Real-time image guiding system

Abstract

The invention relates to a real-time image guiding system, and relates to the technical field of image guiding. The real-time image guiding system comprises image robots, the image robots comprise the first image robot and the second image robot, and the first image robot can dynamically adjust the posture and the position of the first image robot so that an imaging light beam emitted by the first image robot can be received by the second image robot; the second image robot can dynamically adjust the posture and the position of the second image robot, so that an imaging light beam emitted by the second image robot can be received by the first image robot, and the imaging light beam emitted by the first image robot and the imaging light beam emitted by the second image robot intersect at a specified position to obtain a three-dimensional real-time image. According to the real-time image guiding system, the problem that new errors are introduced due to the fact that a patient needs to be positioned for multiple times can be avoided; and the first image robot and the second image robot can dynamically adjust the postures and the positions, so that the treatment efficiency and the patient compliance can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image guidance, and particularly to a real-time imaging guidance system. Background Art

[0002] The CyberKnife imaging guidance system in radiotherapy equipment can clearly display the anatomical structure and tumor location in a patient's body, thereby providing accurate imaging information for subsequent treatment. It generally has a radiation source responsible for emitting X-rays and a digital imaging detector for receiving X-rays. However, since its radiation source is installed on the ceiling of the treatment room, the position of the X-ray convergence point is fixed, resulting in the treatment ray only being able to treat the tumor site near it. Tumors far from this convergence point need to reposition the patient again because they cannot be tracked in real time, so as to move the position to be treated near the convergence point. However, each repositioning of the patient not only introduces new errors and reduces the treatment accuracy, but also greatly reduces the treatment efficiency due to the need for multiple imaging verifications, affecting patient compliance. Summary of the Invention

[0003] The present invention provides a real-time imaging guidance system for solving at least one of the above technical problems.

[0004] The present invention provides a real-time imaging guidance system, comprising: An imaging robot, which includes a first imaging robot and a second imaging robot. The first imaging robot can dynamically adjust its posture and position so that the imaging beam emitted by it can be received by the second imaging robot; the second imaging robot can dynamically adjust its posture and position so that the imaging beam emitted by it can be received by the first imaging robot. The imaging beam emitted by the first imaging robot and the imaging beam emitted by the second imaging robot cross at a specified position to obtain a three-dimensional real-time image; and A treatment robot, which is used to generate a treatment beam. The treatment robot adjusts its posture and displacement according to the three-dimensional registration degree between the three-dimensional real-time image obtained by the imaging robot and the three-dimensional planned image, so that the treatment beam generated by it is aligned with the specified position.

[0005] In one embodiment, the first imaging robot includes a first imaging robotic arm, a first imaging bracket connected to the first imaging robotic arm, and a first X-ray generator and a first X-ray receiver respectively rotatably connected to the first imaging bracket; The second imaging robot includes a second imaging robotic arm, a second imaging bracket connected to the second imaging robotic arm, and a second X-ray generator and a second X-ray receiver respectively rotatably connected to the second imaging bracket; Wherein, the first imaging robotic arm and the second imaging robotic arm can synchronously adjust the distance of their abduction or adduction , the first X-ray generator or the first X-ray receiver can adjust the angle it turns relative to the first imaging bracket , the second X-ray generator or the second X-ray receiver can adjust the angle it turns relative to the second imaging bracket .

[0006] In one embodiment, the first imaging robot and / or the second imaging robot can dynamically adjust their postures and positions, so that the imaging beams emitted by the first X-ray generator and the imaging beams emitted by the second X-ray generator cross at a specified position, and the crossing area can cover the specified position. The area of the crossing area is obtained according to the following relational expression:

[0007] Wherein, A is the area of the crossing area; D is the distance between the first X-ray generator and the second X-ray receiver, or the distance between the second X-ray generator and the first X-ray receiver; α is the angle between the imaging beam emitted by the first X-ray generator and the imaging beam emitted by the second X-ray generator.

[0008] In one embodiment, the angle that the first X-ray generator or the first X-ray receiver turns relative to the first imaging bracket satisfies the following relational expression:

[0009] The angle that the second X-ray generator or the second X-ray receiver turns relative to the second imaging bracket satisfies the following relational expression: ; is the adjustment angle of the first X-ray generator or the first X-ray receiver; is the adjustment angle of the second X-ray generator or the second X-ray receiver; is the angle error between the first imaging bracket and the second imaging bracket at the current moment

[0010] is the proportionality coefficient; is the integral coefficient; is the differential coefficient. In one embodiment, the first imaging robotic arm and the second imaging robotic arm synchronously adjust the distance of their abduction or adduction according to the following relational expression : ); wherein, ΔS is the displacement between the position of the key feature point in the three-dimensional real-time image at the current moment and the position of the feature point at the previous moment; α is the included angle between the imaging beam emitted by the first X-ray generator and the imaging beam emitted by the second X-ray generator. In one embodiment, the first imaging robot and the second imaging robot extract feature points of different scales in the three-dimensional real-time image according to a pre-trained model and screen key feature points, and the first imaging robotic arm and the second imaging robotic arm synchronously adjust the distance of their abduction or adduction according to the key feature points , wherein the feature points are bony landmark points, In one embodiment, a first hinge device and a second hinge device are provided on the first imaging bracket, and the rotating joints of the first hinge device and the second hinge device are respectively connected to the first X-ray generator and the first X-ray receiver, so that the first X-ray generator and the first X-ray receiver can adjust the angles they turn relative to the first imaging bracket; A third hinge device and a fourth hinge device are provided on the second imaging bracket, and the rotating joints of the third hinge device and the fourth hinge device are respectively connected to the second X-ray generator and the second X-ray receiver, so that the second X-ray generator and the second X-ray receiver can adjust the angles they turn relative to the first imaging bracket.

[0011] In one embodiment, the first imaging robot and / or the second imaging robot adjusts its posture and position based on a real-time collision avoidance algorithm and a rapidly-exploring random tree algorithm.

[0012] In one embodiment, the treatment robot includes a treatment robotic arm and a treatment head connected to the treatment robotic arm. When the treatment head generates a treatment beam, the first X-ray generator, the first X-ray receiver, the second X-ray generator, and the second X-ray receiver are all outside the beam range of the treatment beam.

[0013] Compared with the prior art, the advantages of the present invention are that the first imaging robot and the second imaging robot can adjust their postures and positions, so as to be able to obtain three-dimensional real-time images of a specified position in real time, thereby automatically adjusting the direction of the imaging beam, avoiding the problem of introducing new errors due to multiple patient positioning; and since the first imaging robot and the second imaging robot can dynamically adjust their postures and positions, the treatment efficiency and patient compliance can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The present invention will be described in more detail below based on embodiments with reference to the drawings.

[0015] Figure 1a is a schematic perspective view of a real-time imaging guidance system in an embodiment of the present invention; Figure 1b is a top view of a real-time imaging guidance system in an embodiment of the present invention; Figure 2 is a schematic perspective view of a real-time imaging guidance system in an embodiment of the present invention, in which the treatment robot is hidden; Figure 3a and Figure 3b is a schematic perspective view of the first imaging robot in an embodiment of the present invention; Figure 3c is a schematic perspective view of the second imaging robot in an embodiment of the present invention; Figure 4 is Figure 3a a schematic perspective view of the first imaging bracket in Figure 5 and Figure 6 is Figure 4 a schematic perspective view of the first hinge device in Figure 7 is Figure 4 a schematic perspective view of the second hinge device in Figure 8 is a schematic perspective view of a real-time imaging guidance system in an embodiment of the present invention, in which the cone-beam CT scanning trajectory of the imaging robot is shown; Figure 9 and Figure 10 is Figure 8 a view showing the intersection of the first imaging beam and the second imaging beam shown; Figure 11 is Figure 1a a schematic perspective view of the treatment robot in

[0016] Reference numerals: 100, the first imaging robot; 200, the second imaging robot; 300, the treatment robot; 400, the treatment bed; 500, the patient; 110. The first imaging robotic arm; 120. The first imaging support; 130. The first X-ray generator; 140. The first X-ray receiver; 101. The first imaging light beam; 102. The intersection area; 103. The cone-beam CT scanning trajectory; 111. The first arm; 112. The second arm; 113. The third arm; 114. The fourth arm; 115. The base; 116. The flange; 117. The supporting claw; 121. The first hinge device; 122. The second hinge device; 123. The first rotating joint; 124. The second rotating joint; 125. The first rotating support; 126. The second rotating support; 1231. The upper clamping seat; 1232. The lower clamping seat; 210. The second imaging robotic arm; 220. The second imaging support; 230. The second X-ray generator; 240. The second X-ray receiver; 201. The second imaging light beam; 310. The treatment robotic arm; 320. The treatment head; 301. The treatment beam; Detailed implementation manners The present invention will be further described below in conjunction with the accompanying drawings.

[0017] As Figure 1a and Figure 1b shown, the present invention provides a real-time imaging guidance system, including an imaging robot and a treatment robot 300. As Figure 2 shown, the imaging robot includes a first imaging robot 100 and a second imaging robot 200 that are symmetrically arranged with the patient 500 as the center. Among them, the first imaging robot 100 can adjust its posture and position so that the imaging light beam emitted by it can be received by the second imaging robot 200; similarly, the second imaging robot 200 can adjust its posture and position so that the imaging light beam emitted by it can be received by the first imaging robot 100. The imaging light beam emitted by the first imaging robot 100 and the imaging light beam emitted by the second imaging robot 200 intersect at a specified position to obtain a three-dimensional real-time image.

[0018] The treatment robot 300 is used to generate a treatment beam. The treatment robot 300 can adjust its posture and displacement according to the three-dimensional registration degree between the three-dimensional real-time image obtained by the imaging robot and the three-dimensional planned image, so that the generated treatment beam 301 is always aligned with the specified position (lesion position). The three-dimensional real-time image is the DRR (Digital Reconstructed Radiograph) obtained by the imaging robot during treatment, and the three-dimensional planned image is the DRR (Digital Reconstructed Radiograph) obtained by the imaging robot before treatment or the DRR generated by CT examination. Through three-dimensional registration, two images related to the patient's anatomical structure from the same or different sources are accurately aligned in three-dimensional space, so as to better formulate, evaluate and implement radiotherapy plans.

[0019] As Figure 3a and Figure 3b shown, the first imaging robot 100 includes a first imaging robotic arm 110, a first imaging bracket 120 connected to the first imaging robotic arm 110, and a first X-ray generator 130 and a first X-ray receiver 140 respectively rotatably connected to the first imaging bracket 120. As Figure 3c shown, the second imaging robot 200 includes a second imaging robotic arm 210, a second imaging bracket 220 connected to the second imaging robotic arm 210, and a second X-ray generator 230 and a second X-ray receiver 240 respectively rotatably connected to the second imaging bracket 220.

[0020] Among them, the first X-ray generator 130 and the second X-ray receiver 240 are aligned with each other, so that the first X-ray generator 130 emits a first imaging beam 101 (X-ray beam) to the second X-ray receiver 240; similarly, the second X-ray generator 230 and the first X-ray receiver 140 are aligned with each other, so that the second X-ray generator 230 emits a first imaging beam 201 (X-ray beam) to the first X-ray receiver 140. As Figure 2 、 Figure 9 and Figure 10 described, the first imaging beam 101 and the second imaging beam 201 intersect with each other at a specified position (such as the lesion position).

[0021] The first X-ray receiver 140 and the second X-ray receiver 240 can receive the X-rays emitted by the corresponding X-ray emitters and convert them into digital signals to form three-dimensional images.

[0022] The first imaging robotic arm 110 and the second imaging robotic arm 210 can adopt the same configuration, and the first imaging robotic arm 110 will be taken as an example for description below.

[0023] As Figure 3a shown, the first imaging robotic arm 110 includes a first arm 111, a second arm 112, a third arm 113, a fourth arm 114 and a base 115. Among them, the front end of the first arm 111 is connected to the first imaging bracket 120. As Figure 4 shown, a flange 116 with a plurality of support claws 117 is provided on the first imaging bracket 120, and the flange 116 is rotatably connected to the front end of the first arm 111. Therefore, the first arm 111 can rotate relative to the first imaging bracket 120 with the axis of the flange 116 as the rotation axis.

[0024] As Figure 4 shown, the rear end of the first arm 111 is configured as a U-shaped structure, and a rotating shaft perpendicular to the U-shaped opening is provided in the U-shaped structure. Please refer to Figure 3a, the front end of the second arm 112 is rotatably connected to the rotating shaft, so that the second arm 112 can rotate relative to the first arm 111 with the rotating shaft as the rotation axis; in addition, the second arm 112 can also rotate with its own axis as the rotation axis.

[0025] Please continue to refer to Figure 3a , the rear end of the second arm 112 is rotatably connected to the front end of the third arm 113, the rear end of the third arm 113 is rotatably connected to the front end of the fourth arm 114, and the fourth arm 114 is rotatably connected to the base 115. Therefore, relative rotation is possible between the third arm 113 and the second arm 112, between the third arm 113 and the fourth arm 114, and between the fourth arm 114 and the base 115. In addition, the connecting cables in each arm of the first imaging robotic arm 110 can converge in the fourth arm 114 and be connected to external devices such as a control unit or a power supply unit through the base 115.

[0026] As Figure 4 shown, a flange 116 with a plurality of support claws 117 is provided on one side of the first imaging bracket 120. The plurality of support claws 117 are arranged at intervals along the circumferential direction of the flange 116, and the plurality of support claws 117 play a role in support and stability. On the other side of the first imaging bracket 120, a first X-ray generator 130 and a first X-ray receiver 140 are respectively connected.

[0027] Specifically, as Figure 4 and Figure 5 shown, a first hinge device 121 and a second hinge device 122 are provided on the first imaging bracket 120. As Figure 5 shown, the first hinge device 121 includes a first rotating joint 123 and a first rotating support 125. The first rotating support 125 is fixedly connected to the first imaging bracket 120, and the rear end of the first rotating joint 123 is rotatably connected to the first rotating support 125. The first rotating joint 123 and the first rotating support 125 can be connected by a rotating shaft, so that the first rotating joint 123 can rotate with the rotating shaft as the rotation axis; or the first rotating support 125 can be configured as a spherical seat, and a spherical connection is formed between the first rotating joint 123 and the first rotating support 125, so that the first rotating joint 123 can perform universal rotation.

[0028] The front end of the first rotating joint 123 is connected to the first X-ray generator 130. Therefore, when the first rotating joint 123 rotates relative to the first hinge device 121, it can drive the first X-ray generator 130 to transmit, thereby adjusting the angle that the first X-ray generator 130 turns relative to the first imaging bracket 120.

[0029] As Figure 6As shown, the first rotating joint 123 includes an upper clamping seat 1231 and a lower clamping seat 1232. After they are clamped together, a columnar space is formed in the middle to accommodate the first X-ray generator 130.

[0030] The second hinge device 122 has a structure similar to that of the first hinge device 121. As Figure 7 shown, the second hinge device 122 includes a second rotating joint 124 and a second rotating support 126. The second rotating support 126 is fixedly connected to the first imaging bracket 120. The second rotating support 126 and the first rotating support 125 are respectively located at the left and right ends of the first imaging bracket 120. The rear end of the second rotating joint 124 is rotatably connected to the second rotating support 126. The second rotating joint 124 and the second rotating support 126 can be connected by a rotating shaft, so that the second rotating joint 124 can rotate around the rotating shaft; or the second rotating support 126 can be configured as a spherical seat, and a spherical connection is formed between the second rotating joint 124 and the second rotating support 126, so that the second rotating joint 124 can perform universal rotation.

[0031] The front end of the second rotating joint 124 is connected to the first X-ray receiver 140. The first X-ray receiver 140 is configured as a substantially flat plate structure. The second rotating joint 124 can adaptively adjust its structural shape according to the first X-ray receiver 140, so that when the second rotating joint 124 rotates relative to the second hinge device 122, it can drive the first X-ray receiver 140 to transmit, thereby adjusting the angle that the first X-ray receiver 140 turns relative to the first imaging bracket 120.

[0032] The first rotating joint 123 and the second rotating joint 124 can be connected to a power unit (such as a motor), so as to drive them to rotate.

[0033] The second imaging robotic arm 210 has the same structure and configuration as the first imaging robotic arm 110. For example, a third hinge device (which can adopt the same configuration as the first hinge device 121) and a fourth hinge device (which can adopt the same configuration as the second hinge device 122)) are provided on the second imaging bracket 220, and they are respectively connected to the second X-ray generator 230 and the second X-ray receiver 240.

[0034] Therefore, it can be understood that when the first X-ray generator 130 on the first imaging robotic arm 110 adjusts the angle it turns relative to the first imaging support 120, the second X-ray receiver 240 on the second imaging robotic arm 210 also needs to correspondingly adjust the angle it turns relative to the second imaging support 220 to ensure that the first imaging light beam 101 emitted by the first X-ray generator 130 can be accurately received by the second X-ray receiver 240 on the opposite side of the device. Similarly, when the second X-ray generator 230 on the second imaging robotic arm 210 adjusts the angle it turns relative to the second imaging support 220, the first X-ray receiver 140 on the first imaging robotic arm 110 also needs to correspondingly adjust the angle it turns relative to the first imaging support 120 to ensure that the second imaging light beam 201 emitted by the second X-ray generator 230 can be accurately received by the first X-ray receiver 140 on the opposite side of the device.

[0035] When the first imaging robotic arm 110 and the second imaging robotic arm 210 adjust their postures and positions, they move in coordination. As Figure 8 shown, when performing cone beam CT scanning before / during radiotherapy, the first imaging robotic arm 110 and the second imaging robotic arm 210 perform circular motion with a specified position (lesion position) as the center. Therefore, their cone beam CT scanning trajectory is a circle with the lesion position as the center; and they always maintain a parallel position relationship with each other (for example, the first imaging support 120 and the second imaging support 220 are kept parallel to each other) and keep the distance between the first X-ray generator 130 and the second X-ray receiver 240 (source-image distance SID) and the distance between the second X-ray generator 230 and the first X-ray receiver 140 (source-image distance SID) equal. For example, the first imaging robotic arm 110 and the second imaging robotic arm 210 can synchronously adjust the distance of their abduction or adduction , the first X-ray generator 130 or the first X-ray receiver 140 can adjust the angle it turns relative to the first imaging support 120 , the second X-ray generator 230 or the second X-ray receiver 240 can adjust the angle it turns relative to the second imaging support 220 . In addition, by adjusting the distance of their abduction or adduction, the first imaging robotic arm 110 and the second imaging robotic arm 210 change the distance between the first imaging support 120 and the second imaging support 220, and also correspondingly need to adjust the included angle between the first X-ray generator 130 and the second X-ray receiver 240 and the included angle between the second X-ray generator 230 and the first X-ray receiver 140 to ensure that the imaging light beam can be accurately received by the X-ray receiver on the opposite side.

[0036] The source image distance SID is related to the specified position (or target position). For example, if the specified position is at the patient's head, the source image distance SID is smaller (i.e., the distance between the first X-ray generator 130 and the second X-ray receiver 240 and the distance between the second X-ray generator 230 and the first X-ray receiver 140 are relatively close); if the specified position is at the patient's lungs, the source image distance SID is larger (i.e., the distance between the first X-ray generator 130 and the second X-ray receiver 240 and the distance between the second X-ray generator 230 and the first X-ray receiver 140 are relatively far). Therefore, it is necessary to adjust the source image distance SID of the first imaging robotic arm 110 and the second imaging robotic arm 210 according to the specified position.

[0037] The first imaging robot 100 and / or the second imaging robot 200 can adjust its attitude and position so that the first imaging beam 101 emitted by the first X-ray generator 130 and the second imaging beam 201 emitted by the second X-ray generator 230 intersect at the specified position, and the intersection area 102 (as Figure 9 shown) can cover the specified position. The area of the intersection area 102 is obtained according to the following relational expression: .

[0038] Wherein, A is the area of the intersection area; D is the distance between the first X-ray generator 130 and the second X-ray receiver 240 (source image distance SID), or the distance between the second X-ray generator 230 and the first X-ray receiver 140 (source image distance SID). As Figure 10 shown, α is the angle between the first imaging beam 101 emitted by the first X-ray generator 130 and the second imaging beam 202 emitted by the second X-ray generator 230. For example, α can be adjustable within the range of 90°±5°.

[0039] Optionally, whether the intersection area of the first imaging beam 101 and the second imaging beam 201 can cover the specified position (or the coverage of the intersection area can meet the standard) can be judged by the number of key feature points in the intersection area. For example, if the number of key feature points in the intersection area is greater than or equal to 10, it can be considered that the intersection area can cover the specified position or the coverage of the intersection area can meet the standard. If the number of key feature points in the intersection area is less than 10, manually marked feature points can be supplemented.

[0040] Optionally, whether the intersection area of the first imaging beam 101 and the second imaging beam 201 can cover the specified position (or the coverage of the intersection area can meet the standard) can be judged by comparing the area of the intersection area with the area of the specified position. For example, if the area of the intersection area is greater than the area of the specified position (lesion position), it can be considered that the intersection area can cover the specified position or the coverage of the intersection area can meet the standard.

[0041] Conversely, if the intersection area fails to cover the specified position (or the coverage of the intersection area does not meet the standard), secondary parameter optimization can be performed, that is, the postures and positions of the first imaging robot 100 and / or the second imaging robot 200 are adjusted again until the intersection area can cover the specified position (or the coverage of the intersection area meets the standard).

[0042] The first imaging robot 100 and the second imaging robot 200 cooperate in PID control to adjust their postures and positions so that the X-ray generators and X-ray receivers on them can be dynamically balanced. Among them, the angle turned by the first X-ray generator 130 or the first X-ray receiver 140 relative to the first imaging bracket 120

[0043] satisfies the following relational expression: The angle .

[0044] is the adjustment angle of the first X-ray generator 130 or the first X-ray receiver 140. is the adjustment angle of the second X-ray generator 230 or the second X-ray receiver 240.

[0045] is the angle error between the first imaging bracket 120 and the second imaging bracket 220 at the current moment.

[0046] In the above formula, is the proportional term, that is, the control unit adjusts according to the current error amplitude; for example, according to the magnitude of the current displacement deviation, a control quantity proportional to the deviation can be quickly output to quickly correct the current error.

[0047] is the integral term, that is, the control unit accumulates the historical displacement deviation, and gradually eliminates the cumulative error caused by the system static error over time; is the differential term, that is, the control unit predicts the change trend of the error, outputs the control quantity in advance, enhances the stability of the system, and prevents overshoot.

[0048] is the proportional coefficient, which directly affects the response speed of the imaging robotic arm to the current error. The larger it is, the faster the response, but too large will lead to overshoot and oscillation. is the integral coefficient, which can eliminate the steady-state error and adjust the output by accumulating the historical error. However, too strong integration may lead to integral saturation or oscillation. is the differential coefficient, which can predict the trend of error change, suppress overshoot and oscillation, but is sensitive to noise and may amplify high-frequency interference.

[0049] When adjusting the first X-ray generator 130 or the first X-ray receiver 140, determine its response intensity when tilted relative to the first image support 120; correct long-term tilt, prevent violent shaking.

[0050] For or the optimization strategies include: adaptive parameter adjustment or fuzzy fusion control.

[0051] Among them, adaptive parameter adjustment means that in the control process, the PID parameters are dynamically adjusted according to the change of the error. For example, when the error is large, increase the value to speed up the response speed; when the error is close to zero, decrease the and increase the value to eliminate the steady-state error.

[0052] Fuzzy fusion control introduces the fuzzy control theory. According to different ranges of the error and the error change rate, they are divided into different fuzzy sets, and the PID parameters are dynamically adjusted through the fuzzy rule table, so as to improve the robustness and adaptability of the system. And , for example, first adjust , and finally adjust .

[0053] For example, first use only proportional control (i.e., = 0, = 0): step by step increase , until the system shows critical oscillation (such as overshoot or periodic fluctuation), at this time is recorded as the critical value , and the oscillation period is recorded. Subsequently, integral control is introduced (i.e., ≠0): starting from a smaller , gradually increase it to eliminate the steady-state error. Usually is set to of 1 / 10~1 / 2. Subsequently, derivative control is added ( ≠0): = 0.1 starting from, gradually increase it to suppress overshoot, but pay attention to the influence of noise.

[0054] In addition, an automatic tuning method can also be used, such as automatic adjustment by the Ziegler-Nichols critical ratio method.

[0055] In one embodiment, For example, it can be 0.5, It can be 0.1, For example, it can be 0.05.

[0056] The optimal is calculated through the three terms of the proportional term, integral term, and derivative term in the above formula, so that the first X-ray generator 130 or the first X-ray receiver 140 gradually rotates by this adjustment angle, thereby gradually reducing the angle error between the first imaging bracket 120 and the second imaging bracket 220 at the current moment . = the included angle between the first imaging bracket 120 and the second imaging bracket 220 at the current moment - 180°, that is, the target angle (the included angle between the first imaging bracket 120 and the second imaging bracket 220 at the current moment) is 180° (that is, the first imaging robot 100 and the second imaging robot 200 are in a parallel posture).

[0057] In a specific embodiment, the angle adjustment amounts of the first imaging robot 100 and the second imaging robot 200 are obtained based on the displacement of the feature points and The process is as follows.

[0058] First, the parallelism deviation between the corresponding feature points of the first imaging robot 100 and the second imaging robot 200 can be detected through the deep learning automatic feature recognition module. As described above, in the initial state, the first X-ray generator 130 of the first imaging robot 100 and the second X-ray receiver 240 of the second imaging robot 200 (or the first X-ray receiver 140 of the first imaging robot 100 and the second X-ray generator 230 of the second imaging robot 200) are strictly parallel, that is, the theoretical included angle is 180°.

[0059] At a certain moment, it is detected through the deep learning automatic feature recognition module that the current included angle becomes 181°, then the angle error is:

[0060] According to the above PID control formula (1): Among them, Take 0.5, Take 0.1, Take 0.05; and at the current moment, the integral term is relatively small because the error has just started to accumulate (it can be set to 0.1), and the derivative term is set to 0 (because the error change rate is small at this moment).

[0061] Therefore The calculation process of is as follows:

[0062]

[0063] That is, the first imaging robot 100 needs to rotate clockwise by 0.51°, and the second imaging robot 200 needs to rotate counterclockwise by 0.51° to restore the parallelism between the two (vice versa).

[0064] As described above, when the first X-ray generator 130 on the first imaging robotic arm 110 adjusts the angle it turns relative to the first imaging bracket 120, the second X-ray receiver 240 on the second imaging robotic arm 210 also needs to correspondingly adjust the angle it turns relative to the second imaging bracket 220. For example, when the first X-ray generator 130 adjusts by + the second X-ray receiver 240 adjusts by - .

[0065] Sensors (angle sensors or magnetic sensors) are provided on both the first imaging robotic arm 110 and the second imaging robotic arm 210 to obtain the angles (current angles) turned by the above-mentioned X-ray generators and X-ray receivers, or the current angles can also be obtained through the encoders of the motors that drive the above-mentioned X-ray generators and X-ray receivers to rotate.

[0066] The first imaging robotic arm 110 and the second imaging robotic arm 210 synchronously adjust the distance of their abduction or adduction according to the following relational expression : ) (2).

[0067] Wherein, ΔS is the displacement between the position of the key feature point in the three-dimensional real-time image at the current moment and the position of the feature point at the previous moment; α is the included angle between the first imaging beam 101 emitted by the first X-ray generator 130 and the second imaging beam 201 emitted by the second X-ray generator 230. By adjusting the distance of abduction or adduction of the first imaging robotic arm 110 and the second imaging robotic arm 210, a source image distance SID dynamic compensation mechanism is established to ensure that the first imaging beam 101 and the second imaging beam 102 can intersect at a specified position (target point), so as to achieve continuous and precise coverage of the intersection area 102 during the control process.

[0068] The abduction of the first imaging robotic arm 110 means that the first imaging robotic arm 110 extends towards the second imaging robotic arm 210, and the adduction of the first imaging robotic arm 110 means that the first imaging robotic arm 110 contracts in the direction away from the second imaging robotic arm 210. The same applies to the second robotic arm 210.

[0069] For The optimization strategies include: predictive adjustment and multi-objective optimization.

[0070] Predictive adjustment is to analyze the historical data of the displacement of feature points, establish a displacement prediction model to predict in advance ΔS the change trend, so as to make adjustments in advance to reduce the lag of adjustment.

[0071] Multi-objective optimization is when making adjustments , not only the need to cover the target area in the intersection area should be considered, but also the need to avoid collisions with the bed board, the patient or the treatment head 320. Multi-objective optimization algorithms such as genetic algorithms can optimize multiple objective functions simultaneously, so as to find the optimal adjustment plan. In a specific implementation, the adjustment amount of the source image distance SID is calculated based on the displacement of the feature points as follows.

[0072] At the current moment, the deep learning automatic feature recognition module detects the displacement of the feature points in the horizontal direction ΔS = 5mm, and the included angle (i.e., the double-beam included angle) between the first imaging beam 101 emitted by the first X-ray generator 130 and the second imaging beam 201 emitted by the second X-ray generator 230 α is 90°.

[0073] According to the above formula (2):

[0074] That is, the first imaging robotic arm 110 and the second imaging robotic arm 210 need to abduct synchronously (when ΔS is positive) or adduct synchronously (when ΔS is negative) about 7.07mm; at the same time, adjust the angles of the first hinge device and the second hinge device to keep the X-ray beams intersecting at the target point.

[0075] Therefore, the PID control strategy of the first imaging robot 100 and the second imaging robot 200 is PID compensation control based on the displacement between the position of the key feature points in the three-dimensional real-time image at the current moment and the position of the feature points at the previous moment, which can solve the problem that the traditional robotic arms are prone to cumulative errors in synchronous motion.

[0076] As described above, when performing compensation control, the first imaging robot 100 and the second imaging robot 200 respectively adjust the movement of their own joints in real time according to the control parameters output by the control unit (PID controller). In this way, each movement is dynamically corrected based on the actual error at the previous moment, so that the first imaging robot 100 and the second imaging robot 200 will not accumulate the error generated by the previous movement into the next movement during the task execution. Therefore, the problem of easy error accumulation in the synchronous movement of traditional robotic arms can be solved. At the same time, the system will continuously monitor the positions of key feature points and repeat the above process continuously to form a closed-loop control to ensure that the first imaging robot 100 and the second imaging robot 200 can maintain high-precision synchronous movement throughout the working process.

[0077] The first imaging robot 100 and the second imaging robot 200 can extract feature points of different scales in the three-dimensional real-time image according to the pre-trained model and screen key feature points. The feature points can be bony landmark points. Among them, the pre-trained model can be, for example, a bony landmark detection model trained using a deep convolutional neural network (such as the ResNet+Faster R-CNN architecture). The first imaging robot 100 and the second imaging robot 200 can identify more than 20 anatomical landmark points such as cranial sutures, vertebral spinous processes, and rib intersections from the three-dimensional real-time image. For these feature points, the non-maximum suppression (NMS) algorithm is used to remove duplicates, and the most obvious feature points are retained. In addition, only the feature points that meet the threshold conditions (such as confidence > 90%) are finally retained as key feature points, so as to synchronously adjust the postures and positions of the first imaging robot 100 and the second imaging robot 200.

[0078] The first imaging robot 100 and the second imaging robot 200 dynamically adjust their postures and positions based on the feature points and their displacements obtained by deep learning detection, that is, an automatic feature point recognition and imaging robot adjustment linkage mechanism is adopted to realize the closed-loop control of "detection - analysis - adjustment".

[0079] When adjusting its posture and position, the first imaging robot 100 and / or the second imaging robot 200 adjusts its posture and position based on the real-time collision avoidance algorithm and the rapid random tree algorithm (RRT) to avoid collisions with the treatment bed 400, the patient 500 on the treatment bed 400, or the treatment head 320 and other objects that may block the movement of the first imaging robot 100 and / or the second imaging robot 200.

[0080] As Figure 11As shown, the treatment robot 300 includes a treatment arm 310 and a treatment head 320 located at the front end of the treatment arm 310. The treatment arm 310 can adopt the same structure as the above-mentioned imaging robotic arms, and its front end can be connected to the treatment head 320 through a flange, so as to adjust its relative position with the treatment head 320.

[0081] The treatment head 320 includes a treatment beam generating component and a treatment beam shaping component (such as a multi-leaf collimator). The treatment beam generating component is used to generate a conical treatment beam 301, and by controlling the shape of the treatment beam through the treatment beam shaping component, a specific shape suitable for each specific field of clinical treatment can be generated.

[0082] During real-time image-guided control, after each imaging robot adjusts its posture and position, the treatment beam 301 generated by the treatment head 320 is aligned with the designated position (lesion position). Each imaging robot adjusts its position to be outside the beam range of the treatment beam 301 to avoid being damaged by the irradiation of the treatment beam 301. During the control process, each imaging robot continuously acquires the imaging information of the patient 500 and compares and analyzes it with the three-dimensional planning image (such as the DRR generated by preoperative CT examination for radiotherapy), and timely discovers the difference between the lesion position and the planned position. The control unit of the real-time imaging guidance system calculates the direction and displacement amount that the treatment head 320 needs to adjust according to the three-dimensional registration degree between the real-time imaging data (including information such as the lesion position and the change of the patient's body position) and the three-dimensional planning image, and adjusts the posture and position of the treatment 320 in real time through the treatment arm 310, so that the treatment head 320 is always aligned with the lesion. During the adjustment process, each imaging robot continues to monitor the imaging information of the patient 500 to ensure the accuracy of the adjustment and the effectiveness of the control.

[0083] Example 1 Using the real-time imaging guidance system described above, during the imaging-guided control process, the patient lies flat on the treatment bed. The treatment head 320 is installed at the end of the treatment robotic arm 310 and is located at the head position of the treatment bed. The first imaging robot 100 and the second imaging robot 200 are respectively deployed on both sides of the treatment bed.

[0084] During the imaging-guided control process, the first imaging robot 100 and the second imaging robot 200 need to adjust their postures and positions in real time to ensure that the X-ray beam intersection area accurately covers the target position, while avoiding the risk of collision with the bed board of the treatment bed, the patient's body, and the treatment head 320. Therefore, the specific process of adjusting the postures and positions of the first imaging robot 100 and the second imaging robot 200 is as follows.

[0085] First, in the initial state, the positions of each robot are: the initial coordinates of the first imaging robot 100 are , and the initial coordinates of the second imaging robot are .

[0086] In the initial state, the postures of the robots are as follows: the initial angle of the first imaging robot 100 is ; the initial angle of the second imaging robot 100 is .

[0087] In the initial state, the distances between the first X-ray generator 130 and the second X-ray receiver 240 (source-image distance SID) and between the second X-ray generator 230 and the first X-ray receiver 140 (source-image distance SID) are and respectively.

[0088] Obstacle information: Data is collected through sensors such as lidar and vision cameras to construct a three-dimensional obstacle map containing the position and shape information of obstacles such as the bed board of the treatment couch, the patient, and the treatment head 320.

[0089] Secondly, the target state is as follows: Target position: According to the three-dimensional coordinates of the target position, the target coordinates of the first imaging robot 100 are calculated and the target coordinates of the second imaging robot 200 are calculated .

[0090] Target posture: The target angle of the first imaging robot 100 is , and the target angle of the second imaging robot is ; the distances between the first X-ray generator 130 and the second X-ray receiver 240 (source-image distance SID) and between the second X-ray generator 230 and the first X-ray receiver 140 (source-image distance SID) are and respectively. Thirdly, the specific adjustment process is as follows.

[0091] The RRT algorithm is used for path planning, including the following steps: The first step, sampling: Randomly generate points in the three-dimensional space , within the range .

[0092] Among them, the sampling range is (unit: cm), and its purpose is to define the sampling area of the three-dimensional space.

[0093] The second step, finding the nearest node: In the constructed rapidly-exploring random tree (RRT), find the node closest to the sampled point .

[0094] The third step, expanding the tree: Extend and expand the distance from the nearest node towards the sampled point , a new node is obtained .

[0095] Among them, the expansion distance can be 5 cm, which is used to define the length of each tree expansion.

[0096] Step 4, collision detection: Use the real-time collision avoidance algorithm to check whether the new node collides with obstacles such as the bed board, patient, and treatment head 320 in the three-dimensional obstacle map; if not, incorporate the new node into the RRT tree; if so, discard the node and resample.

[0097] Step 5, repeated iteration: Loop through the above steps 1 to 4 until a feasible path from the initial node to the target node is found.

[0098] Among them, the maximum number of iterations is specified as 1000 times to avoid infinite loop of the algorithm.

[0099] The calculation process of the attitude and position adjustment amount is as follows: Attitude adjustment amount : According to the planned path, calculate the attitude adjustment amount of the first imaging robot 100 , and the attitude adjustment amount of the second imaging robot .

[0100] Among them, the angular resolution is 0.1°, that is, the minimum adjustment angle of the attitudes of the first imaging robot 100 and the second imaging robot 200.

[0101] Position adjustment amount : Calculate the source image distance adjustment amount of the first imaging robot 100 , and the source image distance adjustment amount of the second imaging robot 200 .

[0102] Among them, the position resolution is 1 mm, that is, the minimum adjustment distance of the positions of the first imaging robot 100 and the second imaging robot 200.

[0103] Furthermore, the above RRT algorithm optimization strategy is as follows: Adopt the bidirectional RRT algorithm, that is, construct the RRT tree from the initial node and the target node simultaneously. When the two trees meet, a path is quickly generated, which can greatly shorten the search time; and adopt heuristic sampling, that is, according to the position of the target node, set the sampling probability distribution, and give priority to sampling in the area close to the target area to improve the path search efficiency.

[0104] The attitude and position adjustment optimization strategy is optimized by adopting smooth adjustment and implementing dynamic adjustment. Among them, smooth adjustment uses the polynomial interpolation algorithm to generate smooth motion trajectories for the first imaging robot 100 and the second imaging robot 200, preventing violent motion from affecting the control accuracy and equipment. Real-time dynamic adjustment is to use sensors to monitor the position and shape changes of obstacles in real time, and dynamically correct the path and attitude based on the latest data to ensure the safety and accuracy of the control process.

[0105] and and and The corresponding relationships are as follows: After the path planning of the RRT algorithm is completed, according to the attitude and position information of adjacent nodes in the path, calculate the attitude adjustment amount of each step of each imaging robot , and the position adjustment amount .

[0106] For example, the first imaging robot 100 moves from node i to node i +1. The attitude of node i is , the position (source image distance SID) of node i is , the attitude of node i +1 is , and the position is . Then the attitude adjustment amount of this step , and the position adjustment amount .

[0107] The second imaging robot 200 moves from node j to node j +1. The attitude of node j is , the position (source image distance SID) of node i is , the attitude of node j +1 is , and the position is . Then the attitude adjustment amount of this step , and the position adjustment amount .

[0108] Although the present invention has been described with reference to the preferred embodiments, various modifications thereof can be made and components thereof can be replaced with equivalents without departing from the scope of the present invention. In particular, as long as there is no structural conflict, the technical features mentioned in each embodiment can be combined in any way. The present invention is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A real-time image guidance system, characterized in that, Comprising: Imaging robots, including a first imaging robot and a second imaging robot. The first imaging robot can dynamically adjust its attitude and position so that the imaging beam emitted by it can be received by the second imaging robot; the second imaging robot can dynamically adjust its attitude and position so that the imaging beam emitted by it can be received by the first imaging robot. The imaging beam emitted by the first imaging robot and the imaging beam emitted by the second imaging robot cross at a specified position to obtain a three-dimensional real-time image; and A treatment robot for generating a treatment beam. The treatment robot adjusts its attitude and displacement according to the three-dimensional registration degree between the three-dimensional real-time image obtained by the imaging robot and the three-dimensional planned image, so that the treatment beam generated by it is aligned with the specified position.

2. The real-time image guidance system according to claim 1, wherein The first imaging robot includes a first imaging robotic arm, a first imaging bracket connected to the first imaging robotic arm, and a first X-ray generator and a first X-ray receiver respectively rotatably connected to the first imaging bracket; The second imaging robot includes a second imaging robotic arm, a second imaging bracket connected to the second imaging robotic arm, and a second X-ray generator and a second X-ray receiver respectively rotatably connected to the second imaging bracket; Wherein, the first imaging robotic arm and the second imaging robotic arm can synchronously adjust the distance of their abduction or adduction , the first X-ray generator or the first X-ray receiver can adjust the angle it turns relative to the first imaging bracket , the second X-ray generator or the second X-ray receiver can adjust the angle it turns relative to the second imaging bracket .

3. The real-time imaging guidance system according to claim 2, wherein The first imaging robot and / or the second imaging robot can dynamically adjust its attitude and position so that the imaging beam emitted by the first X-ray generator and the imaging beam emitted by the second X-ray generator cross at a specified position, and the crossing area can cover the specified position. The area of the crossing area is obtained according to the following relational expression: Among them, A is the area of the intersection region; D is the distance between the first X-ray generator and the second X-ray receiver, or the distance between the second X-ray generator and the first X-ray receiver; α is the angle between the imaging beam emitted by the first X-ray generator and the imaging beam emitted by the second X-ray generator.

4. The real-time image guidance system according to claim 2, wherein The angle by which the first X-ray generator or the first X-ray receiver rotates relative to the first imaging support satisfies the following relational expression: The angle by which the second X-ray generator or the second X-ray receiver rotates relative to the second imaging bracket satisfies the following relational expression: ； is the adjustment angle of the first X-ray generator or the first X-ray receiver; The adjustment angle for the second X-ray generator or the second X-ray receiver; The angular error between the first imaging bracket and the second imaging bracket at the current moment is the proportionality coefficient; is the integration coefficient; is the differential coefficient.

5. The real-time image guidance system according to claim 2, characterized in that, The first imaging robotic arm and the second imaging robotic arm synchronously adjust the distance of their abduction or adduction according to the following relational expression : )； wherein, ΔS is the displacement between the position of a key feature point in the three-dimensional real-time image at the current moment and the position of the feature point at the previous moment; α is the angle between the imaging light beam emitted by the first X-ray generator and the imaging light beam emitted by the second X-ray generator.

6. The real-time image guidance system according to claim 5, characterized in that, The first imaging robot and the second imaging robot extract feature points of different scales in the three-dimensional real-time image according to a pre-trained model and screen key feature points, and the first imaging robotic arm and the second imaging robotic arm synchronously adjust the distance of their abduction or adduction according to the key feature points , wherein the feature points are bony landmark points.

7. The real-time image guidance system according to claim 2, wherein A first hinge device and a second hinge device are provided on the first imaging bracket. The rotating joints of the first hinge device and the second hinge device are respectively connected to the first X-ray generator and the first X-ray receiver, so that the first X-ray generator and the first X-ray receiver can adjust the angles they turn relative to the first imaging bracket; A third hinge device and a fourth hinge device are provided on the second imaging bracket. The rotating joints of the third hinge device and the fourth hinge device are respectively connected to the second X-ray generator and the second X-ray receiver, so that the second X-ray generator and the second X-ray receiver can adjust the angles they turn relative to the first imaging bracket.

8. The real-time image guidance system according to claim 2, characterized in that, The first imaging robot and / or the second imaging robot adjusts its attitude and position based on a real-time collision avoidance algorithm and a rapid random tree algorithm.

9. The real-time image guidance system according to claim 2, wherein, The treatment robot includes a treatment robotic arm and a treatment head connected to the treatment robotic arm. When the treatment head generates a treatment beam, the first X-ray generator, the first X-ray receiver, the second X-ray generator, and the second X-ray receiver are all outside the beam range of the treatment beam.

Images