Portable image acquisition geometry self-adaptive correction method

Abstract

The application discloses a portable image acquisition geometry self-adaptive correction method and belongs to the technical field of medical images. The method sets a first IMU at the X-ray ball tube end, sets a second IMU and an array type curvature sensor at the flexible imaging plate end, and collects relative postures and flexible imaging plate deformation data in real time; when it is detected that there is an included angle deviation between the ball tube axis and the equivalent main normal line of the imaging plate, a multi-degree-of-freedom mechanical arm is driven to perform motion compensation; meanwhile, a flexible imaging plate three-dimensional curved surface model is constructed according to the deformation data, the original X-ray image is reversely mapped from the curved surface coordinate system back to the standard plane coordinate system, and the corrected image is output. The application realizes self-adaptive correction under a non-standard environment and significantly reduces the dependence on the experience of technicians.

Description

Technical Field

[0001] This invention relates to the field of medical imaging technology, and in particular to a portable image acquisition geometric adaptive correction method. Background Technology

[0002] Portable digital radiography (DR) equipment is widely used in non-standard medical scenarios such as grassroots medical clinics, disaster relief, and battlefield emergency care due to its flexibility and convenience. However, existing portable DR equipment primarily relies on the technician's operational experience for imaging, and the equipment itself only provides basic X-ray exposure functions, lacking the ability to perceive and correct for shooting angles, distances, and the state of the flexible imaging plate in real time.

[0003] In practical applications, existing technologies have the following drawbacks: When used on uneven surfaces such as soft stretchers or grass, the flexible imaging plate is prone to physical bending, and it is difficult to maintain a strictly perpendicular relationship between the incident axis of the X-ray tube and the flexible imaging plate, leading to geometric distortion of the acquired X-ray images. In noisy environments or under obstructed visibility conditions, technicians rely solely on visual estimation to determine distance and angle, resulting in poor voice prompts and a low success rate for imaging. Furthermore, existing equipment lacks adaptive adjustment capabilities for scenarios involving critically ill patients, the elderly, or young children who cannot cooperate with standard positioning. These issues cause image quality to heavily rely on technician experience, making it difficult to guarantee diagnostic accuracy.

[0004] Therefore, this invention proposes a portable image acquisition geometric adaptive correction method. Summary of the Invention

[0005] This invention provides a portable image acquisition geometric adaptive correction method, which realizes automatic correction of imaging status and real-time repair of bending distortion in non-standard environments. This enables the device to stably output standard images without geometric distortion, independent of technician experience and external environment, significantly improving the success rate of one-shot shooting and the accuracy of image diagnosis under harsh conditions.

[0006] This invention provides a portable image acquisition geometric adaptive correction method, comprising: A first inertial measurement unit and a time-of-flight ranging sensor are set at the end of the X-ray tube, and a second inertial measurement unit and an array-type curvature sensor are set at the end of the flexible imaging plate to collect the relative Euler angles between the X-ray tube and the flexible imaging plate in real time. An array-type curvature sensor acquires data on the bending radius and stress point distribution of a flexible imaging plate; When a deviation in the angle between the central axis of the X-ray tube and the normal of the flexible imaging plate is detected based on the relative Euler angle, it is determined whether the degree of bending of the flexible imaging plate exceeds a preset threshold. If the preset threshold is not exceeded, the control unit drives the multi-degree-of-freedom robotic arm to perform motion compensation so that the central axis of the X-ray tube coincides with the normal of the center detection point of the flexible imaging plate. If the preset threshold is exceeded, the control unit constructs a three-dimensional surface model based on the bending radius and fits the equivalent principal normal vector by combining the force distribution data. This drives the multi-degree-of-freedom robotic arm to perform motion compensation, so that the central axis of the X-ray tube is parallel to the corresponding equivalent principal normal. By utilizing the source-image distance and combining it with a three-dimensional curved surface model, a dynamic exposure field model is established. Based on the actual depth difference between different curved areas of the flexible imaging plate and the X-ray tube, local edge exposure compensation or hardware adjustment of the milliampere-second parameters is performed on the generated image. The original X-ray images acquired are reverse-mapped from the surface coordinate system back to the standard plane coordinate system using a three-dimensional surface model, and the corrected images are output.

[0007] Furthermore, the relative Euler angles include pitch angle, roll angle, and yaw angle. The control unit dynamically adjusts the motion trajectory of the multi-degree-of-freedom robotic arm based on the real-time changes in pitch angle, roll angle, and yaw angle.

[0008] Furthermore, it also includes: It has a built-in standard body position database, which contains the standard incident angles corresponding to different shooting parts; Receive the instruction to select the shooting location and call up the corresponding standard incident angle; Based on the standard incident angle and the current attitude of the flexible imaging plate, the control unit calculates the motion path and drives the multi-degree-of-freedom robotic arm to move to the preset incident angle. The method for calculating the equivalent principal normal in the control unit includes: based on the normal vectors of each mesh element on the three-dimensional surface model, combined with the weight coefficients determined by the force point distribution data, an iterative optimization algorithm is used to find a virtual direction vector that minimizes the weighted deviation between the virtual direction vector and the normal vectors of all mesh elements. The determined virtual direction vector is the equivalent principal normal.

[0009] Furthermore, the preset incident angle includes a 40-degree tilt angle corresponding to the calcaneal axis.

[0010] Furthermore, the acquired raw X-ray images are reverse-mapped from the surface coordinate system back to the standard plane coordinate system. Specifically, this includes: calculating the lateral deformation coefficient and longitudinal deformation coefficient based on the bending radius data of each grid cell; establishing a mapping table from surface parameter coordinates to standard plane coordinates; traversing the original image pixels to find the mapping table to obtain the standard plane coordinates; and completing grayscale resampling through an interpolation algorithm.

[0011] Furthermore, the dynamic exposure field model performs local edge exposure compensation or hardware adjustment of milliampere-second parameters on the generated image based on the actual depth difference between different curved areas of the flexible imaging plate and the X-ray tube.

[0012] Furthermore, during motion compensation, the multi-degree-of-freedom robotic arm simultaneously receives real-time data feedback from the first and second inertial measurement units, forming a closed-loop control to continuously maintain the alignment of the X-ray tube's central axis with the normal of the flexible imaging plate's central detection point, or to ensure that the axis is parallel to the equivalent principal normal during surface deformation.

[0013] Furthermore, when the flexible imaging plate is used on a soft stretcher or uneven ground, it undergoes physical bending. An array of curvature sensors collects curvature distribution data in real time under bending conditions, which is used for dynamic updating of the three-dimensional surface model.

[0014] Furthermore, it also includes: In non-cooperative shooting scenarios, keep the flexible imaging plate in its current position; The control unit controls the multi-degree-of-freedom robotic arm to adjust the position of the X-ray tube based solely on the attitude data of the flexible imaging plate collected by the second inertial measurement unit, so that the central axis of the X-ray tube coincides with the normal of the center detection point of the flexible imaging plate. Alternatively, when the flexible imaging plate undergoes surface deformation, the control unit drives the multi-degree-of-freedom robotic arm to perform motion compensation based on the equivalent principal normal vector fitted by the three-dimensional surface model, so that the central axis of the X-ray tube is parallel to the equivalent principal normal.

[0015] The beneficial effects of this invention compared to existing technologies are as follows: It overcomes the technical defects of existing portable DR devices, which rely on technician experience for operation in non-standard environments and suffer from unstable image quality. By sensing the relative posture of the X-ray tube and the flexible imaging plate in real time, as well as the deformation data of the flexible imaging plate, and driving the robotic arm to automatically compensate for angular deviations, while employing curved surface reconstruction technology to reverse the image distortion caused by bending, the device can automatically acquire a standard projection position and output high-quality images without geometric distortion in complex environments such as soft stretchers and uneven ground without manual adjustment by technicians. This completely solves the long-standing industry pain point of image distortion caused by physical bending in emergency imaging.

[0016] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in this application.

[0017] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0018] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of the portable image acquisition geometric adaptive correction method in an embodiment of the present invention; Figure 2 This is a schematic diagram of a portable image acquisition device in an embodiment of the present invention. Detailed Implementation

[0019] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0020] like Figure 1 and Figure 2 As shown, this invention provides an implementation method for a portable image acquisition geometric adaptive correction method, comprising: A first inertial measurement unit and a time-of-flight ranging sensor are installed at the end of the X-ray tube, and a second inertial measurement unit and an array-type curvature sensor are installed at the end of the flexible imaging plate. The first and second inertial measurement units are used to acquire the relative Euler angles between the X-ray tube and the flexible imaging plate in real time. The time-of-flight ranging sensor acquires the source-image distance between the center point of the X-ray tube and the flexible imaging plate in real time by emitting modulated light pulses and measuring the time difference of the reflected light.

[0021] An array-type curvature sensor acquires data on the bending radius and stress point distribution of a flexible imaging plate; the bending radius data is used to construct a three-dimensional surface model, and the stress point distribution data is used to calculate the weight coefficients for subsequent equivalent principal normal fitting.

[0022] When a deviation in the angle between the central axis of the X-ray tube and the normal of the flexible imaging plate is detected based on the relative Euler angle, it is determined whether the degree of bending of the flexible imaging plate exceeds a preset threshold. If the preset threshold is not exceeded, the control unit drives the multi-degree-of-freedom robotic arm to perform motion compensation so that the central axis of the X-ray tube coincides with the normal of the center detection point of the flexible imaging plate. If the preset threshold is exceeded, the control unit constructs a three-dimensional surface model based on the bending radius and fits the equivalent principal normal vector by combining the force distribution data. This drives the multi-degree-of-freedom robotic arm to perform motion compensation so that the central axis of the X-ray tube is parallel to the equivalent principal normal. By utilizing the source-image distance and combining it with a three-dimensional curved surface model, a dynamic exposure field model is established. Based on the actual depth difference between different curved areas of the flexible imaging plate and the X-ray tube, local edge exposure compensation or hardware adjustment of the milliampere-second parameters is performed on the generated image. The original X-ray images acquired are reverse-mapped from the surface coordinate system back to the standard plane coordinate system using a three-dimensional surface model, and the corrected images are output.

[0023] In this embodiment, the first inertial measurement unit and the second inertial measurement unit are used to detect the attitude changes of the X-ray tube and the flexible imaging plate in space in real time, respectively. The relative Euler angles include pitch angle, roll angle and yaw angle, which are used to describe the angular deviation between the two.

[0024] In this embodiment, array-type curvature sensors are distributed on the back of the flexible imaging plate to detect the degree of deformation of the flexible imaging plate when it is compressed or bent. The bending radius is used to describe the bending curvature of the flexible imaging plate, and the distribution of force points is used to determine the compression status of each part of the flexible imaging plate.

[0025] In this embodiment, during implementation, the control unit calculates the overall curvature index of the flexible imaging plate based on the bending radius data collected by the array-type curvature sensors. This index can be the average or minimum bending radius of each detection point, or the maximum curvature value obtained through surface fitting. When this index exceeds a preset threshold, for example, when the bending radius of the central region of the imaging plate is less than 500 mm, the system determines that the flexible imaging plate has undergone significant surface deformation, and at this time, the equivalent principal normal compensation mode is activated; conversely, when the curvature is less than or equal to the preset threshold, the system adopts the center normal compensation mode. This preset threshold can be calibrated and adjusted according to factors such as the physical characteristics of the imaging plate and imaging accuracy requirements.

[0026] In this embodiment, the multi-degree-of-freedom robotic arm can move flexibly in multiple directions. The control unit drives the robotic arm to automatically adjust the position of the X-ray tube according to the detected angular deviation until the central axis of the X-ray tube coincides with the normal of the center detection point of the flexible imaging plate. Or, when the flexible imaging plate undergoes surface deformation, the control unit drives the multi-degree-of-freedom robotic arm to perform motion compensation according to the equivalent principal normal vector fitted by the three-dimensional surface model, so that the central axis of the X-ray tube is parallel to the equivalent principal normal.

[0027] In this embodiment, the three-dimensional curved surface model is a virtual model constructed based on the bending radius data, used to restore the actual bending state of the flexible imaging plate at the moment of exposure.

[0028] In this embodiment, reverse mapping is to remap the distorted image pixels acquired in the bent state back to the standard plane coordinate system according to the geometric relationship of the three-dimensional curved surface model. For example, it can compress the originally elongated bone image on the bent flexible imaging plate back to the normal proportion, thereby eliminating the geometric distortion caused by the bending of the flexible imaging plate.

[0029] Furthermore, the relative Euler angles include pitch angle, roll angle, and yaw angle. The control unit dynamically adjusts the motion trajectory of the multi-degree-of-freedom robotic arm based on the real-time changes in pitch angle, roll angle, and yaw angle, so that the angle deviation between the central axis of the X-ray tube and the normal is always controlled within 0.5 degrees.

[0030] In this embodiment, the pitch angle refers to the angle at which the X-ray tube or flexible imaging plate tilts forward and backward around the horizontal axis. For example, when the tube tilts forward by 10 degrees, the pitch angle is 10 degrees. The roll angle refers to the angle at which the tube tilts left and right around the front and rear axes. For example, when the flexible imaging plate tilts to the right by 5 degrees, the roll angle is 5 degrees. The yaw angle refers to the angle at which the tube rotates horizontally around the vertical axis. For example, when the tube rotates to the left by 15 degrees, the yaw angle is 15 degrees.

[0031] In this embodiment, the control unit receives real-time data on changes in pitch, roll, and yaw angles. Based on the dynamic changes in these three angles, it synchronously adjusts the motion trajectory of the multi-degree-of-freedom robotic arm in multiple directions. For example, when the pitch angle of the X-ray tube is detected to be too large while the roll angle is normal, the robotic arm only adjusts the pitch angle of the X-ray tube, keeping other directions unchanged. When all three angles deviate, the robotic arm performs multi-dimensional composite motion simultaneously to ensure that the central axis of the X-ray tube coincides with the normal of the center detection point of the flexible imaging plate. Alternatively, when the flexible imaging plate undergoes surface deformation, the control unit drives the multi-degree-of-freedom robotic arm to perform motion compensation based on the equivalent principal normal vector fitted by the three-dimensional surface model, so that the central axis of the X-ray tube is parallel to the equivalent principal normal.

[0032] In this embodiment, the control unit dynamically adjusts the movement trajectory of the robotic arm according to the real-time changes in pitch angle, roll angle and yaw angle, so that the angle deviation between the central axis of the X-ray tube and the normal of the flexible imaging plate is always controlled within 0.5 degrees, preferably less than 0.3 degrees, to ensure that the projection angle at the moment of exposure fully meets the standard requirements.

[0033] Furthermore, it also includes: It has a built-in standard body position database, which contains the standard incident angles corresponding to different shooting parts; Receive the instruction to select the shooting location and call up the corresponding standard incident angle; Based on the standard incident angle and the current attitude of the flexible imaging plate, the control unit calculates the motion path and drives the multi-degree-of-freedom robotic arm to move to the preset incident angle. The method for calculating the equivalent principal normal in the control unit includes: based on the normal vectors of each mesh element on the three-dimensional surface model, combined with the weight coefficients determined by the force point distribution data, an iterative optimization algorithm is used to find a virtual direction vector that minimizes the weighted deviation between the virtual direction vector and the normal vectors of all mesh elements. The determined virtual direction vector is the equivalent principal normal.

[0034] In this embodiment, the standard body position database pre-stores standard incident angle data corresponding to multiple common imaging sites. For example, the incident angle corresponding to the chest frontal view is 0 degrees, which is vertical incident, and the incident angle corresponding to the calcaneal axis view is 40 degrees, which is oblique incident.

[0035] In this embodiment, after the technician selects the part to be photographed through the operation interface, the system automatically recognizes the photographing part selection command and calls the standard incident angle corresponding to that part from the standard body position database.

[0036] In this embodiment, the control unit calculates the optimal motion path of the multi-degree-of-freedom robotic arm based on the standard incident angle obtained by calling and the current attitude data of the flexible imaging plate collected in real time by the second inertial measurement unit. For example, when the flexible imaging plate is tilted and a 40-degree incident angle is required, the control unit calculates the angle and direction of movement that the X-ray tube needs to compensate for, and drives the robotic arm to move automatically to the final preset incident angle position so that the X-ray is incident on the flexible imaging plate at a standard angle.

[0037] In this embodiment, the fitting process of the equivalent principal normal is as follows: First, the spatial coordinates and normal directions of each point on the imaging plate surface are obtained based on the three-dimensional surface model; then, with the goal of minimizing the geometric distortion of the entire image, a virtual reference direction is found through iterative calculation, such that the weighted sum of the deviations between this direction and the normals of each point is minimized. This virtual reference direction is the equivalent principal normal, and its physical meaning is the optimal direction for X-ray incident.

[0038] In this embodiment, the method for the control unit to calculate the equivalent principal normal includes the following steps: Step A: Construct a 3D surface model and extract normal information Based on real-time bending radius data acquired by an array of curvature sensors, a three-dimensional surface model of the flexible imaging plate in its current state is constructed. This model consists of several mesh cells, each corresponding to a tiny region of the imaging plate. For each mesh cell, its spatial coordinates and the unit normal vector perpendicular to its surface are calculated; this vector represents the orientation of the tiny region under bending conditions.

[0039] Step B: Determine the weighting coefficients based on the force distribution. Based on the force distribution data synchronously collected by the array curvature sensor, a weight coefficient is assigned to each grid cell. The principle for assigning the weight coefficient is: the greater the force and the more obvious the deformation (such as the area of ​​compression on the patient's body), the higher the weight coefficient; the less force or no force is applied to the area, the lower the weight coefficient. This weight coefficient reflects the degree of contribution of each region to the overall geometric distortion.

[0040] Step C: Construct the deviation metric function and solve for the equivalent principal normal. Based on the normal vectors of each grid cell obtained in step A, and combined with the weight coefficients assigned in step B, a virtual spatial direction vector is found that minimizes the sum of the weighted deviations between the virtual vector and the normal vectors of all grid cells.

[0041] The specific implementation method is as follows: The virtual vector to be found is denoted as the unit vector V. For each grid cell, the angle deviation between vector V and the cell's normal vector Ni is calculated. This deviation value is obtained by calculating the dot product of the two vectors and performing a mathematical transformation. The deviation value is 0 when the two vectors completely coincide, and reaches its maximum when the two vectors are perpendicular to each other.

[0042] Multiply the deviation value of each grid cell by its weight coefficient to obtain the weighted deviation value of that cell. Sum the weighted deviation values ​​of all grid cells to obtain the total deviation value F(V).

[0043] Step D: Determine the equivalent principal normal through iterative optimization. Numerical optimization methods (such as gradient descent or the Levenberg-Marquardt algorithm) are used to iteratively optimize vector V. In each iteration, the direction of vector V is fine-tuned, and the total deviation value F(V) is recalculated until a vector Vopt that minimizes F(V) is found.

[0044] The vector Vopt is the equivalent principal normal vector obtained by fitting. Its physical meaning is: when the central axis of the X-ray tube is incident along this direction, the geometric distortion of the entire imaging area of ​​the flexible imaging plate is minimized, and it can be used as the target direction for robot arm motion compensation.

[0045] Furthermore, the preset incident angle includes a 40-degree tilt angle corresponding to the calcaneal axis.

[0046] In this embodiment, the calcaneal axis view is a specific imaging position for the calcaneus of the foot. This position requires X-rays to be incident on the calcaneal region at a 40-degree angle to clearly show the structural morphology of the lower part of the calcaneus.

[0047] In this embodiment, when the technician selects the calcaneal axis imaging mode, the system automatically calls the 40-degree tilt angle data stored in the standard position database. The control unit calculates the motion path based on the current posture of the flexible imaging plate and drives the multi-degree-of-freedom robotic arm to adjust the X-ray tube to a position with a 40-degree angle to the normal of the flexible imaging plate, so that the X-ray is incident at a preset 40-degree tilt angle.

[0048] Furthermore, the acquired raw X-ray images are reverse-mapped from the surface coordinate system back to the standard plane coordinate system. Specifically, this includes: calculating the lateral deformation coefficient and longitudinal deformation coefficient based on the bending radius data of each grid cell; establishing a mapping table from surface parameter coordinates to standard plane coordinates; traversing the original image pixels to find the mapping table to obtain the standard plane coordinates; and completing grayscale resampling through an interpolation algorithm.

[0049] In this embodiment, the mesh mapping algorithm divides the three-dimensional curved surface model into several mesh units, each mesh unit corresponding to a pixel region in the original X-ray image.

[0050] In this embodiment, the basic principle of the mesh mapping algorithm is as follows: Based on the bending radius and force point distribution data collected by the array curvature sensor, a three-dimensional curved surface model of the flexible imaging plate is constructed. The model consists of multiple spatial triangular facets, each of which corresponds to a tiny region of the imaging plate.

[0051] The 3D surface model is meshed, and a mapping relationship is established between the surface coordinate system and the standard plane coordinate system. Specifically, for any point P(u,v) on the surface model, where u and v are the surface parametric coordinates, the corresponding point Q(x,y) in the standard plane coordinate system is calculated. The mapping relationship satisfies the following geometric constraints: the arc length on the surface is proportional to the chord length on the plane, and the topological relationship between adjacent points on the surface remains unchanged in the plane.

[0052] The algorithm iterates through each pixel of the original X-ray image, determining the 3D surface mesh cell containing that pixel. Based on the spatial coordinates of the four vertices of that mesh cell, bilinear interpolation is used to calculate the theoretical position of that pixel in the standard plane coordinate system. For example, for pixels whose central region of the surface is compressed, their position in the standard plane coordinate system needs to be expanded outwards; for pixels whose edge region of the surface is stretched, their position in the standard plane coordinate system needs to be contracted inwards.

[0053] The grayscale values ​​of the original pixels are assigned to the corresponding pixels in the standard plane coordinate system. For pixels in the standard plane coordinate system that are not directly assigned values, bicubic interpolation or nearest neighbor interpolation algorithms are used to fill the grayscale values, forming a complete corrected image.

[0054] In this embodiment, the specific implementation process of the grid mapping algorithm is as follows: First, a three-dimensional surface model of the flexible imaging plate is constructed based on the bending radius and stress point distribution data collected by the array-type curvature sensor. This model consists of a large number of spatial grid cells, each corresponding to a tiny region of the imaging plate. The spatial position of the grid cell reflects the actual height and tilt angle of that region under bending conditions.

[0055] Secondly, the correspondence between the surface coordinate system and the standard plane coordinate system is established. Specifically, for each mesh cell on the surface model, the proportional relationship between the arc length of the cell in the curved state and its projected length in the ideal plane state is calculated. Based on this proportional relationship, the position in the standard plane that each pixel on the surface should correspond to is determined.

[0056] For example, when the central region of the flexible imaging plate is depressed due to pressure, the mesh cells in that region are compressed in the curved surface model. In the reverse mapping process, the pixels corresponding to that region need to be expanded outward to restore the normal size ratio. When the edge region of the flexible imaging plate is stretched due to suspension, the pixels corresponding to that region need to be shrunk inward to restore the normal size ratio.

[0057] Finally, each pixel in the original X-ray image is traversed, and its theoretical position in the standard plane coordinate system is calculated based on the spatial location and degree of deformation of its grid cell. The gray value of the original pixel is then assigned to that position. For the few missing pixels that may appear in the standard plane, interpolation is performed using the gray values ​​of surrounding pixels to fill the gaps, ultimately forming a complete corrected image without geometric distortion.

[0058] The core of the above mapping process lies in the fact that the mapping relationship depends only on the actual physical deformation data of the imaging plate, rather than the content features of the image itself. Therefore, even in cases of low image contrast or blurred tissue structure, this method can still stably achieve geometric correction.

[0059] In this embodiment, the advantage of the mesh mapping algorithm is that it does not rely on the feature points of the image itself, but performs geometric correction based on the actual physical deformation data of the imaging plate. Therefore, it can still achieve a stable correction effect even when the image contrast is low and the tissue structure is blurred.

[0060] In this embodiment, for each pixel in the curved surface coordinate system, its actual spatial coordinates in the three-dimensional curved surface model are calculated, and then the theoretical position of the pixel in the standard plane coordinate system is found according to the geometric correspondence. For example, the pixels in the central region of the curved flexible imaging plate that are compressed are stretched to normal size, and the pixels in the edge region that are stretched are compressed to normal proportion.

[0061] In this embodiment, after the coordinate transformation is completed pixel by pixel, all pixels are rearranged and combined in the standard plane coordinate system to form a corrected image without geometric distortion.

[0062] The acquired raw X-ray images are mapped back from the curved coordinate system to the standard planar coordinate system. Specifically, a mesh mapping algorithm based on physical deformation parameters is used, which includes the following steps: Step ①: Construct a meshed representation of the surface model: Based on the bending radius and stress point distribution data collected by an array of curvature sensors, a three-dimensional surface model of the flexible imaging plate in its current state is established. This surface model is divided into several continuous mesh units, each corresponding to a small physical region of the imaging plate. Two sets of coordinate parameters are established for each mesh unit: one set is its surface parameter coordinates in the bent state, reflecting the unit's relative position on the flexible imaging plate; the other set is its corresponding ideal plane coordinates, reflecting the unit's standard position in the unbent state.

[0063] Step 2: Calculate the deformation coefficient of each mesh element: Based on the bending radius data collected by the array curvature sensor, the degree of stretching or compression of each grid cell in different directions is calculated, defined as the transverse deformation coefficient and the longitudinal deformation coefficient, respectively. The physical meaning of these two coefficients is the proportional relationship between the actual arc length of the grid cell in the bent state and its projected length in the ideal planar state when the flexible imaging plate bends. For example, for a grid cell in a concave region, its deformation coefficient is less than 1, indicating that the region is compressed; for a grid cell in a convex region, its deformation coefficient is greater than 1, indicating that the region is stretched.

[0064] Step 3: Establish the correspondence between surface coordinates and plane coordinates: Based on the deformation coefficients of each mesh element, a mapping relationship from surface parametric coordinates to standard plane coordinates is established. Specifically, starting from the reference point of the surface model (such as the center point of the imaging plate), the standard plane coordinate position corresponding to each surface parametric coordinate point is calculated row by row and column by column according to the arrangement order of the mesh elements. The calculation principle is: in the horizontal direction, the standard plane distance between two adjacent surface points is equal to their surface arc length divided by the horizontal deformation coefficient; in the vertical direction, the standard plane distance between two adjacent surface points is equal to their surface arc length divided by the vertical deformation coefficient. In this way, a complete mapping relationship table is obtained, which records the standard plane coordinate position corresponding to each surface parametric coordinate point.

[0065] Step 4: Perform pixel-by-pixel coordinate transformation on the original image: Iterate through each pixel of the original X-ray image and determine its corresponding grid cell based on its physical location on the flexible imaging plate. Look up the mapping table generated in step ③ to obtain the pixel's theoretical coordinates in the standard plane coordinate system. For pixels falling within a grid cell, calculate the pixel's precise planar coordinates using a weighted average based on the mapping relationship of the four vertices of its grid cell.

[0066] Step 5: Grayscale resampling and image reconstruction: The grayscale value of the original pixel is assigned its corresponding coordinate position in the standard plane coordinate system. Since the pixel distribution on the standard plane may be uneven after coordinate transformation, some positions may be empty. These empty positions are filled using the following method: First, find the already assigned pixels around the empty position; then, calculate the grayscale value of the empty position by weighting the distance; finally, rearrange all the filled pixels according to the standard plane coordinate system to form a complete, geometrically distortion-free corrected image.

[0067] The core advantage of this algorithm lies in: The mapping relationship depends solely on the actual physical deformation data of the imaging plate, rather than the content features of the image itself. Therefore, even in cases of low image contrast (e.g., unclear soft tissue imaging), blurred tissue structures (e.g., slight blurring caused by patient movement), or indistinct image features (e.g., uniform density bone regions), this method can still stably achieve geometric correction, independent of the accuracy of image recognition algorithms.

[0068] Furthermore, the dynamic exposure field model performs local edge exposure compensation or hardware adjustment of milliampere-second parameters on the generated image based on the actual depth difference between different curved areas of the flexible imaging plate and the X-ray tube.

[0069] In this embodiment, a time-of-flight ranging sensor is installed at the end of the X-ray tube. By emitting light pulses and measuring the time it takes for the reflected light to return, the sensor calculates in real time the straight-line distance between the X-ray tube and the flexible imaging plate, i.e., the source-image distance.

[0070] In this embodiment, when the source-image distance changes due to terrain undulations or equipment movement, for example, when the X-ray tube is raised so that the distance increases from 100 cm to 120 cm, the control unit calculates the exposure adjustment value according to the inverse square law, that is, the exposure is inversely proportional to the square of the distance. When the distance increases, the exposure needs to be increased accordingly, and when the distance decreases, the exposure needs to be decreased accordingly.

[0071] In this embodiment, the milliampere-second (mA / s) parameter is a key parameter for controlling X-ray exposure. It is determined by the product of the tube current (mA) and the exposure time (s). The control unit automatically increases or decreases the mA / s parameter based on the calculated exposure adjustment value. For example, the mA / s value is automatically increased when the source-image distance increases, ensuring that images acquired at different distances have consistent brightness and avoiding images that are too dark or overexposed. The closer the distance, the smaller the exposure range, which may not achieve the diagnostic purpose. The appropriate distance and exposure parameters vary in different parts of the body (tissue thickness, density differences). Both distance and exposure parameters must be considered; the distance is not infinitely close or far, and the mA / s is not infinitely large or small.

[0072] In this embodiment, the dynamic exposure field model refers to a mathematical model constructed based on the source-image distance and a three-dimensional surface model, describing the spatial relationship between different regions of the flexible imaging plate and the X-ray tube. Specifically, this model geometrically correlates the spatial coordinates of each grid cell on the three-dimensional surface model with the focal position of the X-ray tube, calculating the actual distance and incident angle of each grid cell relative to the X-ray tube. Since the distances between different regions and the X-ray tube vary when the flexible imaging plate bends (e.g., concave regions are farther away, convex regions are closer), this model can quantify these spatial differences, providing a precise spatial mapping relationship for subsequent local exposure compensation. Based on this, the control unit calculates the required exposure compensation for each region according to the inverse square law based on the depth difference between different regions, performs local edge exposure compensation on the generated image, or dynamically adapts the overall exposure by adjusting the milliampere-second parameter.

[0073] Furthermore, during motion compensation, the multi-degree-of-freedom robotic arm simultaneously receives real-time data feedback from the first and second inertial measurement units, forming a closed-loop control to continuously maintain the alignment of the X-ray tube's central axis with the normal of the flexible imaging plate's central detection point, or to ensure that the axis is parallel to the equivalent principal normal during surface deformation.

[0074] In this embodiment, closed-loop control refers to the control unit continuously receiving attitude data collected in real time by the first inertial measurement unit and the second inertial measurement unit during the process of driving the multi-degree-of-freedom robotic arm to move, comparing the current attitude with the target attitude, and continuously adjusting the direction and amplitude of the robotic arm's movement based on the deviation value.

[0075] In this embodiment, when the posture of the X-ray tube or flexible imaging plate changes slightly due to external interference during the movement of the robotic arm, such as ground vibration causing the flexible imaging plate to tilt slightly, the control unit immediately calculates a new compensation amount based on the feedback data and drives the robotic arm to make fine adjustments and corrections.

[0076] Furthermore, when the flexible imaging plate is used on a soft stretcher or uneven ground, it undergoes physical bending. An array of curvature sensors collects curvature distribution data in real time under bending conditions, which is used for dynamic updating of the three-dimensional surface model.

[0077] In this embodiment, a soft stretcher refers to a flexible stretcher used for transporting patients. When a flexible imaging plate is placed on the stretcher for imaging, the flexible imaging plate will bend and deform according to the surface shape of the stretcher. Uneven ground refers to uneven ground such as grass or gravel. When the flexible imaging plate is placed on such ground, it will also bend locally due to uneven force.

[0078] In this embodiment, the curvature distribution data refers to the bending degree values ​​collected by the array curvature sensor at each detection point of the flexible imaging plate. For example, the bending radius of the central area of ​​the flexible imaging plate is 50 cm and the bending radius of the edge area is 80 cm, reflecting the actual bending state of the flexible imaging plate at different positions.

[0079] In this embodiment, dynamic updating refers to the system reconstructing the three-dimensional surface model in real time based on the latest curvature distribution data when the bending state of the flexible imaging plate changes. For example, when the bending degree of the flexible imaging plate is aggravated by the pressure of the patient's body or the movement of the stretcher, the three-dimensional surface model is updated synchronously to the latest bending shape to ensure that the model used for image correction is always consistent with the actual state of the flexible imaging plate.

[0080] Furthermore, it also includes: In non-cooperative shooting scenarios, keep the flexible imaging plate in its current position; The control unit controls the multi-degree-of-freedom robotic arm to adjust the position of the X-ray tube based solely on the attitude data of the flexible imaging plate collected by the second inertial measurement unit, so that the central axis of the X-ray tube coincides with the normal of the center detection point of the flexible imaging plate. Alternatively, when the flexible imaging plate undergoes surface deformation, the control unit drives the multi-degree-of-freedom robotic arm to perform motion compensation based on the equivalent principal normal vector fitted by the three-dimensional surface model, so that the central axis of the X-ray tube is parallel to the equivalent principal normal.

[0081] In this embodiment, non-cooperative shooting scenarios refer to situations where critically ill patients, the elderly, or young children are unable to actively adjust their body posture to cooperate with the standard shooting posture. For example, a comatose patient cannot move while lying on their side, or an infant or young child cannot straighten their body when crying.

[0082] In this embodiment, keeping the flexible imaging plate in its current position means not moving or adjusting it under the patient's body or at the point where the body contacts the flexible imaging plate, so as to avoid patient discomfort or positional disruption caused by moving the flexible imaging plate. For example, once the flexible imaging plate is placed under the patient, its position is not moved.

[0083] In this embodiment, the multi-degree-of-freedom robotic arm adopts a six-axis serial structure, with each key component equipped with an absolute encoder. The control unit calculates the target rotation angle of each joint based on the relative Euler angle deviation using an inverse kinematics algorithm, and drives the servo motor to move the end effector of the robotic arm, so that the central axis of the tube is collinear with the equivalent principal normal in space.

[0084] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of this invention and its equivalents, this invention also intends to include these modifications and variations.

Claims

1. A portable image acquisition geometric adaptive correction method, characterized in that, include: A first inertial measurement unit and a time-of-flight ranging sensor are set at the end of the X-ray tube, and a second inertial measurement unit and an array-type curvature sensor are set at the end of the flexible imaging plate to collect the relative Euler angles between the X-ray tube and the flexible imaging plate in real time. An array-type curvature sensor acquires data on the bending radius and stress point distribution of a flexible imaging plate; When a deviation in the angle between the central axis of the X-ray tube and the normal of the flexible imaging plate is detected based on the relative Euler angle, it is determined whether the degree of bending of the flexible imaging plate exceeds a preset threshold. If the preset threshold is not exceeded, the control unit drives the multi-degree-of-freedom robotic arm to perform motion compensation so that the central axis of the X-ray tube coincides with the normal of the center detection point of the flexible imaging plate. If the preset threshold is exceeded, the control unit constructs a three-dimensional surface model based on the bending radius and fits the equivalent principal normal vector by combining the force distribution data. This drives the multi-degree-of-freedom robotic arm to perform motion compensation, so that the central axis of the X-ray tube is parallel to the corresponding equivalent principal normal. By utilizing the source-image distance and combining it with a three-dimensional curved surface model, a dynamic exposure field model is established. Based on the actual depth difference between different curved areas of the flexible imaging plate and the X-ray tube, local edge exposure compensation or hardware adjustment of the milliampere-second parameters is performed on the generated image. The original X-ray images acquired are mapped back from the surface coordinate system to the standard plane coordinate system using a three-dimensional surface model, and the corrected images are output. This also includes: It has a built-in standard body position database, which contains the standard incident angles corresponding to different shooting parts; Receive the instruction to select the shooting location and call up the corresponding standard incident angle; Based on the standard incident angle and the current attitude of the flexible imaging plate, the control unit calculates the motion path and drives the multi-degree-of-freedom robotic arm to move to the preset incident angle. The method for calculating the equivalent principal normal in the control unit includes: based on the normal vectors of each mesh element on the three-dimensional surface model, combined with the weight coefficients determined by the force point distribution data, an iterative optimization algorithm is used to find a virtual direction vector that minimizes the weighted deviation between the virtual direction vector and the normal vectors of all mesh elements. The determined virtual direction vector is the equivalent principal normal.

2. The portable image acquisition geometric adaptive correction method according to claim 1, characterized in that, The relative Euler angles include pitch angle, roll angle, and yaw angle. The control unit dynamically adjusts the motion trajectory of the multi-degree-of-freedom robotic arm based on the real-time changes in pitch angle, roll angle, and yaw angle.

3. The portable image acquisition geometric adaptive correction method according to claim 1, characterized in that, The preset incident angle includes a 40-degree tilt angle corresponding to the calcaneal axis.

4. The portable image acquisition geometric adaptive correction method according to claim 1, characterized in that, The process of reversing the acquired raw X-ray images from the surface coordinate system back to the standard plane coordinate system includes: calculating the lateral and longitudinal deformation coefficients based on the bending radius data of each grid cell; establishing a mapping table from the surface parameter coordinates to the standard plane coordinates; traversing the original image pixels to find the mapping table to obtain the standard plane coordinates; and completing grayscale resampling through an interpolation algorithm.

5. The portable image acquisition geometric adaptive correction method according to claim 1, characterized in that, The dynamic exposure field model performs local edge exposure compensation or hardware adjustment of milliampere-second parameters on the generated image based on the actual depth difference between different curved areas of the flexible imaging plate and the X-ray tube.

6. The portable image acquisition geometric adaptive correction method according to claim 1, characterized in that, During motion compensation, the multi-degree-of-freedom robotic arm simultaneously receives real-time data feedback from the first and second inertial measurement units, forming a closed-loop control. This continuously maintains the alignment of the X-ray tube's central axis with the normal of the flexible imaging plate's central detection point, or parallelism with the equivalent principal normal during surface deformation.

7. The portable image acquisition geometric adaptive correction method according to claim 1, characterized in that, The flexible imaging plate physically bends when used on a soft stretcher or on uneven ground. An array of curvature sensors collects curvature distribution data in real time under bending conditions, which is used for dynamic updating of the three-dimensional surface model.

8. The portable image acquisition geometric adaptive correction method according to claim 1, characterized in that, Also includes: In non-cooperative shooting scenarios, keep the flexible imaging plate in its current position; The control unit controls the multi-degree-of-freedom robotic arm to adjust the position of the X-ray tube based solely on the attitude data of the flexible imaging plate collected by the second inertial measurement unit, so that the central axis of the X-ray tube coincides with the normal of the center detection point of the flexible imaging plate. Alternatively, when the flexible imaging plate undergoes surface deformation, the control unit drives the multi-degree-of-freedom robotic arm to perform motion compensation based on the equivalent principal normal vector fitted by the three-dimensional surface model, so that the central axis of the X-ray tube is parallel to the equivalent principal normal.

Images