Automated process for reusing intermediate digital setups in orthodontic treatment due to treatment plan modifications.

Abstract

A method for generating and reusing digital setups for orthodontic treatment paths. The method receives a digital 3D model of teeth, optionally performs proximal resections on the model, and generates an initial treatment path having stages including an initial setup, a final setup, and intermediate setups. The method segments the initial treatment path into initial steps of feasible tooth movement and results in a final treatment path with setups corresponding to the initial steps. For treatment redesign, the method calculates new steps of feasible tooth movement for only a portion of the initial treatment path and based on the initial steps, and generates the final treatment path with new setups corresponding to the new steps. The setups can be used to fabricate orthodontic appliances, such as clear tray aligners.

Description

[Background technology]

[0001] The staging of teeth from the initial stage to the final stage requires determining the precise movement of each individual tooth so that the teeth move toward their final state without colliding with each other and follow the optimal (preferably short) trajectory. Each tooth has 6 degrees of freedom, and the average arch has approximately 14 teeth, so finding the optimal trajectory of teeth from the initial to the final stage involves a large and complex search space. This optimization problem needs to be simplified by dividing the trajectory into several shorter and easier-to-find trajectories, and by reusing previous trajectories in redesigning the treatment plan. [Overview of the project]

[0002] One embodiment includes a method for generating and reusing a setup for an orthodontic treatment path in order to accept a redesign of the treatment path. The method includes receiving a digital 3D model of teeth and generating an initial treatment path having stages including an initial setup, a final setup, and a number of intermediate setups. The method also includes dividing the initial treatment path into initial steps of feasible tooth movement and resulting in a final treatment path having setups corresponding to the initial steps. For treatment redesign, the method further includes calculating new steps of feasible tooth movement for only a portion of the initial treatment path and based on the initial steps, and generating a final treatment path using new setups corresponding to the new steps. [Brief explanation of the drawing]

[0003] The accompanying drawings are incorporated herein and constitute part of this specification, illustrating the advantages and principles of the present invention together with the description. The drawings are as follows: [Figure 1] This is a diagram of a system for generating digital setups for orthodontic appliances. [Figure 2] This is a flowchart of the method for generating a digital setup for orthodontic appliances. [Figure 3] It is a graph showing the initial sequence of the digital setup. [Figure 4A] It is a graph showing the trajectory refinement of the sequence of the digital setup. [Figure 4B] It is a graph showing the trajectory refinement of the sequence of the digital setup. [Figure 4C] It is a graph showing the trajectory refinement of the sequence of the digital setup. [Figure 4D] It is a graph showing the trajectory refinement of the sequence of the digital setup. [Figure 5A] It is a diagram showing accessible and inaccessible contacts for IPR application. [Figure 5B] It is a diagram showing accessible and inaccessible contacts for IPR application. [Figure 6A] It is a diagram showing the IPR region of accessible contact points. [Figure 6B] It is a diagram showing the IPR region of accessible contact points. [Figure 7] It is a diagram showing the IPR region of accessible contact points. [Figure 8] It is a graph of the IPR accessibility matrix.

Embodiments for Carrying Out the Invention

[0004] Embodiments of the present invention include a partially or fully automated system for generating a set of orthodontic intermediate setups that enable a set of teeth to move from a malocclusion state to a final setup state, or enable partial treatment from one state to another (e.g., from an initial state to a specific intermediate state). Each tooth arrangement ("state" or "setup") is represented as a node in a graph, and a robotic movement plan is used to expand the graph and search for a path to a valid state. These states or setups can be digital representations of tooth arrangements at a particular treatment stage, and the representation is a digital 3D model. Digital setups can be used, for example, to fabricate orthodontic appliances such as clear tray aligners for moving teeth along a treatment path. Clear tray aligners can be fabricated, for example, by converting the digital setup to a corresponding physical model and thermoforming a material sheet on the physical model, or by 3D printing the aligner using the digital setup. Other orthodontic appliances such as brackets and archwires can also be constructed based on the digital setup.

[0005] The system and method may consider interproximal reduction (IPR), which involves removing a portion of the outer surface of a tooth called enamel. IPR is also known as slimming, stripping, enamel reduction, reproximal reduction, and selective reduction, and the term IPR encompasses these. IPR can be used, for example, between contacting teeth to create space for tooth movement in orthodontic treatment. The system and method applies IPR to a digital 3D model of a tooth to simulate the application of IPR to an actual tooth represented by the model. In other embodiments, the system and method is not required to consider IPR. In other embodiments, the system and method may model the inverse of IPR using the same or similar techniques as when applied to a bridge or implant, for example.

[0006] In one embodiment, the method involves: 1) performing IPR in advance on a digital 3D model of the tooth and generating an initial treatment path by digital setup; 2) calculating the IPR accessibility of each tooth at each stage of the initial treatment path and, if applicable, determining when to apply IPR; and 3) applying IPR throughout the treatment path and dividing the treatment path into biologically feasible movement steps, i.e., trajectories having digital setup points for fabricating appliances corresponding to each setup. In other embodiments, the method can proceed without IPR, or with IPR when it is possible to perform IPR.

[0007] Figure 1 shows a diagram of a system 10 for generating a digital setup (21) of an orthodontic appliance. The system 10 includes a processor 20 that receives a digital 3D model (12) of a tooth from an intraoral 3D scan or a scan of a tooth impression, and in other embodiments, the system receives manual input from a user. The system 10 may also include an electronic display device 16, such as a liquid crystal display (LCD) device, and an input device 18 that receives user commands or other information. Systems that generate digital 3D images or models based on a set of images from multiple viewpoints are disclosed in U.S. Patents No. 7,956,862 and No. 7,605,817, both of which are incorporated herein by reference as if they were fully described. These systems may use an intraoral scanner to obtain digital images from multiple views of a tooth or other intraoral structure, and process these digital images to generate a digital 3D model representing the scanned tooth and gum. The system 10 may be implemented using, for example, a desktop, notebook, or tablet computer. The system 10 may receive 3D scans locally or remotely over a network.

[0008] The 3D scans discussed herein are represented as triangular meshes. A triangular mesh is a common representation of a 3D surface and has two components. The first component, called the vertices of the mesh, is simply the coordinates of the 3D points reconstructed on the surface, i.e., the coordinates of the point cloud. The second component, the mesh faces, is an efficient way to encode the connections between points on an object and interpolate between discrete sample points on a continuous surface. Each face is a triangle defined by three vertices, and a surface can be obtained that can be represented as a set of small triangular face patches.

[0009] A. Algorithm Overview The embodiment includes a motion planning-based method for generating intermediate setups (in stages) based on an initial malocclusion setup and a final proposed setup. The goal is to produce a series of setups that result in the final proposed setup from the malocclusion setup, while also satisfying several evaluation criteria, and meeting the limits of tooth movement at each stage. The evaluation criteria may require a collision-free setup, minimal gingival replacement, acceptable movement, how movement may be related, or other conditions.

[0010] The flowchart shown in FIG. 2 presents an overview of the system and method. A sequence of key setups is provided as input, starting with an initial malocclusion state and ending with a final proposed setup, with others being additional setups (the "key setups") that help guide the movement. The key setups are one way to guide the search for intermediate setups, but the key setups are not required for the intermediates. For example, when using the key setups to create space to reduce tooth crowding, the trajectory should typically move the teeth outward and tilt them, apply IPR, and then retract / upright the teeth. In this case, in addition to the first and last, one key setup can be added to specify this flare-out configuration. The intermediate key setups may be generated manually by a technician or by an automated technique. One potential way to generate the intermediate key setups is to code rules for coordinated movement of the teeth (e.g., flare-out) that can be applied to the tooth set. Another possible way is to use a logic controller to create the key intermediate setups based on the inputs, outputs, and rules specified below.

[0011] The input variables of this logic controller are · The difference between the current and target positions of each tooth (i.e., the position error vector) (e x ,e y ,e z ), · The Euler angle error vector of each tooth (e φ ,e θ ,e ψ ), · The penetration depth (Pd) of tooth collisions, and · The possible penetration directions of tooth collisions (c x ,c y ,c z ).

[0012] The output variables or control commands of this logic controller are · The Euler angle vector of the rotation command for each tooth (d φ ,d θ ,d ψ ), and • Translation vector command for each tooth (d x d y d z ) That is the case.

[0013] In one embodiment, there are two sets of control rules governing the keyframe generation system: rules that control each tooth to converge to its final or intermediate state (Table 1), and rules that avoid collisions by moving in the opposite direction to the depth of entry (Table 2).

[0014] [Table 1]

[0015] [Table 2]

[0016] All IPRs are pre-applied to the malocclusion state (step 22). This means that the total IPR prescribed to be performed at the end of treatment is applied to the shape dimensions used throughout the area refinement process. After the IPRs are applied, a graph is formed including key setups S1, S2, S3, and S4 as shown in Figure 3 (step 24), where S1 is the initial malocclusion state, S4 is the final setup at the end of treatment, and S2 and S3 are intermediate setups at different treatment stages. Once the sequence of states is created, multi-resolution trajectory refinement is performed (steps 26, 28, 30, 32, 34, and 36, described below).

[0017] This trajectory refinement process is described for a single resolution in Figures 4A to 4D. First, the shortest path through the graph is linearly interpolated at a resolution δ (Figure 4A), where δ varies with each resolution in the multi-resolution method. After interpolation, each setup in the sequence is evaluated to see if it satisfies a given criterion (Figure 4B). If the evaluation criterion is not met, the refinement process is repeated until the shortest path satisfies the criterion. Invalid states become the target of refinement (setups in region 44 of Figure 4C), and the search begins from these states and their neighborhoods. Here, the search is at the same resolution δ as in the linear interpolation case. Additional edges that enable multiple paths in the graph between the initial and final setups are generated during the search (dashed lines in Figures 4C and 4D). The edges are weighted by the nodes from which they originate. After many new edges have been added, a search is performed to find a new shortest path. If a path is found (region 46 of Figure 4D), the path can be smoothed. Path smoothing involves interpolating new edges between nearby states in the path to find a new shortest path, and edges are added only if they satisfy a given criterion at resolution δ. In other embodiments, smoothing is not required. The refinement process is repeated until all setups in the sequence satisfy a given criterion.

[0018] If the sequence meets the evaluation criteria, the sequence is smoothed again (step 38), and the final path is returned (steps 40 and 42). The path is then fed into the IPR application step, which identifies where in the path IPR application should occur. The stepping then returns a series of setups with a resolution l which may be greater than δ.

[0019] This process also incorporates a treatment redesign function (steps 25 and 27).

[0020] Algorithm 1 details the entire process, excluding the treatment redesign function described below. The design of specific sub-algorithms will be discussed later.

[0021] Algorithm 1 Input: Initial setup v0 and final setup v f , as well as any key intermediate stage. Typically, v0 is a malocclusion setup, and v f This is the final proposed setup. Input: List of step sizes, ResolutionList Input: Step-by-step restriction l Input: Evaluation function E that assigns a score to the setup Output: Starts with v0, v f Ends with a stepped trajectory at resolution l, minimizing undesirable movement. 1. v 0← PreapplyIPR(v 0, v f ) 2. Vertices v0 and v f Forms a graph G including v0 and v f These are connected within G. This could simply be a graph having only these two vertices and a single edge connecting them, a graph having a set of key setups reported by the case's move rules, or a graph resulting from a previous setup suggestion search. 3. P ← ShortestPath(G,v0,v f ) 4. Regarding δ in ResolutionList, 5. P ← AdjustResolution(G,P,δ) 6. Unless P satisfies E, 7. R ← IdentifyRefinementRegion(P,E) 8. RefineRegion(G,v0,v f ,R) 9. P ← ShortestPath(G,v0,v f ) 10. P,G ← Smooth(P,E,δ) Repeat step 11. 12. P,G ← Smooth(P,E,δ) Repeat step 13. 14. P ← IPRBatching(G) 15. Return Stage(P,l)

[0022] B. Sub-algorithm design Algorithm 1 is modular in that different implementations of its sub-algorithms yield different behaviors. This specification provides specific implementations of each module that have been proven to succeed on a given task.

[0023] B1. Pre-application IPR (22) In one embodiment of using IPR, two methods are possible for pre-applying IPR. In the first method, modifications described by the plane and volume of IPR are pre-applied to the tooth shape dimensions, which are used throughout the trajectory refinement period. Alternatively, IPR recognition evaluation methods may be used during the trajectory refinement process. These evaluation methods can accurately evaluate the state describing not only the tooth position and rotation, but also the mesial and distal IPR volumes (the "IPR configuration" portion of the state). In this case, the IPR configuration of the setup state is copied to the IPR configuration of the malocclusion state. These methods yield identical behavior.

[0024] B2. Refinement of multi-resolution trajectories The algorithm refines path P at multiple resolutions, starting at a low resolution (large δ) and then moving to a higher resolution (small δ). In the preferred mode, the algorithm uses the following δ sequence: [8.0, 4.0, 2.0, 1.0]. Each step of the multi-resolution adjustment is discussed individually below.

[0025] B2a.Resolution adjustment (26) This process modifies the paths in the graph to enforce a given resolution. Algorithm 2 describes the implementation.

[0026] [Table 3]

[0027] Algorithm 2 uses an interpolation subroutine with two options: identical finish, where all teeth complete their movement at the same (final) stage, and fast finish, where teeth complete their movement as quickly as possible, and not all teeth may be finished at the same stage. The preferred mode is to use identical finish.

[0028] B2b. Identification of areas for refinement (32) This identifies a subset of trajectories P that require further refinement according to several evaluation criteria E. Several options for the output of this algorithm include returning all parts of P that violate E, the first violation in P, or the worst violation in P. The following techniques are possible implementations for this refinement. • IdRefinementRegionFirst: Returns the first location along P where the score calculated using the evaluation criteria exceeds the threshold score. • IdRefinementRegionAll: Returns all locations along P where the score calculated using the evaluation criteria exceeds the threshold score. • IdRefinementRegionWorst: Returns the worst violation within P, defined as the element that received the worst score using the evaluation criteria. In case of a tie (i.e., multiple worst violations), the last worst violation within P is returned. • IdRefinementRegionWorstAll: Returns all worst violations within P. • IdRefinementRegionWorstNeighbors: Returns the worst violation in P and its N neighbors in P. In case of a tie (i.e., multiple worst violations), the last worst violation in P is returned.

[0029] The preferred mode for the system is to use IdRefinementRegionAll.

[0030] Identifying the refinement area uses evaluation criteria defined by the evaluator subroutine. Other scoring criteria, such as tooth replacement by gingiva or the number of overlapping vertices between teeth in the arch, may also be used instead of or in addition to these described measures. A preferred mode is using the count of collision contact points.

[0031] B2c. Area refinement (34) This refines the region of the input graph G. Algorithm 3 provides this method.

[0032] Algorithm 3 RefineRegion(G,v0,v f ,R) Input: Graph G Input: Initial setup v0 and final setup v f Input: Region of G to refine R (identified using Refinement Region Identification) Output: G is v0~v within G f To obtain a better path, it is refined near R. 1. P ← ShortestPath(G,v0,v f ) 2. s P← ScorePath(P) 3. Let the vertices in R be called Candidates. 4. Let the vertices adjacent to R be called Neighbors. 5. score(P)≧s P As long as that is the case, 6. c ← SelectCandidate(Candidates) 7. c' ← IterativePerturb(c) 8. If score(c') ≤ score(c), 9. C ← Interpolate(c, c', δ) 10. C_valid ← ExtractValidSubset(C) 11. If len(C_valid) < max_length 12. Add C_valid to G, Candidates, and Neighbors 13. For each c neigh ∈ {the k most highly scoring vertices within Neighbors} 14. C ← Interpolate(c', c neigh , δ) 15. C_valid ← ExtractValidSubset(C) 16. Add C_valid to G, Candidates, and Neighbors 17. Repeat 18. Repeat 19. Repeat 20. P ← ShortestPath(G, v0, v f ) 21. P ← Smooth(P, E, δ) 22. Repeat

[0033] This algorithm 3 uses the SelectCandidate subroutine to identify candidates from the list of Candidates. For this subroutine, two methods have been identified. 1. Random selection: Randomly select candidates from the list of Candidates 2. A * Graph search: Select candidate nodes based on heuristic calculations. The heuristic is a quantity that estimates the "remaining cost" of a node (i.e., the quantified effort to move from that node to the end state of the graph). A *The algorithm uses heuristics to efficiently search the graph and selects the next node to explore using heuristics that predict which node is most promising. Because it searches a list of Candidates using heuristics, a single node is selected. The following heuristics are possible:

[0034] Euclidean distance between candidate state and setup state: For each candidate state, calculate the Euclidean distance between all tooth positions in the candidate state and the setup state, then calculate a heuristic as the sum or average of these distances, and select a candidate node by minimizing the heuristic.

[0035] Degree of collision: The degree of collision can be quantified by counting the number of points by one tooth located within the boundary of another tooth, or by calculating the number of contact points between two meshes. A heuristic is calculated as the count of these overlapping points between teeth in the arch. The next node can be selected by maximizing the collision count (to explore the problematic state early) or by minimizing the collision count (for algorithmic methods that attempt to solve the problem by identifying locally optimal (i.e., easily solvable) nodes before moving on to areas where the solution becomes more difficult).

[0036] In the preferred mode, candidates are randomly selected for all iterations except the first, and instead, the candidate with the worst heuristic score is selected. The heuristic is the number of contact points.

[0037] In addition, Algorithm 3 uses the IterativePerturb subroutine to iteratively perturb the state according to the Perturb function, ultimately selecting state c' that achieves the best score by the evaluator. In the preferred mode, IterativePerturb is called 50 times, and the evaluator calculates the number of collision points. The preferred mode uses propagating directional perturbation, which perturbs the colliding teeth and neighboring teeth that are in approximately the opposite direction to the directional entry vector.

[0038] Algorithm 3 calculates the score s of the shortest path P along graph G. P To calculate this, use the ScorePath subroutine. The path score can be calculated using the function: s P =Total([Part inside P) i Regarding (location) i (Number of collisions + 1) 1.25 ])

[0039] The ExtractValidSubset subroutine is used to extract a subset of interpolation paths C that are valid according to several evaluator functions. The preferred mode is to use the number of collision points as the evaluator, and consider only nodes with zero collision points to be valid.

[0040] The preferred mode of the Smooth function is iterative smoothing, as described in "B2d.Trajectory Smoothing" below.

[0041] B2d.Trajectory smoothing (30, 38) Many possible smoothing algorithms exist. One possibility (Algorithm 5) tries all possible edges between any two nodes in P and keeps only those that satisfy E with a resolution δ. However, this method is computationally expensive, especially for long paths. Instead, smoothing can be attempted only for nodes that are within a distance threshold of each other. Another implementation involves an iterative smoother that repeatedly applies this smoothing process until the score is no longer improved by smoothing. This iterative smoother is the preferred mode.

[0042] [Table 4]

[0043] B3. IPR batch processing (40) Because IPR is pre-applied (Section B1), automated intermediate trajectory detection is performed first when the teeth are in a malocclusion state, with all IPRs already performed. However, performing this is practically impossible or impractical because the teeth must be accessible for IPR to be performed, and may not be accessible until later in the treatment. This section provides a strategy for incorporating IPR batch processing into the plan. The method takes as input the graph and path (output from Section 2) generated by the algorithm when IPR is pre-applied. This path is then refined based on the IPR application.

[0044] The method for incorporating IPR is described below. First, we discuss how to determine IPR accessibility. Next, different methods for applying IPR batch processing to a given trajectory are presented. In other embodiments, IPR batch processing is not required.

[0045] B3a. IPR Accessibility In practical IPR applications, the mesial or distal surface of a tooth must be accessible, meaning that the tooth cannot be positioned behind another tooth or intersect with the labial / lingual surface of another tooth (see Figures 5A and 5B). Therefore, IPR accessibility is determined by taking a weighted sum of the posterior tooth distance to neighboring teeth and the tooth intersection with neighboring teeth.

[0046] To measure the anterior or posterior tooth distance of another tooth, the mesial / distal regions of each tooth are identified, as shown by the shaded areas 52 in Figure 6A (occlusal view) and Figure 6B (front view). The distance between the mesial / distal points of neighboring teeth in a direction perpendicular to the arch is then calculated. To measure tooth intersection with neighboring teeth, a ray is guided along the incisal edge of the tooth, and the closest intersection point with the neighboring tooth is identified. The distance between the intersection point of the neighboring tooth and the mesial or distal region is calculated. The following are some other methods for determining whether contact between two neighboring teeth is approved for IPR application: by distance from the mesial / distal surfaces along the incisal edge, by the outward normal direction, and a hybrid method combining these two methods. These methods will be discussed in turn below.

[0047] The contact point is projected onto the incisal edge. The algorithm defines the mesial / distal regions as areas near the endpoints of the incisal edge. See the shaded area 52 in Figure 6A (occlusal view) and Figure 6B (front view). The incisal edge may be defined as the line segment connecting the incisal edge DSL / ABO markers (if markers are provided) and the line segment connecting the minimum and maximum tooth vertices along the local x-axis of the tooth projected onto the local z=0 plane (if no markers are provided). To determine accessibility, the contact point is first projected onto the z=0 plane and then onto the incisal edge line segment. If this projected point is within an absolute distance or within a distance ratio threshold relative to the nearest incisal edge endpoint, the point is accessible for IPR. Otherwise, the point is inaccessible. The absolute distance is likely to be sufficient, as the maximum IPR to be applied is given as input when the final setup is provided. As shown in Figures 6A and 6B, the contact point is within the IPR region 52 if it is near the incisal endpoint 54.

[0048] This is determined by the contact points directed toward the outward normal. The algorithm can also define the mesial / distal regions as areas where the outward normal is directed roughly along an arch (see shaded area 56 in Figure 7). IPR accessibility is determined by comparing outward contact points perpendicular to the local x-axis of the tooth.

[0049] Hybrid methods. Hybrid methods combine the methods presented above. If one method is uncertain, the other method can be used.

[0050] B3b.IPR application If a model exists to determine whether IPR is accessible, that model can be used to determine when IPR is possible. This relaxes the assumption that IPR is always immediately applicable. RefineTrajectoryWithIPR is similar to the trajectory refinement methods presented in Section B2 when only a single resolution level is used.

[0051] [Table 5]

[0052] First, all applicable IPRs are v f It is reset to 0 except for [specific value]. Then, IPR is applied to P according to the IPR strategy (see Section B3c: IPR Strategy). Refinement proceeds as before to remove the remaining invalid portion of the trajectory.

[0053] The ApplyIPR subroutine investigates route P and, only if accessible, applies the policy based on v fApply IPR up to the given limit. (IPRisAccessible in algorithms 2 and 3 implements the desired access model by IPR accessibility in section B3a.) When IPR is applied to a particular vertex, it propagates to all of its children (i.e., the IPR value increases only along the trajectory, which means that no tooth material is added back).

[0054] B3c IPR Strategy The following are two possible IPR batch processing policies: batch processing that is as fast as possible and minimizes the number of IPR sessions.

[0055] As quickly as possible: Algorithm 2 below simply iterates through P and v f Apply the complete IPR value to the pair of neighboring teeth for which the complete value has not yet been applied and which is accessible as reported by IPRisAccessible (see Section B3a. IPR Accessibility).

[0056] [Table 6]

[0057] Batch processing to minimize the number of IPR sessions: IPR batch processing can be viewed as an optimization problem where it is desirable to minimize the number of times IPR is applied. Therefore, IPR should be applied fully when accessible. Applying IPR as early as possible during treatment may also be preferred.

[0058] To optimize batch processing, the algorithm can investigate when all IPR application points are accessible along P. The algorithm can use an accessibility matrix for this function (Figure 8). IPR application points are located along the x-axis in the matrix. Note that teeth that do not require IPR application (e.g., v) fMolars or teeth that have the same IPR at v0 as in the case of ( ) are excluded from this matrix because they are not points where IPR can be applied. The y-axis is from v0 to v along P. f This represents the path steps to be taken. Each cell is shaded if an IPR application point is accessible in that path step (determined using IPRisAccessible by section B6a). Horizontal cuts represented by lines 58 and 60 along this matrix represent the application of IPR to a set of shaded teeth all at once. Thus, the algorithm seeks to minimize the number of cuts required to cover all IPR application points. For a given cut, IPR is applied in the first path step where the maximum number of teeth are accessible.

[0059] Algorithm 3 below outlines how this accessibility matrix can be used to apply batched IPRs. In Algorithm 3, IPRisAccessible is not explicitly called before applying the IPRs, as implied by the construction of accessibility matrix A.

[0060] [Table 7]

[0061] The key step is to find the minimization function for slice S. To do this, the algorithm adds slices to S, starting with the slice that maximizes the number of remaining application points affected.

[0062] B4. Trajectory breakdown (42) The purpose of dividing the trajectory into stages is to create a path P with a resolution l defined based on the tooth movement limits for each stage. The pseudocode for this method is shown below. 1. P_staged=P[0] 2. For each node within P, calculate the tooth position change Δd from the malocclusion state. 3. For each Δd: a. If Δd > tooth movement limit per stage: b. Add P[i-1] (the node immediately preceding the current node) to P_staged. c. Recalculate Δd as the change in tooth position from P[i-1]. 4. Add the setup status (P[end]) to P_staged.

[0063] In the system, trajectory segmentation is not required, but it can be used. Instead, the resolution the system uses to interpolate the path is equal to the tooth movement limit for each stage.

[0064] B5. Redesigning Treatment (25, 27) Digital orthodontic treatment planning follows an iterative process. 1. Manually create the final setup, or evaluate and adjust the automatically generated final setup. 2. Step-by-step: Manually create intermediate steps, or evaluate and adjust a set of automatically generated intermediate steps. 3. For approval, send the final setup and stage breakdown to the practitioner. 4. Make any necessary adjustments; if the final setup is changed, recalculate the two-step process. 5. Repeat steps 3-4 as needed.

[0065] If the final setup is changed, the steps must be adjusted to reflect the new final setup. The steps can be completely recalculated using the adjusted new final setup as the target. However, in many cases the adjustments are minor, and much of the previous step calculation can potentially be reused. Embodiments of the present invention include a method for reusing calculations to improve the responsiveness of automated software for dental technicians. This method can improve the technician's workflow by reducing the amount of time spent waiting for step recalculations. This method can also reduce the amount of context switching between cases, as technicians may not have to move to a new case while waiting for step recalculations. This method can also be used to provide practitioners with an interactive tool for comparing treatment plans and educating patients.

[0066] method The idea is to interpolate the initial setup to a new final setup and create connections (edges) between it and the graph resulting from previous step-by-step calculations. This enhances the solution-finding process from previous efforts. The step-by-step process then proceeds as described above. The following algorithm provides a high-level diagram of the intermediate step-by-step process that incorporates the reuse of pre-calculated graphs. Based on Algorithm 1 in Section A, steps 1, 4, and 5 are further enhanced with support for calculation reuse along with additional inputs. Algorithm:RefineTrajectory(v0,vf,G,d,l,E,G0) Input: Initial setup v0 and final setup vf Input: A graph G containing vertices v0 and vf, where v0 and vf are connected within G. G may simply be a graph with only these two vertices and a single edge connecting them, a graph with a set of key setups reported by a case move rule, or a graph from a previous setup suggestion search. Input: Step size d Input: Step-by-step restriction l Input: Evaluation function E that assigns a score to the setup Input: Run graph G0 using the previous RefineTrajectory. If no previous run has occurred, G0 is null. Input: Small connection constant k Output: Stepped trajectory at resolution l, starting at v0 and ending at vf, minimizing undesirable movement. 1. Add G0 to G 2. P←ShortestPath(G,v0,vf) 3. P←AdjustResolution(G,P,d) 4. For all p within P: Try to find the k closest connections between p and G0. 5. P←ShortestPath(G,v0,vf) 6. Unless P satisfies E, P←Smooth(P,E,d) R←IdentifyRefinementRegion(P,E) RefineRegion(G,v0,vf,R) P←ShortestPath(G,v0,vf) 7. P←Smooth(P,E,d) 8. Return Stage(P,l)

[0067] Before being added to G, the previous graph G0 may optionally be modified to provide an initial estimate of how the graph might need to be modified. This modification may be achieved by scaling movements at each node of the teeth modified in the final setup. For each tooth modified in the final setup, the movement is scaled in proportion to the magnitude and direction of the change in position.

[0068] Furthermore, the perturbation direction learned from the previous graph G0 can be stored and reused during the RefineRegion phase to motivate searching in directions that are more likely to improve the score. The stored perturbation direction may be either the actual perturbation direction or related data such as gradient calculations that can be used to randomly sample or derive new perturbation states.

[0069] To connect the previous graph G0 to the new graph G, additional edges must be added between them. Attempting connections from all G to vertex G0 is computationally expensive (and unnecessary). Instead, only a subset of G containing the candidate path P needs to have the additional connections to vertex G0 that are attempted. For each vertex p in P, connections to vertices in G0 must be attempted, which is typically done by attempting connections between p's k nearest neighbors in G0.

[0070] The RefineRegion process includes a SelectCandidate subroutine used to identify candidates from a list of Candidates. The SelectCandidate subroutine may be modified to preferentially select candidate teeth that were changed in the final setup, as well as teeth in their vicinity.

[0071] For nodes in the new graph, in the rare situation where the nearest neighbor in the previous graph is very far away (a distance exceeding a given threshold), G0 may be discarded, and the adjustment can be treated as a precise region from the start, as if it were a new stepwise calculation. This can potentially be approximated using the distance between the original final setup and the modified final setup. If this distance is too large, the additional cost of calculating the nearest neighbor can be avoided, and the procedure can revert to performing stepwise calculations from the start.

[0072] Interactive system Due to the computational efficiency of this algorithm, it can potentially be deployed as a dashboard that allows a physician or technician to visualize the effect of changing the final setup of intermediate stages in real time. In some other methods, the dashboard may be impractical due to the long computation time required to recalculate the stages, and the method of this embodiment can reduce computation time. Some of the information that may be displayed on this dashboard panel include the total tooth movement per tooth, the number of stages, and the amount of pure extrusion or torque (i.e., movement that is difficult or impossible to achieve by the aligner).

[0073] By providing this immediate feedback, dental technicians and / or physicians can experiment with multiple final setup modifications to understand how the modifications affect treatment time, feasibility of tooth movement, and other factors, ultimately enabling them to plan a desirable treatment with fewer iterations between the treating physician and the dental technician designing the treatment plan.

[0074] An alternative dashboard can be designed that displays the original (pre-modification) final setup, along with key information such as those mentioned above, in either half the display or a smaller inset window, allowing the technician or physician to easily compare the effects of modifications and automatically revert to previously generated setups and steps that have been saved. For example, if applying a few minor modifications to the final setup increases the number of steps beyond what is comfortable for the physician, they can simply select ("click") a "Revert to Original" button to load back to the previous final setup and its steps without having to undo any changes that may have been made in the inset window. This allows for a certain amount of fine-tuning of the final setup. In addition, the views of the teeth and arches in the pre-calculated and modified steps may be locked so that they always move together, making comparison easier.

[0075] C. Additional components and performance C1. Extraction of multiple routes This component allows a set of viable pathways to be presented to the practitioner, enabling the practitioner or patient to select the desired treatment pathway. For the benefit of the technician or practitioner, the multiple pathway options should be distinct. In particular, they should all be feasible but possess different high-level characteristics, such as the number of IPR stages, the amount of reciprocal movement, or other options. To provide this, pathways should be scored or commented based on such characteristics. The algorithm then identifies all feasible pathways, sorts them by score, and reports the best-scoring pathway that does not share the same characteristics as other better-scoring pathways in the list.

[0076] C2. Multiple IPR policies This component presents practitioners with options for which IPR application policy to apply: batch processing to minimize the number of treatments as quickly as possible, or a policy somewhere in between. Practitioners can be presented with options to evaluate on a case-by-case basis whether the different options are desirable or whether those preferred policies do not yield the desired results.

[0077] C3. Single arch vs. multiple arches In one implementation, the upper and lower arches are considered separately. This is because the problem can be solved more easily (and therefore more quickly). In another embodiment, both arches are supported and considered together (for example, if this is preferred).

[0078] C4. Interactive Tools Some library components can be either fully specified by the user, fully automated, or interactive, with the user providing input to the automation method. This category can be seen in the following areas:

[0079] Key setup generation – this can be one of the most difficult processes to automate, and the user may provide input to the method. Selecting a useful key setup may depend on the ability to discriminate which case type (crowding, midline correction, etc.) is being considered. This is something that can be learned over time by an algorithm, but is also something that is easily provided by a dental technician (through quick visual inspection and input such as checkboxes). In addition, the dental technician can also explicitly provide intermediate stages that characterize the orthodontic treatment plan for the case type.

[0080] Perturbational movement—to a lower degree than described above, dental technicians may also be able to provide orthodontic movement plans through approximate setup inputs. This plan can then be applied during inactive automatic perturbations, allowing for a more rapid discovery of clinically effective solutions.

[0081] Identifying refinement regions - If the algorithm is focused on a few local minimums, the user may be able to identify good regions for refinement based on visual inspection and manually perturb the state toward these good regions. In addition to the embodiments described above, the following embodiments are also noted. (Note 1) A computer-based method for generating and reusing setups for orthodontic treatment pathways, The process of receiving a digital 3D model of the teeth, A process for generating an initial treatment pathway having stages including an initial setup, a final setup, and multiple intermediate setups, The process of dividing the initial treatment pathway into initial steps of feasible tooth movement, and resulting in a final treatment pathway having a setup corresponding to the initial steps, A step of calculating new steps of feasible tooth movement based on the initial step and with respect to only a portion of the initial treatment pathway, A step of generating the final treatment pathway using a new setup corresponding to the aforementioned new step, Methods performed by computers, including those mentioned above. (Note 2) The method according to Appendix 1, wherein the step of generating the initial treatment pathway includes generating the initial treatment pathway having a key intermediate setup. (Note 3) The method according to Appendix 1, wherein the segmentation step includes performing trajectory refinement of the initial treatment pathway based on a specific resolution, the resolution comprising a plurality of steps within the treatment pathway. (Note 4) The method according to Appendix 3, wherein the division step includes performing trajectory refinement of the initial treatment pathway based on evaluation criteria for assigning scores to the setup. (Note 5) The method according to Appendix 3 or 4, wherein the trajectory refinement is performed by adjusting the resolution of the treatment pathway. (Note 6) The method according to any one of the appendices 3 to 5, wherein the trajectory refinement is performed, including generating a new treatment route if the treatment route does not meet the evaluation criteria. (Note 7) The method according to Appendix 6, wherein generating the new treatment pathway includes applying a smoothing algorithm to the treatment pathway. (Note 8) The method according to Appendix 6 or 7, wherein generating the new treatment pathway includes applying a region refinement algorithm to the treatment pathway. (Note 9) The method according to any one of the appendices 6 to 8, wherein generating the new treatment pathway includes finding the shortest treatment pathway that satisfies the evaluation criteria. (Note 10) The method according to Appendix 1, wherein the segmentation step includes generating the final treatment pathway based in part on user input. (Note 11) The user input is as described in Appendix 10, including key setup. (Note 12) The user input is as described in Appendix 10 or 11, including intermediate setup. (Note 13) The user input is as described in any one of the appendices 10 to 12, including an approximate setup. (Note 14) The user input is as described in any one of the appendices 11 to 13, including identifying areas for refinement. (Note 15) The method according to Appendix 1, wherein the division step includes dividing the final treatment pathway based on keyframes. (Note 16) The method according to Appendix 1, wherein the aforementioned classification step includes generating a plurality of final treatment pathways. (Note 17) The method described in Appendix 16, further comprising reporting one or more scores for each of the aforementioned multiple treatment routes. (Note 18) The method according to Appendix 16 or 17, further comprising selecting the final treatment route from among the plurality of final treatment routes based on user input. (Note 19) The process involves performing interproximal surface resurfacing (IPR) on the aforementioned model, A step of calculating the IPR accessibility of each tooth at each stage of the initial treatment pathway, The steps include applying IPR to the entire initial treatment pathway based on the calculated IPR accessibility, The method described in Appendix 1, further including the above. (Note 20) The method according to Appendix 19, wherein the step to be applied includes applying an IPR batch processing algorithm to the treatment pathway. (Note 21) The method according to Appendix 20, wherein applying the IPR batch processing algorithm includes applying IPR as early as possible during the treatment. (Note 22) The method described in Appendix 20, wherein applying the aforementioned IPR batch processing algorithm includes minimizing the number of IPR sessions. (Note 23) Applying the aforementioned IPR batch processing algorithm is the method described in Appendix 21 or 22, which includes selecting a batch processing algorithm based on user input. (Note 24) The method according to Appendix 1, wherein the step of calculating the new step includes calculating the new step based on the initial step and the new final setup. (Note 25) The method according to Appendix 24, wherein the step of calculating the new step includes trying the shortest path between the initial step and the new final setup. (Note 26) The method according to Appendix 15, wherein the step of calculating the new step includes calculating the new step based on the changes in the initial step and the keyframes.

Claims

1. A computer-based method for generating and reusing setups for orthodontic treatment pathways, One or more computer processors receive a digital 3D model of a tooth, The process involves one or more computer processors generating an initial treatment path having multiple stages, including an initial setup, a final setup, and a plurality of intermediate setups, wherein each setup includes a digital 3D model of a tooth representing a stage of treatment. The process involves one or more computer processors dividing the initial treatment path into a plurality of trajectories, each representing a feasible movement of the tooth, and resulting in a final treatment path having a plurality of setups corresponding to the plurality of trajectories, wherein the feasible movement of the tooth is determined by the tooth movement limits for each stage. A step in which one or more computer processors calculate a new trajectory of feasible tooth movement based on the plurality of trajectories and with respect to only a portion of the initial treatment path, wherein the calculation of the new trajectory includes adjusting the initial treatment path by reusing at least one setup of the initial treatment path to calculate the new trajectory, The process of generating the final treatment path by having one or more computer processors replace some of the steps included in the initial treatment path with a new setup that reflects the new trajectory, Methods performed by computers, including those mentioned above.

2. The method according to claim 1, wherein the segmentation step includes one or more computer processors performing trajectory refinement of the initial treatment path based on a specific resolution indicating the distance between the setups, the trajectory refinement includes adjusting the setups such that the tooth movement in each setup interpolated for each resolution exceeds the tooth movement limit for each step.

3. The method according to claim 2, wherein the segmentation step includes one or more computer processors performing trajectory refinement of the initial treatment path to satisfy evaluation criteria for assigning a score to the setup, the evaluation criteria being based on at least one of collision-free setup, minimal gingival replacement, and acceptable movement.

Images