Ear - anterior cranial base anchor based on cbct coordinate system construction method, index calculation method and system

Abstract

The application relates to a CBCT coordinate system construction method, an index calculation method and a system based on an inner ear-anterior skull base anchor, the construction method comprising the following steps: acquiring a CBCT image of an oral cavity craniofacial part of a subject, segmenting key structures and corresponding segmentation confidence based on the CBCT image, wherein the key structures comprise left and right stapes crus-foot plate junction points, ethmoidal ridge and sella turcica points; constructing an inner ear-anterior skull base coordinate system based on the left and right stapes crus-foot plate junction points, the ethmoidal ridge or the sella turcica point; judging whether the segmentation confidence of the left and right stapes crus-foot plate junction points meets the standard, if not, verifying the left and right stapes crus-foot plate junction points, and then reconstructing the inner ear-anterior skull base coordinate system after verification, if yes, the inner ear-anterior skull base coordinate system does not need to be reconstructed. Compared with the prior art, the application has the advantages of significantly suppressing posture noise and being applicable to different time points.

Description

Technical Field

[0001] This invention relates to the field of oral and craniofacial examination, and in particular to a method for constructing a CBCT coordinate system based on inner ear-anterior skull base anchoring, a method for calculating indicators, and a system. Background Technology

[0002] Cone-beam computed tomography (CBCT) of the oral and maxillofacial region is a commonly used auxiliary examination for assessing the craniofacial features of patients. The commonly used assessment methods rely on: (1) the device / patient coordinates (the inherent axis of the Digital Imaging and Communications in Medicine (DICOM) standard); (2) the Frankfort level (the plane formed by the infraorbital margin (Or) and the superior auricular margin (Porion, Po)); and (3) the midsagittal plane (MSP) defined by the sella turcica (Sella, S), the nasal root (Nasion, N), the ethmoid ridge (Crista Galli, CG), and the anterior crus–footplate junctions (ACFJ).

[0003] The above method is sensitive to patient head positioning, differences in camera models, and registration errors. It generates non-biological posture noise over time, resulting in systematic drift and increased uncertainty in the root mean square (RMS) readings of rotation, translation, and point displacement. This affects the consistency of results comparisons between different shooting time points and different camera models for the same patient. Summary of the Invention

[0004] The purpose of this invention is to provide a method for constructing a CBCT coordinate system based on inner ear-anterior skull base anchoring, a method for calculating indicators, and a system for reducing non-biological noise.

[0005] The objective of this invention can be achieved through the following technical solutions: A method for constructing a CBCT coordinate system based on inner ear-anterior skull base anchoring includes the following steps: Obtain CBCT images of the oral and maxillofacial region of the subject, and segment key structures and corresponding segmentation confidence based on the CBCT images. The key structures include the left and right anterior stapedial footplate junction, ethmoid ridge, and sella turcica point. Construct an inner ear-anterior skull base coordinate system based on the junction of the left and right anterior stapedial pedicles and the ethmoid crest or sella turcica. Determine whether the segmentation confidence of the left and right anterior foot of the stapes-foot plate junction meets the standard. If not, verify the left and right anterior foot of the stapes-foot plate junction and reconstruct the inner ear-anterior skull base coordinate system after verification. If yes, there is no need to reconstruct the inner ear-anterior skull base coordinate system.

[0006] Furthermore, the segmentation steps of the key structure include: The CBCT images are uniformly reconstructed to obtain three-dimensional voxel data; The three-dimensional voxel data is semi-automatically or automatically segmented using a deep learning segmentation network, and after verification, the final segmented key structures and corresponding segmentation confidence scores are obtained. The deep learning segmentation network has an encoder-decoder structure and is equipped with cross-layer skip connections.

[0007] Furthermore, the key structures also include bilateral vestibules, bilateral cochleas, stapes arch, external auditory meatus, and infraorbital foramen.

[0008] Furthermore, the inner ear-anterior skull base coordinate system is either the BA-cg coordinate system or the BA-s coordinate system. The BA-cg coordinate system is constructed using the junctions of the left and right anterior pedicles of the stapes and the cribriform ridge. When the cribriform ridge is absent from the segmented key structures or its confidence level is below a preset threshold, the sella turcica point is used to replace the cribriform ridge in constructing the BA-s coordinate system. The steps for constructing the BA-cg coordinate system include: Calculate the midpoint between the left and right stapes anterior foot plate junctions. ,in AL The junction of the left stapes, forefoot, and footplate. AR The junction of the right stapes forefoot and footplate; Using the junctions of the left and right anterior stapes and footplates as the bilateral ear anchor points, the vector pointing from the junction of the left anterior stapes and footplate to the junction of the right anterior stapes and footplate is taken as... x axial direction, where x The axis is represented as: , In the formula, The unit direction vector along the junction of the left anterior foot and footplate to the junction of the right anterior foot and footplate is the coordinate system corresponding to the point where the left anterior foot and footplate point is located. x Positive direction of the axis; Midpoint O The vector pointing to the sieve crest is perpendicular to... x The projection of the axis into the plane is used as z axial direction, where z The axis is represented as: , , In the formula, To remove along the x The unit direction vector pointing towards the sieve ridge, obtained after normalization and axial direction component processing, corresponds to the coordinate system of... z Positive axis direction From the midpoint O The vector pointing to the sieve CG, CG For sieve ridges; Then determine the right-hand rule. y Construct the BA-cg coordinate system along the axis direction, where y The axis is represented as: , In the formula, To satisfy the right-hand rule, and the aforementioned x shaft and the z A unit direction vector that is orthogonal to all axes corresponds to the coordinate system of... y Positive direction of the axis.

[0009] Furthermore, the step of constructing the BA-s coordinate system includes: Calculate the midpoint between the left and right stapes anterior foot plate junctions. ,in AL The junction of the left stapes, forefoot, and footplate. AR The junction of the right stapes forefoot and footplate; Using the junctions of the left and right anterior stapes and footplates as the bilateral ear anchor points, the vector pointing from the junction of the left anterior stapes and footplate to the junction of the right anterior stapes and footplate is taken as... x axial direction, midpoint O The vector pointing to the saddle point is x The projection of the axis into the plane is used as z The axis direction is then determined using the right-hand rule. y The BA-s coordinate system is constructed along the axis direction.

[0010] Furthermore, it also includes: performing orientation consistency control during the construction of the BA-cg coordinate system or BA-s coordinate system. The steps of the direction consistency control include the following: Change the current BA-cg coordinate system or BA-s coordinate system x The axial direction is compared with a preset reference direction to determine if the angle formed by the two is greater than 90°. If so, then... x Axial direction, y Axial direction and z The axes are reversed to obtain the final BA-cg or BA-s coordinate system. If not, there is no need to reverse them, thus completing the direction consistency control.

[0011] Furthermore, before constructing the BA-cg coordinate system or the BA-s coordinate system, it is first determined whether the resolution of the CBCT image meets the standard. If it does, the BA-cg coordinate system or the BA-s coordinate system is constructed. If not, the CBCT image is re-acquired. Once the CBCT image meets the standard, the BA-cg coordinate system or the BA-s coordinate system is constructed.

[0012] This invention also provides a method for calculating oral craniofacial indices, which calculates oral craniofacial indices based on the BA-cg coordinate system or BA-s coordinate system described above. The specific calculation steps include: Any one of the multiple follow-up time points of the same subject is taken as the baseline time point, the CBCT image corresponding to the baseline time point is taken as the baseline image, and the CBCT images corresponding to the other time points of the same subject are taken as the images to be registered. The BA-cg coordinate system or the BA-s coordinate system are constructed based on the baseline image and the image to be registered, respectively. Based on the segmentation results of the baseline image and each image to be registered, the stable bony region of the skull base is selected as the registration region; The baseline image and the three-dimensional voxel data in each image to be registered are transformed to the corresponding BA-cg coordinate system or BA-s coordinate system. Then, at each time point, the voxel set corresponding to the registration region and / or the three-dimensional surface point set generated by the registration region are extracted. Each image to be registered is translated as a whole so that the origin of each image to be registered is aligned with the baseline image. O Overlap the images, then adjust the direction of the line connecting the left and right anterior foot of the stapes to the footplate in each image to be registered so that it is consistent with the direction of the line connecting the left and right anterior foot of the stapes to the footplate in the baseline image, and then adjust the line from the origin in each image to be registered. O The direction pointing towards the sieve ridge and / or saddle point is adjusted to align with the origin in the baseline image. O The directions pointing towards the sieve crest and / or the saddle point are consistent, thus obtaining a set of initial rigidity transformations; Using the initial rigid transformation as the starting value, the voxel set and / or the three-dimensional surface point set are used as the registration objects. The iterative nearest point algorithm and / or the optimization algorithm based on similarity metric are used to iteratively solve each image to be registered, so as to obtain the final rigid body transformation that minimizes the distance metric between the baseline image and the image to be registered in the registration area. Based on the final rigid body transformation applied to all three-dimensional voxel data in the image to be registered and the pre-selected key structure coordinates, the relative pose of each remaining time point relative to the baseline time point is obtained. The relative pose of each of the remaining time points with respect to the baseline time point is represented as a three-dimensional rigid body transformation, which is decomposed into a three-dimensional rotation part and a three-dimensional translation part. From the three-dimensional rotating part, an equivalent three-dimensional rotation axis is obtained, and then the total rotation angle around the three-dimensional rotation axis is obtained. The total rotation angle is used as the axis-angle rotation index. The three-dimensional translation component is directly used as the translation vector, and the three components of the translation vector correspond to the lower edge of the BA-cg coordinate system or the BA-s coordinate system, respectively. x axis, y shaft and z The translation component along the axis is used to obtain the translation vector of each other time point relative to the baseline time point; Within the anterior skull base and / or petrous bone region, a set of relatively stable bony points or a set of surface points uniformly sampled within the petrous bone region are selected based on the segmentation results of the baseline image and each image to be registered, as stable points or a set of stable surface points. For each point in the set of stable points or stable surface points, its coordinates after rigid body transformation at each of the remaining time points and the baseline time point are recorded, and its three-dimensional Euclidean distance between any two time points is calculated as its displacement. The root mean square value of the displacement of each point in the set of stable points or stable surface points is obtained by summing the squares of the displacements and taking the average value, and then taking the square root of the average value.

[0013] This invention also provides a CBCT coordinate system construction system based on inner ear-anterior skull base anchoring, comprising: Image acquisition and segmentation module: used to acquire CBCT images of the patient's oral and maxillofacial region, and segment key structures and corresponding segmentation confidence based on the CBCT images, wherein the key structures include the left and right anterior stapedial footplate junction, cribriform ridge and sella turcica point; Coordinate system construction module: used to construct an inner ear-anterior skull base coordinate system based on the left and right anterior stapes foot-foot plate junction, ethmoid ridge or sella turcica point; Verification and Reconstruction Module: Used to determine whether the segmentation confidence of the left and right anterior foot of the stapes-foot plate junction meets the standard. If not, the left and right anterior foot of the stapes-foot plate junction is verified, and the inner ear-anterior skull base coordinate system is reconstructed after verification. If yes, the inner ear-anterior skull base coordinate system does not need to be reconstructed.

[0014] This invention also provides a system for calculating oral and maxillofacial indices, which calculates oral and maxillofacial indices based on the inner ear-anterior skull base coordinate system or the BA-s coordinate system described above, including: The registration module is used to take any one of the multiple follow-up time points of the same subject as the baseline time point, the CBCT image corresponding to the baseline time point as the baseline image, and the CBCT images corresponding to the other time points of the same subject as the images to be registered; it constructs the BA-cg coordinate system or BA-s coordinate system based on the baseline image and the images to be registered respectively; it selects the stable bony region of the skull base as the registration region according to the segmentation results of the baseline image and each image to be registered; it transforms the three-dimensional voxel data in the baseline image and each image to be registered to the corresponding BA-cg coordinate system or BA-s coordinate system, and then extracts the voxel set corresponding to the registration region and / or the three-dimensional surface point set generated by the registration region at each time point; it performs a global translation of each image to be registered so that the origin of each image to be registered is aligned with the baseline image. O Overlap the images, then adjust the direction of the line connecting the left and right anterior foot of the stapes to the footplate in each image to be registered so that it is consistent with the direction of the line connecting the left and right anterior foot of the stapes to the footplate in the baseline image, and then adjust the line from the origin in each image to be registered. O The direction pointing towards the sieve ridge and / or saddle point is adjusted to align with the origin in the baseline image. O The directions pointing towards the sieve ridge and / or saddle point are aligned to obtain an initial rigid transformation. Using the initial rigid transformation as the starting value, the voxel set and / or 3D surface point set are used as the registration objects. The iterative nearest point algorithm and / or optimization algorithm based on similarity metric are used to iteratively solve each image to be registered to obtain the final rigid body transformation that minimizes the distance metric between the baseline image and the image to be registered within the registration area. Based on all 3D voxel data in the image to be registered and the pre-selected key structure coordinates, the relative pose of each remaining time point relative to the baseline time point is obtained. The index calculation module is used to represent the relative pose of each remaining time point with respect to the baseline time point as a three-dimensional rigid body transformation, which is decomposed into a three-dimensional rotation component and a three-dimensional translation component. An equivalent three-dimensional rotation axis is derived from the three-dimensional rotation component, and the total rotation angle around the three-dimensional rotation axis is calculated, which is used as the axis-angle rotation index. The three-dimensional translation component is directly used as the translation vector, and the three components of the translation vector correspond to the lower edge of the BA-cg coordinate system or the BA-s coordinate system, respectively. x axis, y shaft and zThe translation component in the axial direction is used to obtain the translation vector of each remaining time point relative to the baseline time point; in the anterior skull base and / or petrous bone region, a set of relatively stable bony points or a set of surface points uniformly sampled in the petrous bone region are selected based on the segmentation results of the baseline image and each image to be registered, as stable points or stable surface point set. For each point in the stable point or stable surface point set, its coordinates after rigid body transformation at each remaining time point and the baseline time point are recorded, and its three-dimensional Euclidean distance between any two time points is calculated as its displacement; the displacement of each point in the stable point or stable surface point set is squared, summed, averaged, and then the square root of the average value is taken to obtain the root mean square value of the point displacement of the stable point or stable surface point set.

[0015] Compared with the prior art, the present invention has the following beneficial effects: (1) The present invention is based on the fact that key structures (such as the cribriform ridge and the sella turcica point) are all located on bony structures in the midline region of the anterior skull base. These structures are highly stable in morphology after individual development and are not easily displaced by age or slight changes in head posture. Furthermore, by mapping the junction of the left and right anterior footplates of the stapes to a stable coordinate system, it is equivalent to establishing a robust ruler for CBCT images, reducing noise from non-biological factors such as patient head position, machine model differences, and sensitivity to registration errors.

[0016] (2) In the quality control and backoff strategy, the present invention also designs a backoff step when the ridge is absent or the confidence of the ridge is not up to standard, or when the resolution of the CBCT image is not up to standard, so as to replace the ridge with the saddle point S to construct the BA-s coordinate system or re-acquire CBCT image data to construct the coordinate system, which can further improve the accuracy of coordinate system construction. Compared with the existing DICOM inherent coordinates, the BA framework of the present invention significantly reduces the longitudinal drift and interval width of rotation / translation / RMS, and significantly improves the noise suppression effect.

[0017] (3) By designing the relocation and registration steps, the present invention rearranges the three-dimensional voxel matrix based on the constructed coordinate system, so that each time point is compared in a unified reference system, thereby eliminating the registration error, improving the accuracy of relative pose calculation, and thus improving the accuracy of the index.

[0018] (4) This invention introduces orientation consistency control during the coordinate system construction process, which ensures that all subjects and all time points are aligned in the coordinate system. x The axes are aligned in the left and right directions to avoid inconsistent registration results caused by left and right flipping.

[0019] (5) The present invention can ensure the consistency of the results of different imaging time points and different imaging models for the same patient, which is helpful for the efficacy evaluation and multi-center / different model follow-up studies of the same patient by comparing information at different imaging time points.

[0020] (6) Under the BA framework, the annualized rotation of patients of different ages and treatment methods has no significant interaction with RMS, which is closer to the biological reality. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the coordinate system construction method of the present invention; Figure 2 This is a schematic diagram of the inner ear-anterior skull base coordinate system of the present invention; Figure 3 This is a schematic diagram of the oral craniofacial index calculation process of the present invention. Detailed Implementation

[0022] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0023] Example 1 This embodiment provides a method for constructing a CBCT coordinate system based on inner ear-anterior skull base anchoring, such as... Figure 1 As shown, the method includes the following steps: S1. Image Processing and Segmentation: First, CBCT images are acquired, and then the acquired raw CBCT data is uniformly reconstructed to obtain three-dimensional voxel data. The three-dimensional voxel data is semi-automatically or automatically segmented, and manually checked by the operator. The segmentation includes at least the bilateral vestibules, bilateral cochleas, stapes arch, left and right anterior stapes footplate junction (ACFJ, corresponding to AL and AR), ethmoid crest (Crista Galli, CG), sella turcica point (Sella, S), external auditory meatus and infraorbital foramen, etc. (also known as anatomical structures).

[0024] In this embodiment, the semi-automatic or automatic segmentation of key structures can be performed by a deep learning segmentation network (U-Net or nnU-Net). In automatic segmentation, the deep learning segmentation network has an encoder-decoder structure and cross-layer skip connections to receive 3D voxel data and output the probability of each 3D voxel belonging to different anatomical structures (i.e., segmentation confidence). This deep learning segmentation network is trained on a CBCT image dataset with manually annotated key structures. During the inference phase, it generates the segmentation confidence of each anatomical point or structure based on the category probability. The generated segmentation confidence can be used for subsequent quality control and backtracking steps.

[0025] S2. Construction of the inner ear-anterior skull base coordinate system: In this embodiment, the inner ear-anterior skull base coordinate system includes two types: the BA-cg (inner ear-ethmoid crest anterior skull base) coordinate system and the BA-s (inner ear-sellar alar base) coordinate system. Generally, the BA-cg coordinate system is constructed. When the BA-cg coordinate system cannot be established, the BA-s coordinate system is constructed (this situation is described in the quality control and rollback steps in step 3).

[0026] This step involves constructing the BA-cg coordinate system, as detailed below: Calculate the midpoint using AL and AR , will be AL point to AR The vector as x Axial direction, x The axis is represented as: , In the formula, Let be the unit direction vector along the junction of the left anterior foot and footplate, pointing to the junction of the right anterior foot and footplate, corresponding to the coordinate system. x Positive direction of the axis; by O point to CG The vector perpendicular to x The projection of the axis into the plane is used as z Axial direction, z The axis is represented as: , , In the formula, To remove along x The unit direction vector pointing towards the sieve ridge, obtained after normalization and axial direction component processing, corresponds to the coordinate system of... z Positive axis direction From the midpoint O A vector pointing to the sieve CG; Then determine according to the right-hand rule y axial direction, thus constructing as Figure 2 The inner ear-anterior skull base (BA–cg) coordinate system shown is as follows: y The axis is represented as: , In the formula, To satisfy the right-hand rule, and x shaft and z A unit direction vector that is orthogonal to all axes corresponds to the coordinate system of... y Positive direction of the axis.

[0027] In addition, this step also performs orientation consistency control between different scanned images, so that... x The axis remains consistent left and right across different subjects and at different times.

[0028] S3. Quality Control and Returns: This step assesses data quality based on the segmentation confidence of each anatomical point or structure obtained from the segmentation and the spatial resolution of the CBCT image. When the segmentation confidence of the bilateral ACFJ and / or CG is lower than a preset threshold (0.7) and / or the voxel size (0.4 mm) is greater than a preset threshold, the quality is deemed substandard, triggering a rollback strategy, including at least one of the following: reconstructing the BA–cg coordinate system using manually corrected or manually labeled ACFJ, or rolling back to the BA–s coordinate system based on ACFJ and S. If the quality requirements are still not met after rollback, the corresponding scan data is marked as needing review or rescanning.

[0029] Specifically, the process includes the following: (1) Determine whether the segmentation confidence of the left and right anterior foot of the stapes-foot plate junction meets the standard. If not, the left and right anterior foot of the stapes-foot plate junction should be manually corrected or manually marked for verification. After verification, the inner ear-anterior skull base coordinate system should be reconstructed. If yes, the inner ear-anterior skull base coordinate system does not need to be reconstructed.

[0030] (2) When the cribriform ridge is absent in the segmented key structures or the confidence level of the cribriform ridge is lower than the preset threshold, the sella turcica point S is used to replace the cribriform ridge to construct the BA-s coordinate system. The construction process of this BA-s coordinate system is similar to that of the BA-cg coordinate system mentioned above, that is, the coordinate system is constructed with the junction of the anterior crus of the inner ear and the footplate as the anchor points of both ears and the sella turcica point S as the reference point of the anterior skull base. The specific construction steps include: Calculate the midpoint between the left and right stapes anterior foot plate junctions. ,in AL The junction of the left stapes, forefoot, and footplate. AR The junction of the right stapes forefoot and footplate; Using the junctions of the left and right anterior stapes and footplates as the anchor points for both ears, the vector pointing from the junction of the left anterior stapes and footplate to the junction of the right anterior stapes and footplate is taken as... x axial direction, midpoint O The vector pointing to the saddle point is x The projection of the axis into the plane is used as z The axis direction is then determined using the right-hand rule. y In the axial direction, construct the BA–s coordinate system.

[0031] Preferably, in this embodiment, before establishing the BA-cg coordinate system or the BA-s coordinate system, the quality control process first determines whether the resolution of the CBCT image meets the standard. If it does, the BA-cg coordinate system or the BA-s coordinate system is constructed. If not, the CBCT image is re-acquired. Once the CBCT image meets the standard, the BA-cg coordinate system or the BA-s coordinate system is constructed.

[0032] Preferably, this embodiment further proposes to perform orientation consistency control during the construction of the BA-cg coordinate system or BA-s coordinate system in this step of quality control. The orientation consistency control steps include the following: Change the current BA–cg coordinate system or BA–s coordinate system x The axis direction is compared with a preset reference direction. If the angle formed by the two is greater than 90°, the current scan is... x axis, y shaft and z The entire axis is reversed to ensure that all subjects and all time points are within the range of the axis being examined. x The axes should be aligned in the left and right directions to avoid inconsistent registration results caused by left and right flipping; otherwise, there is no need to reverse them.

[0033] In addition, if the above rollback strategy is still not up to standard, the corresponding scan data will be marked as needing to be reviewed or rescanned, and a quality warning will be given in the results report.

[0034] Example 2 This embodiment provides a method for calculating oral and maxillofacial indices. This method utilizes the BA-cg coordinate system or BA-s coordinate system constructed in Embodiment 1 to calculate oral and maxillofacial indices, such as... Figure 3 As shown, the specific calculation steps include: Step 1, Redirection and Registration: This step mainly transforms the voxel coordinates of the 3D voxel data in the CBCT images at each time point to the BA-cg or BA-s coordinate system. Rigid body or surface registration is then performed on the 3D voxel data at different time points in the BA-cg or BA-s coordinate system to ensure that the data at each time point is reconciled. O The points coincide, and the directions (AL–AR) of the two points at the junction of the left and right stapes anterior foot and footplate are aligned with the midpoint. O The direction pointing towards the sieve ridges ( O →CG) The orientation is aligned between each time point to obtain the relative pose of each time point relative to the baseline time point. Specifically, rigid body or surface registration includes the following steps: (1) Selection of reference time points: Any one of the multiple follow-up time points for the same subject is selected as the baseline time point, and the CBCT image corresponding to that time point is recorded as the baseline image; the CBCT images of the same subject at other time points are recorded as images to be registered.

[0035] (2) Coordinate unification: The baseline image and each image to be registered were constructed into either a BA–cg coordinate system or a BA–s coordinate system according to the method in Example 1, and their respective three-dimensional voxel data were transformed to the corresponding coordinate system, so that the origins of the images at different time points were aligned. O as well as x axis, z The axial direction is within a unified anatomical reference frame.

[0036] (3) Selection of stable areas at the skull base: In the baseline image and each image to be registered, a stable bony region of the skull base is selected as the registration region based on the segmentation results. The stable bony region of the skull base may include the sphenoid body, cribriform plate, petrous bone, etc. Thus, at each time point, the voxel set corresponding to the region and / or the three-dimensional surface point set generated by the segmentation results of the region are extracted.

[0037] (4) Initial alignment: First, translate the entire image to be registered so that the origin of the image is aligned. O Origin in baseline image O Overlap; in O Based on the point overlap, the direction of the line connecting the left and right anterior foot of the stapes to the footplate in the image to be registered is adjusted to be as consistent as possible with the direction of the line connecting the left and right anterior foot of the stapes to the footplate in the baseline image. The line in the image to be registered is then adjusted to... O The direction pointing to the sieve crest and / or saddle point is adjusted to be as consistent as possible with the corresponding direction in the baseline image, thereby obtaining a set of initial rigid body transformations.

[0038] (5) Rigid body registration optimization: Using the initial rigid body transformation obtained in step (4) as the starting value, the set of voxels and / or the set of surface points in the stable bony region of the skull base are selected as the registration objects. The iterative nearest point algorithm and / or the optimization algorithm based on similarity measure (e.g., mutual information, correlation or mean square error) are used to iteratively solve the image to be registered, so as to obtain the final rigid body transformation that minimizes the distance metric between the baseline image and the image to be registered in the stable bony region of the skull base.

[0039] (6) Optional implementation of surface registration: In one alternative implementation, registration can be achieved based on a three-dimensional surface, that is, the segmentation result of the stable bony region of the skull base is converted into a three-dimensional surface mesh, the surface corresponding to the image to be registered is coarsely aligned with the surface corresponding to the baseline image under the initial rigid body transformation obtained in step (4), and then the iterative nearest point algorithm or other registration algorithm based on surface distance is used to optimize the rigid body registration of the three-dimensional surface at the two time points to obtain the final rigid body transformation.

[0040] (7) Calculation of relative pose and longitudinal indices: The final rigid body transformation obtained in step (5) and / or step (6) is applied to all voxels in the image to be registered and the coordinates of the pre-selected anatomical landmarks to obtain the relative pose of each time point relative to the baseline time point. Based on this, the root mean square values ​​of axis-angle rotation, translation vector and point displacement of stable point or stable surface are calculated according to step 2 in Example 2, and the indicators are annualized according to the follow-up time length.

[0041] Step 2, Longitudinal Index Output: This step mainly calculates the longitudinal indices between time points in the BA-cg or BA-s coordinate system based on the relative pose, serving as oral craniofacial indices. These include: axis-angle rotation, translation vectors, and the root mean square (RMS) values ​​of point displacements at stable points or surfaces located in the anterior cranial base and / or petrous bone region. The rotation, translation, and RMS values ​​are then annualized according to the follow-up time. The specific calculation process includes: (1) Rigid body transformation decomposition: The rigid body registration result of each time point obtained in step 1 relative to the baseline time point is represented as a three-dimensional rigid body transformation. The rigid body transformation can be decomposed into a three-dimensional rotation part and a three-dimensional translation part, wherein the rotation part is used to characterize the rotation of the overall attitude and the translation part is used to characterize the movement of the overall position.

[0042] (2) Calculation of axis-angle rotation: The three-dimensional rotation component is converted into an axis-angle form, that is, a unit direction vector is used as the rotation axis, and a scalar angle is used as the rotation amount around the rotation axis to represent the overall rotation. Specifically, an equivalent three-dimensional rotation axis can be obtained from the rotation component, and then the total rotation angle around the rotation axis can be calculated. This rotation angle is used as the axis-angle rotation index. If necessary, the direction of the rotation axis in the BA-cg coordinate system or the BA-s coordinate system can be recorded simultaneously.

[0043] (3) Calculation of translation vector: The translation component of the rigid body transformation is directly taken as the translation vector, and its three components correspond to the lower edge of the BA–cg coordinate system or the BA–s coordinate system, respectively. x axis, y shaft and zThe translation components in the axial direction are used to obtain the three-dimensional translation vectors of each time point relative to the baseline time point.

[0044] (4) Selection of stable point / stable surface and displacement calculation: Within the anterior skull base and / or petrous bone region, a set of bony points considered relatively stable over time, or a set of surface points uniformly sampled within this region, are selected based on the segmentation results in Example 1, as stable points or a set of stable surface points. For each point in the set, its coordinates after rigid body registration at the baseline time point and the follow-up time point are recorded, and the three-dimensional Euclidean distance between the two time points is calculated as the displacement of the point.

[0045] (5) Calculation of root mean square (RMS) displacement of point: The displacements of all points in the set of stable points or stable surface points are statistically analyzed. The squares of the displacements at each point are summed, and the average is taken. The square root of the average is then taken to obtain the root mean square (RMS) value of the point displacements in the set. This value is used to characterize the overall stability and residual displacement of the anterior skull base and / or petrous bone region. If necessary, the RMS indexes for different anatomical subregions or different sides can be calculated separately.

[0046] (6) Annualized processing of indicators: Based on the actual time interval between each follow-up time point and the baseline time point, the above-mentioned axis-angle rotation index, translation vector, and RMS index are divided by the time interval (in years) to obtain the corresponding annualized axis-angle rotation, annualized translation, and annualized RMS. These are used to describe the average rotation change, position change, and residual displacement change in the stable region per unit time, thereby more accurately reflecting the comparability of different subjects or different treatment regimens at different follow-up periods.

[0047] Experiments have verified that the annualized RMS is approximately 0.29 mm / year; the median annualized angle for the four coordinate systems T0→T1 (from the initial examination T0 to the follow-up examination T1) is: BA–cg 0.48° / year < BA–s 0.77° / year < FH_R (Right-sided Frankfort Horizontal-based reference coordinate system) 1.11° / year ≪ MSP 9.25° / year. When the follow-up period exceeds 1 year, the stability advantage of the BA frame in this embodiment is more pronounced.

[0048] Example 3 This embodiment provides a CBCT coordinate system construction system based on inner ear-anterior skull base anchoring, including: Image acquisition and segmentation module: used to acquire CBCT images of the patient's oral and maxillofacial region, and segment key structures and corresponding segmentation confidence based on the CBCT images. The key structures include the left and right anterior stapedial footplate junction, ethmoid ridge, and sella turcica point. Coordinate system construction module: used to construct an inner ear-anterior skull base coordinate system based on the left and right anterior stapes-foot plate junction, ethmoid crest, or sella turcica point; Verification and Reconstruction Module: Used to determine whether the segmentation confidence of the left and right anterior foot of the stapes-foot plate junction meets the standard. If not, the left and right anterior foot of the stapes-foot plate junction is verified, and the inner ear-anterior skull base coordinate system is reconstructed after verification. If yes, the inner ear-anterior skull base coordinate system does not need to be reconstructed.

[0049] The rest are as in Example 1.

[0050] Example 4 This embodiment provides an oral and maxillofacial index calculation system, including: Registration module: Used to perform rigid or surface registration of 3D voxel data from CBCT images at various time points in the inner ear-anterior skull base coordinate system or BA-S coordinate system, so that the zero points of each time point coincide, and the directions and midpoints of the two points at the junction of the left and right anterior stapedius and footplate are aligned. O The directions pointing to the ridges are aligned between each time point to obtain the relative pose of each time point relative to the baseline time point; The index calculation module is used to calculate various index values ​​based on relative pose and to annualize each index value according to the follow-up time. The indexes include axis-angle rotation and translation vectors between each time point, as well as the root mean square value of point displacement of stable points / stable surfaces located in the anterior skull base and / or petrous bone region.

[0051] The rest are as in Example 2.

[0052] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0053] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0054] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0055] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0056] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0057] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0058] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for constructing a CBCT coordinate system based on inner ear-anterior skull base anchoring, characterized in that, Includes the following steps: Obtain CBCT images of the oral and maxillofacial region of the subject, and segment key structures and corresponding segmentation confidence based on the CBCT images. The key structures include the left and right anterior stapedial footplate junction, ethmoid ridge, and sella turcica point. Construct an inner ear-anterior skull base coordinate system based on the junction of the left and right anterior stapedial pedicles and the ethmoid crest or sella turcica. Determine whether the segmentation confidence of the left and right anterior foot of the stapes-foot plate junction meets the standard. If not, verify the left and right anterior foot of the stapes-foot plate junction and reconstruct the inner ear-anterior skull base coordinate system after verification. If yes, there is no need to reconstruct the inner ear-anterior skull base coordinate system.

2. The method for constructing a CBCT coordinate system based on inner ear-anterior skull base anchoring according to claim 1, characterized in that, The segmentation steps of the key structure include: The CBCT images are uniformly reconstructed to obtain three-dimensional voxel data; The three-dimensional voxel data is semi-automatically or automatically segmented using a deep learning segmentation network, and after verification, the final segmented key structures and corresponding segmentation confidence scores are obtained. The deep learning segmentation network has an encoder-decoder structure and is equipped with cross-layer skip connections.

3. The method for constructing a CBCT coordinate system based on inner ear-anterior skull base anchoring according to claim 1, characterized in that, The key structures also include bilateral vestibules, bilateral cochleas, stapes arch, external auditory meatus, and infraorbital foramen.

4. The method for constructing a CBCT coordinate system based on inner ear-anterior skull base anchoring according to claim 1, characterized in that, The inner ear-anterior skull base coordinate system is either the BA-cg coordinate system or the BA-s coordinate system. The BA-cg coordinate system is constructed using the junctions of the left and right anterior pedicles of the stapes and the cribriform ridge. When the cribriform ridge is absent from the segmented key structures or its confidence level is below a preset threshold, the sella turcica point is used to replace the cribriform ridge in constructing the BA-s coordinate system. The steps for constructing the BA-cg coordinate system include: Calculate the midpoint between the left and right stapes anterior foot plate junctions. ,in AL The junction of the left stapes, forefoot, and footplate. AR The junction of the right stapes forefoot and footplate; Using the junctions of the left and right anterior stapes and footplates as the bilateral ear anchor points, the vector pointing from the junction of the left anterior stapes and footplate to the junction of the right anterior stapes and footplate is taken as... x axial direction, where x The axis is represented as: ， In the formula, The unit direction vector along the junction of the left anterior foot and footplate to the junction of the right anterior foot and footplate is the coordinate system corresponding to the point where the left anterior foot and footplate point is located. x Positive direction of the axis; Midpoint O The vector pointing to the sieve crest is perpendicular to... x The projection of the axis into the plane is used as z axial direction, where z The axis is represented as: ， ， In the formula, To remove along the x The unit direction vector pointing towards the sieve ridge, obtained after normalization and axial direction component processing, corresponds to the coordinate system of... z Positive axis direction From the midpoint O The vector pointing to the sieve CG, CG For sieve ridges; Then determine the right-hand rule. y Construct the BA-cg coordinate system along the axis direction, where y The axis is represented as: ， In the formula, To satisfy the right-hand rule, and the aforementioned x shaft and the z A unit direction vector that is orthogonal to all axes corresponds to the coordinate system of... y Positive direction of the axis.

5. The method for constructing a CBCT coordinate system based on inner ear-anterior skull base anchorage according to claim 4, characterized in that, The steps for constructing the BA-s coordinate system include: Calculate the midpoint between the left and right stapes anterior foot plate junctions. ,in AL The junction of the left stapes, forefoot, and footplate. AR The junction of the right stapes forefoot and footplate; Using the junctions of the left and right anterior stapes and footplates as the bilateral ear anchor points, the vector pointing from the junction of the left anterior stapes and footplate to the junction of the right anterior stapes and footplate is taken as... x axial direction, midpoint O The vector pointing to the saddle point is x The projection of the axis into the plane is used as z The axis direction is then determined using the right-hand rule. y The BA-s coordinate system is constructed along the axis direction.

6. A method for constructing a CBCT coordinate system based on inner ear-anterior skull base anchoring according to claim 1, 4, or 5, characterized in that, Also includes: During the construction of the BA-cg coordinate system or BA-s coordinate system, directional consistency control is performed. The steps of the direction consistency control include: the following: Change the current BA-cg coordinate system or BA-s coordinate system x The axial direction is compared with a preset reference direction to determine if the angle formed by the two is greater than 90°. If so, then... x Axial direction, y Axial direction and z The axes are reversed to obtain the final BA-cg or BA-s coordinate system. If not, there is no need to reverse them, thus completing the direction consistency control.

7. A method for constructing a CBCT coordinate system based on inner ear-anterior skull base anchorage according to claim 1, 4, or 5, characterized in that, Before constructing the BA-cg coordinate system or BA-s coordinate system, it is first determined whether the resolution of the CBCT image meets the standard. If it does, the BA-cg coordinate system or BA-s coordinate system is constructed. If not, the CBCT image is re-acquired. Once the CBCT image meets the standard, the BA-cg coordinate system or BA-s coordinate system is constructed.

8. A method for calculating oral and craniofacial indices, characterized in that, The calculation of oral craniofacial indices using the BA-cg coordinate system or BA-s coordinate system according to any one of claims 1-7 includes the following specific calculation steps: Any one of the multiple follow-up time points of the same subject is taken as the baseline time point, the CBCT image corresponding to the baseline time point is taken as the baseline image, and the CBCT images corresponding to the other time points of the same subject are taken as the images to be registered. The BA-cg coordinate system or the BA-s coordinate system are constructed based on the baseline image and the image to be registered, respectively. Based on the segmentation results of the baseline image and each image to be registered, the stable bony region of the skull base is selected as the registration region; The baseline image and the three-dimensional voxel data in each image to be registered are transformed to the corresponding BA-cg coordinate system or BA-s coordinate system. Then, at each time point, the voxel set corresponding to the registration region and / or the three-dimensional surface point set generated by the registration region are extracted. Each image to be registered is translated as a whole so that the origin of each image to be registered is aligned with the baseline image. O Overlap the images, then adjust the direction of the line connecting the left and right anterior foot of the stapes to the footplate in each image to be registered so that it is consistent with the direction of the line connecting the left and right anterior foot of the stapes to the footplate in the baseline image, and then adjust the line from the origin in each image to be registered. O The direction pointing towards the sieve ridge and / or saddle point is adjusted to align with the origin in the baseline image. O The directions pointing towards the sieve crest and / or the saddle point are consistent, thus obtaining a set of initial rigidity transformations; Using the initial rigid transformation as the starting value, the voxel set and / or the three-dimensional surface point set are used as the registration objects. The iterative nearest point algorithm and / or the optimization algorithm based on similarity metric are used to iteratively solve each image to be registered, so as to obtain the final rigid body transformation that minimizes the distance metric between the baseline image and the image to be registered in the registration area. Based on the final rigid body transformation applied to all three-dimensional voxel data in the image to be registered and the pre-selected key structure coordinates, the relative pose of each remaining time point relative to the baseline time point is obtained. The relative pose of each of the remaining time points with respect to the baseline time point is represented as a three-dimensional rigid body transformation, which is decomposed into a three-dimensional rotation part and a three-dimensional translation part. From the three-dimensional rotating part, an equivalent three-dimensional rotation axis is obtained, and then the total rotation angle around the three-dimensional rotation axis is obtained. The total rotation angle is used as the axis-angle rotation index. The three-dimensional translation component is directly used as the translation vector, and the three components of the translation vector correspond to the lower edge of the BA-cg coordinate system or the BA-s coordinate system, respectively. x axis, y shaft and z The translation component along the axis is used to obtain the translation vector of each other time point relative to the baseline time point; Within the anterior skull base and / or petrous bone region, a set of relatively stable bony points or a set of surface points uniformly sampled within the petrous bone region are selected based on the segmentation results of the baseline image and each image to be registered, as stable points or a set of stable surface points. For each point in the set of stable points or stable surface points, its coordinates after rigid body transformation at each of the remaining time points and the baseline time point are recorded, and its three-dimensional Euclidean distance between any two time points is calculated as its displacement. The root mean square value of the displacement of each point in the set of stable points or stable surface points is obtained by summing the squares of the displacements and taking the average value, and then taking the square root of the average value.

9. A CBCT coordinate system construction system based on inner ear-anterior skull base anchoring, characterized in that, include: Image acquisition and segmentation module: used to acquire CBCT images of the patient's oral and maxillofacial region, and segment key structures and corresponding segmentation confidence based on the CBCT images, wherein the key structures include the left and right anterior stapedial footplate junction, cribriform ridge and sella turcica point; Coordinate system construction module: used to construct an inner ear-anterior skull base coordinate system based on the left and right anterior stapes foot-foot plate junction, ethmoid ridge or sella turcica point; Verification and Reconstruction Module: Used to determine whether the segmentation confidence of the left and right anterior foot of the stapes-foot plate junction meets the standard. If not, the left and right anterior foot of the stapes-foot plate junction is verified, and the inner ear-anterior skull base coordinate system is reconstructed after verification. If yes, the inner ear-anterior skull base coordinate system does not need to be reconstructed.

10. A system for calculating oral and craniofacial indices, characterized in that, Calculating oral and maxillofacial indices using the inner ear-anterior skull base coordinate system or the BA-s coordinate system according to any one of claims 1-7 includes: The registration module is used to take any one of the multiple follow-up time points of the same subject as the baseline time point, the CBCT image corresponding to the baseline time point as the baseline image, and the CBCT images corresponding to the other time points of the same subject as the images to be registered; it constructs the BA-cg coordinate system or BA-s coordinate system based on the baseline image and the images to be registered respectively; it selects the stable bony region of the skull base as the registration region according to the segmentation results of the baseline image and each image to be registered; it transforms the three-dimensional voxel data in the baseline image and each image to be registered to the corresponding BA-cg coordinate system or BA-s coordinate system, and then extracts the voxel set corresponding to the registration region and / or the three-dimensional surface point set generated by the registration region at each time point; it performs a global translation of each image to be registered so that the origin of each image to be registered is aligned with the baseline image. O Overlap the images, then adjust the direction of the line connecting the left and right anterior foot of the stapes to the footplate in each image to be registered so that it is consistent with the direction of the line connecting the left and right anterior foot of the stapes to the footplate in the baseline image, and then adjust the line from the origin in each image to be registered. O The direction pointing towards the sieve ridge and / or saddle point is adjusted to align with the origin in the baseline image. O The directions pointing towards the sieve ridge and / or saddle point are aligned to obtain an initial rigid transformation. Using the initial rigid transformation as the starting value, the voxel set and / or 3D surface point set are used as the registration objects. The iterative nearest point algorithm and / or optimization algorithm based on similarity metric are used to iteratively solve each image to be registered to obtain the final rigid body transformation that minimizes the distance metric between the baseline image and the image to be registered within the registration area. Based on all 3D voxel data in the image to be registered and the pre-selected key structure coordinates, the relative pose of each remaining time point relative to the baseline time point is obtained. The index calculation module is used to represent the relative pose of each remaining time point with respect to the baseline time point as a three-dimensional rigid body transformation, which is decomposed into a three-dimensional rotation component and a three-dimensional translation component. An equivalent three-dimensional rotation axis is derived from the three-dimensional rotation component, and the total rotation angle around the three-dimensional rotation axis is calculated, which is used as the axis-angle rotation index. The three-dimensional translation component is directly used as the translation vector, and the three components of the translation vector correspond to the lower edge of the BA-cg coordinate system or the BA-s coordinate system, respectively. x axis, y shaft and z The translation component in the axial direction is used to obtain the translation vector of each remaining time point relative to the baseline time point; in the anterior skull base and / or petrous bone region, a set of relatively stable bony points or a set of surface points uniformly sampled in the petrous bone region are selected based on the segmentation results of the baseline image and each image to be registered, as stable points or stable surface point set. For each point in the stable point or stable surface point set, its coordinates after rigid body transformation at each remaining time point and the baseline time point are recorded, and its three-dimensional Euclidean distance between any two time points is calculated as its displacement; the displacement of each point in the stable point or stable surface point set is squared, summed, averaged, and then the square root of the average value is taken to obtain the root mean square value of the point displacement of the stable point or stable surface point set.

Images