X-ray fluoroscopy device

Abstract

The X-ray fluoroscopic imaging device (100) comprises a photographing unit (1), an X-ray image acquiring unit (20) for acquiring an X-ray image (10), an object distribution learning and recognizing unit (21) for outputting the distribution of an object captured in the X-ray image using a learning model (7), an image quality improving processing unit (22), and a display unit (3), wherein the image quality improving processing unit is configured to be able to switch between a first image processing mode and a second image processing mode, wherein in the first image processing mode, image quality improving processing is performed on the X-ray image using a learning and recognition result (7a) of the object distribution learning and recognizing unit, and in the second image processing mode, image quality improving processing is performed on the X-ray image without using the learning and recognition result.

Description

Technical Field

[0001] The present invention relates to an X-ray fluoroscopy apparatus, and more particularly to an X-ray fluoroscopy apparatus that uses a learned model to recognize a device introduced into a subject's body. Background Art

[0002] Conventionally, there is a known X-ray fluoroscopy device that uses a learned model to identify a device inserted into a subject's body. This type of X-ray fluoroscopy device is disclosed in, for example, Japanese Patent Application Laid-Open No. 2017-185007.

[0003] The X-ray fluoroscopy device disclosed in Japanese Patent Application Laid-Open No. 2017-185007 includes an irradiation unit, a radiation detection unit, an image generation unit, and an object detection unit. The irradiation unit is configured to irradiate a subject with radiation. The radiation detection unit is configured to detect radiation that has passed through the subject. The image generation unit is configured to generate a radiographic image based on the detection signal from the radiation detection unit. The object detection unit is configured to identify an object from a radiographic image through image recognition based on learning result data for image recognition previously acquired through machine learning. Furthermore, Japanese Patent Application Laid-Open No. 2017-185007 is configured to perform processing to emphasize the identified object. Japanese Patent Application Laid-Open No. 2017-185007 discloses a structure for identifying and emphasizing a device such as a guidewire as an object. Furthermore, Japanese Patent Application Laid-Open No. 2017-185007 discloses an object detection unit that is configured to identify an object from a radiographic image in the form of a moving image.

[0004] Prior art literature

[0005] Patent Literature

[0006] Patent Document 1: Japanese Patent Application Laid-Open No. 2017-185007 Summary of the Invention

[0007] Problems to be solved by the invention

[0008] Here, although not disclosed in Japanese Patent Application Laid-Open No. 2017-185007, the learning algorithm of the learning model generated by machine learning and the training data used for learning affect the recognition accuracy of the object (device). In addition, in order to reduce the burden on the patient, the shape of the device used in the operation of introducing the device into the patient's body is being improved. Therefore, when using a device (object) that has been improved to reduce the burden on the patient, due to the difference between the shape of the object (device) in the training data used in the learning of the learning model and the shape of the device (object) actually used for the operation, there is a situation where the recognition accuracy of the object is reduced by the learning model. In this case, the accuracy of the recognition result of the learning model does not reach sufficient accuracy and there is a situation where the object is misdetected. In the case of misdetection of the object, there is the following undesirable situation: by emphasizing the part that is not the object based on the learning recognition result output by the learning model, the visual recognition of the object in the X-ray image is reduced. Therefore, an X-ray fluoroscopy apparatus is desired that can suppress a reduction in the visual recognition of an object due to a learning recognition result output by the learning model, in a configuration that uses a learning model to recognize an object in an X-ray image. Furthermore, the learning recognition result includes the distribution of the object (device position information) output by the learning model.

[0009] The present invention is made to solve the above-mentioned problems and aims to provide an X-ray fluoroscopy apparatus that can suppress the reduction in the visual recognition of the device due to the learning recognition results output by the learning model in a structure that uses a learning model to recognize objects in X-ray images.

[0010] Solutions for solving problems

[0011] In order to achieve the above-mentioned purpose, an X-ray fluoroscopy apparatus according to one aspect of the present invention comprises: a photographing unit, which includes an X-ray source for irradiating X-rays to a subject and an X-ray detector for detecting the X-rays irradiated from the X-ray source; an X-ray image acquisition unit, which acquires the X-ray image captured by the photographing unit; an object distribution learning and recognition unit, which uses a learned learning model to output the distribution of objects captured in the X-ray image; an image quality improvement processing unit, which performs image quality improvement processing to improve the image quality of the X-ray image; and a display unit, which displays the X-ray image, wherein the image quality improvement processing unit is configured to be able to switch between a first image processing mode and a second image processing mode, in which the image quality improvement processing is performed on the X-ray image using the learning and recognition results of the object distribution learning and recognition unit in the first image processing mode and the image quality improvement processing is performed on the X-ray image without using the learning and recognition results in the second image processing mode.

[0012] Effects of the Invention

[0013] In the X-ray fluoroscopy apparatus according to the aforementioned aspect, as described above, the apparatus includes an image quality improvement processing unit configured to switch between a first image processing mode in which image quality improvement processing is performed on X-ray images using the learned recognition results, and a second image processing mode in which image quality improvement processing is performed on X-ray images without using the learned recognition results. This allows switching between the first image processing mode in which the learned recognition results are used and the second image processing mode in which the learned recognition results are not used, depending on whether the learned recognition results have high accuracy or low accuracy. Specifically, when the learned recognition results have high accuracy, the first image processing mode can be used to improve the visibility of the object, while when the learned recognition results have low accuracy, the image quality improvement processing is performed on the X-ray images in the second image processing mode. This allows suppressing any degradation in the visibility of the object due to the learned recognition results. Consequently, in a configuration in which a learned model is used to identify devices in X-ray images, it is possible to suppress any degradation in the visibility of the object due to the learned recognition results output by the learning model. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 This is a diagram showing the overall configuration of an X-ray fluoroscopy apparatus according to one embodiment.

[0015] Figure 2 This is a schematic diagram for explaining a configuration in which an image quality improvement processing unit according to one embodiment generates an enhanced image after image quality improvement processing based on a first image processing mode.

[0016] Figure 3 This is a schematic diagram for explaining a configuration in which an image quality improvement processing unit generates an X-ray image after image quality improvement processing based on a second image processing mode according to one embodiment.

[0017] Figure 4 It is a schematic diagram for explaining switching between the first image processing mode and the second image processing mode.

[0018] Figure 5 This is a flowchart for explaining image quality improvement processing according to one embodiment.

[0019] Figure 6 This is a flowchart for explaining image processing mode switching processing according to one embodiment.

[0020] Figure 7 It is a diagram showing the overall configuration of an X-ray fluoroscopy apparatus according to a modified example.

[0021] Figure 8This is a flowchart for explaining the image quality improvement process according to a modification.

[0022] Figure 9 This is a flowchart for explaining image processing mode switching processing according to a modification example. DETAILED DESCRIPTION

[0023] (Structure of X-ray Fluoroscopic Device)

[0024] Reference Figure 1 The structure of an X-ray fluoroscopy apparatus 100 according to one embodiment of the present invention will be described.

[0025] like Figure 1 As shown, the X-ray fluoroscopy apparatus 100 of this embodiment includes an imaging unit 1, a computer 2, a display unit 3, an input receiving unit 4, a storage unit 5, a top plate 6, and an apparatus control unit 8. In this embodiment, the X-ray fluoroscopy apparatus 100 performs imaging of a subject 90 as a subject. The X-ray fluoroscopy apparatus 100 is used, for example, to perform imaging of a subject 90 using a device 80 (see Figure 2 ) is used in a method for treating a stenotic site in a blood vessel of a subject 90. The device 80 includes at least one of a catheter, a stent, and a guide wire that is introduced into a blood vessel of the subject 90.

[0026] The imaging unit 1 includes an X-ray source 1 a , an X-ray detector 1 b , and an arm 1 c arranged so that the X-ray source 1 a and the X-ray detector 1 b face each other.

[0027] The X-ray source 1a is configured to irradiate a subject with X-rays. Specifically, the X-ray source 1a irradiates the subject by applying a voltage to the source 1a via a driver (not shown). The X-ray source 1a includes a collimator that adjusts the irradiation field, which is the range of the X-rays. In this embodiment, the X-ray source 1a is mounted on the front end of one side of the arm 1c.

[0028] The X-ray detector 1b is configured to detect X-rays emitted from the X-ray source 1a. In this embodiment, the X-ray detector 1b is mounted on the other end of the arm 1c. Specifically, the X-ray detector 1b is located on the side opposite the X-ray source 1a, separated by the top plate 6. Furthermore, the X-ray detector 1b is configured to detect X-rays. For example, the X-ray detector 1b is an FPD (Flat Panel Detector). The X-ray detector 1b is configured to detect X-rays transmitted through the subject and output a detection signal based on the detected X-rays.

[0029] The computer 2 is configured to include a first processor 2a, a second processor 2b, a ROM (Read Only Memory) and a RAM (Random Access Memory), wherein the first processor 2a is a CPU (Central Processing Unit), a GPU (Graphics Processing Unit) or an FPGA (Field-Programmable Gate Array) configured for image processing, etc., and the second processor 2b is a CPU, a GPU or an FPGA configured for image processing, etc.

[0030] like Figure 1 As shown, the first processor 2a includes an X-ray image acquisition unit 20 and an image quality improvement processing unit 22. In addition, in the present embodiment, the first processor 2a also includes a learning and recognition result utilization switching unit 23. In addition, in the present embodiment, the first processor 2a also includes a display control unit 24. The X-ray image acquisition unit 20, the image quality improvement processing unit 22, the learning and recognition result utilization switching unit 23, and the display control unit 24 are configured in software as functional blocks implemented by the first processor 2a executing various programs. The X-ray image acquisition unit 20, the image quality improvement processing unit 22, the learning and recognition result utilization switching unit 23, and the display control unit 24 can also be configured in hardware by providing a dedicated processor (processing circuit).

[0031] The second processor 2b is provided separately from the first processor 2a. The second processor 2b includes an object distribution learning and recognition unit 21. The object distribution learning and recognition unit 21 is implemented as software as a functional block implemented by the second processor 2b executing various programs. The object distribution learning and recognition unit 21 can also be implemented as hardware using a dedicated processing circuit.

[0032] The X-ray image acquisition unit 20 is configured to acquire the X-ray image 10 captured by the imaging unit 1. In this embodiment, the X-ray image acquisition unit 20 is configured to acquire the X-ray image 10 as a moving image. In other words, the X-ray image acquisition unit 20 is configured to acquire the X-ray image 10 frame by frame.

[0033] The object distribution learning and recognition unit 21 is configured to output the distribution of objects captured in the X-ray image 10 using the learned learning model 7. In this embodiment, the object includes at least one of the device 80, blood vessels, and bones captured in the X-ray image 10. That is, in this embodiment, the object includes at least one of a stent, a guide wire, a catheter, a blood vessel, and a bone. In addition, in this embodiment, an example in which the object is the device 80 is used for explanation. In addition, the distribution of the object is the position information of the device 80. The learning model 7 is pre-generated by learning to recognize the device 80 captured in the X-ray image 10. In addition, the learning model 7 is stored in the storage unit 5.

[0034] In addition, the image quality improvement processing unit 22 is configured to perform image quality improvement processing to improve the image quality of the X-ray image 10. In addition, the learning recognition result utilization switching unit 23 is configured to switch whether to use the learning recognition result 7a of the device 80 output by the object distribution learning recognition unit 21. In addition, the display control unit 24 is configured to display the image processing mode together with the X-ray image 10. The detailed structure of the object distribution learning recognition unit 21, the image quality improvement processing unit 22, the learning recognition result utilization switching unit 23, and the display control unit 24 will be described later. In addition, the learning recognition result 7a of the device 80 includes the position information of the device 80 captured in the X-ray image 10.

[0035] The display unit 3 is configured to display the X-ray image 10. In this embodiment, the display unit 3 is configured to display the X-ray image 10a (see FIG. 1 ) as a moving image after image quality improvement processing. Figure 3 ) or the enhanced image 11a as a moving image after image quality improvement processing (refer to Figure 2 The display unit 3 is a monitor included in the X-ray fluoroscopy apparatus 100 .

[0036] The input receiving unit 4 is configured to receive an operation input from an operator. The input receiving unit 4 includes, for example, an input device such as a mouse and a keyboard.

[0037] The storage unit 5 is configured to store the X-ray image 10 acquired by the X-ray image acquisition unit 20, the X-ray image 10a after image quality improvement processing, and the enhanced image 11 obtained by enhancing the device 80 in the X-ray image 10 (see FIG. Figure 2 ), enhanced image 11a after image quality improvement processing, etc. Furthermore, the storage unit 5 is configured to store various programs executed by the first processor 2a and the second processor 2b. The storage unit 5 includes a non-volatile memory such as an HDD (Hard Disk Drive) or an SSD (Solid State Drive).

[0038] like Figure 1 As shown, the top plate 6 is formed into a rectangular flat plate in a plan view. The subject 90 is placed on the top plate 6 such that the head and feet of the subject 90 are along the long sides of the rectangle and the left and right directions of the subject 90 are along the short sides of the rectangle.

[0039] A moving mechanism (not shown) is provided on the top plate 6. The X-ray fluoroscopy apparatus 100 can image a subject while changing the relative position between the top plate 6 and the imaging unit 1 by moving the top plate 6 in the longitudinal direction using the moving mechanism.

[0040] The device control unit 8 is configured to control the X-ray fluoroscopy apparatus 100. Specifically, the device control unit 8 is configured to control the X-ray source 1a, the arm 1c, the top plate 6, and the like. Furthermore, the device control unit 8 is configured to control the X-ray source 1a in response to input from an operator, thereby controlling the dose of X-rays output from the X-ray source 1a.

[0041] like Figure 1 As shown, a surgeon (doctor, technician, etc.) administers a contrast agent to a subject 90 placed on a tabletop 6 and captures a plurality of X-ray images 10 while changing the relative position of the imaging unit 1 and the tabletop 6. Furthermore, the surgeon captures X-ray images 10 while moving the device 80, which has been introduced into a blood vessel of the subject 90, to a predetermined position. In this embodiment, the surgeon introduces the device 80 into the cardiac blood vessels of the subject 90. To reduce the radiation dose to the subject 90, it is preferable to reduce the dose of X-rays emitted from the X-ray source 1a.

[0042] However, if the X-ray dose is reduced, then Figure 2 As shown in FIG. 1 , noise is generated in the X-ray image 10. When noise is generated in the X-ray image 10, the visual recognition of the device 80 in the X-ray image 10 is reduced. Therefore, it may be difficult for the operator to perform surgery using the device 80. Figure 2 In the example shown, noise in the X-ray image 10 is represented graphically by hatching.

[0043] (First image processing mode)

[0044] Therefore, in this embodiment, if Figure 2 As shown, the image quality improvement processing unit 22 is configured to improve the visibility of the device 80 by performing image quality improvement processing to improve the image quality of the X-ray image 10. In this embodiment, the image quality improvement processing includes at least noise reduction processing.

[0045] In addition, if Figure 2As shown, a human body structure 90a of the subject 90 may be included in the X-ray image 10. In this case, it may be difficult to distinguish between the device 80 and the human body structure 90a. Furthermore, the human body structure 90a may be, for example, a blood vessel of the heart other than the vessel into which the device 80 is to be introduced. Furthermore, the human body structure 90a includes not only blood vessels but also the edges of organs such as the heart and the edge of the diaphragm.

[0046] Therefore, the image quality improvement processing unit 22 is configured to generate an enhanced image 11 by performing an enhancement process on the device 80 in the X-ray image 10 in the first image processing mode. Furthermore, in this embodiment, the image quality improvement processing unit 22 uses the learned recognition result 7a to distinguish the device 80 from the background portion that is not the device 80 when performing the enhancement process. The image quality improvement processing unit 22 then performs a process to increase the pixel value of the device 80 to enhance the device 80. Furthermore, in this specification, the increase in the pixel value of the device 80 is represented by varying the thickness of the device 80 when depicted. That is, as the pixel value of the device 80 increases, the device 80 is depicted in a thicker manner.

[0047] Furthermore, in the present embodiment, the image quality improvement processing unit 22 is configured to perform noise reduction processing on the enhanced image 11 in the first image processing mode using the learned recognition result 7 a .

[0048] In the first image processing mode, the image quality improvement processing unit 22 generates an enhanced image 11a after image quality improvement by performing noise reduction processing on the enhanced image 11. Furthermore, in this embodiment, the image quality improvement processing unit 22 is configured to perform at least a recursive filtering process that adds the pixel values ​​of predetermined pixels in each frame of the X-ray image 10 as noise reduction processing. In the recursive filtering process, the pixel values ​​of predetermined pixels in each frame of the X-ray image 10 are weighted and then added. However, in a surgical procedure involving the insertion of a device 80 into a cardiac blood vessel, for example, as in this embodiment, the position of the device 80 in each frame is different.

[0049] Therefore, in the present embodiment, when processing using recursive filtering is performed, the position of the device 80 is aligned for each frame, and the position of the pixels to be added is aligned.

[0050] When the learning recognition result 7a has sufficient accuracy, the position alignment accuracy of the device 80 in each frame is high, so the enhanced image 11a after the image quality improvement process is similar to the X-ray image 10a after the image quality improvement process in the second image processing mode described later (see Figure 3), the visual recognition of the device 80 is high. Therefore, in this embodiment, the first image processing mode is set to the standard image processing mode.

[0051] (Second image processing mode)

[0052] Here, if the accuracy of the device 80 recognition result obtained based on the learned recognition result 7a is low, blood vessels, etc. other than the device 80, may be mistakenly detected as the device 80 in the X-ray image 10. In this case, blood vessels, etc. located at a different position from the device 80 may be emphasized in some frames of the X-ray image 10, which is a moving image. Furthermore, if the accuracy of the device 80 recognition result obtained based on the learned recognition result 7a is low, the accuracy of the device 80 position alignment will also be reduced when noise reduction processing using recursive filtering is performed on the X-ray image 10, which is a moving image. In this case, the processing using recursive filtering results in a shift in the position of the pixels to be added, resulting in an afterimage of the device 80. Consequently, the visual recognition of the device 80 in the X-ray image 10, which is a moving image, is reduced.

[0053] Therefore, in this embodiment, if Figure 3 As shown, the image quality improvement processing unit 22 is configured to perform noise reduction processing on the X-ray image 10 that has not undergone emphasis processing in the second image processing mode that does not use the learned recognition result 7a. In other words, in the second image processing mode, the image quality improvement processing unit 22 performs noise reduction processing on the X-ray image 10 acquired by the X-ray image acquisition unit 20. Specifically, in the second image processing mode, the image quality improvement processing unit 22 performs positional alignment of the device 80 in each frame through pattern matching. Thus, in the process using recursive filtering, the generation of residual images is suppressed while reducing noise. In addition, in the case of low X-ray dose, even if image quality improvement processing is performed in the second image processing mode, the contrast of the X-ray image 10 after image quality improvement processing may be low. In this case, it is conceivable that the operator performs an operation input to increase the X-ray dose. Therefore, the device control unit 8 is configured to be able to accept an operation input to increase the X-ray dose irradiated from the X-ray source 1a.

[0054] (Switching image processing mode)

[0055] Some doctors may not tolerate even a slight decrease in the visibility of the device 80 due to the accuracy of the learned recognition result 7a. Furthermore, some doctors may prefer to display the X-ray image 10a obtained by performing image quality improvement processing in the second image processing mode from the outset.

[0056] Therefore, in the present embodiment, the image quality improvement processing unit 22 is configured to be able to switch between a first image processing mode and a second image processing mode, wherein in the first image processing mode, the image quality improvement processing is performed on the X-ray image 10 using the learning recognition result 7a, and in the second image processing mode, the image quality improvement processing is performed on the X-ray image 10 without using the learning recognition result 7a. Specifically, the image quality improvement processing unit 22 is configured to switch between the first image processing mode and the second image processing mode by switching the image processing mode using the learning recognition result switching unit 23. In the present embodiment, the learning recognition result switching unit 23 is configured to switch between the first image processing mode and the second image processing mode based on the image processing mode of the input receiving unit 4 (refer to Figure 1 ) accepts input to switch between the first image processing mode and the second image processing mode.

[0057] Specifically, if Figure 4 As shown, the learning recognition result utilizing switching unit 23 is configured to switch between the first image processing mode and the second image processing mode based on an operation input inputted by a doctor or the like by operating a switching button 4a via the input receiving unit 4. The switching button 4a is a button on a GUI (Graphical User Interface) displayed on the display unit 3 along with the X-ray image 10a after image quality improvement processing or the enhanced image 11a after image quality improvement processing.

[0058] Furthermore, in this embodiment, the object distribution learning and recognition unit 21 is configured to perform object recognition (device 80) regardless of whether the unit is in the first image processing mode or the second image processing mode. Specifically, although the learning and recognition results 7a are not used in the second image processing mode, the object distribution learning and recognition unit 21 continuously outputs the learning and recognition results 7a even in the second image processing mode.

[0059] (Image processing mode display)

[0060] like Figure 4 As shown, in this embodiment, the display control unit 24 is configured to display the message that the second image processing mode is in progress on the display unit 3 together with the X-ray image 10 at least during the execution of the second image processing mode. Specifically, when the X-ray image 10a or the X-ray image 10b generated in the second image processing mode is displayed on the display unit 3, the display control unit 24 is configured to display the message 30 that the second image processing mode is in progress together with the X-ray image 10a or the X-ray image 10b.

[0061] In addition, in this embodiment, if Figure 4As shown in FIG. 1 , the first image processing mode is set to the standard image processing mode. Therefore, the display control unit 24 is configured not to display the fact that the first image processing mode is in progress on the display unit 3 during the execution of the first image processing mode.

[0062] (Switching of shooting position or shooting conditions)

[0063] The X-ray fluoroscopy apparatus 100 is configured to switch imaging positions and imaging conditions by accepting an operation input from a doctor or the like via the input accepting unit 4. The imaging conditions include, for example, the dose of X-rays output from the X-ray source 1a and the arrangement of the X-ray source 1a and the X-ray detector 1b via the arm 1c.

[0064] Here, the doctor or the like intentionally switches between the first image processing mode and the second image processing mode by operating the switch button 4a. For example, if the doctor or the like changes at least one of the imaging position and imaging conditions after switching between the first and second image processing modes, if the image processing mode is restored to the pre-switching state, the doctor or the like will need to switch back to the second image processing mode, which would require unnecessary operations.

[0065] Therefore, if Figure 4 As shown, in this embodiment, the learning recognition result switching unit 23 is configured to maintain the first image processing mode or the second image processing mode being executed even when at least one of the imaging site and the imaging conditions is changed.

[0066] exist Figure 4 The example shown shows a case where a doctor or the like switches imaging conditions. Specifically, Figure 4 The example shown shows a case where a doctor or the like changes the dose of X-rays. Figure 4 , an example is shown in the figure where the X-ray dose is lower under the second imaging condition than under the first imaging condition. Under the second imaging condition, since the X-ray dose is lower, the pixel values ​​of the device 80 and the human body structure 90a captured in the X-ray image 10b after image quality improvement processing obtained under the second imaging condition are smaller than the pixel values ​​of the device 80 and the human body structure 90a captured in the X-ray image 10a after image quality improvement processing obtained under the first imaging condition. Figure 4 In the figure, the device 80 and the human body structure 90a captured in the X-ray image 10b captured according to the second imaging condition are illustrated in greater detail, thereby showing that the pixel values ​​of the device 80 and the human body structure 90a captured in the X-ray image 10a captured according to the first imaging condition are small.

[0067] In addition, the device 80 and the human body structure 90a captured in the enhanced image 11b obtained after the image quality is improved according to the second photographic condition are also illustrated in a thinner manner than the device 80 and the human body structure 90a captured in the enhanced image 11a obtained after the image quality is improved according to the first photographic condition, thereby showing that the pixel value of the device 80 and the human body structure 90a captured in the enhanced image 11b obtained after the image quality is improved according to the second photographic condition is smaller.

[0068] Next, refer to Figure 5 The image quality improvement process of the X-ray image 10 performed by the X-ray fluoroscopy apparatus 100 will be described. The X-ray fluoroscopy apparatus 100 performs the image quality improvement process of the X-ray image 10 every time each frame of the X-ray image 10 is acquired.

[0069] In step 101, the X-ray image acquisition unit 20 acquires the X-ray image 10. Specifically, the X-ray image acquisition unit 20 acquires the X-ray image 10 as a moving image for each frame.

[0070] In step 102 , the learning recognition result is used to obtain the image processing mode currently being executed by the switching unit 23 .

[0071] In step 103, the learning recognition result is used to determine whether the image processing mode currently being executed is the first image processing mode using the switching unit 23. If the image processing mode currently being executed is the first image processing mode, the process proceeds to step 104. If the image processing mode currently being executed is the second image processing mode, the process proceeds to step 107.

[0072] In step 104 , the image quality improvement processing unit 22 uses the learned recognition result 7 a to emphasize the device 80 in the X-ray image 10 , thereby acquiring an emphasized image 11 .

[0073] In step 105, the image quality improvement processing unit 22 performs image quality improvement processing on the enhanced image 11. Specifically, the image quality improvement processing unit 22 performs noise reduction processing on the enhanced image 11 to obtain an enhanced image 11a after the image quality improvement processing.

[0074] In step 106 , the display control unit 24 displays the enhanced image 11 a after the image quality improvement process on the display unit 3 .

[0075] When the process proceeds from step 102 to step 107, in step 107, the image quality improvement processing unit 22 performs image quality improvement processing on the X-ray image 10. Specifically, the image quality improvement processing unit 22 performs noise reduction processing on the X-ray image 10 to obtain an X-ray image 10a after image quality improvement processing.

[0076] In step 108 , the display control unit 24 displays the X-ray image 10 a after the image quality improvement process on the display unit 3 .

[0077] In step 109, the display control unit 24 displays the second image processing mode on the display unit 3. Specifically, the display control unit 24 displays the message 30 indicating the second image processing mode together with the X-ray image 10a after image quality improvement processing on the display unit 3. The processing then ends.

[0078] Next, refer to Figure 6 The following describes a process of switching between the first image processing mode and the second image processing mode by the X-ray fluoroscopy apparatus 100. When the operator operates the switching button 4a using the input receiving unit 4, the following process starts.

[0079] In step 201 , the learning recognition result is used to obtain an image processing mode by the switching unit 23 .

[0080] In step 202, the learning recognition result switching unit 23 determines whether the first image processing mode is being executed. If the first image processing mode is not being executed, the process proceeds to step 203. If the first image processing mode is being executed, the process proceeds to step 204.

[0081] In step 203, the image processing mode is switched to the first image processing mode using the learning recognition result switching unit 23. Thereafter, the process ends.

[0082] When the process proceeds from step 202 to step 204, in step 204, the image processing mode is switched to the second image processing mode using the learning recognition result switching unit 23. Thereafter, the process ends.

[0083] (Effects of this embodiment)

[0084] In this embodiment, the following effects can be obtained.

[0085] In the present embodiment, as described above, the X-ray fluoroscopy apparatus 100 includes: an imaging unit 1 including an X-ray source 1a for irradiating an object (subject 90) with X-rays and an X-ray detector 1b for detecting the X-rays irradiated from the X-ray source 1a; an X-ray image acquisition unit 20 for acquiring the X-ray image 10 captured by the imaging unit 1; an object distribution learning and recognition unit 21 for outputting the distribution of objects captured in the X-ray image 10 (position information of the device 80) using the learned learning model 7; an image quality improvement processing unit 22 for performing image quality improvement processing to improve the image quality of the X-ray image 10; and a display unit 3 for displaying the X-ray image 10, wherein the image quality improvement processing unit 22 is configured to be able to switch between a first image processing mode in which the image quality improvement processing is performed on the X-ray image 10 using the learned recognition result 7a and a second image processing mode in which the image quality improvement processing is performed on the X-ray image 10 without using the learned recognition result 7a.

[0086] Thus, depending on whether the accuracy of the learned recognition result 7a is high or low, it is possible to switch between a first image processing mode that uses the learned recognition result 7a and a second image processing mode that does not use the learned recognition result 7a. Specifically, when the accuracy of the learned recognition result 7a is high, the visual recognizability of the object (device 80) can be improved using the first image processing mode, while when the accuracy of the learned recognition result 7a is low, the image quality of the X-ray image 10 can be improved using the second image processing mode. This can suppress any reduction in the visual recognizability of the object due to the learned recognition result 7a. Consequently, when the learned model 7 is used to identify the structure of the object in the X-ray image 10, any reduction in the visual recognizability of the object due to the learned recognition result 7a output by the learning model 7 can be suppressed.

[0087] Furthermore, in the above-described embodiment, by configuring as follows, further effects as described below can be obtained.

[0088] Specifically, in this embodiment, as described above, image quality improvement processing includes at least noise reduction processing. The image quality improvement processing unit 22 is configured to perform noise reduction processing on an enhanced image 11 obtained by performing an enhancement process on the object (device 80) in the X-ray image 10 in the first image processing mode using the learned recognition result 7a, and to perform noise reduction processing on the unenhanced X-ray image 10 in the second image processing mode not using the learned recognition result 7a. Thus, by performing noise reduction processing in the first image processing mode in addition to the enhancement process on the device 80, the visibility of the device 80 can be further improved compared to an image obtained by performing image quality improvement processing in the second image processing mode not using the learned recognition result 7a. Furthermore, even in the second image processing mode, where the learned recognition result 7a is not used due to its low accuracy, noise reduction processing can be performed to improve the visibility of the device 80 compared to an X-ray image 10 that has not undergone noise reduction processing. As a result, it is possible to suppress a decrease in the visibility of the device 80 due to the learned recognition result 7 a and improve the visibility of the device 80 compared to the X-ray image 10 on which the noise reduction process is not performed.

[0089] Furthermore, as described above, this embodiment further includes a display control unit configured to display the information that the second image processing mode is in progress on the display unit 3 together with the X-ray image 10 at least during the execution of the second image processing mode, and not display the information that the first image processing mode is in progress on the display unit 3 during the execution of the first image processing mode. Thus, since the information that the second image processing mode is in progress is displayed on the display unit 3 only during the execution of the second image processing mode, the operator can easily grasp the state of the learning recognition result 7a without using the learning model 7.

[0090] Furthermore, as described above, this embodiment further includes a learning and recognition result utilization switching unit 23 that switches whether to use the learning and recognition result 7a from the object distribution learning and recognition unit 21. Thus, by switching whether to use the learning and recognition result 7a by the learning and recognition result utilization switching unit 23, switching between the first image processing mode and the second image processing mode can be easily performed.

[0091] Furthermore, in this embodiment, as described above, the learned recognition result utilizing switching unit 23 is configured to maintain the first or second image processing mode in execution even if at least one of the imaging site or imaging conditions changes. This maintains the currently executed image processing mode, thereby preventing the image processing mode intentionally changed by the operator from reverting to the pre-change image processing mode even if at least one of the imaging site or imaging conditions changes. Consequently, compared to a configuration that returns the image processing mode to the standard image processing mode when at least one of the imaging site or imaging conditions changes, unnecessary operator operations can be prevented, thereby improving operator convenience (usability).

[0092] Furthermore, in this embodiment, as described above, an input accepting unit 4 is further provided for accepting an operator's input, and the learning recognition result-using switching unit 23 is configured to switch between the first image processing mode and the second image processing mode based on the input accepted by the input accepting unit 4. This allows the operator to switch between the first image processing mode and the second image processing mode at any time, thereby further improving operator convenience (usability).

[0093] Furthermore, as described above, this embodiment includes a first processor 2a including an X-ray image acquisition unit 20, an image quality improvement processing unit 22, and a learned recognition result utilization switching unit 23; and a second processor 2b, provided separately from the first processor 2a and including an object distribution learned recognition unit 21. The X-ray image acquisition unit 20 is configured to acquire an X-ray image 10 as a moving image, and the object distribution learned recognition unit 21 is configured to perform processing for recognizing the device 80 regardless of whether the device is in the first image processing mode or the second image processing mode. Thus, the processing for acquiring the learned recognition result 7a of the device 80 and the processing for improving the image quality of the X-ray image 10 are performed by separate processors. Therefore, unlike a configuration in which the X-ray image acquisition unit 20, the image quality improvement processing unit 22, the learned recognition result utilization switching unit 23, and the object distribution learned recognition unit 21 are included in a single processor, an increase in the processing load on the processor can be suppressed regardless of whether the object (device 80) recognition processing is performed in the first image processing mode or the second image processing mode. Furthermore, the object distribution learning and recognition unit 21 included in the second processor 2b allows for continuous acquisition of the learning and recognition results 7a even during the execution of the second image processing mode. Therefore, compared to a configuration in which the learning and recognition results 7a are not acquired during the execution of the second image processing mode, even when switching from the second image processing mode to the first image processing mode, the increase in the time required for switching the image quality improvement process can be suppressed. Consequently, the present invention is particularly effective when applied to X-ray images 10 that are moving images.

[0094] Furthermore, in this embodiment, as described above, the image quality improvement processing unit 22 is configured to perform, as noise reduction processing, at least recursive filtering, which adds the pixel values ​​of predetermined pixels in each frame of the X-ray image 10. Consequently, in the first image processing mode, the learned recognition result 7a of the device 80 can be used when performing processing using recursive filtering. As a result, the position of the device 80 can be accurately acquired in each image, thereby reducing residual images of the device 80 caused by the recursive filtering process. Furthermore, in the second image processing mode, the use of the device 80 captured in a position where it does not actually exist, which is caused by the learned recognition result 7a, can be suppressed during the recursive filtering process, thereby enhancing the noise reduction effect achieved by the recursive filtering process.

[0095] Furthermore, in this embodiment, as described above, the object includes at least one of the stent, guidewire, catheter, blood vessel, and bone captured in the X-ray image 10. Thus, image quality enhancement processing can be performed on the operator's desired object among the stent, guidewire, catheter, blood vessel, and bone captured in the X-ray image 10 in either the first or second image processing mode. As a result, image quality enhancement processing can be performed on the desired object in the desired image processing mode, thereby improving operator convenience (usability).

[0096] (Variation)

[0097] The embodiments disclosed herein are illustrative in all respects and are not restrictive. The scope of the present invention is not defined by the description of the embodiments described above, but by the claims, and includes all modifications (variations) within the meaning and scope equivalent to the claims.

[0098] For example, in the above embodiment, the learning recognition result switching unit 23 switches whether to use the learning recognition result 7a based on the input of the input receiving unit 4, but the present invention is not limited to this. In the present invention, for example, Figure 7 As in the illustrated modification, the learned recognition result utilization switching unit 120 may be configured to switch whether to utilize the learned recognition result 7 a based on the learned recognition result 7 a .

[0099] Figure 7 The X-ray fluoroscopy apparatus 200 of the illustrated modified example differs from the X-ray fluoroscopy apparatus 100 of the aforementioned embodiment in that it includes a computer 12 in place of the computer 2. The computer 12 differs from the computer 2 of the aforementioned embodiment in that it includes a first processor 12a in place of the first processor 2a. Furthermore, the first processor 12a differs from the first processor 2a of the aforementioned embodiment in that it includes a learned recognition result use switching unit 120 in place of the learned recognition result use switching unit 23.

[0100] In a modified example, the learning recognition result utilizing the switching unit 120 is configured to switch between the first image processing mode and the second image processing mode based on the learning recognition result 7a output from the learning model 7. Specifically, the learning recognition result utilizing the switching unit 120 is configured to switch between the first image processing mode and the second image processing mode based on the accuracy of the learning recognition result 7a. More specifically, the learning recognition result utilizing the switching unit 120 is configured to set the first image processing mode when the accuracy of the learning recognition result 7a is above a threshold value, and to set the second image processing mode when the accuracy of the learning recognition result 7a is less than the threshold value. In addition, in this embodiment, the accuracy of the learning recognition result 7a includes a numerical value indicating the accuracy of the output result output by the learning model 7 together with the learning recognition result 7a.

[0101] Next, refer to Figure 8 The image quality improvement process of the modified example will be described. The same processes as those of the image quality improvement process of the above embodiment are denoted by the same reference numerals, and detailed descriptions thereof will be omitted.

[0102] In step 101 , the X-ray image acquisition unit 20 acquires the X-ray image 10 .

[0103] In step 110, the learned recognition result switching unit 120 acquires the learned recognition result 7a. Specifically, the learned recognition result switching unit 120 acquires the accuracy of the learned recognition result 7a.

[0104] In step 111, the learned recognition result switching unit 120 determines whether the accuracy of the learned recognition result 7a is greater than a predetermined threshold. If the accuracy of the learned recognition result 7a is greater than the predetermined threshold, the process proceeds to steps 104 through 106 and then ends. If the accuracy of the learned recognition result 7a is less than the predetermined threshold, the process proceeds to steps 107 through 109 and then ends.

[0105] Next, refer to Figure 9 The image processing mode switching process of the modified example will be described. The same processes as those of the image processing mode switching process of the above embodiment are denoted by the same reference numerals, and detailed descriptions thereof will be omitted.

[0106] In step 206, the learned recognition result switching unit 120 acquires the learned recognition result 7a. Specifically, the learned recognition result switching unit 120 acquires the accuracy of the learned recognition result 7a.

[0107] In step 207, the learned recognition result switching unit 120 determines whether the accuracy of the learned recognition result 7a is greater than a predetermined threshold. If the accuracy of the learned recognition result 7a is greater than the predetermined threshold, the process proceeds to steps 202 and 203 before terminating. If the accuracy of the learned recognition result 7a is less than the predetermined threshold, the process proceeds to step 204 before terminating.

[0108] In the modified example, as described above, the learned recognition result-utilizing switching unit 120 is configured to switch between the first image processing mode and the second image processing mode based on the learned recognition result 7a output from the learning model 7. Thus, by switching between the first and second image processing modes based on the learned recognition result 7a, even if, for example, the operator (doctor or technician) has low proficiency and is unable to grasp the occurrence of a false detection of the device 80, the first and second image processing modes can still be switched. Consequently, regardless of the user's proficiency, it is possible to suppress a reduction in the visual recognition of the device 80 due to the learned recognition result 7a.

[0109] Furthermore, in the above embodiment, an example of a configuration is shown in which the image quality improvement processing unit 22 performs enhancement processing on the X-ray image 10 to obtain the enhanced image 11. However, the present invention is not limited thereto. For example, the present invention may also be configured such that the enhanced image 11 is obtained by causing a learning model to learn to identify the device 80 and to enhance the identified device 80.

[0110] Furthermore, in the above embodiment, the display control unit 24 displays the information indicating that the second image processing mode is in progress while the second image processing mode is in progress, and does not display the information indicating that the first image processing mode is in progress while the first image processing mode is in progress. However, the present invention is not limited thereto. For example, the display control unit may also be configured to display the information indicating that the first image processing mode is in progress while the first image processing mode is in progress.

[0111] In the above embodiment, an example configuration in which the X-ray fluoroscopy apparatus 100 includes a first processor 2a and a second processor 2b is shown, but the present invention is not limited thereto. For example, the X-ray fluoroscopy apparatus may also include only a single processor. However, if the X-ray fluoroscopy apparatus includes only a single processor, the processing load on the processor increases due to the recognition processing of the device 80 by the object distribution learning and recognition unit. Therefore, the X-ray fluoroscopy apparatus is preferably configured to include a first processor 2a and a second processor 2b.

[0112] Furthermore, in the above embodiment, an example configuration is shown in which the object distribution learning and recognition unit 21 performs recognition processing for the device 80 captured in the X-ray image 10 regardless of whether the first image processing mode or the second image processing mode is used. However, the present invention is not limited thereto. For example, the object distribution learning and recognition unit may be configured to perform recognition processing for the device 80 captured in the X-ray image 10 only during the execution of the first image processing mode. In other words, the object distribution learning and recognition unit may be configured not to perform recognition processing for the device 80 captured in the X-ray image 10 during the execution of the second image processing mode.

[0113] Furthermore, in the above embodiment, the image quality improvement processing unit 22 performs noise reduction as an example of image quality improvement processing, but the present invention is not limited thereto. For example, the image quality improvement processing unit may be configured to perform processing to enhance the contrast of the X-ray image 10 in addition to noise reduction processing. The image quality improvement processing unit may perform any processing as long as it improves the visibility of the device 80 captured in the X-ray image 10.

[0114] Furthermore, in the above embodiment, an example configuration is shown in which the image quality improvement processing unit 22 performs processing using recursive filtering as noise reduction processing, but the present invention is not limited thereto. For example, the image quality improvement processing unit may be configured to perform processing other than recursive filtering, such as processing using low-pass filtering, as noise reduction processing. The image quality improvement processing unit may be configured to perform any processing as long as it can reduce the noise of the X-ray image 10.

[0115] Furthermore, in the above embodiment, an example of a configuration is shown in which the learning and recognition result utilization switching unit maintains the first or second image processing mode in execution even when imaging conditions change. However, the present invention is not limited thereto. For example, the learning and recognition result utilization switching unit may be configured to maintain the image processing mode in execution even when the imaging site changes.

[0116] Furthermore, in the above embodiment, an example of a configuration is shown in which the learning and recognition result utilization switching unit maintains the first image processing mode or the second image processing mode being executed even when the X-ray dose, which is an imaging condition, is changed. However, the present invention is not limited to this configuration. For example, the learning and recognition result utilization switching unit may be configured to maintain the first image processing mode or the second image processing mode being executed even when imaging conditions other than the X-ray dose, such as a change in the angle of the arm 1c, are changed.

[0117] Furthermore, in the above embodiment, an example of a configuration is shown in which the learning and recognition result utilization switching unit maintains the first or second image processing mode in execution even when the imaging site or imaging conditions change. However, the present invention is not limited thereto. For example, the learning and recognition result utilization switching unit may be configured to switch to the standard image processing mode when the imaging site or imaging conditions change.

[0118] In the above embodiment, the X-ray fluoroscopy apparatus 100 is shown as including the learning recognition result switching unit 23 , but the present invention is not limited thereto.

[0119] Furthermore, in the above embodiment, an example configuration is shown in which the X-ray image acquisition unit 20 acquires the X-ray image 10 as a moving image, and the image quality improvement processing unit 22 performs image quality improvement processing on the X-ray image 10 as a moving image in either the first image processing mode or the second image processing mode. However, the present invention is not limited thereto. For example, a configuration may also be employed in which the X-ray image acquisition unit acquires the X-ray image as a still image, and the image quality improvement processing unit performs image quality improvement processing on the X-ray image as a still image in either the first image processing mode or the second image processing mode.

[0120] In the above embodiment, the switch button 4a is displayed as a button on the GUI on the display unit 3, but the present invention is not limited thereto. For example, the switch button 4a may be included in the input receiving unit as a physical button.

[0121] Furthermore, in the above embodiment, an example of a configuration is shown in which the image quality improvement processing unit 22 targets the device 80 captured in the X-ray image 10 for image quality improvement. However, the present invention is not limited thereto. For example, the image quality improvement processing unit 22 may be configured to target blood vessels or bones captured in the X-ray image 10 for image quality improvement.

[0122] Furthermore, in the above embodiment, the learning model 7 is configured to recognize at least one of a catheter, a stent, and a guidewire as the device 80, but the present invention is not limited thereto. For example, the learning model may also be configured to recognize a coil used in the treatment of an aneurysm, etc., as a device. The device to be recognized by the learning model may be any device, as long as it is used while being imaged by an X-ray fluoroscopy device.

[0123] [Implementation Method]

[0124] Those skilled in the art will appreciate that the above-described exemplary embodiments are specific examples of the following aspects.

[0125] (Item 1)

[0126] An X-ray fluoroscopy device, comprising:

[0127] an imaging unit including an X-ray source for irradiating an object with X-rays and an X-ray detector for detecting the X-rays irradiated from the X-ray source;

[0128] an X-ray image acquiring unit configured to acquire the X-ray image captured by the imaging unit;

[0129] an object distribution learning and recognition unit that uses the learned learning model to output a distribution of objects captured in the X-ray image;

[0130] an image quality improvement processing unit that performs image quality improvement processing to improve the image quality of the X-ray image; and

[0131] a display unit that displays the X-ray image,

[0132] In which, the image quality improvement processing unit is configured to be able to switch between a first image processing mode and a second image processing mode. In the first image processing mode, the image quality improvement processing is performed on the X-ray image using the learning and recognition results of the object distribution learning and recognition unit. In the second image processing mode, the image quality improvement processing is performed on the X-ray image without using the learning and recognition results.

[0133] (Item 2)

[0134] The X-ray fluoroscopy apparatus according to item 1, wherein:

[0135] The image quality improvement process includes at least noise reduction process,

[0136] The image quality improvement processing unit is configured to perform the noise reduction processing on the emphasized image obtained by performing the emphasis processing on the object in the X-ray image in the first image processing mode using the learned recognition result, and to perform the noise reduction processing on the X-ray image that has not performed the emphasis processing in the second image processing mode not using the learned recognition result.

[0137] (Item 3)

[0138] The X-ray fluoroscopy apparatus according to item 1 or 2, wherein:

[0139] It also includes a display control unit, which is configured to: at least during the execution of the second image processing mode, display the meaning of being in the second image processing mode together with the X-ray image on the display unit; during the execution of the first image processing mode, not display the meaning of being in the first image processing mode on the display unit.

[0140] (Item 4)

[0141] The X-ray fluoroscopy apparatus according to any one of items 1 to 3, wherein:

[0142] The system further includes a learning and recognition result use switching unit configured to switch whether to use the learning and recognition result of the object distribution learning and recognition unit.

[0143] (Item 5)

[0144] The X-ray fluoroscopy apparatus according to item 4, wherein:

[0145] The learning recognition result utilization switching unit is configured to maintain the first image processing mode or the second image processing mode being executed even when at least one of an imaging site and imaging conditions is changed.

[0146] (Item 6)

[0147] The X-ray fluoroscopy apparatus according to item 4 or 5, wherein:

[0148] It also includes an input receiving unit that receives an operation input from an operator.

[0149] The learning recognition result switching unit is configured to switch between the first image processing mode and the second image processing mode based on the input accepted by the input accepting unit.

[0150] (Item 7)

[0151] The X-ray fluoroscopy apparatus according to any one of items 4 to 6, wherein:

[0152] The learning recognition result switching unit is configured to switch between the first image processing mode and the second image processing mode based on the learning recognition result output from the learning model.

[0153] (Item 8)

[0154] The X-ray fluoroscopy apparatus according to any one of items 4 to 7, comprising:

[0155] a first processor including the X-ray image acquisition unit, the image quality improvement processing unit, and the learning recognition result utilization switching unit; and

[0156] A second processor is provided separately from the first processor and includes the object distribution learning and recognition unit.

[0157] The X-ray image acquisition unit is configured to acquire the X-ray image as a moving image.

[0158] The object distribution learning and recognition unit is configured to perform a process of recognizing the object in either the first image processing mode or the second image processing mode.

[0159] (Item 9)

[0160] The X-ray fluoroscopy apparatus according to item 2, wherein:

[0161] The image quality improvement processing unit is configured to perform, as the noise reduction processing, processing using at least recursive filtering for adding pixel values ​​of predetermined pixels in each frame of the X-ray image.

[0162] (Item 10)

[0163] The X-ray fluoroscopy apparatus according to any one of items 1 to 9, wherein:

[0164] The object includes at least any one of a stent, a guide wire, a catheter, a blood vessel, and a bone captured in the X-ray image.

[0165] Description of Reference Numerals

[0166] 1: Photographic unit; 1a: X-ray source; 1b: X-ray detector; 2a, 12a: First processor; 2b: Second processor; 3: Display unit; 4: Input receiving unit; 7: Learning model; 7a: Learning recognition result; 10, 10a, 10b: X-ray image; 11, 11a, 11b: Enhanced image; 20: X-ray image acquisition unit; 21: Object distribution learning recognition unit; 22: Image quality improvement processing unit; 23, 120: Learning recognition result utilization switching unit; 24: Display control unit; 80: Device (object); 100, 200: X-ray fluoroscopy device.

Claims

1. An X-ray fluoroscopy device comprising: an imaging unit including an X-ray source for irradiating an object with X-rays and an X-ray detector for detecting the X-rays irradiated from the X-ray source; an X-ray image acquiring unit configured to acquire the X-ray image captured by the imaging unit; an object distribution learning and recognition unit that uses the learned learning model to output a distribution of objects captured in the X-ray image; an image quality improvement processing unit that performs image quality improvement processing to improve the image quality of the X-ray image; and a display unit that displays the X-ray image, in, The image quality improvement processing unit is configured to be able to switch between a first image processing mode and a second image processing mode. In the first image processing mode, the image quality improvement processing is performed on the X-ray image using the learning and recognition results of the object distribution learning and recognition unit. In the second image processing mode, the image quality improvement processing is performed on the X-ray image without using the learning and recognition results.

2. The X-ray fluoroscopy apparatus according to claim 1, wherein: The image quality improvement process includes at least noise reduction process, The image quality improvement processing unit is configured to perform the noise reduction processing on the emphasized image obtained by performing the emphasis processing on the object in the X-ray image in the first image processing mode using the learned recognition result, and to perform the noise reduction processing on the X-ray image that has not performed the emphasis processing in the second image processing mode not using the learned recognition result.

3. The X-ray fluoroscopy apparatus according to claim 1, wherein: It also includes a display control unit, which is configured to: at least during the execution of the second image processing mode, display the meaning of being in the second image processing mode together with the X-ray image on the display unit; during the execution of the first image processing mode, not display the meaning of being in the first image processing mode on the display unit.

4. The X-ray fluoroscopy apparatus according to claim 1, wherein: The system further includes a learning and recognition result use switching unit configured to switch whether to use the learning and recognition result of the object distribution learning and recognition unit.

5. The X-ray fluoroscopy apparatus according to claim 4, wherein: The learning recognition result utilization switching unit is configured to maintain the first image processing mode or the second image processing mode being executed even when at least one of an imaging site and imaging conditions is changed.

6. The X-ray fluoroscopy apparatus according to claim 4, wherein: It also includes an input receiving unit that receives an operation input from an operator. The learning recognition result switching unit is configured to switch between the first image processing mode and the second image processing mode based on the input accepted by the input accepting unit.

7. The X-ray fluoroscopy apparatus according to claim 4, wherein: The learning recognition result switching unit is configured to switch between the first image processing mode and the second image processing mode based on the learning recognition result output from the learning model.

8. The X-ray fluoroscopy apparatus according to claim 4, wherein: have: a first processor including the X-ray image acquisition unit, the image quality improvement processing unit, and the learning recognition result utilization switching unit; as well as A second processor is provided separately from the first processor and includes the object distribution learning and recognition unit. The X-ray image acquisition unit is configured to acquire the X-ray image as a moving image. The object distribution learning and recognition unit is configured to perform a process of recognizing the object in either the first image processing mode or the second image processing mode.

9. The X-ray fluoroscopy apparatus according to claim 2, wherein: The image quality improvement processing unit is configured to perform, as the noise reduction processing, processing using at least recursive filtering for adding pixel values ​​of predetermined pixels in each frame of the X-ray image.

10. The X-ray fluoroscopy apparatus according to claim 1, wherein: The object includes at least any one of a stent, a guide wire, a catheter, a blood vessel, and a bone captured in the X-ray image.

Images