flow ratio from the input original chest image and lung field mask. These sets include a chest radiograph (original image), an image in the original image with lung field masks annotated by a physician or the like (this may also be the result of automatically extracting the lung field regions from the original image), and an image in the original image with lung field masks for blood flow ratios annotated by a physician or the like (the correct image of the lung field mask for blood flow ratio). Then, in step S14, a representative frame image and the lung field region extracted from that frame image (the lung field mask extracted in step S12) are input to the trained model M2 to obtain a lung field mask for blood flow ratio, and the region of the obtained lung field mask for blood flow ratio is set as the analysis region.

[0075] (C) As shown in Figure 9, a trained model M3 is stored in the memory unit 32. This trained model is generated by training the model using deep learning with sets of training data consisting of chest radiographs (original images) and images in which a physician or other person has annotated areas affected by either the brachiocephalic artery or the subclavian artery. The trained model M3 outputs likelihood maps of the brachiocephalic artery and subclavian artery from the input original chest image. In step S14, a representative frame image is input to the trained model M3 to obtain likelihood maps of the brachiocephalic artery and subclavian artery. In post-processing, the lung field area (lung field mask) extracted in step S12 of the representative frame image is modified to remove areas where the likelihood of the brachiocephalic artery and subclavian artery is above a predetermined threshold. This area is then obtained as a lung field mask for blood flow ratio, and the obtained area is set as the analysis area.

[0076] (D) As shown in Figure 10, a trained model M4 is stored in the memory unit 32 by training the model using deep learning with sets (multiple sets) of training data, which include a chest radiograph (original image), an image in the original image with lung field masks annotated by a physician or the like (or the result of automatically extracting lung field regions from the original image), and an image in the original image with regions affected by one or more of the brachiocephalic arteries and subclavian arteries annotated by a physician or the like. The trained model M4 outputs likelihood maps of the brachiocephalic artery and subclavian artery from the input chest original image and lung field mask. Then, in step S14, the representative frame image and the lung field region extracted from that frame image (the lung field mask extracted in step S12) are input to the trained model M4 to obtain likelihood maps of the brachiocephalic artery and subclavian artery. In post-processing, the region obtained by removing the area where the likelihood of the brachiocephalic artery and subclavian artery is above a predetermined threshold from the lung field mask extracted in step S12 is obtained as a lung field mask for blood flow ratio, and the obtained region is set as the analysis region.

[0077] (E) It is known that the brachiocephalic artery and subclavian artery are located on the medial side of the upper part of the lung field. Therefore, as shown in Figure 11, a trained model M5 is stored in the memory unit 32, which is generated by training a deep learning model using sets (multiple sets) of training data: a chest radiograph (original image) and a lung field mask, with the region including the brachiocephalic artery and subclavian artery in the medial side of the upper part of the lung field trimmed from the lung field mask, and an image in which a lung field mask for blood flow ratio has been annotated by a physician or the like in the original image. The trained model M5 outputs a lung field mask for blood flow ratio from the input original chest image and lung field mask (trimmed). Then, in step S14, the representative frame image and the lung field region extracted from that frame image (the lung field mask extracted in step S12), trimmed to include the brachiocephalic artery and subclavian artery in the medial upper part of the lung field, are input to the trained model M5 to obtain likelihood maps of the brachiocephalic artery and subclavian artery. In post-processing, the region obtained by removing the area where the likelihood of the brachiocephalic artery and subclavian artery is above a predetermined threshold from the lung field mask extracted in step S12 is obtained as a lung field mask for blood flow ratio, and the obtained region is set as the analysis region.

[0078] (F) Based on the density information of the lung field region in the representative frame image, the regions of the brachiocephalic artery and subclavian artery are identified, and the region obtained by excluding the brachiocephalic artery and subclavian artery from the lung field region is set as the analysis region. For example, as shown in Figure 12, in the upper half of the lung field region extracted from the representative frame image (the lung field region extracted in step S12), the area of ​​reduced density (the area where the density value (signal value) is below a predetermined threshold) is reduced inward to set the analysis area.

[0079] (G) The regions of the brachiocephalic artery and subclavian artery are identified from the pulmonary blood flow summarized image and lung field mask, and the region obtained by excluding the brachiocephalic artery and subclavian artery from the lung field region is set as the analysis region. For example, as shown in Figure 13, in the upper half of the lung field region in the pulmonary blood flow summary image, the region where the inter-frame difference value exceeds a predetermined threshold is reduced inward to set the analysis region. Here, the medial part of the upper lung field should have low pulmonary blood flow (small inter-frame difference value). However, if the brachiocephalic artery or subclavian artery is present, the inter-frame difference value is expected to be larger compared to the pulmonary blood flow. Therefore, the region excluding the area where the inter-frame difference value exceeds a predetermined threshold (a value corresponding to pulmonary blood flow) is set as the analysis region.

[0080] The analysis region is larger than the block region and does not include the lung area that overlaps with other organs in dynamic images of the chest.

[0081] Next, the left-right ratio of pulmonary blood flow, including behind the organs (the ratio of pulmonary blood flow in the left lung field to that in the right lung field), is calculated (Step S15).

[0082] Traditionally, when calculating blood flow features using dynamic images of the chest, lung regions that overlap with other organs (e.g., heart, diaphragm, thoracic vertebrae, lumbar vertebrae, vertebral bodies, aorta, pulmonary artery, aortic arch, etc.) (lung regions behind organs) were either excluded from the analysis and ignored, or, while acknowledging that noise from the heart and diaphragm would be present, it was determined that signal changes related to pulmonary blood flow could still be extracted, and blood flow features were calculated in the same way as for lung regions that did not overlap with other organs.

[0083] However, it has been found that conventional methods have problems when measuring the left-right ratio of pulmonary blood flow. For example, if the lung field region behind organs is excluded from the analysis area, the pulmonary blood flow in the entire lung field is not considered, resulting in discrepancies with the measurement results of the left-right ratio of pulmonary blood flow using currently available pulmonary perfusion scintigraphy. Furthermore, if features related to pulmonary blood flow are extracted from the lung field region behind organs in the same way as the lung field region not overlapping with organs, and the left-right ratio of pulmonary blood flow is measured, the noise in the features related to pulmonary blood flow calculated from behind organs is large, resulting in discrepancies with the measurement results of the left-right ratio of pulmonary blood flow using currently available pulmonary perfusion scintigraphy.

[0084] To address these problems, the inventors of this application have found that, regarding the lung field region behind organs, instead of actually measuring it, it is more efficient to assume that the same pulmonary blood flow conditions exist in the lung field region behind organs as in the lung field region not overlapping with other organs. By using a correction value (called an organ-back correction coefficient) to correct the left-right ratio of pulmonary blood flow to a ratio that takes into account the blood flow in the lung field region behind organs, and then correcting the characteristic quantities related to blood flow measured in the lung field region not overlapping with other organs to calculate the left-right ratio, the results differ less from the measurement results of the left-right ratio of pulmonary blood flow using currently widespread pulmonary perfusion scintigraphy. Therefore, in step S15, an organ-back correction coefficient is obtained, and the left-right ratio of pulmonary blood flow including the organ-back area is calculated using the organ-back correction coefficient.

[0085] The following explains how to calculate the organ correction coefficient. In dynamic images of the chest, the left lung field overlaps with other organs more than the right lung field. Therefore, if the left-right ratio of pulmonary blood flow is measured based on the lung field area that does not overlap with organs, the blood flow in the left lung field will be underestimated compared to the right lung field. In this embodiment, a correction coefficient to correct this is calculated as an organ-based correction coefficient. The organ-back correction coefficients described below are derived based on the assumption that the lung blood flow in the lung field behind the organ is the same as that in the lung field area that does not overlap with other organs in dynamic images, and that the area and volume of the lung field are proportional to the blood flow rate.

[0086] • When using volume (a) As shown in Figure 14, first, a CT image of the lungs from the apex to the base of the lungs of the subject is obtained as a dynamic image of the chest, and for each slice of the CT image, the area of ​​the region that does not overlap with other organs and the region that overlaps with other organs (hidden region) in the dynamic image of the chest (in the direction of radiation irradiation when the dynamic image of the chest is taken) are measured for the right lung field region and the left lung field region, respectively. (b) Add up the areas of each slice and calculate the volume of the right lung field region and the left lung field region, respectively, including the area that does not overlap with other organs in the dynamic image of the chest and the area that overlaps with other organs (hidden area). (c) Using the obtained volume, calculate the organ back correction coefficient. As shown in Figure 15, if A is the volume of the right lung field that does not overlap with other organs in a dynamic image of the chest, and B is the volume of the right lung field that overlaps with other organs, then the coefficient A', which represents the ratio of the total volume of the right lung field to A, is: A' = (A + B) / A ... (Equation 3) Similarly, if C is the volume of the left lung field that does not overlap with other organs in a dynamic image of the chest, and D is the volume of the left lung field that overlaps with other organs, then the coefficient C', which represents the ratio of the total volume of the left lung field to C, is: C´=(C+D) / C...(Formula 4) If the organ-back correction coefficient for the right lung field is set to 1, then the organ-back correction coefficient α for the left lung field is: α=C´ / A´···(Formula 5)

[0087] When using area The following areas are measured from a frontal chest image (or a single frame image from a dynamic chest image) (see Figure 15). A: Area of ​​the right lung field that does not overlap with other organs B: Area of ​​the right lung field region that overlaps with other organs C: Area of ​​the left lung field that does not overlap with other organs D: Area of ​​the left lung field region that overlaps with other organs. The organ correction coefficient is determined using the above equations (3) to (5).

[0088] For example, using the volume or area method described above, organ correction coefficients are calculated from multiple case data of various physical characteristics without distinguishing between age, sex, height, weight, etc., and the average value or median value of these coefficients is pre-stored in the memory unit 32. In step S15, the organ correction coefficients (representative values) stored in the memory unit 32 are read and obtained. Alternatively, organ correction coefficients may be calculated for each physical characteristic such as age, sex, height, and weight, or for each combination thereof, and the average or median values ​​of these coefficients may be stored in the memory unit 32 in advance. In step S15, the organ correction coefficients corresponding to the subject's physical characteristics stored in the memory unit 32 may be read and obtained. Alternatively, if data exists that allows for the calculation of the right and left lung field regions that overlap with other organs, and the right and left lung field regions that do not overlap with other organs, such as CT images of the subject with dynamic chest images, then the individual organ-based correction coefficient for the subject is calculated and obtained.

[0089] Once the organ-back correction coefficient is obtained, the analysis region is divided into the left lung field analysis region and the right lung field analysis region. Representative values ​​(integrated value, mean, minimum, maximum, or median) of the interframe difference values ​​within each analysis region of the left and right lung blood flow summary images, which have been corrected for upper limits, are calculated, and these calculated representative values ​​are corrected using the organ-back correction coefficient. Specifically, the organ-back correction coefficient for the right lung field region is multiplied by the representative value of the right lung field analysis region, and the organ-back correction coefficient for the left lung field region is multiplied by the representative value of the left lung field analysis region. This allows the calculated left-right ratio of lung blood flow to be corrected to a ratio that takes into account the blood flow in the lung field region behind the organs. The ratio of the representative values ​​of the left and right analysis regions after correction is then calculated as the left-right ratio of lung blood flow. Alternatively, the left-right ratio of pulmonary blood flow may be calculated without correcting for representative values, and the calculated left-right ratio of pulmonary blood flow may be corrected using an organ-based correction coefficient.

[0090] Then, the analysis results screen 341 showing the left-right ratio of pulmonary blood flow is displayed on the display unit 34 (step S16), and the pulmonary blood flow analysis process is completed.

[0091] Figure 16 shows an example of the analysis results screen 341. As shown in Figure 16, the analysis results screen 341 displays the calculated left-right ratio of pulmonary blood flow 341a, a pulmonary blood flow summary image 341b, the applied upper limit 341c, information 341d indicating whether the lung field mask for blood flow ratio was automatically or manually generated, information 341e indicating whether the organ back correction coefficient was applied, and the value of the organ back correction coefficient 341f if it was applied. The pulmonary blood flow summary image is an image colored according to the inter-frame difference value (i.e., the feature quantity related to blood flow). In addition, the lung field mask for blood flow ratio 341g is displayed on the pulmonary blood flow summary image 341b. The lung field mask for blood flow ratio 341g is displayed with the lung field region extended to include the organ back, compared to the lung field mask for blood flow ratio 341g1 created in step S14, so that it can be seen that the lung field region behind the organ is taken into consideration in the calculation of the blood flow ratio. In this case, as shown in Figure 16, the region of the lung field mask 341g1 created in step S14 may be displayed within the lung field mask 341g for blood flow ratio so that it is possible to determine whether or not it is behind an organ, or, as shown in Figure 17, only the lung field mask 341g for blood flow ratio may be displayed without distinguishing between the region behind an organ and the region behind an organ.

[0092] On the analysis results screen 341, the upper limit 341c, automatic / manual generation of lung field masks for blood flow ratio 341d, and whether or not to apply the organ back correction coefficient 341e can be changed. For example, as shown in Figure 18, if the upper limit is changed by the user operating the control unit 33, the control unit 31 changes the upper limit to the input value and recreates the pulmonary blood flow summary image. At the same time, the left-right ratio of pulmonary blood flow is recalculated, and the value of the left-right ratio of pulmonary blood flow 341a and the coloring of the pulmonary blood flow summary image 341b on the analysis results screen 341 are changed.

[0093] Furthermore, for example, as shown in Figure 19, when the user changes the generation of the lung field mask for blood flow ratio from automatic to manual by operating the control unit 33, the control unit 31 displays coordinate points (points indicated by black circles in Figure 19) and a cursor 341i on the lung field mask 341g for blood flow ratio. When the user moves the cursor 341i to the coordinate points using the control unit 33, the left-right ratio is recalculated in accordance with the movement of the coordinate points, and the recalculated result is displayed as the left-right ratio of lung blood flow 341a.

[0094] Furthermore, as shown in Figure 20, for example, if the application of the organ back correction coefficient is changed from enabled to disabled by the user's operation of the control unit 33, the control unit 31 changes the shape of the lung field mask 341g for blood flow ratio to one that does not take organ back into consideration. In addition, the left-right ratio of pulmonary blood flow is recalculated, and the recalculation result is displayed as the left-right ratio of pulmonary blood flow 341a. It will be displayed.

[0095] It should be noted that the left-right ratio of pulmonary blood flow after applying the organ-overlay correction coefficient is a value based on the assumption that "the same blood flow as in the lung field area not overlapping with other organs exists in the lung field area behind the organ," and therefore may contain errors. For this reason, in order to allow for flexibility in the judgment of medical professionals when making a diagnosis, it is preferable to include the assumed error of 341h for the left-right ratio of pulmonary blood flow, as shown in Figure 21. The expected error can be calculated, for example, as shown in (Equation 6) below. (Right lung ratio before applying organ correction coefficient - Right lung ratio after applying organ correction coefficient) × 1.5 ... (Equation 6) The value 1.5 was added to provide a margin of error. Therefore, you can set a different value other than 1.5.

[0096] As explained above, the control unit 31 of the diagnostic console 3 acquires dynamic images of the chest obtained by dynamic radiography, extracts the lung field region from the acquired dynamic images, and calculates blood flow-related feature quantities (inter-frame difference values) from the extracted lung field region. Then, it determines an upper limit for the calculated blood flow-related feature quantity and imposes a restriction based on that upper limit. Therefore, from dynamic images of the chest, it is possible to calculate characteristic features related to blood flow, excluding the concentration changes caused by the major blood vessels, which have high values. In other words, it becomes possible to accurately calculate characteristic features related to pulmonary blood flow excluding the major blood vessels.

[0097] Furthermore, the control unit 31 sets an analysis region in the lung field area and performs analysis based on the blood flow characteristics calculated in the set analysis region. For example, the analysis is performed by setting the analysis region as the region that has been removed from the lung field area, specifically the region that is affected by concentration changes due to one or more of the brachiocephalic artery and subclavian artery. Therefore, by removing the region overlapping with blood flow outside the lungs from the lung field region of the dynamic chest image and using that region as the analysis area, we can perform an analysis based on blood flow features, thus enabling a highly accurate analysis of blood flow.

[0098] Furthermore, for example, the control unit 31 divides the analysis area into a left lung field region and a right lung field region, each containing multiple sub-regions. For each divided region, it calculates a representative value of the blood flow feature quantities calculated within that region. The calculated representative value is then corrected based on a correction value that includes the blood flow feature quantities of regions overlapping with other organs within that region. Based on the corrected representative value, the control unit calculates the pulmonary blood flow ratio between the left and right lung fields. Therefore, the pulmonary blood flow ratio can be calculated with high accuracy, reflecting the pulmonary blood flow behind the organs.

[0099] The above description of the embodiment is merely an example of a preferred dynamic image analysis system according to the present invention, and is not limited thereto.

[0100] For example, in the above embodiment, a pulmonary blood flow summary image is generated, then an analysis region is set for calculating the left-right ratio of pulmonary blood flow, and the left-right ratio of pulmonary blood flow is calculated using the inter-frame difference values ​​within the analysis region. However, after setting the analysis region, a pulmonary blood flow summary image is generated by aggregating the inter-frame difference values ​​within the analysis region, and the left-right ratio of pulmonary blood flow is calculated from that image.

[0101] Furthermore, although the above embodiment described an example in which the present invention is applied to calculate the left-right ratio of pulmonary blood flow, it is not limited to this. For example, the blood flow ratio of six regions, the right upper lung field, the right middle lung field, the right lower lung field, the left upper lung field, the left middle lung field, and the left lower lung field, may be calculated. In this case, the organ correction coefficient is calculated for each region from multiple case data, and the average value or median value or other representative value is stored in the memory unit 32 in advance, and the correction is performed using the organ correction coefficient corresponding to each region.

[0102] Furthermore, although the above embodiment described an example where the inter-frame difference value with a reference frame image is used as a feature quantity related to blood flow, the invention is not limited to this. For example, the cross-correlation coefficient showing the correlation between the pulsation signal waveform and the signal waveform (signal waveform for each block region), as described in Japanese Patent Application Publication No. 2012-239796, may be used as a feature quantity related to blood flow.

[0103] Furthermore, while the above description discloses examples in which a hard disk or semiconductor non-volatile memory is used as a computer-readable medium for the program according to the present invention, the invention is not limited to these examples. Other computer-readable media include CD-R Portable recording media such as OM can be used. Furthermore, carrier waves can also be used as a medium for providing program data according to the present invention via a communication line.

[0104] Furthermore, the detailed configuration and operation of each device constituting the dynamic image analysis system 100 can also be modified as appropriate without departing from the spirit of the present invention. [Explanation of symbols]

[0105] 100 Dynamic Image Analysis System 1. Imaging device 11 Radiation source 12. Radiation irradiation control device 13. Radiation detection unit 14. Reading control device 2. Shooting console 21 Control Unit 22 Memory section 23 Control section 24 Display 25 Communications Department 26 bus 3. Diagnostic console 31 Control Unit 32 Storage section 33 Operation section 34 Display section 35 Communications Department 36 bus

Claims

1. An acquisition unit that acquires dynamic images of the chest obtained by dynamic imaging using radiation, An extraction unit for extracting lung field regions from the aforementioned dynamic image, A blood flow feature calculation unit calculates concentration change amounts, which represent changes in blood thickness and density within blood vessels, as feature quantities related to blood flow from the lung field region, An analysis region setting unit sets an analysis region by removing the region that overlaps with blood flow outside the lungs from the aforementioned lung field region, An analysis unit that calculates the blood flow ratio of multiple regions in the analysis region based on the characteristic quantities related to blood flow calculated in the analysis region, A dynamic image analysis device equipped with the following features.

2. The dynamic image analysis device according to claim 1, wherein the region to be removed by the analysis region is a region affected by the density change caused by one or more of the brachiocephalic artery and the subclavian artery.

3. The dynamic image analysis device according to Claim 2, wherein the analysis region setting unit sets the analysis region using a trained model that takes a radiographic image of the chest as input and outputs an analysis region excluding the region affected by the density change due to one or more of the brachiocephalic artery and subclavian artery.

4. The dynamic image analysis device according to claim 2, wherein the analysis region setting unit sets the analysis region by removing from the lung region regions regions affected by density changes due to one or more of the brachiocephalic artery and subclavian artery, based on the density information of the lung field region in the frame image of the dynamic image, or the characteristic quantities relating to the blood flow.

5. The dynamic image analysis device according to Claim 1, wherein the analysis unit divides the analysis area into a plurality of regions including a plurality of sub-regions, calculates a representative value of the characteristic quantity related to the blood flow calculated in each of the divided plurality of regions, calculates the ratio of the calculated representative values ​​of the plurality of regions as the blood flow ratio of the plurality of regions, obtains a correction value to correct the ratio to a ratio that takes into account the blood flow of lung field regions that overlap with other organs, and corrects the ratio based on the obtained correction value.

6. The dynamic image analysis device according to claim 5, wherein the correction value is a value calculated based on the volume or area of ​​the lung field region that overlaps with other organs in the direction of radiation irradiation at the time of capturing the dynamic image.

7. Computers Acquisition unit that acquires dynamic images of the chest obtained by dynamic imaging using radiation, An extraction unit for extracting the lung field region from the aforementioned dynamic image, A blood flow feature calculation unit calculates the amount of concentration change, which indicates the change in blood thickness and density of blood in the blood vessels, as a feature quantity related to blood flow from the lung field region. An analysis region setting unit sets an analysis region by removing the region that overlaps with blood flow outside the lungs from the aforementioned lung field region. An analysis unit calculates the blood flow ratio of multiple regions in the analysis region based on the characteristic quantities related to blood flow calculated in the analysis region. A program designed to function as such.

Images