Calibration phantoms, methods and systems
By using multiple radiopaque plates and checkerboard-patterned calibration phantoms in the CBCT system, the problem of component instability under gravity in imaging equipment was solved, achieving high-precision and low-cost calibration results.
Patent Information
- Application Number
- CN202280053852.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Priority Date
- 2021-06-01
- Filing Date
- 2022-06-01
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2042-06-01
AI Technical Summary
In existing CBCT systems, the relative positions and orientations of imaging equipment components are unstable under the influence of gravity, leading to a decline in image reconstruction quality. Existing phantom calibration methods are complex and lack precision, especially in the semi-fan mode where calibration robustness is poor.
Multiple plates made of radiopaque material are used, each plate is positioned at a certain angle relative to the other plates by a support structure, and each plane calibration pattern faces a different direction. Combined with a checkerboard pattern, the bending parameters of the imaging scanner component are determined by multi-angle projection images.
It improves the calibration accuracy and robustness of imaging scanners, reduces manufacturing complexity and cost, and is suitable for calibration of imaging devices under gravity.
Smart Images

Figure CN117769390B_ABST
Abstract
Description
Technical Field
[0001] This disclosure relates to calibration phantoms and associated systems and methods for calibrating imaging scanners. Background Technology
[0002] Radiation therapy apparatuses typically include imaging capabilities provided by imaging systems configured to provide images of the patient. These images aid in treatment planning and patient localization. Imaging systems may include, for example, a kilovolt (kV) energy radiation source (e.g., X-rays) and a detector (such as a flat panel detector). Such imaging systems are typically mounted at an angle relative to the treatment apparatus on a rotatable gantry. An exemplary imaging modality that can be incorporated into a radiation therapy apparatus is cone-beam computed tomography (CBCT). CBCT images are used in the field of radiation therapy to determine the accuracy of the patient's position on the treatment surface before treatment, compared to images that form the basis of the patient's treatment plan.
[0003] CBCT is an imaging modality that involves using an imaging radiation source, such as X-rays. The imaging radiation is emitted in a cone shape. These diverging imaging radiation beams pass through the patient, are detected by detectors, and projected images are acquired. A three-dimensional image of the object can be generated from multiple such projected images acquired at different angles around the object.
[0004] In a perfectly rigid, mathematically ideal system, the relative distances between each component of an imaging apparatus would be fixed and stable, and the relative orientation of these components would not change. However, in practice, the relative distances and orientations of the components of an imaging apparatus are not fixed and unstable. As the imaging system rotates through the gantry, the imaging system, the gantry, and the mechanical devices connecting the imaging system to the gantry experience small mechanical offsets or "bending" as a function of the gantry angle. In embodiments where the imaging system forms part of a radiotherapy apparatus, the heavy treatment apparatus exacerbates the bending of the gantry and imaging system components under gravity. This treatment apparatus may include a large therapeutic radiation source and may, for example, take the form of a linear accelerator. This heavy treatment apparatus causes additional bending of the gantry as a function of the rotation angle, which affects the relative positions of the imaging system components.
[0005] To address this curvature and ensure good image reconstruction quality, the positions of the detector and kV source must be accurately determined based on the gantry rotation angle. This can be achieved by capturing images of a phantom with known characteristics at different gantry rotation angles and using the resulting projected images to calculate information about the position and / or orientation of the detector and kV source as a function of the gantry angle. Figure 2 depicts a phantom 300 according to the prior art, which has been used to date in such calibration methods.
[0006] The phantom 300 includes a set of radiopaque ball bearings 330 arranged in a specific pattern, comprising an upper ring 310 and a lower ring 320 of ball bearings spaced circumferentially around the phantom. The phantom 300 is functional; however, using such a phantom 300 to calibrate an imaging device presents several drawbacks. For example, the ball bearings 330 must be aligned with the pattern with very high accuracy, making the manufacture of the phantom 300 complex and expensive. Even slight misalignments in the ball bearing positions can lead to suboptimal calibration. Furthermore, in the projected image of the ball bearings 330, each ball bearing occupies a small number of pixels, and the sub-pixel position of each ball bearing is determined based on a small number of pixels at the edges of the balls. The small number of pixels per ball bearing adds additional uncertainty to the method of determining the sub-pixel position of the ball bearing center.
[0007] Furthermore, some modern CBCT systems allow the detector to be moved to an offset "half-fan" position to increase the field of view of the imaged object. When imaging in half-fan mode, if the phantom 300 is placed in the center of the imaging system's field of view, the projected image may not capture the entire phantom 300. However, the robustness of calibration methods using the phantom 300 is affected when only a portion of the phantom 300 is visible in each projected image. Therefore, alternatives to the phantom 300 are desired to ensure the calibration accuracy required for imaging modes with offset detectors. Additionally, if the entire phantom is not included in the projected image for each gantry angle, different bearing balls 300 can be observed in different projected images when scanning in half-fan mode, thereby increasing sensitivity to small deviations in the relative positions of the ball bearings.
[0008] This invention seeks to address these and other drawbacks encountered in the prior art. Summary of the Invention
[0009] The invention is described in the claims.
[0010] According to one aspect, a calibration phantom for calibrating an imaging scanner is provided. The imaging scanner includes a rotatable gantry and imaging radiation sources and detectors, both configured to rotate with the gantry. The calibration phantom includes a support structure and a plurality of plates, each plate including a planar calibration pattern formed of a radiopaque material, wherein the plates are positioned along a first axis. Each plate is positioned by the support structure at an angle relative to each of the other plates, such that each planar calibration pattern faces a different direction, wherein the direction faced by each planar calibration pattern is substantially perpendicular to the first axis.
[0011] Optionally, the imaging radiation source and detector are configured to rotate with the rack due to their attachment to the rack.
[0012] Optionally, the plurality of plates includes at least three plates, each of which is positioned at an angle relative to the other plates by a support structure, such that each planar calibration pattern faces a different direction.
[0013] Optionally, each plate is positioned by a support structure such that when the calibration phantom is imaged via imaging radiation incident from any direction perpendicular to the first axis, at least a portion of at least two planar calibration patterns are included in the image.
[0014] Optionally, the plurality of plates includes at least: a first plate including a first planar alignment pattern; a second plate including a second planar pattern; and a third plate including a third planar pattern. The first planar pattern is positioned facing a first direction by a support structure, the second planar pattern is positioned facing a second direction by a support structure, and the third planar pattern is positioned facing a third direction by a support structure. The first direction, the second direction, and the third direction all intersect each other at an inclined angle.
[0015] Optionally, the plates in the plurality of plates are separated from each other along a first axis.
[0016] Optionally, each of the plurality of plates includes a substrate, and the planar calibration pattern is formed by metal laminated on the substrate. Each of the calibration patterns may be formed via a subtractive manufacturing process. The subtractive manufacturing process may include one or more of the following processes: chemical etching, physical milling, digital lithography, and photolithography.
[0017] Optionally, the support structure has a central axis aligned with or parallel to the first axis.
[0018] Optionally, the support structure includes a plurality of stackable elements, each stackable element being stacked along a first axis, and each stackable element including at least one recess configured to hold one of the plurality of plates in place.
[0019] Optionally, each planar calibration pattern comprises multiple intersecting lines. The planar calibration pattern can be a checkerboard pattern comprising repeating diamond patterns.
[0020] This document discloses a calibration phantom for calibrating an imaging scanner. The imaging scanner includes a rotatable frame and imaging radiation sources and detectors, both configured to rotate with the frame. The calibration phantom includes a support structure and a plurality of plates, each plate comprising a planar calibration pattern formed of a radiopaque material, wherein the plates are stacked relative to each other along a stacking axis. Each plate is positioned by the support structure at an angle to each of the other plates such that each planar calibration pattern faces a different direction, wherein the direction faced by each planar calibration pattern is substantially perpendicular to the stacking axis.
[0021] According to one aspect, a method for calibrating an imaging scanner is provided, the imaging scanner including a rotatable gantry and imaging radiation sources and detectors configured to rotate with the gantry. The method includes receiving a plurality of projected images of a phantom positioned within the field of view of the imaging scanner, the projected images being captured from a plurality of gantry rotation angles. The phantom includes a plurality of plates, each plate including a calibration pattern formed of a transmissive material, and each plate is positioned by a support structure at an angle to each of the other plates, such that each planar calibration pattern faces a different direction. The method further includes determining at least one calibration parameter based on the projected images. The at least one calibration parameter indicates the difference between the nominal position or orientation of a component of the imaging scanner and the actual position or orientation of the component of the imaging scanner.
[0022] At least one calibration parameter can indicate the bending of an imaging radiation source or detector under gravity. For example, the difference between the nominal position or orientation of a component and its actual position or orientation can indicate the bending of the component under gravity.
[0023] Optionally, the method further includes acquiring projected images by capturing multiple projected images at multiple rack rotation angles. The method also includes identifying the locations of edge points in each projected image and associating the identified edge point locations with lines of a calibration pattern.
[0024] Optionally, the calibration pattern also includes a plurality of intersecting lines, wherein the lines intersect to form intersections, and the method further includes: identifying intersections in the projected images; and determining at least one calibration parameter based on the intersections identified in each projected image.
[0025] According to another aspect, a system for calibrating an imaging scanner is provided, the imaging scanner including a rotatable gantry and imaging radiation sources and detectors configured to rotate with the gantry. The system includes a phantom comprising a plurality of plates, each plate including a calibration pattern formed of a radiopaque material, and each plate being positioned at an angle to each of the other plates by a support structure such that each planar calibration pattern faces a different orientation. The system also includes a computer-readable medium including computer-executable instructions that, when executed by a processor, cause the processor to receive projected images of the phantom taken by the imaging scanner from the plurality of gantry rotation angles, and based on the projected images, determine at least one calibration parameter. The at least one calibration parameter indicates the difference between the nominal position or orientation of a component of the imaging scanner and the actual position or orientation of the component of the imaging scanner.
[0026] At least one calibration parameter can indicate the bending of an imaging radiation source or detector under gravity. For example, the difference between the nominal position or orientation of a component and its actual position or orientation can indicate the bending of the component under gravity.
[0027] Alternatively, the phantom is the phantom described as above or elsewhere in this document.
[0028] Optionally, at least one calibration parameter relates to at least one of the location of the imaging radiation source, the location of the detector, and the orientation of the detector.
[0029] Optionally, at least one calibration parameter is determined as a function of the rack rotation angle.
[0030] According to one aspect, a computer-readable medium is provided, the computer-readable medium including computer-executable instructions that, when executed by a processor, cause the processor to perform the methods described above or elsewhere herein.
[0031] This paper discloses a method for calibrating an imaging scanner. The imaging scanner includes a rotatable gantry and imaging radiation sources and detectors, both configured to rotate with the gantry. The method includes: positioning a phantom comprising multiple plates within the field of view of the imaging scanner, each of the multiple plates including a calibration pattern formed of a radiopaque material, and each plate being positioned at an angle to each of the other plates by a support structure such that each planar calibration pattern faces a different direction. The method includes: acquiring projected images of the phantom from the multiple gantry rotation angles; and determining, based on the projected images, at least one calibration parameter indicating the bending of the imaging radiation source or detector under gravity. Attached Figure Description
[0032] Specific embodiments will now be described by way of example only and with reference to the accompanying drawings, in which:
[0033] Figure 1 A radiotherapy apparatus or device according to this disclosure is described;
[0034] Figure 2a It is a schematic diagram depicting a configuration suitable for "full sector" operation. Figure 2b It is a schematic diagram depicting a configuration suitable for "semi-sector" operation;
[0035] Figure 3 A physical model including ball bearings is depicted according to the prior art;
[0036] Figure 4a and Figure 4b A phantom according to this disclosure is depicted, wherein the supporting structure of the phantom is depicted as transparent to aid understanding;
[0037] Figure 5a Depicting Figure 4a and Figure 4b A partially disassembled phantom, or a phantom in an unfinished manufacturing stage. Figure 5b Depicting Figure 4a and Figure 4b phantom;
[0038] Figure 6 A calibration pattern according to this disclosure is depicted;
[0039] Figure 7a and Figure 7b Depicting different fields of view Figure 4a and Figure 4b The projected image of the phantom;
[0040] Figure 8 The method according to this disclosure is described;
[0041] Figure 9a and Figure 9b Different mathematical coordinate systems that can be used in the methods of this disclosure are described;
[0042] Figure 10 This is a block diagram depicting an embodiment of a computing device configured to perform the methods of this disclosure;
[0043] Figure 11 This is another embodiment of the phantom according to the present disclosure. Detailed Implementation
[0044] In general, this disclosure relates to a calibration phantom for calibrating an imaging scanner. The imaging scanner may be a CBCT scanner and may be part of a radiotherapy apparatus. The imaging scanner includes a rotatable gantry. The imaging scanner includes an imaging radiation source and a detector, both of which are configured to rotate together with the gantry. Therefore, the imaging radiation source and detector can rotate around the patient to different angles to acquire projected images from multiple angles around the patient. These images can be used to form a 3D image via a reconstruction algorithm.
[0045] The calibration phantom itself comprises a support structure and multiple plates. Each plate includes a planar calibration pattern formed of a radiopaque material, thus visible in the projected image acquired by an imaging scanner. These plates are positioned relative to each other along a first axis. This first axis may be the central axis of the support structure. Each plate is positioned by the support structure at an angle to each of the other plates, such that each planar calibration pattern faces a different direction, wherein the direction faced by each planar calibration pattern is substantially perpendicular to the first axis. This has the technical effect of ensuring that when the phantom is placed in the imaging scanner and the first axis is parallel to the gantry rotation axis, at least a portion of the calibration pattern is visible in the projected image regardless of the gantry rotation angle. In a particularly advantageous embodiment, at least three plates are present, and each plate is positioned at an angle to the other plates. This arrangement ensures that calibration patterns from at least two different plates are visible in each projection, and that the same plate is visible in each pair of projections separated by a 180° rotation of the gantry.
[0046] This disclosure also relates to a system for calibrating an imaging scanner, the system comprising a phantom and a computer-readable medium including computer-executable instructions that, when executed by a processor, cause the processor to acquire (or otherwise receive) projected images of the phantom from multiple gantry rotation angles, and determine, based on the projected images, at least one parameter related to sag under the influence of gravity of an imaging radiation source or detector. The system may also include the scanner itself.
[0047] The method disclosed herein seeks to calibrate nine parameters for an imaging scanner (which has components subjected to sagging and bending under gravity): the position of the imaging radiation source (x_s, y_s, z_s) and the position (x_d, y_d, z_d) and orientation / rotation (xr_d, yr_d, zr_d) of the detector. These parameters vary as a function of the gantry angle due to bending in the mechanical system during gantry rotation, and this calibration method identifies the bending parameters as a function of the gantry angle.
[0048] Some existing methods for calibrating optical cameras have utilized checkerboard patterns. In such calibration tasks, it is typically desirable to determine three to five internal camera parameters and six external parameters. However, these existing methods and calibration phantoms use a single checkerboard visible to the optical camera, which is unsuitable for use with imaging scanners where bending or sagging under gravity is a problem, such as CBCT systems. This is because these existing methods rely on constant internal camera parameters between different images. This is not the case with CBCT scanners, where bending and sagging cause the relative positions and orientations between imaging components to vary as a function of the gantry rotation angle. The phantom of this disclosure overcomes this problem by specifically positioning the calibration patterns (along an axis and each calibration pattern facing a different direction in the manner described below). This means that at least a portion of the two planar calibration patterns will be visible in each projected image, which allows for the determination of all nine calibration parameters. Furthermore, this calibration method does not rely on the relative positions of the plates / patterns, as the relative position information is determined as part of the calibration.
[0049] Electronic circuit board manufacturing techniques, such as etching and photolithography, can be used to produce boards that include radiopaque calibration patterns. Therefore, the board can include a substrate with a metallic coating (such as a copper coating) that has undergone subtractive manufacturing processes to produce the appropriate calibration pattern. Such PCB technology is a highly precise process that can produce planar (2D) patterns with micron-level accuracy at low cost. Copper is the most commonly used material in circuit boards and has the advantage of being radiopaque. Therefore, the patterns on the circuit board are highly visible in X-ray imaging.
[0050] The geometric calibration of an optical camera is a different and slightly simpler problem compared to the geometric calibration of an imaging scanner, which may undergo bending under gravity. This sag / bending is not a problem for optical camera calibration, and therefore fewer parameters need to be determined. For example, in the case of an optical camera, fewer parameters change from one projected image to another. This checkerboard defines lines that can be identified and parameterized with high precision and accuracy. Their numerous intersections on the image board, corresponding to the imaging corners of the checkerboard squares in the 3D checkerboard object, are then used to calculate nine desired system parameters.
[0051] In the currently disclosed embodiments of the phantom, the phantom may therefore include a circuit board with a checkerboard pattern or other well-defined intersecting line pattern, and the calibration method may involve taking X-ray images from multiple gantry rotation angles. The X-ray images of the board are used to define projection points, which in turn are used to determine the X-ray source and detector geometry at each gantry angle. The method uses information from several or all images taken at different gantry angles during calibration to determine the relative position and rotation of the circuit board. Therefore, the calibration result does not depend on the precision or accuracy of the phantom components, but rather on the precision or accuracy (higher) of the pattern making and the flatness of the circuit board, which is ensured by the rigidity of the board and the supporting structure.
[0052] Figure 1 A radiotherapy apparatus 100 is depicted, adapted to deliver a radiation beam to a patient during radiotherapy and configured to deliver a radiation beam to a patient during radiotherapy. The apparatus 100 and its components will be described generally for the purpose of providing useful supplementary information to aid in understanding the currently disclosed calibration phantoms and methods. Figure 1 The apparatus 100 depicted is an apparatus according to the present disclosure and is suitable for use with the disclosed calibration phantom and calibration method.
[0053] Radiation therapy can be described as treating a human or animal body using ionizing radiation (such as high-energy X-rays). Radiation therapy is commonly used to treat tumors in a patient or subject. In such treatment, ionizing radiation is used to irradiate the body, thereby destroying or damaging the cells that form part of the tumor.
[0054] Figure 1 The device 100 depicted includes a therapeutic radiation source 116 and a radiation detector 118. The radiation source 116 may include a beam generation system. For a linear accelerator, the beam generation system may include an RF energy source, an electron gun, and a waveguide. The therapeutic radiation source 116 is attached to a rotatable gantry 120 so as to rotate together with the gantry 120. In this way, the radiation source 116 can rotate around the patient, allowing a therapeutic beam to be applied from different angles around the gantry 120.
[0055] Radiation detector 118 is positioned opposite radiation source 116. Radiation detector 118 is adapted to generate and is configured to generate radiation intensity data. Specifically, radiation detector 118 is positioned and configured to detect the intensity of therapeutic radiation that has passed through the patient. Radiation detector 118 can also be described as a radiation detection device and can form part of a portal imaging system.
[0056] Figure 1The device 100 depicted also includes an imaging system comprising components configured to rotate with a gantry 120. These components include an imaging radiation source 112 and a detector 114 fixed to the gantry 120. The imaging radiation source 112 may be a kV X-ray source. Together with other components such as connecting cables, onboard processors, etc., the imaging radiation source 112 and imaging detector 114 comprise an imaging device or imaging system rotatable around a patient. The imaging system provides the device 100 with imaging capabilities, and in particular X-ray, CBCT, and / or CT imaging capabilities. By rotating the imaging system around the patient, projected images can be acquired from multiple gantry angles, and three-dimensional images can be reconstructed based on these projected images using known tomographic techniques. The imaging capabilities of the device 100 mean that the device 100 is an imaging scanner.
[0057] Imaging radiation source 112 and imaging detector 114 are coupled to gantry 120. Components 112 and 114 are heavy, and gravity acts on these components as gantry 120 rotates, resulting in a bending or "sagging" effect. The reconstructed images acquired by the imaging scanner may not be optimal unless this bending is accounted for. These bending effects are further exacerbated when the imaging scanner is also a radiotherapy device including detector 118 and a heavy treatment head containing radiation source 116.
[0058] The device 100 also includes a controller (not shown) in the form of a computer, processor, or other processing device. The controller may be formed from several discrete processors. For example, the controller may include: an imaging processor that controls an imaging component such as an imaging radiation source 112 and is configured to receive data and images acquired by an imaging detector 114; an RT device processor that controls the operation of an RT component such as a therapeutic radiation source 116 and is configured to receive data and images acquired by a detector 118; and a patient support surface processor that controls the operation and actuation of a patient support surface (not shown). The controller is communicatively coupled to a memory, such as a computer-readable medium. The controller may employ... Figure 10 The form shown and described below.
[0059] As those skilled in the art will understand, device 100 also includes several other components and systems. For example, appropriate shielding is provided to ensure that the linear accelerator does not leak radiation. In use, device 100 will form part of a radiotherapy system that also includes a patient positioning system for positioning the patient for imaging and / or treatment. Such a patient positioning system can also be used to position a phantom within the field of view of device 100 for QA and calibration purposes.
[0060] In some imaging scanners, the field of view can be adjusted. This function is available in... Figure 2a and Figure 2b Depicting in the middle. For example Figure 2a and Figure 2b As shown, the X-ray source 202 of the CBCT imaging scanner (which can also be a radiotherapy device) is configured to emit a fan-shaped beam. The fan-shaped beam... Figure 2a and Figure 2b The beam is shown in dashed lines, and the central axis 208 of the beam is shown in solid lines. The size of the fan-shaped beam can be varied depending on the size of the patient and / or the area to be imaged.
[0061] Figure 2a It is a schematic diagram depicting a configuration suitable for "full sector" operation, and Figure 2b This is a schematic diagram depicting a configuration suitable for "semi-sector" operation. For example... Figure 2a As shown, for full-sector operation, panel detector 204 is centered on the central axis 208 of the X-ray beam, that is, on the central axis 208 of the "sector" formed by the radiation. Panel detector 204 can be described as being located at a first position, first orientation, or first positioning suitable for full-sector scanning.
[0062] In the “full-sector” configuration, the panel detector 204 and the X-ray source 202 can rotate 180 degrees around the patient to acquire projection data. When the panel detector 204 and the X-ray source 202 are rotated 180 degrees plus the sector angle, imaging software and known reconstruction techniques can be used to generate a complete image of the target localization.
[0063] like Figure 2b As shown, for "half-sector" operation, detector panel 204 is positioned such that it is not centered on the central axis 208 of the X-ray sector. Instead, panel detector 204 is moved laterally to one side. In other words, panel detector 204 is laterally offset or shifted to one side relative to its position during full-sector operation. Panel detector 204 may also be referred to as "off-center" panel detector 204. Panel detector 214 can be described as being located in a second position, second orientation, or second positioning suitable for half-sector scanning.
[0064] Existing methods for calibration systems used in both full-sector and half-sector operations benefit from displaying the entire phantom (such as...) at each projection angle. Figure 3The phantom shown is used to ensure that the resulting projected image contains enough points to perform calibration. However, this is problematic for half-fan imaging because only half of the phantom is imaged at each angle. When these existing phantoms are imaged in a half-fan scanning mode, only about half of the ball bearings will be visible in the resulting projected image. When half of the ball bearings cannot be found in the image, calibration methods that rely on fitting geometry (e.g., ellipse) based on a known arrangement of the ball bearings will not provide accurate calibration. Furthermore, relative projected images, i.e., those taken from rack positions approximately or precisely 180° apart, may not depict the same bearing balls. This places higher demands on knowing the precise phantom geometry, for example, by precisely manufacturing the phantom. Therefore, when using these known phantoms for calibration, geometric parameters (e.g., geometric parameters describing the bending and sagging of the system under gravity) may not be determined to achieve the accuracy expected in a half-fan / offset imaging mode.
[0065] The phantom of this application mitigates and / or solves these problems by using a phantom comprising a planar calibration pattern consisting of intersecting lines of a transmissive material. At least some of the lines constituting the intersecting line pattern have components extending along the width and length of the pattern. By providing lines of transmissive material extending to a certain extent along the width and length of the calibration pattern, the projected images from a half-sector scan may comprise pairs of opposing projected images in which different portions of the same line of the calibration pattern can be seen. This allows the opposing projected images to be matched and aligned relative to each other more easily in a way that would be impossible when using patterns including ball bearings. By identifying when different portions of the same line are imaged in this manner, the calibration method is able to provide robust calibration even when the entire phantom is not visible in each projected image. An example of a suitable calibration pattern is a diamond checkerboard pattern.
[0066] Figure 4a and Figure 4b A phantom 400 according to this disclosure is depicted. The phantom includes a support structure 410 and a plurality of plates 422, 424, 426. The support structure 410 comprises a transmissive material or a material that is at least more transmissive than the calibration pattern, and is configured to hold / support the plates 422, 424, 426 in an arrangement such that each plate faces a different orientation. The support structure 410 also holds / supports the plates 422, 424, 426 such that they are positioned relative to each other along a first axis 430. The plates are positioned in such a way that when the phantom 400 is imaged from a direction perpendicular to the first axis 430, the overlap of the calibration pattern is minimized when projected onto a detector.
[0067] Plates 422, 424, and 426 are flat and elongated, and have rectangular faces. Each plate 422, 424, and 426 includes a calibration pattern. The calibration pattern is formed of a radiopaque material. Figure 4a , Figure 4b The opaque pattern shown is a checkerboard pattern. The checkerboard pattern includes repeating diamond shapes. The pattern is depicted on... Figure 6 The transmissive pattern extends substantially along the entire surface of one of the main rectangular faces of the plate, as described in more detail below. Figure 3 The use of a checkerboard pattern is advantageous compared to the ball bearings employed in the prior art phantom 300 depicted in the diagram, because the checkerboard pattern defines lines and dots. Regardless of whether the imaging scanner is in full-sector or half-sector scanning mode, portions of the same line are visible at all gantry angles. In this method, not all lines need to be visible. The accuracy used to determine the position of lines in the calibration pattern is also higher than the accuracy used to determine the position of ball bearings, because in most cases, the lines will include more pixels in the projected image.
[0068] The support structure 410 is cylindrical and has a diameter similar in length to that of each plate 422, 424, 426. The depicted calibration pattern spans the length of each plate, and each plate spans the diameter of the cylindrical support structure.
[0069] Support structure 410 supports and positions plates 422, 424, and 426, such that plates 422, 424, and 426 are positioned relative to each other along the first axis 430. (When viewing...) Figure 4a At this time, the first axis 430 enters the plane of the page. The first axis 430 is aligned with the central axis of the cylindrical support structure 410. Plates 422, 424, and 426 are positioned such that the centers of each plate are spaced apart in a direction parallel to the first axis 430. The support structure 410 positions plates 424, 426, and 428 at an angle to each other, such that each plane alignment pattern faces a different direction. One or more angles ensure that the plates are not coplanar (the angle is zero in the case of coplanarity) or perpendicular to each other (the angle is 90° in this case). Therefore, plates 424, 426, and 428 are positioned / stacked / supported in a non-coplanar arrangement.
[0070] Each planar calibration pattern includes a central length axis and a central width axis. Plates 424, 426, and 428 are positioned / stacked along a first axis 430 such that the central width axes of the plates are aligned with each other and are parallel to the central axis 430 of the support structure 410. Each plate 424, 426, and 428 is translated along and rotated about the central axis 430 compared to the previous plate. The first axis 430 may be referred to herein as the "stack axis" along which plates 424, 426, and 428 may be stacked relative to each other.
[0071] exist Figure 4a and Figure 4b In the example shown, the first plate 422 includes a first calibration pattern facing a first direction 432; the second plate 424 includes a second calibration pattern facing a second direction 434; and the third plate 426 includes a third calibration pattern facing a third direction 436. Each of the first direction 432, the second direction 434, and the third direction 436 intersects each other at an angle relative to each other. Figures 4a to 4b In the embodiment shown, plates 424, 426, and 428 are positioned at 60 degrees or approximately 60 degrees relative to each other. Each of the first direction 432, the second direction 434, and the third direction 436 towards which the planar calibration pattern faces is substantially perpendicular to the first axis.
[0072] In use, the calibration phantom 400 is positioned such that its first axis (or, equivalently, its central axis for the embodiment depicted in the figure) is substantially aligned with the axis of rotation of the gantry. This positioning relative to the rotatable gantry of the imaging scanner... Figure 4a The direction is depicted by a curved arrow. Each of the directions 432, 434, and 436 that the calibration pattern faces is an example direction from which the imaging direction can be directed toward the phantom 400.
[0073] During the calibration process, projected images of the phantom are acquired from multiple gantry rotation angles, and based on these images, parameters indicating the bending of the imaging scanner's components under gravity are obtained. The use of planar calibration patterns makes calibration more accurate compared to existing methods. Checkerboard patterns are particularly advantageous because they provide a large number of points defined by the corners of squares, rectangles, or rhombuses. Calibration can then rely on the fact that the straight lines formed by the squares, rectangles, or rhombuses remain straight after the projection transformation.
[0074] Furthermore, by positioning (stacking) the boards relative to each other, such that each planar calibration pattern faces a different direction, and these directions are substantially perpendicular to the first (stacking) axis, it is ensured that at least a portion of the pattern is visible from each rack rotation angle. Moreover, due to the way the boards are positioned, there is no confusion about which board the pattern in the projected image is associated with. Figures 4a to 4bIn the embodiment shown, at least three plates are used and each plate is positioned at an angle relative to each other, and at least two portions of different plates will be found in all projected images regardless of the rack rotation angle.
[0075] As mentioned above, Figure 3 The existing phantom 300 depicted in the previous work must have ball bearings for very high accuracy in aligning with the pattern, making phantom manufacturing complex and expensive. In contrast, the only accuracy required for the phantom of this disclosure is accuracy in the 2D board pattern. Ensuring high accuracy in planar alignment of the pattern is much easier; for example, high accuracy is an inherent characteristic of circuit board manufacturing. PCB manufacturing provides high accuracy at a lower manufacturing cost.
[0076] The relative positioning of patterns 310 and 320 must also be highly precise, or known to be of high precision and accuracy, to ensure accurate results when using existing phantoms 300. However, using currently disclosed calibration patterns, particularly those with intersecting lines (such as checkerboard patterns), as part of the calibration process, the precise position of the plate itself relative to other plates or system geometry can be easily determined, making phantom 400 easier to manufacture compared to known phantoms. This is because the pattern on a single plate allows for the determination of eight degrees of freedom, while the position of one plate relative to another can be fully characterized by six degrees of freedom (three translations and three rotations). Thus, projected images showing two plates can be used to determine up to 16 parameters, while the configuration itself has 15 degrees of freedom (nine projection parameters plus six relative translations and rotations of the second plate relative to the first). In the case where three plates are visible in the projected images, there are 21 degrees of freedom to characterize the configuration, but the images provide the ability to determine 24 parameters. Therefore, regardless of the number of projected images, the projection parameters and phantom geometry can be fully determined in the same optimization process.
[0077] Another way to discuss the aforementioned advantageous properties of the phantom and method is to discuss the gold standard used in the calibration method. For the phantom of this disclosure, the checkerboard pattern itself serves as the gold standard. The relative positioning of the patterns with respect to each other can be determined. A calibration process can be performed to determine calibration parameters as part of the same process. In contrast, for existing ball bearing phantoms, what is required is a gold standard with high accuracy in the known positioning of the ball bearings relative to each other, or, as part of the optimization method, the ball bearing geometry must be determined. However, in order to perform the determination of the ball bearing geometry as part of optimization, numerous images of the ball bearing from a highly stable or pre-calibrated imaging geometry must be provided to provide sufficient accuracy for the determination, because a single projected image of the ball bearing does not provide as many degrees of freedom as an image of the calibration pattern according to this disclosure.
[0078] Figure 5a and Figure 5b A phantom 500 is depicted, having a support structure including stackable support elements 512, 514, 516, and 518. The phantom 500 is related to the above-mentioned... Figure 4a and Figure 4b The phantom 400 is described in the same manner. Each stackable element 512, 514, 516, 518 includes at least one recess configured to receive plates 522, 524, 526. The stackable elements 512, 514, 516, 518 are configured to stack along a first (stack) axis. The recess extends across the diameter of the support structure in a direction perpendicular to the stack axis. The length of each recess is substantially equal to the length of each plate 522, 524, 526. The depth of each recess in a direction parallel to the stack axis (or equivalently, in a direction parallel to the length / central axis of the phantom) is approximately half the width of each plate 522, 524, 526.
[0079] Each stackable support element 512, 514, 516, 518 includes two opposing faces. (For...) Figure 5a , Figure 5b The phantom 500 depicted has circular opposing faces. Two of the stackable support elements 512 and 518 are end elements and are configured to be positioned at the ends of a complete stack of stacked support elements. These end support elements include recesses in which a plate is positioned in the assembled phantom 500. The recess is formed in one of its multiple faces. The remaining stackable support elements 514 and 516 are inner stackable elements and include two recesses, each formed on one of the opposing circular faces of the inner stackable support element. Thus, the calibration phantom 500 can be easily assembled from its components. During phantom assembly, a first plate 522 is positioned in the recess of the first end stackable element 512. A first inner stackable support element 514 is stacked on top of the end stackable support element 512, and the plate 522 is positioned in one of the recesses of the first inner stackable support element 514. Therefore, plate 522 is supported in the two recesses of two stackable support elements 512, 514.
[0080] The recess on the opposite surface of the first internal stackable support element 514 can now accommodate another plate 514. Figure 5a The components depicted show the phantom at this assembly stage. The process of stacking stackable support elements and positioning the plates such that they are held in place by the recesses of two different stackable support elements continues until the desired number of plates are reached, at which point the second end stackable element 518 completes the stacking to form the assembled phantom 500.
[0081] Figure 5a , Figure 5bA phantom with a support structure is depicted, comprising a plurality of stackable elements 512, 514, 516, and 518. Each stackable element 512, 514, 516, and 518 is stacked along a stacking direction. Each stackable element 512, 514, 516, and 518 includes at least one recess configured to hold one of the plurality of plates in place. Using stackable support elements in this manner is advantageous because it is more efficient than manufacturing processes such as… Figure 3 Compared to the complex manufacturing methods required for existing phantoms, the phantom described herein represents a simple manufacturing method. Furthermore, the number of plates can be adjusted for different applications (e.g., by adding more plates along the stacking axis to allow for calibration helical scans, where the imaged object moves parallel to the axis of rotation during scanning), or plates with different calibration patterns can be swapped in and out. This advantage cannot be provided by, for example, existing ball bearing phantoms, because the ball bearing geometry is a known quantity for these phantoms and must remain identical to provide gold standard information to the calibration method. Using recesses in the stackable support elements, depending on the requirements of the calibration method, is a simple and effective way to ensure that the plates remain flat and planar.
[0082] Figure 6 An example calibration pattern for use with a currently disclosed phantom is depicted. The pattern comprises multiple intersecting lines that define points at their intersections. The calibration pattern includes two sets of parallel lines. Each set of parallel lines extends in a direction not parallel to either the length or width axis of the phantom. The first set of parallel lines extends in a first direction, while the second set extends in a different second direction. Intersection points are defined where these lines intersect. The intersecting lines define elongated rhombuses, but not all rhombuses have the same opacity linearity. Rhombuses with different opacity linearities are formed using an opacity material to create an alternating dark and light opacity effect in the projected image. There are two sets of rhombuses: dark rhombuses and light rhombuses, where the dark rhombuses have a higher degree of opacity linearity compared to the light rhombuses.
[0083] A calibration pattern is formed by creating two sets of intersecting lines that extend in different directions, neither of which are parallel to the width and length axes of the calibration pattern. If the projected image cannot capture the entire calibration pattern, the lines in the two different directions will be interrupted. This is in... Figure 7bThe calibration pattern is illustrated using an example of a projection image obtained through a half-fan scan. Approximately half of the phantom is depicted. One of the calibration patterns has two interrupted lines 710 extending away from the edge of the projection image, each with two different directions. In this way, opposing projection images obtained through a half-fan scan may capture a greater number of these interrupted lines. These opposing projection images will contain different portions of the same line. By ensuring that the same set of interrupted lines 710 is captured in different projection images in this manner, the calibration method is able to provide robust correspondence between opposing projections, even when the same portion of the phantom is not visible in either projection.
[0084] The pattern is a checkerboard pattern. Checkerboard patterns, or similar intersecting line patterns, are beneficial for finding 3D-2D correspondences because these lines are precisely defined and provide many intersection points. This means that the determination of features in the phantom is highly accurate. The pattern does not contain squares, but rather elongated rhombuses, with the major axis of the elongated rhombuses parallel to the length of the plate. The reason for using elongated patterns is to facilitate pattern recognition on plates imaged at narrow angles. The calibration pattern of this disclosure includes rhombuses instead of rectangles or squares, and this provides several advantages. For example, it forms a greater number of interrupted lines in a half-sector scan, such as... Figure 7b As shown in number 710 and discussed above. Furthermore, the projected lengths of the two line segments defining the corner portion are equalized to reduce the influence of bias and noise in edge detection. Additionally, edge detection may have slight biases, causing detected edges to systematically shift towards dark or bright areas. Examples include this due to the algorithm, blurred projected edges, or etched copper edges not being perpendicular to the surface. A characteristic of alternating light and dark patterns is that the lines identified in the image will typically include several boundaries in alternating directions, thus reducing the influence of arbitrary biases.
[0085] Figure 7a , Figure 7b Example phantom projections of small and large field-of-view images taken at the same projection angle are depicted. The x-axis and y-axis depict the position on the detector (in mm). Figure 7a It is a full-sector configuration (such as the above regarding) Figure 2a Images with a smaller field of view (described in the configuration) captured by a CBCT imager. Figure 7b It is a semi-sector configuration (such as the above regarding Figure 2b The large field-of-view images are captured by a CBCT imager (described in the configuration). The phantom in the projected images is based on... Figure 4a , Figure 4b , Figure 5a , Figure 5b The phantoms depicted in the text include those with Figure 6 The board depicting the pattern.
[0086] The boards used in the phantoms disclosed herein can be manufactured using circuit board manufacturing techniques. Manufacturing techniques for electronic circuit boards (such as etching and photolithography) are highly precise processes that can create planar (2D) patterns with micron-level accuracy. Furthermore, these methods are inexpensive. Copper is the most commonly used material in circuit boards and has the advantage of being radiopaque. Therefore, patterns on circuit boards containing copper will be highly visible in X-ray imaging. Calibration patterns can be formed on PCB substrates using subtractive manufacturing methods such as chemical etching, physical milling, digital lithography, and / or photolithography.
[0087] Using etching techniques, highly attenuated copper patterns with heights of up to several hundred micrometers and lateral resolutions as low as tens of micrometers can be etched onto epoxy resin backings of almost any size.
[0088] The support structure of the phantom is preferably, but not necessarily, a solid cylinder of low-density rigid foam. This allows the board to be supported along its entire length to avoid bending or warping, while also preventing abrupt gradients in the projected path lengths of different parts of the phantom. The support structure can be formed of a translucent material, such as homogeneous foam. This material can be polystyrene foam.
[0089] In an embodiment where the support structure surrounds and encloses the plate, the support structure is formed of a material that is more transparent to X-rays than the calibration pattern, allowing for the acquisition of... Figure 7a and Figure 7b The image depicts the form. However, the supporting structure can take any of several different suitable forms.
[0090] Once projected images of the phantom have been acquired from multiple gantry rotation angles, the processing steps involve using algorithms to determine various geometric parameters, such as the location of the imaging radiation source (x). s ,y s ,z s ) and the position of the detector of the imaging scanner (x d ,y d ,z d ) and rotation (xr d ,yr d ,zr d Once determined, these parameters can be used to improve the accuracy of reconstructing future 3D images from projected images. The imaging scanner can then generate and display 3D tomographic images calibrated by the calibration parameters. These parameters will be referred to herein as calibration parameters, and each parameter indicates the bending of the imaging radiation source or detector under gravity.
[0091] The currently disclosed phantom comprises multiple plates, each with its own calibration pattern formed on it. The six degrees of freedom (three translations and three rotations) of each additional pattern relative to the first pattern can be determined based on the additional information provided by having two or more patterns in each projected image. Since each pattern defines homography with eight degrees of freedom, in the example using two patterns, a single projection (1×2×8 = 16 equations in total) is sufficient to fully determine both the CBCT parameters and the relative orientation of the two plates (1×9+6 = 15 degrees of freedom). However, more projections will improve the estimation of the relative pattern orientation (i.e., the phantom geometry) in the presence of noise.
[0092] The calibration algorithm locates edges in the projected image, fits lines to the edge set, and identifies these lines using lines on the phantom's plates. Line identification is based on the phantom's numerical model, which, given nine projection parameters and six relative position parameters for each plate, outputs the expected positions of lines and intersections in the projected image. Once the lines in the projection are identified, the coordinates of their intersections are calculated. Then, the optimal projection parameters and plate positions are obtained by solving an optimization problem that minimizes the distances between the intersections and / or lines in the projection and their corresponding intersections and / or lines given by the numerical model.
[0093] At a high level, the calibration method includes receiving projected images of a phantom taken from multiple gantry rotation angles and determining at least one calibration parameter based on the projected images to indicate the bending of the imaging radiation source or detector under gravity.
[0094] Calibration parameters can be described in many different ways. Figure 9a and Figure 9b An example geometric parameterization of an imaging scanner, including a rotatable gantry, a radiation source, and a detector, is depicted. Figure 9a The position and orientation of the local source and detector system are shown in the local coordinate system parameterization. Figure 9b The parameters of the actual source location are shown, along with the position and orientation of the local detector coordinate system within the source coordinate system parameterization. Both sides include the fixed coordinate system and the gantry coordinate system.
[0095] Although this method is described in terms of these coordinate systems and geometric parameterizations, those skilled in the art will understand that some of the choices made in establishing these systems and parameterizations are either arbitrary or have been chosen to make mathematical operations more efficient. Therefore, Figure 9a and Figure 9b The specific coordinate system and geometric parameterization described herein are presented only for the purpose of understanding the method and are not fundamental features of the method disclosed herein.
[0096] Figure 9a and Figure 9bDifferent coordinate systems were depicted, especially the fixed coordinate system X relative to the treatment room. f Y f and Z f and the frame coordinate system X, which is rotated and fixed relative to the nominal frame. g Y g and Z g (And therefore it can rotate relative to a fixed coordinate system). For an ideal gantry with a perfectly circular detector and source trajectory, it would be natural to have the origins of both systems coincide with the imaging isocentre. However, as will be explained below, any fixed point in the chamber can be chosen as the origin, and the only effect is that the calibrated gantry parameters are translated and / or rotated. When calibration is performed, the isocentre is initially set as a point relative to the calibration phantom, such as the central rhombus in the calibration pattern. Once these parameters are obtained, new parameters relative to the isocentre position can be defined (these new parameters are independent of the phantom position in the gantry), and this is generally desirable.
[0097] exist Figure 9a The text also describes: the local source coordinate system X s Y s and Z s The origin is located along the positive Z-axis. g The nominal source position is located at 905, at the nominal source isocenter distance (SID) from the isocenter, and its axis is parallel to the rack coordinate system; and the local detector coordinate system X. d Y d and Z d The origin is located at the nominal detector position 915, found at the nominal isocentric detector distance (IDD) along the negative Zg axis, and its axis is parallel to the rack coordinate system.
[0098] Figure 9a The rotation angle φ of the specific rack is described. g The nominal source location 905 (i.e., the location the radiation source would have in a mathematically ideal system) and the actual source location 910 (i.e., the actual location of the source). Due to the sag / bending effect caused by gravity, the actual source location 910 and the nominal source location 905 are partially or completely different from each other. The actual source location 910 can be represented using coordinates X... s Y s Z s The local source coordinate system is used to describe this. The nominal puncture point location 915 is the point on the detector projected onto the isocenter of the nominal geometry, i.e., the point on the detector projected onto the isocenter from the nominal source location 905. The nominal puncture point location 915 can be used as a coordinate system with X... d Y d Z d The origin of the local detector coordinate system is as described above.
[0099] Figure 9b An alternative parameterization of the projected geometry is described. Here, the expected location 925 of the nominal puncture point is a point located at a nominal IDD distance from the isocenter along a direction given by a vector from the true source location 910 to the isocenter. This expected location 925 of the nominal puncture point can be used as another local detector coordinate system X'. d Y' d Z' d The origin, where Z' d The axis is parallel to the vector from the isocenter to the true source location, X' d The axis lies in the XZ plane of a fixed coordinate system. In this parameterization, the true source position is determined by ΔSID (describing the relationship between the nominal source position 905 and the true source position 910 along the Z-axis). g The distance, Δφ (in-plane angle difference between nominal source position 905 and real source position 910) and Δψ (out-of-plane angle difference between nominal source position 905 and real source position 910) are given.
[0100] The true location 910 of the source can be described in a fixed system, a rack system, its local coordinate system, or by parameters ΔSID, Δφ, and Δψ. Similarly, the true location 920 of the nominal puncture point can be described in... Figure 9a and Figure 9b The detector can be described in any of the following coordinate systems: a fixed system, a rack system, or a local detector coordinate system. Finally, the true orientation of the detector must be considered. The true detector orientation can be determined using three rotation parameters X. rd Y rd and Z rd To describe, these three rotation parameters X rd Y rd and Z rd Any of the coordinate systems described above can be used for representation. As mentioned above, different parameterizations can be used according to the method of this disclosure.
[0101] Algorithm 1 provides a suitable algorithm based on Figure 4a and Figure 4b This document provides a detailed overview of one implementation of an algorithm for determining appropriate calibration parameters using a calibration phantom as depicted in the text.
[0102] Algorithm 1
[0103] enter:
[0104] A set of N-projected images P
[0105] The nine projection parameters α i And 18 plate parameters β are mapped to a set Projection lines (and their intersections) Model M
[0106] 9N nominal projection parameters α nom and 18 nominal plate parameters β nom
[0107] Output:
[0108] 9N optimized projection parameters α opt and 18 optimized plate parameters β opt
[0109] For each of the N projected images:
[0110] E i ←Find the projected image P i A set of edge coordinates
[0111] ←Create projection P i The smooth dark image of the expected line set in the image
[0112] t←Found in E i Minimize its value Rigid two-dimensional translation
[0113]
[0114] E i The edge in the middle is projected onto the line F.
[0115] G(E i ,L i )← E i Each edge in L i All lines within the distance and direction thresholds are associated
[0116] According to G(E) i ,L i Find out what makes E i Each edge in Minimize the shortest distance between any associated lines in the 18 plate parameters β +
[0117] L i ←Based on the previous step, create an updated set of expected line segments.
[0118] E i The edge in the middle is projected onto the line F.
[0119] F(E i ,L i )←If it is within the distance and nominal threshold, then E iEach edge in L is mapped to each edge in L. i The nearest line in
[0120] L i ←Targeting L i For each line in F(E), select each line in F(E). i ,L i Find the nearest edge in the graph and fit a new line.
[0121] F * (E i ,L i )← If within the distance and nominal threshold, then map each edge in Ei to that edge in L. i The nearest line in
[0122] L i ←Targeting L i For each line in F, fit a line to that line in F. * (E i ,L i All edges in )
[0123] X i ←Find L i The set of all intersections between lines
[0124] α opt ,β opt ← By minimizing X simultaneously for all projection angles i i Points in the modeling points Find 9N+18 calibration parameters for the 2D distance between them.
[0125] Figure 8 A calibration method 800 is described. This method is suitable for calibrating imaging scanners including rotatable gantry systems. A system for calibrating such a 3D scanner may include a 3D scanner, a phantom as described elsewhere herein, and a computer-readable medium including computer-executable instructions that, when executed by a processor, cause the processor to perform… Figure 1 The method described in the text.
[0126] In frame 801, a processor receives multiple projected images captured from multiple gantry rotation angles. This frame may also include image acquisition, for example, sending instructions to an imaging scanner to perform a scan to obtain images. Each projected image depicts a phantom placed within the imaging scanner's field of view. The phantom includes multiple calibration patterns as described above. In use, the phantom is positioned within the imaging scanner, with its first axis substantially parallel to the gantry rotation axis. The received images may have… Figure 7a and Figure 7b The forms depicted in it.
[0127] In box 802, the positions of edge points in the projected image are located for each projected image. Edge points can be described as features of the image. Depending on the algorithm used, edge points may appear at points in the image where brightness changes abruptly, or at points where the projected image displays other discontinuities. These edge points are likely to appear at the boundaries of objects and at the boundaries of different non-transmissive linear regions. Therefore, at least some of the edge points detected in box 802 may be associated with the calibration pattern of the phantom, some edge points may be associated with the boundaries of the supporting structure, and some edge points may not be associated with the phantom at all, but rather with the "background" of structures such as the patient positioning surface on which the phantom is placed. Edge detection algorithms are known to those skilled in the art, such as the Sobel edge detector. The Sobel edge detector has the advantage of simplicity, making it easy to implement and fast to run.
[0128] In block 803, each identified edge point location is associated with at most one of the lines in the calibration pattern. Therefore, the method includes associating each identified edge point location with at most one of the lines defined by intersecting lines in the calibration pattern; for example, with lines defined by… Figure 6 At most one of the lines defined by the rhombus shape in the calibration pattern is associated. Not all edges in the image need to be associated with lines of the calibration pattern, because, for example, some of the edges identified in box 802 may be associated with the boundaries of the plate or with discontinuities in the supporting structure. Box 803 relates to using a known calibration pattern and may include assigning the identified edge points to possible or trial lines of the known pattern, and iteratively updating the assignment of the identified edge points to lines of the known calibration pattern as more edge point locations are identified and assigned. The processing in this box can be implemented using known algorithms.
[0129] In a specific implementation, the step of associating identified edge points with lines of the calibration pattern from each projected image can be a multi-step process. In such an implementation, at each step, the line pattern guess is improved, and based on the improved line pattern guess, edge points in the projected image are associated with lines in the known calibration pattern. In such a multi-step process, firstly, given a nominal gantry geometry (i.e., a gantry geometry of a mathematically ideal system in which there is no bending or sagging), an image of the expected line pattern on the detector, the gantry angle, and the phantom geometry are calculated and optionally blurred. Then, a rigid offset of the image that minimizes the distance between the pattern and the detected edge points is approximated by minimizing the square of the sum of the values of the (optionally blurred) image at the edge point locations. Next, the edge points are projected onto translated segments of the expected projected line pattern. Edge points are temporarily associated with line segments whose distance is below a given threshold and whose edge normals are sufficiently close to the normals of the projected lines. These associations can be described as “trial” associations. After this initial step, identified edges not associated with any lines of the calibration pattern do not participate in the next step.
[0130] Following this, the position and translation of each plate are found to minimize the sum of the squares of the distances from each participating edge point to its associated projection line. The resulting expected line pattern becomes a guess of the true line pattern. Note that this step is used to find a good guess of the true line pattern, so the plate orientation calculated here is different for each rack angle and may not be used in subsequent steps. All edge points are again projected onto the improved line pattern guess. This time, edge points are associated only with their nearest projection lines, provided these projection lines are close enough and their gradients are sufficiently parallel. In this step, the minimum distance threshold is reduced compared to the previous step.
[0131] In box 804, for each projected image, lines are fitted to all edge points associated with the same line in the calibration pattern. This stage includes fitting line segments to the nearest edge point associated with each line in the calibration pattern, provided that the line has enough associated edge points. These line segments constitute the final estimate of the line pattern of the projection. Each edge point that is sufficiently close to the closest line segment in this estimate is permanently associated with the corresponding line in the calibration pattern. The optimal line fit for the edges associated with each line is obtained.
[0132] In box 805, the intersections of the fitted lines are identified. These intersections are now expected to correspond to the corner positions of the diamond shape in the projection. Boxes 804 and 805 together include fitting lines in one or more projected images to all points associated with the same line in one or more calibration patterns and calculating the intersections between the fitted lines in the images. Each intersection uniquely corresponds to a corner in the checkerboard pattern. For a 1024×1024 pixel projected image, steps 802 through 805 typically take a few seconds to complete.
[0133] At box 806, at least one parameter indicating the bending under gravity is determined. This determination is based on the processing performed in boxes 802 to 805, and specifically on the intersections of the identified fitted lines. The determined parameter indicates the bending of the imaging radiation source or detector under gravity, for example, one of the following: the position of the imaging radiation source (x... d ,y d ,z d ) and the position of the imaging scanner detector (x d ,y d ,z d ) and rotation (xr d ,yr d ,zr d In the preferred method, these nine projection / calibration parameters are found simultaneously for each projection angle. In addition, six relative position parameters of the two plates are also found by minimizing the sum of the squares of the distances between the corner points in the image and the patterns projected by the phantom.
[0134] In the example, at box 806, calibration parameter values can be obtained through nonlinear optimization of the sum of the squares of the 2D distances between the identified intersections and their corresponding points predicted by the numerical model. This is accomplished, for example, by iteratively changing the projection parameters and the parameters describing the pattern plate position in a systematic manner, such as by using the Levenberg-Marquardt algorithm. The result is: six orientation parameters, each for both in the pattern plate; and nine projection (calibration) parameters for each rack angle represented by the projected image received at box 801. Thus, each of the nine calibration parameters is determined as a function of the rack angle.
[0135] In summary, the optimization process then feeds the parameters (projection parameters and those parameters describing the relative positions of the plates of the phantom) into the numerical model, uses the numerical model to predict the positions of the intersections, calculates the distance between the expected intersections and the identified intersections, and then iterates until this distance between the expected intersections and the detected intersection positions based on the parameters is very small, for example, below a threshold.
[0136] When using a phantom that includes three plates, such as Figure 4a and Figure 4b The phantom 400 depicted in the diagram has its center point of the central calibration pattern defined as the origin of a fixed coordinate system, with the long side, short side, and normal direction of the plate defining the x, y, and z axes of this coordinate system, respectively. While the calibration method could be designed to use different points to define the origin of the fixed coordinate system, the central rhombus of the central checkerboard pattern is particularly useful because, in an ideal setup and with an ideal phantom placement, this central rhombus will coincide with the isocenter of the CBCT scanner and the radiotherapy machine.
[0137] The calibration results are contained within an object of CBCT geometry that maintains three distinct parameters of the same geometry. Fixed coordinate system parameterization provides the absolute (or “true”) source and detector coordinates, as well as the rotation angle of the detector relative to the fixed coordinate system defined by the phantom. For example, for each gantry angle, the true source position 910 is represented by its position along the X-axis. f Axis, Y f Axis and Z f The coordinates of the axes are used to represent the coordinates. Local coordinate system parameterization gives the deviation from the nominal values of the source position, detector position, and detector rotation, in order to... Figure 9a The respective local coordinate systems are shown. Similarly, the source coordinate system parameterization uses ΔSID, Δφ, and Δψ to describe and represent the true source location. Figure 9b The detector position and rotation in the local coordinate system are shown. (As mentioned above regarding...) Figure 9a and Figure 9b The local coordinate systems described are those with their origin at the expected source / puncture point location, given a nominal SID / IDD and a reported rack angle φ, which are rotated such that their coordinate axes are parallel to the axes of the rack coordinate system. In the source coordinate system parameterization, the source position is parameterized by deviations from the nominal geometry in the SID, and by deviations from the in-plane angle (i.e., rack angle φ) and out-of-plane angle ψ relative to the origin / isocenter defined by the phantom. The origin of the local detector coordinate system in this parameterization is then taken as a point at a distance IDD from the isocenter (or the origin defined by the phantom) along the direction from the true source location to the isocenter. The local detector coordinate system is oriented such that the z-axis is parallel to the vector from the isocenter to the source, and the x-axis is parallel to the xz plane of the fixed coordinate system. The detector parameters are then defined as deviations from the nominal position and rotation in this coordinate system.
[0138] Defining the phantom's center point as the origin of a fixed coordinate system is somewhat arbitrary and makes the obtained parameter values dependent on the phantom's orientation within the gantry. While this doesn't affect image reconstruction, it's undesirable when comparing calibrations or matching coordinate systems performed at different time points. Instead, calibrations can be represented in a coordinate system independent of the phantom's orientation. One possible option is to find a coordinate system in which the deviation of the projection parameters is, in some sense, as small as possible from their nominal values. However, for two subsequent calibrations with different source and detector trajectories, there is never a guarantee that the origin of such a system will not shift relative to room coordinates. Therefore, even for reproducible curvature, the calculated origin may differ between different detector positions or gantry rotation directions. To have a common origin between different calibrations and the MV system, the calibration coordinates must be relative to a fixed reference, such as on a table.
[0139] In a specific implementation, the "best-fit" fixed coordinate system is obtained in a separate step after calibration. First, the average radius and coordinates of the source and detector trajectories in their respective centroid systems are calculated. Then, the average rotation axis is determined as the normal to the plane that fits the union of these source and detector positions and scaled to make their average radii the same. This axis defines the y-axis direction in the new fixed coordinate system. For each gantry angle, the expected center point is given by taking the vector from the source to the detector, dividing it by the average magnification factor (given by the source and detector radii), and adding it to the source position. The isocenter of the new fixed coordinate system is then found by calculating the average center point and finding the intersection of the source trajectory plane on one hand and the line defined by the average center point and the y-axis direction on the other hand. The remaining rotational degrees of freedom of the new coordinate system are ultimately determined by minimizing the difference between the reported gantry angle and the angle of the source position. This new fixed coordinate system is unaffected by the global rotation and translation of the initial fixed coordinate system and therefore unaffected by the body modulus. Once the parameters in the best-fit fixed coordinate system are obtained, the parameters in the corresponding local and source coordinate systems can be calculated.
[0140] While the discussion in this application focuses on CBCT imaging scanners that form part of a radiotherapy device, it should be understood that the phantoms and calibration methods disclosed herein are equally applicable to calibrating other X-ray flat panel systems, such as fusion tomography and stereotactic imaging.
[0141] Because the calibration patterns for projections at certain rack angles are nearly translationally invariant, the correct association of edges and lines is highly sensitive to the initial estimation of the phantom's position. Therefore, in some embodiments of the phantom disclosed herein, the calibration patterns on each plate include markings, such as circles (not shown), on the central rhombus of each pattern. These markings are readily detectable in the projection, and an improved initial estimate of the phantom's position can be obtained from them. If a very large detector offset is desired, for example, to create a large difference in the field of view, markings on rhombuses at fixed distances on either side of the central rhombus can be used to further improve the accuracy of the initial position estimation. Another method to improve the initial position estimation is to find the projection of one of the plates that is imaged close to the edge. The projection of the plate then becomes a dark line, from which the phantom's rotation and translation can be estimated.
[0142] The above is about Figure 8 Algorithm 1 and the described calibration method describe the use of an edge detection algorithm to assign identified edges to lines of a calibration pattern, fit lines to all points assigned to the same line, identify intersections, and determine calibration parameters based on the identified intersections and knowledge of the expected calibration pattern. However, other calibration algorithms and techniques can be used. For example, image registration and / or pattern fitting techniques can be used to find the best fit between each projected image and the pattern of the expected projection based on existing knowledge of the calibration pattern. Calibration parameters can be determined by evaluating the desired image transformation.
[0143] The method described herein is suitable for finding calibration parameters that describe the difference between the nominal and actual positions of a radiation source and the difference between the nominal and actual positions and orientations of a detector. The difference between nominal / orientation and actual / orientation is primarily caused by the bending of heavy components under gravity; therefore, the calibration parameters indicate bending under gravity. However, it should be understood that the calibration parameters can further indicate other possible sources of the difference between nominal / orientation and actual / orientation. Other sources of the difference between the nominal and actual values of source and detector parameters include: manufacturing tolerances, rack wobbling (rotation of the rack outside its plane), various mechanical misalignments, differences between the reported rack position and the actual rack position due to hysteresis effects, varying engine loads or accelerations, and vibration.
[0144] Figure 4a , Figure 4bThe phantom 400 depicted includes multiple plates positioned along a first axis. Here, the first axis 430 is a line passing through each plate. However, the first axis need not be an axis passing through each plate; rather, it simply defines the direction along which the plates are positioned relative to each other. The first axis can be considered an axis of a coordinate system, which can be used to define the relative position of each plate, and the plates are positioned along this axis, for example, separated or spaced apart along this axis. For example, the center of each plate would have different coordinate values in a coordinate system using the first axis. It is not necessary to draw a single axis or a single line that will contact each plate, as is the case in the accompanying figure. Similarly, although... Figure 4b The first axis 430 depicted is aligned with the central axis of the cylindrical support structure 410, but this is not necessary. The support structure can have various shapes and may not have a central axis. Therefore, it should be understood that the first axis is an axis that can be defined in space and is not necessarily limited by the symmetry properties of the support structure.
[0145] To illustrate this point, Figure 11 Alternative embodiments of the phantom according to this disclosure are described. Figure 11 It is similar in many ways to the phantom 400 depicted in Figure 4, and Figure 11 A similar illustration is shown, viewed from the base of the phantom (or equivalently from the top). The support structure 1110 of the phantom is depicted as transparent except for its outline to aid in understanding the arrangement of the plates 1122, 1124, and 1126. Like phantom 44, phantom 1100 includes the support structure 1110 and a plurality of plates 1122, 1124, and 1126. Each plate 1122, 1124, and 1126 includes a planar alignment pattern formed of a translucent material. The plates are positioned by the support structure for arrangement along a first axis. The first axis enters... Figure 11 The first axis is aligned with the plane of the page and the central axis of the support structure 1110. However, it is crucial that the first axis does not pass through any of the plates 1122, 1124, and 1126.
[0146] Similar to phantom 400, each plate 1122, 1124, 1126 is positioned at an angle relative to each of the other plates 1122, 1124, 1126 by support structure 1110, such that each planar calibration pattern faces a different direction. The direction faced by each planar calibration pattern is substantially perpendicular to the first axis.
[0147] As described above, the currently disclosed one or more phantoms and associated one or more calibration methods are superior to known methods. By providing a calibration phantom in which a plate including calibration patterns is positioned at an angle to each of the other plates by a support structure, such that each planar calibration pattern faces a different direction, wherein the direction facing each planar calibration pattern is substantially perpendicular to a first axis along which the plates are stacked, it is ensured that at least a portion of the calibration pattern is visible in each projected image when the calibration phantom is imaged. In an arrangement comprising three or more plates, by positioning each plate at an angle to the other plates, it is ensured that at least a portion of the calibration patterns of two different plates is visible in each projected image.
[0148] Because the disclosed calibration method can determine the relative positions of the plates in the phantom, the only required manufacturing accuracy relates to the calibration pattern. Such a pattern can be formed on a board using circuit board manufacturing technology, which is both highly accurate and inexpensive.
[0149] Figure 10 A block diagram illustrating one embodiment of a computing device 1000 is shown, in which a set of instructions is executable to cause the computing device to perform any or more of the methods discussed herein. The computing device may form part of a radiotherapy apparatus. In this embodiment, the computing device may be connected (e.g., networked) to other machines in a local area network (LAN), intranet, extranet, or the Internet. The computing device may operate as a server or client machine in a client-server network environment, or as a peer machine in a peer-to-peer (or distributed) network environment. The computing device may be a personal computer (PC), tablet computer, set-top box (STB), personal digital assistant (PDA), cellular phone, network appliance, server, network router, switch, or bridge, or any machine capable of executing (sequentially or otherwise) a set of instructions specifying the operations to be performed by the machine. Furthermore, although only a single computing device is shown, the term "computing device" should also be considered to include any collection of machines (e.g., computers) that individually or jointly execute one or more sets of instructions to perform any or more of the methods discussed herein.
[0150] Example computing device 1000 includes processing device 1002, main memory 1004 (e.g., read-only memory (ROM), flash memory, dynamic random access memory (DRAM) such as synchronous DRAM (SDRAM) or Rambus DRAM (RDRAM)), static memory 1006 (e.g., flash memory, static random access memory (SRAM), etc.) and auxiliary memory (e.g., data storage device 1018), which communicate with each other via bus 1030.
[0151] Processing device 1002 represents one or more general-purpose processors, such as microprocessors, central processing units, etc. More specifically, processing device 1002 may be a Complex Instruction Set Computing (CISC) microprocessor, a Reduced Instruction Set Computing (RISC) microprocessor, a Very Long Instruction Word (VLIW) microprocessor, a processor implementing other instruction sets, or a processor implementing a combination of instruction sets. Processing device 1002 may also be one or more special-purpose processing devices, such as application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), network processors, etc. Processing device 1002 is configured to execute processing logic (instructions 1022) to perform the operations and steps discussed herein.
[0152] The computing device 1000 may also include a network interface device 1008. The computing device 1000 may also include a video display unit 1010 (e.g., a liquid crystal display (LCD) or a cathode ray tube (CRT)), an alphanumeric input device 1012 (e.g., a keyboard or a touch screen), a cursor control device 1014 (e.g., a mouse or a touch screen), and an audio device 1016 (e.g., a speaker).
[0153] Data storage device 1018 may include one or more machine-readable storage media (or more specifically, one or more non-transitory computer-readable storage media) 1028 on which one or more instruction sets 1022 are stored, which embody any or more of the methods or functions described herein. The instructions 1022 may also reside wholly or at least partially within main memory 1004 and / or processing device 1002 during execution by computer system 1000, which also constitute computer-readable storage device media.
[0154] The various methods described above can be implemented by a computer program. A computer program may include computer code configured to instruct a computer to perform one or more of the functions described above. The computer program and / or code for performing such methods may be provided on one or more computer-readable media or more generally on a computer program product to a device such as a computer. The computer-readable media may be transient or non-transient. One or more computer-readable media may be, for example, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, or a propagation medium for data transmission (e.g., for downloading code via the Internet). Alternatively, one or more computer-readable media may take the form of one or more physical computer-readable media, such as semiconductor or solid-state memory, magnetic tape, removable computer floppy disks, random access memory (RAM), read-only memory (ROM) such as CD-ROM, CD-R / W, or DVD, hard disks, and optical disks.
[0155] In one implementation, the modules, components, and other features described herein may be implemented as discrete components or integrated into the functionality of hardware components such as ASICs, FPGAs, DSPs, or similar devices.
[0156] A “hardware component” is a tangible (e.g., non-transitory) physical component (e.g., a group of one or more processors) capable of performing certain operations and which can be physically configured or arranged in some way. A hardware component may include dedicated circuitry or logic permanently configured to perform certain operations. A hardware component may be or include dedicated processors, such as field-programmable gate arrays (FPGAs) or ASICs. A hardware component may also include programmable logic or circuitry temporarily configured by software to perform certain operations.
[0157] Therefore, the phrase “hardware component” should be understood to encompass tangible entities that can be physically constructed, permanently configured (e.g., hardwired) or temporarily configured (e.g., programmed) to operate in some way or perform certain operations described herein.
[0158] Furthermore, modules and components can be implemented as firmware or functional circuitry within a hardware device. Additionally, modules and components can be implemented as any combination of hardware devices and software components, or solely as software (e.g., code stored or otherwise embodied in a machine-readable medium or transmission medium).
[0159] Unless otherwise specifically stated, it should be understood, as is apparent from the following discussion, that throughout this specification, the use of terms such as “receive,” “determine,” “locate,” “associate,” and “identify” refers to the actions and processing of a computer system or similar electronic computing device that manipulates data represented as physical (electronic) quantities in the registers and memory of the computer system and converts it into other data similarly represented as physical quantities in the memory or registers or other such information storage, transmission, or display devices of the computer system.
[0160] It should be understood that the above description is intended to be illustrative and not restrictive. Many other embodiments will be apparent to those skilled in the art upon reading and understanding the above description. While this disclosure has been described with reference to specific exemplary embodiments, it will be appreciated that this disclosure is not limited to the described embodiments but can be practiced with modifications and changes within the spirit and scope of the appended claims. Therefore, the specification and drawings should be considered illustrative rather than restrictive. Consequently, the scope of this disclosure should be determined by reference to the appended claims and the full scope of their equivalents.
Claims
1. A calibration phantom for calibrating an imaging scanner, the imaging scanner comprising a rotatable gantry, and an imaging radiation source and a detector, both configured to rotate together with the gantry; the calibration phantom comprising: A support structure and multiple plates, wherein each plate includes a planar alignment pattern formed of a radiopaque material, wherein the plates are positioned along a first axis; and Each of the plates is positioned at an angle relative to each of the other plates by the support structure, such that each planar calibration pattern faces a different direction, wherein the direction in which each planar calibration pattern faces is substantially perpendicular to the first axis, and wherein the plates of the plurality of plates are separated from each other along the first axis.
2. The calibration phantom of claim 1, wherein, The plurality of plates includes at least three plates, and each of the plates is positioned at an angle relative to the other plates by the support structure, such that each planar calibration pattern faces a different direction.
3. The calibration phantom of claim 2, wherein, Each of the plates is positioned by the support structure such that when the calibration phantom is imaged via imaging radiation incident from any direction perpendicular to the first axis, at least a portion of at least two planar calibration patterns are included in the image.
4. The calibration phantom of any one of claims 1 to 3, wherein, The plurality of plates includes at least: a first plate including a first planar calibration pattern; a second plate including a second planar calibration pattern; and a third plate including a third planar calibration pattern. The first planar calibration pattern is positioned facing a first direction by the support structure, the second planar calibration pattern is positioned facing a second direction by the support structure, and the third planar calibration pattern is positioned facing a third direction by the support structure. The first direction, the second direction, and the third direction all intersect each other at an angle relative to each other.
5. The calibration phantom of any one of claims 1 to 3, wherein, Each of the plurality of plates includes a substrate, and the planar calibration pattern is formed by metal laminated on the substrate.
6. The calibration phantom of claim 5, wherein, Each of the calibration patterns is formed via a subtractive manufacturing process.
7. The calibration phantom according to claim 6, wherein, The subtractive manufacturing process includes one or more of the following processes: chemical etching, physical milling, digital lithography, and photolithography.
8. The calibration phantom according to any one of claims 1 to 3, wherein, The support structure has a central axis aligned with or parallel to the first axis.
9. The calibration phantom according to any one of claims 1 to 3, wherein, The support structure includes a plurality of stackable elements, each stackable element being stacked along the first axis, and each stackable element including at least one recess configured to hold one of the plurality of plates in place.
10. The calibration phantom according to any one of claims 1 to 3, wherein, Each of the planar calibration patterns includes multiple intersecting lines.
11. The calibration phantom according to claim 10, wherein, The planar calibration pattern is a checkerboard pattern that includes a repeating pattern of rhombuses.
12. A method for calibrating an imaging scanner, the imaging scanner comprising a rotatable gantry and components configured to rotate together with the gantry, the components including an imaging radiation source and a detector, the method comprising: Receive multiple projected images of a phantom positioned within the field of view of the imaging scanner, the projected images being captured from multiple gantry rotation angles, and the phantom comprising multiple plates, each plate comprising a calibration pattern formed of a transmissive material, and each plate being positioned by a support structure at an angle to each of the other plates, such that each planar calibration pattern faces a different direction; as well as At least one calibration parameter is determined based on the projected image. Wherein, the at least one calibration parameter indicates the difference between the nominal position or orientation of a component of the imaging scanner and the actual position or orientation of the component of the imaging scanner; and Wherein, the phantom is the calibration phantom according to any one of claims 1 to 11.
13. The method according to claim 12, wherein, The difference between the nominal position or orientation of the component and the actual position or orientation of the component indicates the bending of the component under the influence of gravity.
14. The method according to claim 12, wherein, The method further includes identifying the location of edge points in each projected image and associating the identified edge point locations with lines of the calibration pattern.
15. The method according to claim 12, wherein, The calibration pattern further includes multiple intersecting lines that intersect to form intersection points, and the method further includes: Identify the intersections in the projected image; and The at least one calibration parameter is determined based on the intersection points identified in each of the projected images.
16. The method according to any one of claims 12 to 15, wherein, The at least one calibration parameter relates to at least one of the location of the imaging radiation source, the location of the detector, and the orientation of the detector.
17. The method according to any one of claims 12 to 15, wherein, The at least one calibration parameter is determined as a function of the rack rotation angle.
18. A system for calibrating an imaging scanner, the imaging scanner including a rotatable gantry and components configured to rotate with the gantry, the components including an imaging radiation source and a detector, the system comprising: The calibration phantom according to any one of claims 1 to 11; as well as A computer-readable medium comprising computer-executable instructions that, when executed by a processor, cause the processor to: Receive projected images of the phantom captured by the imaging scanner from multiple gantry rotation angles, and determine at least one calibration parameter based on the projected images; The at least one calibration parameter indicates the difference between the nominal position or orientation of a component of the imaging scanner and the actual position or orientation of the component of the imaging scanner.
19. The system according to claim 18, wherein, The difference between the nominal position or orientation of the component and the actual position or orientation of the component indicates the bending of the component under the influence of gravity.
20. The system according to claim 18 or 19, wherein, The at least one calibration parameter relates to at least one of the location of the imaging radiation source, the location of the detector, and the orientation of the detector.
21. The system according to claim 18 or 19, wherein, The at least one calibration parameter is determined as a function of the rack rotation angle.
22. A computer-readable medium comprising computer-executable instructions that, when executed by a processor, cause the processor to perform the method according to any one of claims 12 to 17.
Citation Information
Patent Citations
Calibration phantom
US20180014809A1
