Predicting the manufacturing results of additive manufacturing objects

Abstract

Methods and systems for predicting manufacturing outcomes are provided. In some embodiments, a method includes receiving at least one image representing a target geometry of an object to be manufactured using an additive manufacturing process. The method may include generating at least one modified image by inputting the at least one image into a machine learning algorithm. The machine learning algorithm may be trained to determine one or more modifications to the at least one image, wherein the one or more modifications are configured to compensate for predicted deviations from the target geometry of the object when the object is manufactured via the additive manufacturing process based on the at least one image. The method may further include generating instructions for manufacturing the object using the additive manufacturing process based on the at least one modified image.

Description

Cross-references to one or more related applications

[0001] This application claims priority to U.S. Provisional Application No. 63 / 617,649, filed January 4, 2024, the entire contents of which are incorporated herein by reference. Technical Field

[0002] This technology generally relates to additive manufacturing, and specifically to predicting the manufacturing results of additive manufacturing objects. Background Technology

[0003] Additive manufacturing encompasses a variety of techniques involving building 3D objects from multiple layers of material. Typically, the manufacturing process involves creating a digital model of the object, converting the model into a series of slices, and then sequentially printing the slices to build the object layer by layer. However, the actual geometry of the printed object may not match the initial object design. For example, issues such as over-curing and resin contamination during vat polymerization can affect the dimensional accuracy of the final printed object. Furthermore, post-processing conditions such as centrifugation, heating, and solvent washing can cause warping of the object's geometry. This deviation between the object's actual and intended geometry can adversely affect the functionality and properties of the printed object. Attached Figure Description

[0004] Many aspects of this disclosure can be better understood by referring to the following figures: the components in the figures are not necessarily drawn to scale. Rather, the focus is on clearly illustrating the principles of this disclosure.

[0005] Figure 1 This is a flowchart providing a general overview of a method for fabricating and post-processing additively manufactured objects according to embodiments of the present technology.

[0006] Figure 2 This is a partial schematic diagram providing a general overview of an additive manufacturing process according to embodiments of the present technology.

[0007] Figure 3 This is a block diagram providing a general overview of a workflow for additive manufacturing of objects according to embodiments of the present technology.

[0008] Figure 4 This is a flowchart illustrating a method for generating instructions for additive manufacturing of an object according to an embodiment of the present technology.

[0009] Figure 5 This is a block diagram illustrating a prediction algorithm for predicting manufacturing results according to an embodiment of the present technology.

[0010] Figures 6A to 6CThe illustration shows the determination of an experimental calibration threshold for filtering predicted images according to an embodiment of the present technology.

[0011] Figure 7 This is a schematic illustration of the divide-and-conquer approach according to an embodiment of the present technology.

[0012] Figure 8 This is a block diagram illustrating an optimization algorithm for generating modified images according to an embodiment of the present technology.

[0013] Figures 9A to 9C This is a schematic illustration of an inverse optimization process for modifying the geometry of an object according to an embodiment of the present technology.

[0014] Figure 10 This is a block diagram illustrating a workflow for generating instructions for additive manufacturing of an object according to an embodiment of the present technology.

[0015] Figure 11 This is a block diagram illustrating the workflow for training a surrogate algorithm according to an embodiment of the present technology.

[0016] Figure 12 This is a flowchart illustrating a method for generating instructions for additive manufacturing of an object according to an embodiment of the present technology.

[0017] Figure 13 This is a flowchart illustrating a workflow for evaluating a modified image of an object according to an embodiment of the present technology.

[0018] Figure 14 This is a block diagram illustrating the workflow for training a reverse algorithm according to an embodiment of the present technology.

[0019] Figure 15A The illustration shows a representative example of a tooth repositioning appliance configured according to an embodiment of the present technology.

[0020] Figure 15B The illustration shows a tooth repositioning system comprising multiple appliances according to an embodiment of the present technology.

[0021] Figure 15C A method for orthodontic treatment using multiple appliances according to an embodiment of the present technology is illustrated.

[0022] Figure 16 The illustration depicts a method for designing orthodontic appliances according to an embodiment of the present technology.

[0023] Figure 17The illustration depicts a method for digitally planning orthodontic treatment and / or designing or fabricating an appliance according to an embodiment of the present technology. Detailed Implementation

[0024] This technology relates to methods and systems for additive manufacturing of objects, such as dental appliances. In some embodiments, for example, this technology provides a method comprising: receiving at least one image representing a target geometry of an object to be manufactured using an additive manufacturing process. The method may include: generating at least one modified image by inputting the at least one image into a machine learning algorithm. The machine learning algorithm may be trained to determine one or more modifications to the at least one image, wherein one or more are configured to: compensate for predicted deviations from the target geometry of the object when the object is manufactured via an additive manufacturing process based on the at least one image. For example, the machine learning may be a convolutional neural network trained on an initial object image and an optimized object image to predict how to modify the image so that the actual printed object geometry more closely conforms to the target geometry. The method may also include: generating instructions for manufacturing the object using an additive manufacturing process based on the at least one modified image.

[0025] As another example, this technology can provide a method comprising: receiving at least one image representing a target geometry of an object to be manufactured using an additive manufacturing process. The method may include: determining a predicted geometry of the object after manufacturing based on the at least one image. For example, the predicted geometry can be determined using a machine learning algorithm, such as a convolutional neural network that receives at least one image as input and generates at least one output image representing the predicted geometry of the object. The method may further include: identifying a deviation between the target geometry and the predicted geometry; and modifying the at least one image based on the identified deviation. The method may further include: generating instructions for manufacturing the object using the additive manufacturing process based on the at least one modified image.

[0026] Compared to conventional methods for additive manufacturing of objects, this technology offers numerous advantages. For example, conventional methods simply instruct the additive manufacturing system to print the object as designed, without considering geometric deviations that may occur during the additive manufacturing and / or post-processing of the object. If the deviation is significant and / or occurs in a critical part of the object, the object may not be suitable for its intended function. For instance, dimensional inaccuracies can affect the ability of dental appliances to properly fit onto teeth and / or apply forces accurately. The methods and systems disclosed herein overcome these and other challenges by predicting how the geometry of the object will deviate from its intended geometry during additive manufacturing and / or post-processing and by adjusting instructions sent to the additive manufacturing system to compensate for the predicted deviations (“deviation prediction and compensation”). Predictions can be made with a high degree of accuracy even for objects with customized (e.g., patient-specific) geometries. Therefore, the approach described herein can improve manufacturing outcomes, particularly for objects requiring high precision and / or highly variable shapes, and reduce wasted time and materials when reprinting objects that do not meet accuracy standards. Moreover, the approach described herein can be applied to many different types of additive manufacturing processes and post-processing operations.

[0027] Embodiments of this disclosure are described more fully below with reference to the accompanying drawings, in which like reference numerals denote like elements in several figures, and exemplary embodiments are illustrated. However, the embodiments of the claims may be implemented in many different forms and should not be construed as limited to the embodiments set forth herein. The examples set forth herein are non-limiting examples and are merely examples of other possible examples.

[0028] As used herein, the terms “vertical,” “lateral,” “upper,” “lower,” “left,” “right,” etc., can refer to the relative orientation or position of a feature of the embodiments disclosed herein, given the orientation shown in the figures. For example, “upper” or “topmost” can refer to a feature positioned closer to the top of the page than another feature. However, these terms should be interpreted broadly to include embodiments with other orientations, such as inverted or tilted orientations, where top / bottom, above / below, above / below, up / down, and left / right can be interchanged according to orientation.

[0029] The headings provided herein are for convenience only and do not define the scope or meaning of the claimed technology. Embodiments under any heading may be used in conjunction with embodiments under any other heading. I. Prediction of Additive Manufacturing Results

[0030] Figure 1This is a flowchart providing a general overview of a method 100 for fabricating and post-processing additively manufactured objects according to embodiments of the present technology. Method 100 can be used to produce many different types of additively manufactured objects, such as orthodontic appliances (e.g., braces, palatal expanders, retainers, attachment placement devices, attachments), restorative objects (e.g., crowns, veneers, implants), and / or other dental appliances (e.g., oral sleep apnea appliances, oral protectors). Additional examples of dental appliances and associated methods applicable to the present technology are described in Section II below.

[0031] Method 100 begins at box 102, in which an object is created using an additive manufacturing process. Additive manufacturing processes can implement any suitable technique known to those skilled in the art. Additive manufacturing (also referred to herein as “3D printing”) includes a variety of techniques for creating 3D objects directly from a digital model using additive processes. In some embodiments, additive manufacturing includes depositing precursor material onto a build platform. The precursor material may be cured, polymerized, melted, sintered, fused, and / or otherwise solidified to form a portion of the object and / or combined with previously formed portions of the object. In some embodiments, the additive manufacturing techniques provided herein build the object geometry in a layer-by-layer manner, wherein successive layers are formed in discrete build steps. Alternatively or in combination, the additive manufacturing techniques described herein may allow for the continuous construction of the object geometry.

[0032] Examples of additive manufacturing technologies include, but are not limited to, the following: (1) tank photopolymerization, in which the object is constructed from a tank or other volumetric source of liquid photopolymer resin, including techniques such as stereolithography (SLA), digital light processing (DLP), continuous liquid interface production (CLIP), two-photon induced photopolymerization (TPIP), and volumetric additive manufacturing; (2) material jetting, in which material is jetted onto a build platform using a continuous or drop-on-demand (DOD) approach; (3) binder jetting, in which alternating layers of build material (e.g., powder-based material) and binder material (e.g., liquid binder) are deposited via a printhead; (4) material extrusion. In this process, materials are extracted through a nozzle, heated, and deposited layer by layer, such as fused deposition modeling (FDM) and direct ink writing (DIW); (5) powder bed fusion, including techniques such as direct metal laser sintering (DMLS), electron beam melting (EBM), selective thermal sintering (SHS), selective laser melting (SLM), and selective laser sintering (SLS); (6) sheet lamination, including techniques such as layered object fabrication (LOM) and ultrasonic additive manufacturing (UAM); and (7) directional energy deposition, including techniques such as laser engineered net-shape forming, directional light fabrication, direct metal deposition, and 3D laser cladding. Optionally, the additive manufacturing process may use a combination of two or more additive manufacturing techniques.

[0033] For example, additively manufactured objects can be fabricated using a trench photopolymerization process, in which light is used to selectively cure a trench or other volumetric source of a curable material (e.g., a polymeric resin). Each layer of curable material can be selectively exposed via a single exposure (e.g., DLP) or by scanning a beam across that layer (e.g., SLA). Trench photopolymerization can be performed in a "top-down" or "bottom-up" manner depending on the relative positions of the material source, light source, and build platform.

[0034] As another example, additive manufacturing objects can be fabricated using high-temperature lithography (also known as "thermal lithography"). High-temperature lithography can include any photopolymerization process that involves heating a photopolymerizable material (e.g., a polymeric resin). For example, high-temperature lithography can include heating the material to a temperature of at least 30°C, 40°C, 50°C, 60°C, 70°C, 80°C, 90°C, 100°C, 110°C, or 120°C. In some embodiments, the material is heated to temperatures in the range of 50°C to 120°C, 90°C to 120°C, 100°C to 120°C, 105°C to 115°C, or 105°C to 110°C. Heating can reduce the viscosity of the photopolymerizable material before and / or during curing, and / or increase the reactivity of the photopolymerizable material. Thus, high-temperature lithography can be used to fabricate objects from materials with high viscosity and / or poor flowability, which, when cured, are less likely to exhibit improved mechanical properties (e.g., stiffness, strength, stability) compared to other types of materials. For example, high-temperature lithography can be used to fabricate objects from materials with a viscosity of at least 5 Pa⁻¹, 10 Pa⁻¹, 15 Pa⁻¹, 20 Pa⁻¹, 30 Pa⁻¹, 40 Pa⁻¹, or 50 Pa⁻¹ at 20 °C. Representative examples of high-temperature lithography processes that may be incorporated herein are described in whole in each of the publications incorporated herein by reference in their international publication numbers WO 2015 / 075094, WO 2016 / 078838, WO 2018 / 032022, WO 2020 / 070639, WO 2021 / 130657, and WO 2021 / 130661.

[0035] In some embodiments, additively manufactured objects are made using continuous liquid-phase production (also referred to as "continuous liquid-phase printing"), wherein the object is continuously constructed from a reservoir of photopolymerizable resin by forming a gradient of partially cured resin between the object's build surface and a polymerization inhibition "dead zone". In some embodiments, a semi-permeable membrane is used to control the delivery of photopolymerization inhibitors (e.g., oxygen) into the dead zone to form a polymerization gradient. Representative examples of continuous liquid-phase production processes that may be incorporated herein are described in the disclosure of each of these documents, the entire contents of which are incorporated herein by reference in U.S. Patent Publications 2015 / 0097315, 2015 / 0097316, and 2015 / 0102532.

[0036] As another example, continuous additive manufacturing methods can achieve continuous construction of object geometry through the continuous movement of a construction platform (e.g., along the vertical or Z-direction) during an irradiation phase, such that the curing depth of the irradiated photopolymer is controlled by the movement speed. Thus, continuous polymerization of the material on the construction surface can be achieved. Such methods are described in their entirety in U.S. Patent No. 7,892,474, which is incorporated herein by reference. In another example, continuous additive manufacturing methods may include: extruding a composite material consisting of a curable liquid material surrounding a solid strand. The composite material can be extruded along a continuous three-dimensional path to form an object. Such methods are described in their entirety in U.S. Patent No. 10,162,624 and U.S. Patent Publication No. 2014 / 0061974, which are incorporated herein by reference. In yet another example, continuous additive manufacturing methods may utilize a “spiral lithography” approach, in which a liquid photopolymer is cured using focused radiation while the construction platform continuously rotates and rises. Thus, object geometry can be continuously constructed along a spiral construction path. Such methods are described in their entirety in U.S. Patent No. 10,162,264 and U.S. Patent Publication No. 2014 / 0265034, which are incorporated herein by reference.

[0037] In another example, additively manufactured objects can be fabricated using a volumetric additive manufacturing (VAM) process, where the entire object is produced from a 3D volume of resin in a single printing step, without layer-by-layer construction. During the VAM process, the entire build volume is irradiated with energy, but the configuration of the projected pattern ensures that only certain voxels accumulate a sufficient dose of energy for curing. Representative examples of VAM processes that can be incorporated into this technology include tomographic volumetric printing, holographic volumetric printing, multiphoton volumetric printing, and xolography. For instance, a tomographic VAM process can be performed by projecting a 2D optical pattern onto a rotating volume of photosensitive material with perpendicular and / or angular incidence to produce a cured 3D structure. A holographic VAM process can be performed by projecting a holographic light pattern onto a fixed reservoir of photosensitive material. A xolography process can initiate local polymerization within a volume of photosensitive material using a photo-switchable photoinitiator after linear excitation by interleaving beams of different wavelengths. Additional details of the VAM process suitable for use with this technology are described in the entire disclosure herein and are incorporated herein by reference in U.S. Patent No. 11,370,173, U.S. Patent Publication No. 2021 / 0146619, U.S. Patent Publication No. 2022 / 0227051, International Publication No. WO 2017 / 115076, International Publication No. WO2020 / 245456, International Publication No. WO 2022 / 011456 and U.S. Provisional Patent Application No. 63 / 181,645.

[0038] In yet another example, additive manufacturing objects can be created using a powder bed fusion process (e.g., selective laser sintering), which involves selectively fusing layers of powder material according to a desired cross-sectional shape using a laser beam to construct the object geometry. As another example, additive manufacturing objects can be created using a material extrusion process (e.g., fused deposition modeling), which involves selectively depositing filaments of material (e.g., thermoplastic polymers) in a layer-by-layer manner to form the object. In yet another example, additive manufacturing objects can be created using a material jetting process, which involves jetting or extruding one or more materials onto a build surface to form continuous layers of the object geometry.

[0039] Additively manufactured objects can be made from any suitable material or combination of materials. As discussed above, in some embodiments, additively manufactured objects are made partly or entirely of polymeric materials, such as curable polymeric resins. The resin may consist of one or more monomeric components initially in a liquid state. The resin may be in a liquid state at room temperature (e.g., 20°C) or at elevated temperatures (e.g., temperatures in the range of 50°C to 120°C). When exposed to energy (e.g., light), the monomeric components may undergo a polymerization reaction, causing the resin to solidify into the desired object geometry. Representative examples of curable polymeric resins and other materials suitable for use with the additive manufacturing techniques described herein are described in the entirety of the disclosures in each of these publications, which are incorporated herein by reference in International Publications WO 2019 / 006409, WO 2020 / 070639 and WO 2021 / 087061.

[0040] Optionally, an additively manufactured object can be made from multiple different materials (e.g., at least two, three, four, five, or more different materials). These materials may differ from each other in composition, curing conditions (e.g., curing energy wavelength), material properties before curing (e.g., viscosity), and material properties after curing (e.g., stiffness, strength, transparency). In some embodiments, the additively manufactured object is formed from many materials in a single manufacturing step. For example, a multi-head extruder can be used to selectively dispense many types of materials from different material supply sources to manufacture an object from multiple different materials. Examples of such methods are described in their entirety in U.S. Patent Nos. 6,749,414 and 11,318,667, which are incorporated herein by reference. Alternatively or in combination, the additively manufactured object can be formed from many materials in multiple successive manufacturing steps. For example, a first portion of the object may be formed from a first material according to any of the manufacturing methods described herein, then a second portion of the object may be formed from a second material according to any of the manufacturing methods described herein, and so on, until the entire object has been formed.

[0041] After an additively manufactured object is created, the object may undergo one or more additional processing steps, also referred to herein as "post-processing". As described in detail below with respect to boxes 104 to 108, post-processing may include: removing residual material from the object; performing post-curing of the object; and / or additional post-processing operations.

[0042] For example, at box 104, method 100 may continue to remove residual material from the object. Excess material may include excess precursor material (e.g., uncured resin) and / or other unwanted material (e.g., debris) remaining on or within the object after the additive manufacturing process. Residual material can be removed in many different ways, such as by exposing the object to a solvent (e.g., via spraying, immersion), heating or cooling the object, applying a vacuum to the object, blowing pressurized gas onto the object, applying mechanical forces to the object (e.g., vibration, agitation, centrifugation, tumbling, brushing), and / or other suitable techniques. Optionally, residual material may be collected and / or processed for reuse.

[0043] At block 106, method 100 may optionally include post-curing the object. Post-curing is an additional curing process that can be used when the object remains in a partially cured, "green" state after fabrication. For example, the energy used to fabricate the object in block 102 may only partially polymerize the precursor material forming the object. Therefore, a post-curing step may be needed to fully cure (e.g., fully polymerize) the object to its final usable state. Post-curing can provide various benefits, such as improved mechanical properties (e.g., stiffness, strength) and / or temperature stability of the object. Post-curing can be performed by heating the object, applying radiation (e.g., UV, visible light, microwaves), or a suitable combination thereof. However, in other embodiments, the post-curing process at block 106 is optional and may be omitted.

[0044] At box 108, method 100 may include one or more additional post-processing operations, such as removing sacrificial parts that are not intended to be part of the final object (e.g., support structure), cleaning the object (e.g., washing, solvent extraction), annealing the object, separating the object from the build platform, performing surface modifications and / or treatments, and / or packaging the object for transport.

[0045] Figure 1 The illustrated method 100 can be modified in many different ways. For example, although the steps of method 100 described above are for a single object, method 100 can be used to sequentially or concurrently manufacture and post-process any suitable number of objects, such as dozens, hundreds, or thousands of additively manufactured objects. As another example, Figure 1 The order of the processes shown can be changed. Some processes in method 100 can be omitted, and / or method 100 may include... Figure 1 Additional processes not shown in the diagram.

[0046] Figure 2 This is a partial schematic diagram providing a general overview of an additive manufacturing process according to embodiments of the present technology. (As shown...) Figure 2As shown, object 202 is manufactured on build platform 204 by a series of curable material layers, each layer having a geometry corresponding to a respective cross-section of object 202. To manufacture individual object layers, a layer of curable material 206 (e.g., a polymerizable resin) is brought into contact with build platform 204 (when manufacturing the first layer of object 202) or with a previously formed portion of object 202 on build platform 204 (when manufacturing subsequent layers of object 202). In some embodiments, curable material 206 is formed on and supported by a substrate (not shown), such as a film. Energy 208 (e.g., light) from energy source 210 (e.g., a laser, projector, or light engine) is then applied to curable material 206 to form a cured material layer 212 on build platform 204 or object 202. The remaining curable material 206 can then be removed from the build platform 204 (e.g., by lowering the build platform 204, by laterally moving the build platform 204, by raising the curable material 206, and / or by laterally moving the curable material 206), thereby leaving the cured material layer 212 in the appropriate position on the build platform 204 and / or the object 202. This manufacturing process can then be repeated using a new cured material layer 206 to build the next layer of the object 202.

[0047] The illustrated embodiment shows a "top-down" configuration where the energy source 210 is located above the build platform 204 and directs energy 208 downwards to the build platform 204, such that the object 202 is formed on the upper surface of the build platform 204. Therefore, the build platform 204 can be gradually lowered relative to the energy source 210 as successive layers of the object 202 are formed. However, in other embodiments, Figure 2 The additive manufacturing process can be performed using a bottom-up configuration, where the energy source 210 is located below the build platform 204 and directs energy 208 upwards to the build platform 204, so that the object 202 is formed on the lower surface of the build platform 204. Therefore, as continuous layers of the object 202 are formed, the build platform 204 can gradually rise relative to the energy source 210.

[0048] although Figure 2 The illustrations represent representative examples of additive manufacturing processes, but are not intended to be limiting, and the embodiments described herein can be adapted to other types of additive manufacturing systems (e.g., trench-based systems) and / or other types of additive manufacturing processes (e.g., material jetting, binder jetting, material extrusion, powder bed fusion, sheet lamination, directional energy deposition).

[0049] During additive manufacturing and / or post-processing, deviations may occur between the intended and actual geometry of an object. For example, over-building and / or over-curing may occur when the precursor material used to form the object cures to a greater extent than intended in the horizontal (e.g., x and y) and / or vertical (e.g., z) dimensions, resulting in the actual geometry (e.g., area and / or height) of the cured object portion being larger than the intended geometry. As another example, excess material not completely removed from the object may be incorporated into the object during post-processing, altering the object's geometry. After fabrication, highly viscous materials (e.g., resin) may be difficult to completely remove from the object, and / or material may be trapped on or within the object based on local geometry (e.g., material may accumulate at object portions with high curvature, such as holes, corners, recesses, cavities, etc.). As yet another example, the object may deform due to mechanical stress, temperature, solvents, and / or other conditions during additive manufacturing and / or post-processing. In some instances, large centrifugal forces are used to remove highly viscous resin from an object, which can lead to expansion, warping, and / or other deformations, particularly in relatively thin sections of the object. As another example, during additive manufacturing and / or post-processing, material may be lost from one or more parts of the object due to solvents, mechanical abrasion, and / or other conditions.

[0050] Deviations between the intended and actual geometry of an object can impair its function and properties. For example, certain types of dental appliances may have small and / or detailed features with tight manufacturing tolerances. Areas of dental appliances that are important or necessary for certain functions (e.g., clinical efficacy, proper positioning, ergonomics, mechanical properties, aesthetics) may also be subject to tight tolerances. For example, tolerances for certain features and / or areas of a dental appliance may be less than or equal to 500 μm, 200 μm, 100 μm, 50 μm, 20 μm, or 10 μm. If the actual size, shape, and / or location of a feature and / or area deviates significantly from the intended size, shape, and / or location (e.g., deviations exceeding tolerances), the appliance may not be suitable for its intended function; for example, the appliance may not fit properly onto the teeth and / or may not be able to apply the correct forces to the teeth.

[0051] To address these and other challenges, this technology provides a method for predicting the geometry of an object after additive manufacturing and / or post-processing. For example, the method described herein can be used (e.g., using predictive algorithms such as machine learning algorithms) to predict the geometry of the object to be produced after additive manufacturing and / or post-processing. The predicted geometry can then be compared with a target geometry to determine if there are significant deviations. If a significant deviation is detected, the method can (e.g., using optimization algorithms) modify the instructions sent to the additive manufacturing system so that the actual geometry of the produced object more closely conforms to the target geometry.

[0052] Figure 3 This is a block diagram providing a general overview of a workflow 300 for additive manufacturing of an object according to embodiments of the present technology. Workflow 300 may begin by receiving and / or generating a 3D model (box 302) of an object to be manufactured via an additive manufacturing process. The 3D model can be any digital representation of the target 3D geometry of the object, such as a surface model, mesh model, parametric model, non-parametric model, etc. The 3D model can be provided in any suitable file format, such as CAD files, STL files, OBJ files, AMF files, 3MF files, etc.

[0053] In embodiments where the object is a dental appliance, a 3D model of the dental appliance can be generated using a computing system or device that executes software for designing the appliance according to a treatment plan. The appliance can be designed based on a treatment prescription received from a clinician and data on the patient's teeth received from an intraoral state capture system (box 304). The intraoral state capture system can be configured to acquire sensor data of the patient's dentition, intraoral cavity, and / or other relevant anatomical structures (e.g., craniofacial anatomy). The sensor data can depict the patient's dentition in any suitable arrangement, such as an initial arrangement before the start of treatment planning, an intermediate arrangement after the start of treatment, or a final arrangement after the completion of treatment. The sensor data can be generated via any suitable means and can include photographs, videos, scan data (e.g., intraoral and / or extraoral scans), magnetic resonance imaging (MRI) data, radiographic data (e.g., standard X-ray data, such as bite wing X-ray data, panoramic X-ray data, cephalometric X-ray data, computed tomography (CT) data, cone-beam computed tomography (CBCT) data, fluoroscopy data), and / or motion data. Sensor data may include 2D data (e.g., 2D photographs or videos), 3D data (e.g., 3D photographs, intraoral and / or extraoral scans, digital models), 4D data (e.g., fluorescence fluoroscopy data, dynamic joint data, hard and / or soft tissue motion capture data), or suitable combinations thereof.

[0054] For example, in some embodiments, the intraoral state capture system includes or is operatively coupled to a scanner configured to acquire a 3D digital representation (e.g., image, surface topography data) of a patient's teeth, for example, via direct intraoral scanning or indirectly via castings, impressions, models, etc. The scanner may include a probe (e.g., a handheld probe) for optically capturing 3D structures (e.g., via confocal focusing of a beam array). Examples of scanners include, but are not limited to, the iTero® intraoral digital scanner manufactured by AlignTechnology, Inc., the 3M True Definition Scanner, and the Cerec Omnicam manufactured by Sirona®.

[0055] In other embodiments (e.g., if the object to be made is not a dental appliance), the intraoral state capture system is optional and can be omitted from workflow 300.

[0056] A 3D model of an object can be used to generate multiple images (box 306). These images may be referred to herein as “slices,” and the process of generating slices from a 3D model may be referred to herein as “slice processing.” Images may represent multiple cross-sections (e.g., layers) of an object to be manufactured via a layer-by-layer additive manufacturing process. Layer-by-layer additive manufacturing processes may include any of the techniques described herein. For example, a layer-by-layer additive manufacturing process may include: using energy to cure resin in a layer-by-layer manner to form an object, such as DLP or SLA. As another example, a layer-by-layer additive manufacturing process may include: using energy to melt powder in a layer-by-layer manner to form an object, such as SLS. Images may be any 2D digital representation suitable for generating instructions for applying energy to a precursor material (e.g., resin or powder) to create an object according to the layer-by-layer additive manufacturing process control. For example, each pixel of an image may represent a corresponding voxel of a corresponding cross-section of the object, where the pixel value indicates whether the corresponding voxel is part of the object and therefore energy should be applied to solidify the precursor material at that voxel, or whether the corresponding voxel is an empty space and therefore the precursor material should remain unsolidified at that voxel. Images can be provided in any suitable file format, such as BMP or PNG.

[0057] In some embodiments, the slicing process includes determining the positions of a plurality of slicing planes along a 3D model. For example, the slicing planes may be spaced apart from each other at multiple different vertical positions (e.g., z-positions) along the 3D model. An image can then be generated based on the 2D cross-sectional geometry of the 3D model at each slicing plane. The slicing process may be based on specific parameters of the additive manufacturing system (box 314) to be used to fabricate the object. For example, the spacing between the slicing planes (e.g., slice height or thickness) may be at least the minimum layer height of the additive manufacturing system.

[0058] Workflow 300 may include generating a prediction of the object's geometry after manufacturing, based on some or all of the images (box 308). As indicated by the dashed arrow, the prediction may be a digital representation of the object's expected geometry after fabrication by an additive manufacturing system (box 314) and / or after processing by a post-processing system (box 316), thus taking into account the specific process types, parameters, conditions, etc., associated with these systems. For example, the prediction may be or include one or more images representing the predicted geometry of each cross-section (e.g., layer) of the object. Alternatively or in combination, the prediction may be or include a 3D model representing the overall 3D geometry of the object, which may be generated by combining images or may be generated independently of the images.

[0059] In some embodiments, a software prediction algorithm is used to generate a prediction of the object geometry. This algorithm receives an input dataset of one or more images from an image including box 306 and produces an output dataset comprising digital representations of the predicted geometry of one or more corresponding object cross-sections. The output dataset may include, for example, one or more images depicting the predicted geometry of multiple cross-sections of the object, a 3D model depicting the predicted overall geometry of the object, or a combination thereof. In some embodiments, the prediction algorithm is or includes a machine learning algorithm (e.g., a convolutional neural network) trained to generate a prediction of the object geometry after manufacturing based on input images corresponding to instructions provided to the additive manufacturing system for fabricating the object.

[0060] The prediction algorithm can be customized for a specific additive manufacturing process implemented by the additive manufacturing system (box 314) and / or a specific post-processing operation implemented by the post-processing system (box 316). For example, the prediction algorithm can take into account some or all of the following conditions and / or parameters associated with additive manufacturing and / or post-processing: the type of additive manufacturing process (e.g., DLP, SLA, SLS), the printing parameters of the additive manufacturing system (e.g., curing time, gray level, printing speed, light intensity, minimum feature size, minimum layer height, printing resolution, printing cell shape, printing orientation, printing offset, expected over-curing and / or over-build amount), the properties of the precursor material used to create the object (e.g., viscosity, optical properties, transmittance, light scattering), and / or post-processing conditions (e.g., conditions resulting from centrifugation, washing, post-curing, etc., such as temperature, applied force, exposure to solvents and / or other chemicals). The prediction algorithm can be customized through the design of the functions used by the prediction algorithm and / or the training of the prediction algorithm. Examples below are related to... Figures 4 to 8 The description includes additional details and examples of the prediction algorithms that can be used.

[0061] Subsequently, the predicted geometry can be compared with the target geometry to determine the accuracy of the prediction manufacturing, such as whether there are any deviations between the predicted and target geometries. For example, in embodiments where the prediction includes one or more images depicting the predicted geometry (“predicted images” or “predicted slices”), the predicted images can be compared with one or more images generated from the 3D model in box 306 (“initial images” or “initial slices”) to identify the location and extent of the deviation. As described herein, deviations between the predicted and target geometries can be attributed to conditions during additive manufacturing and / or post-processing, such as over-build, over-curing, excess material on the object, deformation, material loss, etc.

[0062] If the identified deviation is large enough and / or occurs in a significant part of the object (e.g., a part of a dental appliance that applies force to teeth), optimization algorithms can be used to modify some or all of the images to reduce, prevent, or otherwise mitigate the deviation (box 310). This modification can be configured to improve manufacturing accuracy by reducing the deviation between the target geometry of the object and the predicted geometry of the object after manufacturing. For example, the modification can change the size, shape, and / or position of some or all of the object cross-sections represented in the image, such that an object manufactured based on the modified image is more similar to the target geometry than an object manufactured based on the initial image. Examples below are related to... Figures 4 to 8 The description includes additional details and examples of the optimization algorithms that can be used.

[0063] The output of the optimization algorithm can be one or more modified images (box 306), which are then fed into a prediction algorithm (box 308) to generate an updated prediction of the object geometry after manufacturing. The updated prediction can then be analyzed to determine if the predicted manufacturing accuracy is acceptable, for example, whether deviations between the predicted geometry and the target geometry still exist. For example, one or more predicted images representing the updated prediction can be compared with one or more initial images to identify the location and extent of the deviations. If significant deviations still exist, the optimization routine can be executed again to further modify the images to reduce, prevent, or otherwise mitigate the deviations (box 310). This process can be repeated to iteratively modify the images until the predicted manufacturing accuracy is satisfactory (e.g., the predicted deviations are significantly small and / or do not appear in significant parts of the object).

[0064] The image generated by the optimization routine can be used to generate fabrication instructions (box 312) for controlling the additive manufacturing system to form an object (box 314). As described herein, the additive manufacturing system can be configured to apply energy to a precursor material (e.g., resin or powder) to solidify, polymerize, melt, sinter, fuse, or otherwise solidify the precursor material into an individual cross-section (e.g., layer) of the object. The energy can be applied based on data in a corresponding image of the cross-section. For example, the pixel value at a specific location in the image can indicate whether energy should be applied to a corresponding voxel in the precursor material to form a part of the object, and optionally, parameters of the energy to be applied to that location (e.g., intensity, exposure time, dose, wavelength). In some embodiments, the image is a black and white image, where white pixels indicate that energy should be applied and black pixels indicate that energy should not be applied, or vice versa. In other embodiments, the image can be a grayscale image, where the grayscale value of a pixel corresponds to the desired dose of energy to be applied (e.g., intensity and / or exposure time). For example, if the object is intended to have heterogeneous properties (e.g., different degrees of curing may result in variations in properties such as modulus, glass transition temperature), a grayscale image can be used. The fabrication instructions can be any data type that the additive manufacturing system can use to fabricate the object. For example, fabrication instructions may include images, and / or may include other data generated based on images, such as toolpath files (e.g., G-code files).

[0065] Fabrication instructions can be transmitted to an additive manufacturing system to fabricate an object (box 314). The additive manufacturing system may include an energy source (e.g., a laser, projector, light engine), a precursor material source (e.g., a tank, carrier film, powder bed), and / or other equipment configured to perform the various additive manufacturing processes described herein. Optionally, after fabrication, the object may be transferred to a post-processing system for post-processing (box 316). For example, the post-processing system may include one or more centrifuges, solvent baths, post-curing and / or annealing furnaces, trimming systems, and / or other equipment configured to perform the various post-processing operations described herein.

[0066] The various elements of workflow 300 can be implemented using any suitable combination of hardware and software components. For example, the intraoral state capture system (box 304), the additive manufacturing system (box 314), and the post-processing system (box 316) can each include a computing system or device (e.g., a controller) having one or more processors and memory configured to control the respective hardware components (e.g., controlling the scanner, printer, or post-processing device, respectively) to perform the various operations described herein. 3D models (box 302), images (box 306), prediction algorithms (box 308), optimization algorithms (box 310), and fabrication instructions (box 312) can be generated and / or implemented by software components of one or more computing systems (e.g., a treatment planning system and / or an instrument design system). Some or all of the systems in workflow 300 can be implemented as a distributed “cloud” server across any suitable combination of hardware and / or virtual computing resources. The various systems in Workflow 300 can communicate with each other via one or more communication networks, such as wired networks, wireless networks, metropolitan area networks (MANs), local area networks (LANs), wide area networks (WANs), virtual local area networks (VLANs), the Internet, extranets, intranets, and / or any other suitable type of network or a combination thereof.

[0067] Figure 3 The configuration of the illustrated workflow 300 can vary in many ways. For example, in Figure 3 Any component of the workflow 300, shown as different components, may be combined and / or include related code. Any component of the workflow 300 may be implemented as a single software fragment and / or related software fragments, or different software fragments. Any component of the workflow 300 may be embodied on a single machine or any combination of multiple machines. Some components of the workflow 300 may be omitted (e.g., intraoral state capture system, post-processing system), and / or the workflow 300 may include... Figure 3 Additional components not shown.

[0068] Figure 4 This is a flowchart illustrating a method 400 for generating instructions for additive manufacturing of an object according to an embodiment of the present technology. Method 400 can be used to generate manufacturing instructions for any object (such as one or more dental appliances) described herein. In some embodiments, some or all of the processes in method 400 are implemented as computer-readable instructions (e.g., program code) configured to be executed by one or more processors of a computing device (e.g., an appliance design system). Method 400 can be combined with any of the other methods described herein. For example, method 400 can be used as… Figure 3It is part of the 300 workflow execution.

[0069] Method 400 may begin at block 402: receiving at least one image representing a target geometry of an object to be manufactured using an additive manufacturing process. The image may include one or more slices representing multiple cross-sections of the object to be manufactured via a layer-by-layer additive manufacturing process (e.g., DLPA, SLA, SLS, inkjet) as described herein. In some embodiments, the image corresponds to manufacturing instructions for controlling the application of energy to a precursor material (e.g., resin or powder) to manufacture the object via the layer-by-layer additive manufacturing process. For example, pixels within the image may indicate whether energy should be applied to a corresponding location in the precursor material to form a portion of the object, and optionally, parameters of the energy to be applied to that location (e.g., intensity, exposure time, dose, wavelength) may indicate this energy. The image may be a black-and-white or grayscale image and may be provided in any suitable file format (e.g., BMP file, PNG file).

[0070] At box 404, method 400 may include: determining a predicted geometry of the object after it has been fabricated using an additive manufacturing process. The predicted geometry may be generated using a prediction algorithm configured to receive at least one image representing the target geometry as input and generate a digital representation of the predicted geometry of the object as output. For example, the digital representation may include one or more images of the predicted geometry, a 3D model of the predicted geometry, or a suitable combination thereof.

[0071] In some embodiments, the prediction algorithm is or includes at least one machine learning algorithm, such as at least one of the following: regression algorithms (e.g., ordinary least squares regression, linear regression, logistic regression, stepwise regression, multivariate adaptive regression splines, local estimation scatter plot smoothing), instance-based algorithms (e.g., k-nearest neighbors, learned vector quantization, self-organizing graph, locally weighted learning), regularization algorithms (e.g., ridge regression, minimum absolute shrinkage and selection operator, elastic net, minimum angle regression), decision tree algorithms (e.g., iterative binary divider). 3 (ID3), C4.5, C5.0, Classification and Regression Trees, Chi-square Automatic Interaction Detection, Decision Stubs, M5), Bayesian Algorithms (e.g., Naive Bayes, Gaussian Naive Bayes, Multinomial Naive Bayes, Average Single Correlation Estimator, Bayesian Belief Network, Bayesian Network, Hidden Markov Model, Conditional Random Field), Clustering Algorithms (e.g., k-means, Single Link Clustering, k-median, Expectation-Maximization, Hierarchical Clustering, Fuzzy Clustering, Density-Based Noise Applied Spatial Clustering (DBSCAN), Ranking for Identifying Cluster Structures The algorithms used include: Point (OPTICS), Non-negative Matrix Factorization (NMF), Latent Dirichlet Allocation (LDA), Gaussian Mixture Model (GMM), Association Rule Learning Algorithms (e.g., Prior Algorithms, Eclat Algorithms, Frequent Pattern (FP) Growth), Artificial Neural Network Algorithms (e.g., Perceptron, Neural Networks, Backpropagation, Hopfield Networks, Autoencoders, Boltzmann Machines, Restricted Boltzmann Machines, Spiral Neural Networks, Radial Basis Function Networks), Deep Learning Algorithms (e.g., Deep Boltzmann Machines, Deep Belief Networks, Convolutional Neural Networks, Stacked Autoencoders), Dimensionality Reduction Algorithms (e.g., Principal Component Analysis (PCA), Independent Component Analysis (ICA), Principal Component Regression (PCR), Partial Least Squares Regression (PLSR), Salmon Mapping, Multidimensional Scaling, Projective Pursuit, Linear Discriminant Analysis, Mixture Discriminant Analysis, Quadratic Discriminant Analysis, Flexible Discriminant Analysis), Ensemble Algorithms (e.g., Augmentation, Bootstrap Aggregation, AdaBoost, Hybrid, Gradient Boosting Machines, Gradient Boosting Regression Trees, Random Forests), or suitable combinations thereof.

[0072] Machine learning algorithms can be trained to predict the geometry of an object after it has been manufactured. For example, a machine learning algorithm can be trained to predict the geometry of an object after it has been manufactured via an additive manufacturing process (e.g., SLA, DLP, SLS) based on instructions generated from at least one image representing the geometry of the target object. Optionally, the machine learning algorithm can also be trained to predict the geometry of an object after it has undergone one or more post-processing operations (such as material removal (e.g., centrifugation), post-curing, cleaning, etc.). In some embodiments, the machine learning algorithm is trained to predict deviations from the target geometry of the object due to over-curing of the material used to manufacture the object, over-construction of the material used to manufacture the object, retention of material on the surface of the object, loss of material from the object, deformation of the object, or combinations thereof.

[0073] Training data for machine learning algorithms can include data from other objects manufactured using the same additive manufacturing process and / or that have undergone the same post-processing operations. For example, in embodiments where the object is a dental appliance, the training data can include data from other dental appliances, which may include dental appliances of the same type as the dental appliance, dental appliances of a different type than the dental appliance, or suitable combinations thereof. Alternatively or in combination, training data for machine learning algorithms can include data from reference objects manufactured using the same additive manufacturing process and / or that have undergone the same post-processing operations. The reference object can be a specimen with standardized shape and geometry (e.g., a block, rod, cylinder, etc., with known dimensions and shape).

[0074] Training data may include any suitable number of objects (such as at least 5, 10, 20, 50, 100, 500, or 1000 objects) and / or no more than 1000, 500, 100, 50, 20, 10, or 5 objects. For each object, the training data may include a first digital representation of the object's target geometry (e.g., a first set of images representing the target geometry) and a second digital representation of the object's actual geometry after additive manufacturing and / or post-processing (e.g., a second set of images representing the actual geometry). Training may be performed using any suitable method (such as supervised learning, unsupervised learning, or reinforcement learning).

[0075] Figure 5 This is a block diagram illustrating a representative example of a prediction algorithm 502 for predicting manufacturing outcomes according to an embodiment of the present technology. Prediction algorithm 502 can be used in the process of block 404 to determine the predicted geometry of an object after additive manufacturing and / or post-processing. Figure 5In this embodiment, prediction algorithm 502 is or includes a convolutional neural network (CNN) trained to perform prediction. A CNN is a machine learning algorithm that can be used to process images and / or other array-like data structures. A CNN consists of multiple layers, each of which includes one or more neurons to which the operations described herein are applied. A CNN can transform input data (e.g., data received at an input layer) into output data (e.g., data output by an output layer) through a network architecture that includes multiple intermediate layers. In some embodiments, the multiple intermediate layers include one or more convolutional layers. Each convolutional layer of a CNN can apply at least one filter (also referred to as a “kernel”) to the input data from the previous layer via a convolution operation. In some embodiments, the kernel includes one or more functions, as further discussed herein. The parameters of the kernel (e.g., kernel size, weights, biases, parameters of one or more kernel functions) can be learned (e.g., using backpropagation) from training data. A CNN may optionally include multiple convolutional layers, wherein the input data of each convolutional layer includes output data from the previous layer (e.g., another convolutional layer or another type of layer). In some embodiments, a CNN may include one or more additional layers in addition to one or more convolutional layers, such as at least one pooling layer and / or at least one fully connected layer. Furthermore, a CNN may include any layer arrangement that forms a custom network architecture. The predictions produced by a CNN may include output data determined from the convolutional layers or any other layers of the CNN.

[0076] In some embodiments, the CNN includes a kernel designed based on physical and / or chemical phenomena associated with additive manufacturing processes and / or post-processing operations. Specifically, the kernel may include at least one function representing physical and / or chemical phenomena associated with additive manufacturing processes and / or post-processing operations. For example, in embodiments where the additive manufacturing process includes applying light energy to cure precursor materials (e.g., photopolymerization processes such as SLA, DLP, inkjet, etc.), the kernel may include one or more functions representing one or more of the following phenomena: light absorption through the precursor material (e.g., Beer-Lambert Law), light scattering within the precursor material (e.g., scattering modeled by a Gaussian distribution), curing kinetics, diffusion, and / or surface tension. As another example, in embodiments where the additive manufacturing process includes applying energy to melt a precursor material (e.g., powder bed melting processes such as SLS), the kernel may include one or more functions representing one or more of the following phenomena: the thermal conductivity of the powder material, (e.g., modeled via Fourier's laws) heat transfer and / or distribution within the powder bed, phase transitions of the powder material (e.g., melting and / or consolidation kinetics), laser-material interactions (e.g., absorption and / or reflection properties of the powder material), particle size distribution, particle packing density effects, residual stress formation and / or relaxation, and / or (e.g., modeled via a Gaussian distribution or other suitable spatial distribution to reflect the focusing and energy dispersion properties of the laser) the size and intensity distribution of the laser spot. These functions may take any suitable form (e.g., depending on the type of physical and / or chemical phenomenon being modeled), such as Gaussian functions, exponential functions, polynomial functions, etc. In embodiments where the kernel includes multiple functions, the functions may be combined via operations such as summation, multiplication, etc. The parameters of the functions may be determined by training a CNN, as described in detail below. In some embodiments, kernels designed based on physical and / or chemical phenomena reduce the amount of training data required for a CNN to generate accurate predictions, as described below.

[0077] For example, in embodiments where the additive manufacturing system includes forming an object by applying light energy to a curable material (e.g., resin), the kernel may include a composite equation representing the light intensity at each pixel within the curable material. Specifically, a composite equation representing the sum of three modified Gaussian beams can be used to represent the light intensity at each pixel in the xy plane: in and These represent the beam widths in the x and y directions, respectively. and These are the coordinates of the pixel center; and Represents peak intensity.

[0078] The kernel can be expressed using the Beer-Lambert law equation to represent the light intensity in the z-direction: in This represents the strength at the surface of the curable material, and μ is the permeability constant.

[0079] Various parameters of the functions in equations (1) to (5) can be determined through training. These parameters include, but are not limited to, those of the functions in equations (1) to (5). and (which defines the light range in the x and y directions), μ (which defines the light attenuation in the z direction), and / or the kernel size (x, y, z), as further described below.

[0080] The prediction algorithm 502 can be configured to receive an input tensor 504 (e.g., a 3D tensor) generated from an input dataset 506, which includes one or more images representing the target geometry of an object. For example, the images can be multiple slices corresponding to multiple cross-sections of the object, where pixels in each image correspond to voxels in the corresponding object cross-section. Pixel values ​​can indicate whether the corresponding voxels are intended to be solidified (e.g., cured, sintered) within the target object geometry, and optionally, indicate the desired degree of solidification (e.g., degree of curing). Images can be converted into input tensor 504 by transforming each image into a numerical grid representing image pixels, and then layering or stacking the grids to create a multidimensional structure (3D tensor). Other image preprocessing that can be performed includes: cropping the image (e.g., removing parts that are irrelevant or less relevant to the prediction, such as parts corresponding to blank spaces), resizing the image (e.g., adjusting to a predetermined image length and width), adjusting the image color (e.g., converting to black and white or grayscale), and suitable combinations thereof.

[0081] Based on input tensor 504, prediction algorithm 502 can generate output tensor 508 representing the manufacturing result (e.g., predicted geometry) of the object after additive manufacturing and / or post-processing. Output tensor 508 can be the direct output of the convolution performed by prediction algorithm 502 and can represent the comprehensive prediction data generated by prediction algorithm 502. For example, output tensor 508 can represent the predicted consolidation of precursor material after additive manufacturing and / or post-processing, as described herein, and / or it can indicate the predicted cumulative light exposure received at each voxel location. Optionally, prediction algorithm 502 can include the effects of islands (e.g., portions of the object not connected to any other part of the object or otherwise unsupported) that are reduced or eliminated as part of the computational graph, such that the object geometry represented by output tensor 508 includes few or no islands.

[0082] In some embodiments, the output tensor 508 is generated by performing convolutions (e.g., 3D convolutions) on the input tensor using a trained CNN. For example, the output tensor can be generated using the following convolution equation: in input The input tensor is 504. out The output tensor is 508. kernel It is the kernel of CNN. k This represents the position index within the tensor. To improve computational efficiency, convolution can be performed using computing systems or devices that include graphics processing units (GPUs), tensor processing units (TPUs), or suitable combinations thereof.

[0083] In some embodiments, the output tensor 508 is then converted into an output dataset 510 providing a digital representation of the predicted geometry of the object. The output dataset 510 can be any suitable file format, for example, for enhancing visualization and / or potential subsequent processing. For example, the output dataset 510 can include multiple predicted images corresponding to multiple predicted cross-sections of the object, where pixels in each image represent the prediction of the corresponding voxel in the corresponding object cross-section. Pixel values ​​can indicate whether the corresponding voxel is fixed in the predicted object geometry and, optionally, the degree of prediction fixation. Pixel values ​​can be binary values ​​(e.g., black or white) or can be grayscale values. Alternatively or in combination, the output dataset 510 can be in the VTK (Visualization Toolkit) XML Image Data (VTI) format, which may be useful for detailed 3D visualization and can allow for a more detailed examination of the predicted object geometry. Optionally, the output dataset 510 can include a 3D digital model (e.g., in a 3D file format such as STL). The flexibility of the output dataset 510 format not only broadens the applicability of the prediction algorithm 502, but also facilitates seamless integration into all stages of the manufacturing and post-processing workflow.

[0084] In some embodiments, the output tensor 508 and / or output dataset 510 generated by prediction algorithm 502 undergo post-processing (e.g., in an optimization routine, as described elsewhere herein). For example, in embodiments where output tensor 508 includes multiple scalar values ​​representing the cumulative optical dose derived from the convolution result, a threshold may be applied to output tensor 508 to filter the result. The threshold may be determined based on experimental data. For example, Figures 6A to 6C The illustration shows the determination of an experimental calibration threshold for filtering prediction results according to an embodiment of the present technology. Figure 6AThis refers to a cross-sectional view of the target geometry 602 of an object (e.g., a calibration specimen) and the actual geometry 604 of the object after fabrication. For example, the target geometry 602 may correspond to the object's CAD design file (e.g., an STL file), and the actual geometry 604 may correspond to the scan data of the fabricated object. Figure 6A As shown, the actual geometry 604 may deviate from the target geometry 602. Figure 6B This is an image representing a cross-sectional view of the predicted geometry 606 of the object. As described herein, prediction algorithms (e.g., Figure 5 The prediction algorithm 502 generates a predicted geometry 606 from the target geometry 602. The predicted geometry 606 can be compared with the actual geometry 604 of the object being produced to determine a threshold that, when applied to an image, (e.g., by filtering out pixels in the predicted geometry 606 that deviate from the actual geometry 604) makes the predicted geometry 606 match the actual geometry 604. Figure 6C It is an image representing a cross-sectional view of the predicted geometry 608 after the threshold has been applied.

[0085] Refer again Figure 5 Training dataset 512 can be used to train prediction algorithm 502 to generate predictions of manufacturing results. Training dataset 512 may include image data of other objects previously manufactured, for example, using the same or similar additive manufacturing process and / or the same or similar post-processing operations. For example, in an embodiment where the object to be manufactured is a dental appliance, training dataset 512 may include data of previously manufactured dental appliances, which may be dental appliances of the same type as the dental appliance to be manufactured, dental appliances of a different type than the dental appliance to be manufactured, or suitable combinations thereof. Alternatively or in combination, training dataset 512 may include image data of reference objects previously manufactured, for example, using the same or similar additive manufacturing process and / or the same or similar post-processing operations. Reference objects may be specimens with standardized shapes and geometries (e.g., blocks, rods, cylinders, etc. with known sizes and shapes). In some embodiments, because prediction algorithm 502 is designed based on physical and / or chemical phenomena included in additive manufacturing processes and / or post-processing operations, training dataset 512 may be relatively small compared to conventional datasets used to train CNNs. For example, training dataset 512 may include image data of no more than 50, 25, 20, 15, 10, or 5 objects. Alternatively or in combination, a larger amount of training data may be used; for example, training dataset 512 may include image data of at least 50, 100, 200, 500, or 1000 objects.

[0086] In some embodiments, for each object, training dataset 512 includes one or more first images representing the target geometry of the object and one or more second images representing the actual geometry of the object after additive manufacturing and / or post-processing. The one or more first images may be slices representing cross-sections of the object (e.g., slices generated from the object's CAD file), which are used to generate instructions for fabricating the object in a layer-by-layer additive manufacturing process. The one or more second images may be photographs, scan data (e.g., micro-CT data), and / or other data providing a digital representation of the actual cross-sectional geometry of the object after fabrication. In some embodiments, the training process includes: providing the first images to prediction algorithm 502 as input dataset 506; using prediction algorithm 502 to generate an output dataset 510 including one or more prediction images; and then comparing the prediction images with the second images to determine whether the predicted geometry matches the actual geometry. This comparison can be used (e.g., using loss functions such as mean squared error (MSE), mean absolute error (MAE), binary cross-entropy loss, classification cross-entropy loss, DICE loss, structural similarity index (SSIM), Huber loss, L1 loss, L2 loss, etc., or suitable combinations thereof)) to calculate the error between the predicted geometry and the actual geometry. This error can be used to update the parameters of prediction algorithm 502. For example, backpropagation can be used to update the parameters of a CNN (e.g., corresponding to the parameters of one or more kernel functions of the CNN) based on the calculated error. Training can be performed until prediction algorithm 502 achieves sufficient accuracy. Optionally, a portion of the training dataset 512 can be retained to validate the accuracy of the trained prediction algorithm 502.

[0087] The prediction algorithm 502 can be customized for specific additive manufacturing processes and / or post-processing operations based on the training dataset 512 used. For example, the prediction algorithm 502 can be trained to consider some or all of the following conditions and / or parameters associated with additive manufacturing and / or post-processing: the type of additive manufacturing process (e.g., DLP, SLA, SLS), printing parameters of the additive manufacturing system (e.g., curing time, gray level, printing speed, light intensity, minimum feature size, minimum layer height, printing resolution, printing cell shape, printing orientation, printing offset, expected over-curing and / or over-build amount), properties of the precursor material used to create the object (e.g., viscosity, optical properties, transmittance, light scattering), and / or post-processing conditions (e.g., conditions resulting from centrifugation, washing, post-curing, etc., such as temperature, applied force, exposure to solvents and / or other chemicals, etc.). In such embodiments, the training dataset 512 may include data on one or more objects created using the specific conditions and / or parameters to be considered by the prediction algorithm 502.

[0088] In some embodiments, prediction algorithm 502 uses a divide-and-conquer approach to determine the predicted object geometry. A divide-and-conquer approach may be advantageous (e.g., using parallel processing) in reducing processing requirements and / or increasing processing speed. For example, the object geometry may be divided into multiple smaller parts, and prediction algorithm 502 may (e.g., sequentially or in parallel) generate a corresponding prediction for each smaller part, and may combine the predictions to produce a single prediction of the entire object geometry. In embodiments where prediction algorithm 502 is or includes a CNN, the divide-and-conquer approach may include: generating multiple input tensors (e.g., based on images of multiple smaller parts of the object); performing convolutions on the multiple input tensors using a CNN to generate multiple corresponding output tensors; and then combining the output tensors to produce a single output tensor representing the predicted geometry of the entire object. The object can be divided into smaller parts in any suitable manner; for example, the object can be divided into smaller parts of the same or different sizes, wherein the size of each part is determined based on the type of the object, the relative importance of the part, the expected amount of deviation at the part, the geometric complexity of the part, processing constraints, and / or other relevant considerations. In some embodiments, it may not be necessary to predict certain parts of the object, for example, if those parts are not functionally important to the object, if those parts correspond to empty space in or around the object, or if no significant deviation is expected at those parts. However, in other embodiments, prediction algorithm 502 may generate a prediction of the entire object geometry immediately, rather than using a divide-and-conquer approach.

[0089] Figure 7This is a schematic illustration of a divide-and-conquer approach that can be used by prediction algorithm 502 according to an embodiment of the present technology. The divide-and-conquer approach can be applied to input tensor 702 (e.g., representing the target geometry of an object). Input tensor 702 can be divided into multiple smaller portions 704a to 704i, some or all of which may include overlapping regions. In the illustrated embodiment, for example, portion 704a includes regions overlapping with adjacent portions 704b, 704e, and 704d; portion 704b includes regions overlapping with adjacent portions 704a, 704c, 704d, 704e, and 704f; smaller portions 704a to 704i can be input into the prediction algorithm respectively to generate a corresponding prediction for each portion. Subsequently, overlapping regions can be removed from the predictions of each smaller portion 704a to 704i, and the “cropped” predictions can be combined with each other to generate a single output tensor 706 (e.g., representing the predicted geometry of an object). Using a divide-and-conquer approach with overlapping regions in the partitioned portions can offer various advantages, such as mitigating potential problems related to kernel size and / or edge effects that may occur in CNN processing, resulting in more accurate and consistent predictions. This approach can be used to replicate in a single instance the results that might be obtained using CNNs across the entire object, thus maintaining the integrity and accuracy of predictions across the entire geometry of the object.

[0090] Refer again Figure 5 Although this embodiment has been described relative to CNN-based prediction algorithm 502, in other embodiments, prediction algorithm 502 may be alternatively or additionally modified to utilize other types of machine learning algorithms, such as recurrent neural networks (RNN), generative adversarial networks (GAN), capsule networks (CapsNet), graph neural networks (GNN), autoencoders, visual transformers (ViT), other types of artificial neural networks (ANN), or any other type of machine learning algorithm described herein.

[0091] Refer again Figure 4 At box 406, method 400 may include: identifying the deviation between the target geometry and the predicted geometry. This can be achieved by comparing an initial image of the target geometry received at box 402 with (e.g., via...) Figure 5The prediction algorithm 502 identifies biases by generating a predicted image at box 404. Specifically, each initial image can be compared with its corresponding predicted image to determine the location where the object geometry shown in the initial image differs from the object geometry shown in the predicted image, and optionally, the magnitude of the difference (e.g., distance) can be determined. In some embodiments, biases are calculated using loss functions such as mean squared error (MSE), mean absolute error (MAE), binary cross-entropy loss, classification cross-entropy loss, DICE loss, structural similarity index (SSIM), Huber loss, L1 loss, L2 loss, etc., or suitable combinations thereof.

[0092] At box 408, the method 400 may continue to modify at least one image based on the identified deviations. In some embodiments, modifications are made only when the deviation is significant (e.g., exceeding a predetermined threshold and / or occurring at a significant portion of the object), while in other embodiments, modifications are made if any deviation is detected, regardless of its significance. The acceptable amount of deviation may be uniform across the entire object, or the acceptable amount of deviation may be different for different parts of the object (e.g., a larger deviation may be acceptable for parts of the object that are less important for the object's proper function).

[0093] Some or all of the initial images of the initial object received at box 402 can be modified to reduce or otherwise mitigate the deviation between the target geometry and the predicted geometry. For example, the modification can change the size, shape, position, and / or grayscale values ​​of some or all of the object cross-sections represented in the initial image, such that an object created based on the modified image is more similar to the target geometry than an object created based on the initial image. In some embodiments, modifications are made at the pixel or voxel level, such that each pixel or voxel (and / or corresponding tensor) in any image can be changed individually.

[0094] Figure 8 This is a block diagram illustrating a representative example of an optimization algorithm 802 for generating modified images according to embodiments of the present technology. The optimization algorithm 802 can be used in the process of block 408 to determine one or more modified images to reduce or mitigate the deviation between the target geometry and the predicted geometry of an object. Figure 8As shown, the input to optimization algorithm 802 includes one or more input images 804 (e.g., an initial image of the target geometry of the object). The input may also include a deviation 806 identified between the target geometry of the object and the predicted geometry of the object when made according to the geometry specified by the input image 804 (e.g., the output of a loss function). The output of optimization algorithm 802 is one or more modified images 808 configured to produce an object having a geometry more similar to the target geometry than the predicted geometry produced by the input image 804. In some embodiments, the input image 804 and the deviation 806 are provided to optimization algorithm 802 as corresponding input tensors, and the modified image 808 generated by optimization algorithm 802 is provided as an output tensor.

[0095] The optimization algorithm 802 can implement an optimization process to adjust the size, shape, position, grayscale values, etc., of the geometry of some or all of the objects depicted in the input image 804 to produce a modified image 808. As indicated above, adjustments can be made to individual pixels or voxels in the input image 804. The optimization process may include one or more optimization functions, such as stochastic gradient descent (SGD), Adam (adaptive moment estimation), RMSprop (root mean square propagation), Adagrad (adaptive gradient algorithm), Adadelta, Nadam (Nesterov accelerated adaptive moment estimation), etc. Optionally, the learning rate of the optimization function can be selected to avoid instability and / or overshoot while maintaining a sufficiently short processing time. The learning rate can determine the size of the step taken by the optimization function toward the minimum of the loss function. An appropriate learning rate can vary, for example, depending on the location of the deviation, the type of object, the type of additive manufacturing process, and / or post-processing operations. In some embodiments, the learning rate has a value of approximately 0.001X, 0.0002X, 0.003X, 0.005X, 0.0075X, 0.01X, 0.02X, 0.03X, 0.05X, 0.075X, 0.1X, 0.2X, 0.3X, 0.4X, or 0.5X.

[0096] In some embodiments, optimization algorithm 802 is an inverse optimization algorithm. The inverse optimization algorithm can modify the input image 804 to generate a modified image 808 by reversing the bias 806 (e.g., changing a positive distance to a negative distance, or vice versa), and then applying the reversed bias to the input image 804.

[0097] For example, Figures 9A to 9C A representative example of a reverse optimization process for modifying the geometry of object 900 according to an embodiment of the present technology is illustrated schematically. Specifically, Figure 9A The illustration shows the target geometry 902 of object 900. Figure 9B The illustration shows the predicted geometry 904 of object 900, and Figure 9C The illustration shows a modified geometry 906 used to create object 900, which can be achieved via a reverse optimization algorithm (e.g., Figure 8 The optimized algorithm (802) is generated. For example... Figure 9A and Figure 9B As shown, the predicted geometry 904 of object 900 deviates at the first portion 908 of object 900 (in Figure 9B The target geometry 902 (described by dashed lines) is at a distance of X1 and deviates by a distance Y1 at the second part 910 of the object 900. The distance X1 can be negative (indicating that the predicted geometry 904 is less than the target geometry 902 at the first part 908), and the distance Y1 can be positive (indicating that the predicted geometry 904 is greater than the target geometry 902 at the second part 910).

[0098] Next reference Figure 9C The inverse optimization algorithm can generate a modified geometry 906 of object 900 by reversing the deviations identified at the first part 908 and the second part 910. For example, the modified geometry 906 can be generated by applying a positive distance X2 to the first part 908 of object 900 (resulting in the modified geometry 906 being larger than the target geometry 902 at the first part 908) and applying a negative distance Y2 to the second part 900 of object 900 (resulting in the modified geometry 906 being smaller than the target geometry 902 at the second part 910). The magnitude of distance X2 can be the same as the magnitude of distance X1, or it can be a multiple of the magnitude of distance X1 (e.g., 0.25X, 0.5X, 0.75X, 1X, 1.25X, 1.5X, 1.75X, or 2X). Similarly, the value of distance Y2 can be the same as the value of distance y1, or it can be a multiple of the value of distance Y1 (e.g., 0.25X, 0.5X, 0.75X, 1X, 1.25X, 1.5X, 1.75X or 2.X).

[0099] Refer again Figure 8 The modifications made by the optimization algorithm 802 to the input image 804 may be subject to one or more constraints. For example, constraints may limit the magnitude of changes made to the object geometry, for example, to ensure that the optimization algorithm 802 converges to a stable result. Other constraints that may be applied include: limiting the minimum feature size in the modified geometry to be greater than the minimum feature size of the additive manufacturing system; avoiding potentially unmanufacturable feature shapes (e.g., islands, overhangs); limiting the type of changes made to important parts of the object (e.g., to avoid interfering with the object's functionality) and / or suitable combinations thereof. For example, the optimization algorithm may penalize the output that generates islands.

[0100] In some embodiments, optimization algorithm 802 uses a divide-and-conquer approach to determine a modified object geometry that may be advantageous (e.g., using parallel processing) in reducing processing requirements and / or increasing processing speed, as discussed herein. For example, the object geometry may be divided into multiple smaller parts, and optimization algorithm 802 may (e.g., sequentially or in parallel) determine a corresponding modification for each smaller part, and the modified parts may be combined to form the modified geometry of the entire object. The object can be divided into smaller parts in any suitable manner; for example, the object may be divided into smaller parts of the same or different sizes, wherein the size of each part is determined based on the type of the object, the relative importance of the part, the amount of deviation at the part, the geometric complexity of the part, processing constraints, and / or other relevant considerations. Optionally, the smaller parts may include overlapping regions, for example, as described above. Figure 7 In conjunction with the description, in some embodiments, it may not be necessary to modify certain parts of the object through the optimization algorithm 802, for example, if those parts are not important to the functionality of the object, if those parts correspond to empty space in or around the object, if no significant deviation is observed at those parts, etc. However, in other embodiments, the prediction algorithm 802 may generate the modified geometry of the entire object immediately rather than using a divide-and-conquer approach.

[0101] The modified image 808 generated by the optimization algorithm 802 can represent an object geometry that, when used as a basis for object fabrication, is expected to produce improved manufacturing accuracy, such that the actual geometry of the resulting object is expected to be the same as or sufficiently similar to the target geometry (e.g., there is no significant deviation between the actual geometry and the target geometry and / or any deviations that exist do not occur at significant parts of the object). In some embodiments, the modified object geometry represented in the modified image 808 differs from the target geometry represented in the input image 804, but an object fabricated based on the modified image 808 is expected to have a geometry closer to the target geometry than an object fabricated based on the input image 804.

[0102] Refer again Figure 4Method 400 can then return to box 404 to determine the predicted geometry of the object based on the modified image generated in box 408. The prediction process can be the same as described above, except that one or more modified images representing the modified object geometry are used as input to the prediction algorithm. Method 400 can then proceed to box 406: identifying any deviations between the target geometry and the predicted geometry based on the modified image, as previously described. If the deviation is significant, method 400 can continue to box 408 to further modify the modified image to reduce or mitigate the deviation. The modification process can be the same as described above, except that the input image to the optimization algorithm is the modified image instead of the initial image of the target geometry. In some embodiments, the processes of boxes 404, 406, and 408 are repeated to iteratively modify the image until the prediction manufacturing accuracy is satisfactory (e.g., the prediction deviation is small enough and / or does not appear in significant parts of the object).

[0103] At block 410, method 400 may include generating instructions for fabricating an object using an additive manufacturing process based on the modified image generated in block 408. These instructions may be configured to control an additive manufacturing system (e.g., SLA, DLP, SLS, inkjet, or hybrid printing system) to apply energy to solidify precursor material according to the object geometry represented in the modified image. For example, the modified image may indicate the location where energy is to be applied (and optionally, parameters of the energy application) to form multiple object cross-sections according to a layer-by-layer additive manufacturing process sequence, as described elsewhere herein.

[0104] In some embodiments, the process at block 410 includes converting the modified image into a format suitable for controlling the additive manufacturing system (e.g., a G-code file). Optionally, some or all of the modified images may undergo image post-processing before and / or during the conversion process. For example, image post-processing may include: removing artifacts (e.g., islands, spikes, holes, and / or other discontinuities) from the modified image, resizing the modified image (e.g., adjusting to a predetermined image length and width), adjusting the color of the modified image (e.g., converting to black and white or grayscale), adjusting the modified image to improve printability (e.g., ensuring a minimum feature thickness greater than a minimum printability thickness, increasing the thickness of the object's underlying layer), or suitable combinations thereof. Image post-processing may also include: adding components to the object geometry, such as support structures (e.g., struts, blocks, crossbars, etc.) to stabilize the object during additive manufacturing and / or post-processing, identifiers for tracking the object (e.g., labels, tags, barcodes, QR codes, etc.), and so on.

[0105] Figure 4The illustrated method 400 can be modified in many different ways. For example, although the steps of method 400 described above are for a single object, method 400 can be used to generate instructions sequentially or concurrently for creating any suitable number of objects (such as dozens, hundreds, or thousands of objects). As another example, Figure 4 The order of the processes shown can be changed. Some processes in method 400 can be omitted, and / or method 400 may include... Figure 4 Additional processes not shown. For example, method 400 may additionally include displaying a graphical representation of the predicted geometry, identified deviations, and / or modified image to a user (e.g., a technician or clinician) via a suitable display device (e.g., a monitor, screen, etc. of a computing system or device). In such embodiments, the user may provide feedback, if appropriate, to approve or adjust the modified image.

[0106] Figure 10 This is a block diagram illustrating a workflow 1000 for generating instructions for additive manufacturing of an object according to an embodiment of the present technology. Workflow 1000 can be combined with any other process described herein (e.g., with...). Figure 3 Workflow 300 and / or Figure 4 This is achieved by combining the 400 methods.

[0107] Workflow 1000 includes an optimization routine 1002 for determining (e.g., optimizing) a set of slices for manufacturing an object (e.g., a dental appliance). Figure 10 As shown, multiple initial slices 1004 are provided to the image preprocessing model 1006. The initial slices 1004 may be multiple images (e.g., black and white images, grayscale images) representing the target geometry of an object. The slices may correspond to multiple cross-sections (e.g., layers) of an object produced via a layer-by-layer additive manufacturing process (e.g., SLA, DLP, SLS, inkjet). In some embodiments, the slices are generated from a 3D digital model (e.g., a CAD model) of the object via a slicing process, as described elsewhere herein.

[0108] An initial slice 1004 may be provided to an image preprocessing model 1006, which includes one or more software algorithms configured to perform image preprocessing on the slice 1004. Image preprocessing may include any of the operations disclosed herein, such as cropping the slice, resizing the slice, adjusting the color of the slice, converting the slice to a 3D tensor, and / or suitable combinations thereof.

[0109] The output of image preprocessing model 1006 can be provided to forward prediction model 1008. Forward prediction model 1008 includes one or more software algorithms configured to generate a predicted manufacturing result 1010 based on initial slices. For example, the predicted manufacturing result 1010 can be a prediction of the object geometry after additive manufacturing based on initial slices and / or post-processing. Forward prediction model 1008 can include, for example, those described herein (e.g., with...). Figure 4 and / or Figure 5 The prediction algorithm described in the box 404 (in conjunction with the prediction algorithm) is any prediction algorithm, such as a machine learning algorithm. In some embodiments, the forward prediction model 1008 is or includes a CNN configured to apply convolution operations (e.g., Equation 6) to the input slices to generate a plurality of prediction slices representing the geometry of the predicted object.

[0110] The forward prediction model 1008 may include one or more experimental calibration parameters 1012 determined based on experimental data (e.g., data from other objects fabricated using the same or similar additive manufacturing processes and / or post-processing operations). For example, experimental data may be used to determine parameters of a CNN kernel (e.g., kernel size in x, y, and / or z; the form of one or more kernel functions; parameters of one or more kernel functions (e.g., in equations (1) through (5)). , And / or μ). As another example, experimental data can be used to determine thresholds to be applied to the convolution results, for example, to remove artifacts, avoid printability issues, etc. In some embodiments, some or all of the experimental calibration parameters in experimental calibration parameters 1012 are determined by training a CNN (or other machine learning algorithm implemented by the forward prediction model 1008), as described elsewhere herein.

[0111] The predicted object geometry (represented in the predicted manufacturing result 1010) can be compared with the target object geometry (represented in the preprocessed slices) to identify any deviations and, if present, whether those deviations are acceptable (box 1014). For example, deviations may be considered unacceptable if they exceed a predetermined threshold (e.g., manufacturing tolerance) and / or if they occur in parts of the object that are important to the object's characteristics and / or function (e.g., the part of a dental appliance that applies force to teeth). In some embodiments, a loss function is used to determine the deviations.

[0112] If the deviation is unacceptable, workflow 1000 can use inverse optimization model 1016 to generate modified slices to reduce, prevent, or otherwise mitigate the deviation. Inverse optimization model 1016 includes components configured to adjust the size, shape, position, grayscale values, etc., of the object geometry depicted in the initial slice to produce, as described herein (e.g., with...). Figure 4, Figure 8 and / or Figures 9A to 9C One or more software algorithms describe the modified slices (in conjunction with box 406). The modified slices can then be fed to the forward prediction model 1008 to generate an updated predicted manufacturing result 1010. The optimization routine 1002 can be repeated to iteratively modify the slices until the deviation between the predicted geometry of the object and the target geometry is acceptable, for example, the deviation is below a predetermined threshold and / or does not occur in significant parts of the object.

[0113] The modified slices generated by optimization routine 1002 can be provided to image post-processing model 1018, which includes one or more software algorithms configured to perform image post-processing on the modified slices. Image post-processing can include any of the operations disclosed herein, such as removing artifacts (e.g., islands, spikes, holes, and / or other discontinuities), resizing the slices, adjusting the color of the slices, adjusting the grayscale values ​​of the slices, modifying the slices to improve printability (e.g., via filtering, thresholding, smoothing, increasing the thickness of the underlying layer of the object), and / or suitable combinations thereof.

[0114] The output of the image post-processing model 1018 can be a plurality of output slices 1022, which can be used to generate instructions for additive manufacturing of an object. For example, the output slices 1022 can be a plurality of images (e.g., black and white images, grayscale images) that indicate how energy should be applied to the precursor material to create a plurality of cross-sections of the object via a layer-by-layer additive manufacturing process.

[0115] Optionally, optimization routine 1002 can be periodically recalibrated to reflect changes in additive manufacturing processes and / or post-processing operations, such as changes in system type, material type, process parameters, etc. Recalibration can be performed alternatively or additionally to improve the results produced by optimization routine 1002 over time, for example, to ensure that output slice 1022 produces object geometries tending towards the middle of the desired tolerance range. Recalibration can be performed by fabricating one or more reference objects with a known target geometry (e.g., a specimen) using a new process, and then measuring the actual geometry of the reference objects after fabrication. The target geometry, the actual geometry, and / or the deviation between the target geometry and the actual geometry can be stored in a recalibration dataset (e.g., a recalibration file), which can then be used to update the parameters of optimization routine 1002 (e.g., the parameters of forward prediction model 1008 and / or backward optimization model 1016).

[0116] The various elements of Workflow 1000 can be implemented using any suitable combination of hardware and software components. For example, some or all of the processes in Workflow 1000 can be implemented using one or more computing systems or devices having one or more processors and memories configured to perform the various operations described herein. The computing system or device may include components configured to improve the efficiency of convolution operations, such as GPUs or TPUs.

[0117] and, Figure 10 The configuration of the illustrated workflow 1000 can vary in many ways. For example, in Figure 10 Any component of the workflow 1000, shown as different components, can be combined and / or include related code. Any component of the workflow 1000 can be implemented as a single software fragment and / or related software fragments, or different software fragments. Any component of the workflow 1000 can be embodied on a single machine or any combination of multiple machines. Although the workflow 1000 is described above with respect to a single object, the workflow 1000 can be used to sequentially or concurrently generate instructions for creating any suitable number of objects (such as tens, hundreds, or thousands of objects). Some components of the workflow 1000 can be omitted (e.g., image preprocessing model 1006 and / or image postprocessing model 1018), and / or the workflow 1000 may include... Figure 10 Additional components not shown. Furthermore, any process in workflow 1000 can optionally be implemented using a divide-and-conquer approach.

[0118] In some embodiments, this technology provides a method for determining instructions for fabricating an object with an actual geometry that more closely conforms to a target geometry, wherein the method can be performed with reduced computation time and / or resources. Executing computationally intensive algorithms may not be well-suited for mass production of a large number of objects with unique geometries. This technology provides an algorithm that can generate a set of optimized object slices and / or fabrication instructions for a single object within a relatively short time period (e.g., no more than 10 minutes, 5 minutes, 2 minutes, 1 minute, or 30 seconds). This algorithm can be a “surrogate algorithm” that receives an image of the object to be fabricated and determines modifications to the image to produce better fabrication accuracy, without performing iterative prediction and optimization workflows to obtain the modified image. Thus, shorter processing and fabrication times can be achieved while still ensuring that the final printed object exhibits high fidelity to the intended design.

[0119] Figure 11This is a block diagram illustrating a workflow 1100 for training a proxy algorithm 1102 according to an embodiment of the present technology. The proxy algorithm 1102 can be configured to receive one or more input images 1104 (or tensors representing input images 1104) representing a target geometry of an object to be manufactured via an additive manufacturing process, and output one or more modified images 1106 (or tensors representing modified images 1106) configured to produce an object having a geometry more similar to the target geometry than a predicted geometry produced from the input images 1104. The input images 1104 and the modified images 1106 can be the same as or substantially similar to other embodiments described herein.

[0120] In some embodiments, the surrogate algorithm 1102 is or includes at least one machine learning algorithm, such as at least one of the following: regression algorithms (e.g., ordinary least squares regression, linear regression, logistic regression, stepwise regression, multivariate adaptive regression spline, local estimation scatter plot smoothing), instance-based algorithms (e.g., k-nearest neighbors, learned vector quantization, self-organizing graph, locally weighted learning), regularization algorithms (e.g., ridge regression, minimum absolute shrinkage and selection operator, elastic net, minimum angle regression), decision tree algorithms (e.g., iterative...). Divider 3 (ID3), C4.5, C5.0, Classification and Regression Trees, Chi-square Automatic Interaction Detection, Decision Stubs, M5), Bayesian algorithms (e.g., Naive Bayes, Gaussian Naive Bayes, Multinomial Naive Bayes, Average Single Correlation Estimator, Bayesian Belief Network, Bayesian Network, Hidden Markov Model, Conditional Random Field), Clustering algorithms (e.g., k-means, Single Linked Clustering, k-median, Expectation Maximization, Hierarchical Clustering, Fuzzy Clustering, Density-Based Noise Applied Spatial Clustering (DBSCAN), Identifying Cluster Structures Sort Points (OPTICS), Nonnegative Matrix Factorization (NMF), Latent Dirichlet Allocation (LDA), Gaussian Mixture Model (GMM)), Association Rule Learning Algorithms (e.g., Prior Algorithms, Eclat Algorithms, Frequent Pattern (FP) Growth), Artificial Neural Network Algorithms (e.g., Perceptron, Neural Networks, Backpropagation, Hopfield Networks, Autoencoders, Boltzmann Machines, Restricted Boltzmann Machines, Spiral Neural Networks, Radial Basis Function Networks), Deep Learning Algorithms (e.g., Deep Boltzmann Machines, Deep Belief Networks, Convolutional Neural Networks, Stacked Autoencoders), Dimensionality Reduction Algorithms (e.g., Principal Component Analysis (PCA), Independent Component Analysis (ICA), Principal Component Regression (PCR), Partial Least Squares Regression (PLSR), Salmon Mapping, Multidimensional Scaling, Projective Pursuit, Linear Discriminant Analysis, Mixture Discriminant Analysis, Quadratic Discriminant Analysis, Flexible Discriminant Analysis), Ensemble Algorithms (e.g., Augmentation, Bootstrap Aggregation, AdaBoost, Hybrid, Gradient Boosting Machines, Gradient Boosting Regression Trees, Random Forests) or suitable combinations thereof. For example, the surrogate algorithm 1102 can be or includes CNN, RNN, GAN, GNN, capsule network, autoencoder, ViT, etc.

[0121] In some embodiments, the surrogate algorithm 1102 is trained using input images 1112 and modified images 1114 of multiple objects associated with the full optimization routine 1110. The full optimization routine 1110 may include a prediction algorithm that receives the input images 1112 of the objects and determines the predicted geometry of the objects after fabrication, and an optimization algorithm that generates the modified image 1114 to compensate for any deviation between the predicted geometry and the target geometry of the objects. The full optimization routine 1110, including the prediction algorithm and the optimization algorithm, may be or includes the above-described... Figures 3 to 10Any of the embodiments described herein.

[0122] Input image 1112 and the corresponding modified image 1114 generated by the fully optimized routine 1110 can be used as training data for surrogate algorithm 1102. Training data can include any suitable number of objects (such as at least 5, 10, 20, 50, 100, 500, or 1000 objects) and / or no more than 1000, 500, 100, 50, 20, 10, or 5 objects. Training can be performed using any suitable method (such as supervised learning, unsupervised learning, or reinforcement learning). Thus, surrogate algorithm 1102 can learn the correlation between input image 1112 and the corresponding modified image 1114, and therefore can directly determine appropriate modifications to the geometry of a particular object without requiring the full prediction and optimization process implemented by the fully optimized routine 1110. In some embodiments, training penalties for surrogate algorithm 1102 result in outputs addressing adaptation problems (such as island generation).

[0123] Figure 12 This is a flowchart illustrating a method 1200 for generating instructions for additive manufacturing of an object according to an embodiment of the present technology. Method 1200 can be used to generate manufacturing instructions for any object (such as one or more dental appliances) described herein. In some embodiments, some or all of the processes in method 1200 are implemented as computer-readable instructions (e.g., program code) configured to be executed by one or more processors of a computing device (e.g., an appliance design system). Method 1200 can be combined with any of the other methods described herein.

[0124] Method 1200 may begin at block 1202: receiving at least one image representing a target geometry of an object to be manufactured using an additive manufacturing process. The image may include one or more slices representing multiple cross-sections of the object to be manufactured via a layer-by-layer additive manufacturing process (e.g., DLPA, SLA, SLS, inkjet) as described herein. In some embodiments, the image corresponds to manufacturing instructions for controlling the application of energy to a precursor material (e.g., resin or powder) to manufacture the object via the layer-by-layer additive manufacturing process. For example, pixels within the image may indicate whether energy should be applied to a corresponding location in the precursor material to form a portion of the object, and optionally, parameters of the energy to be applied to that location (e.g., intensity, exposure time, dose, wavelength) may indicate this energy. The image may be a black-and-white or grayscale image and may be provided in any suitable file format (e.g., BMP file, PNG file).

[0125] At box 1204, the method 1200 may include generating at least one modified image using a proxy algorithm. The proxy algorithm may be a machine learning algorithm (e.g., a CNN) configured to receive at least one image as input and determine one or more modifications to the at least one image, the modifications being configured to compensate for predicted deviations from the target geometry of the object when the object is manufactured via an additive manufacturing process based on the at least one image (e.g., the at least one image is used to generate fabrication instructions for implementing an additive manufacturing process). In some embodiments, the proxy algorithm (e.g., [image data]) is trained based on initial image data (e.g., input image 1112) and corresponding modified image data (e.g., modified image 1114) generated by another algorithm (e.g., fully optimized routine 1110). Figure 11 The proxy algorithm 1102).

[0126] The modified image may include one or more modifications relative to the initially received image, such as changing the size, shape, position, and / or grayscale values ​​of some or all of the object cross sections represented in the image. The modifications may be configured to compensate for prediction biases arising from the additive manufacturing and / or post-processing of the object, such as over-curing of the material used to make the object, over-construction of the material used to make the object, material retention on the surface of the object, material loss from the object, deformation of the object, or combinations thereof.

[0127] The modified image 808 generated by the proxy algorithm can represent the object geometry, which, when used as a basis for object fabrication, is expected to produce improved manufacturing accuracy, such that the actual geometry of the resulting object is expected to be the same as or sufficiently similar to the target geometry (e.g., there is no significant deviation between the actual geometry and the target geometry and / or any deviations that exist do not occur at significant parts of the object). In some embodiments, the modified object geometry represented in the modified image differs from the target geometry represented in the initial image, but an object fabricated based on the modified image is expected to have a geometry closer to the target geometry than an object fabricated based on the initial image.

[0128] At block 1206, method 1200 may include generating instructions for fabricating an object using an additive manufacturing process based on at least one modified image generated in block 1204. These instructions may be configured to control an additive manufacturing system (e.g., SLA, DLP, SLS, inkjet, or hybrid printing system) to apply energy to solidify precursor material according to the object geometry represented in the modified image. For example, the modified image may indicate the location where energy is to be applied (and optionally, parameters of the energy application) to form multiple object cross-sections according to a layer-by-layer additive manufacturing process sequence, as described elsewhere herein.

[0129] In some embodiments, the process at block 1206 includes converting the modified image into a format suitable for controlling the additive manufacturing system (e.g., a G-code file). Optionally, the modified image may undergo image post-processing before and / or during the conversion process. For example, image post-processing may include: removing artifacts (e.g., islands, spikes, holes, and / or other discontinuities) from the modified image, resizing the modified image (e.g., adjusting to a predetermined image length and width), adjusting the color of the modified image (e.g., converting to black and white or grayscale), adjusting the modified image to improve printability (e.g., ensuring a minimum feature thickness greater than a minimum printability thickness, increasing the thickness of the object's underlying layer), or suitable combinations thereof. Image post-processing may also include: adding components to the object geometry, such as support structures (e.g., struts, blocks, crossbars, etc.) to stabilize the object during additive manufacturing and / or post-processing, identifiers for tracking the object (e.g., labels, tags, barcodes, QR codes, etc.), and so on.

[0130] Refer again Figure 12 Method 1200 can be modified in many different ways. For example, although the steps of method 1200 described above are for a single object, method 1200 can be used to generate instructions sequentially or concurrently for creating any suitable number of objects (such as dozens, hundreds, or thousands of objects). As another example, Figure 12 The order of the processes shown can be changed. Some processes in method 1200 can be omitted, and / or method 1200 may include... Figure 12 Additional processes not shown. For example, method 1200 may additionally include displaying a graphical representation of the modified image to a user (e.g., a technician or clinician) via a suitable display device (e.g., a monitor, screen, etc. of a computing system or device). In such embodiments, the user may provide feedback, if appropriate, to approve the modified image or make adjustments to the modified image.

[0131] Figure 13 This is a flowchart illustrating a workflow 1300 for evaluating a modified image of an object according to an embodiment of the present technology. Workflow 1300 can be combined with any other process described herein (e.g., with...). Figure 11 Workflow 1100 and / or Figure 12 This is achieved by combining the methods of 1200.

[0132] Modified images of objects can be generated using optimization algorithms and / or surrogate algorithms as described herein (box 1302). The modified images can undergo quality control evaluation (box 1304) to check for problems that may affect the accuracy and manufacturability of the object. For example, the quality control evaluation may include evaluating whether the modified image meets one or more quality parameters, such as detecting whether the image includes artifacts (e.g., spikes, holes), broken features (e.g., islands, discontinuities), undersupported features, and / or features smaller than the minimum feature size (e.g., for printability and / or support purposes). If such problems are detected, the modified image can be adjusted to correct the problems (e.g., removing artifacts, connecting broken features, adding support to undersupported features, increasing the size of undersized features, increasing the thickness of the object's underlying layer).

[0133] Subsequently, the predicted geometry of the object after additive manufacturing is determined based on the modified image (box 1306). Any prediction algorithm described herein can be used to determine the predicted geometry. The predicted geometry of the object can then be compared with the target geometry to identify any deviations (box 1308). The identified deviations can optionally be displayed to the user, for example, via a graphical representation (e.g., a 2D image, a 3D model), text, alerts, or any other indication. The user can then review the identified deviations to determine if corrective actions are appropriate, such as making further adjustments to the modified image.

[0134] Figure 14 This is a block diagram illustrating a workflow for training an inverse algorithm according to an embodiment of the present technology. In some embodiments, the present technology provides a method for using an "inverse algorithm" to determine instructions for creating an object with an actual geometry that more closely conforms to a target geometry, the "inverse algorithm" being the reverse of the forward prediction algorithm described herein (e.g., Figure 5 Prediction algorithm 502 Figure 10 The predictive relationship is established by the forward prediction model (1008). Specifically, the forward prediction algorithm can be trained to predict the relationship in a given input image slice. x Predict output under the following circumstances y (For example, f(x→y The goal of the inverse algorithm can be to reconstruct a slice of the original input image given the output y predicted by the forward model. x For example, the inverse algorithm is trained to learn the mapping. g(y)→x This can be achieved through the following: first, using a forward prediction algorithm. f(x→y Generate known xy The dataset is used to predict y, and then the forward prediction algorithm is used to output the result. y As input and input image slices xThe target output is used to train the inverse algorithm. g(y) When applying this inverse algorithm, the design goal represented as the input slice can be considered as the result of the forward prediction, and the inverse algorithm can optimize the input slice to determine the image slice suitable for printing. x Therefore, reverse engineering algorithms can be considered as proxy algorithms that can be used to replace fully optimized routines.

[0135] In some embodiments, the inverse algorithm is or includes at least one machine learning algorithm, such as at least one of the following: regression algorithms (e.g., ordinary least squares regression, linear regression, logistic regression, stepwise regression, multivariate adaptive regression splines, local estimation scatter plot smoothing), instance-based algorithms (e.g., k-nearest neighbors, learned vector quantization, self-organizing graph, locally weighted learning), regularization algorithms (e.g., ridge regression, minimum absolute shrinkage and selection operator, elastic net, minimum angle regression), decision tree algorithms (e.g., iterative binary divider). 3 (ID3), C4.5, C5.0, Classification and Regression Trees, Chi-square Automatic Interaction Detection, Decision Stubs, M5), Bayesian Algorithms (e.g., Naive Bayes, Gaussian Naive Bayes, Multinomial Naive Bayes, Average Single Correlation Estimator, Bayesian Belief Network, Bayesian Network, Hidden Markov Model, Conditional Random Field), Clustering Algorithms (e.g., k-means, Single Link Clustering, k-median, Expectation-Maximization, Hierarchical Clustering, Fuzzy Clustering, Density-Based Noise Applied Spatial Clustering (DBSCAN), Ranking for Identifying Cluster Structures The algorithms used include: Point (OPTICS), Non-negative Matrix Factorization (NMF), Latent Dirichlet Allocation (LDA), Gaussian Mixture Model (GMM), Association Rule Learning Algorithms (e.g., Prior Algorithms, Eclat Algorithms, Frequent Pattern (FP) Growth), Artificial Neural Network Algorithms (e.g., Perceptron, Neural Networks, Backpropagation, Hopfield Networks, Autoencoders, Boltzmann Machines, Restricted Boltzmann Machines, Spiral Neural Networks, Radial Basis Function Networks), Deep Learning Algorithms (e.g., Deep Boltzmann Machines, Deep Belief Networks, Convolutional Neural Networks, Stacked Autoencoders), Dimensionality Reduction Algorithms (e.g., Principal Component Analysis (PCA), Independent Component Analysis (ICA), Principal Component Regression (PCR), Partial Least Squares Regression (PLSR), Salmon Mapping, Multidimensional Scaling, Projective Pursuit, Linear Discriminant Analysis, Mixture Discriminant Analysis, Quadratic Discriminant Analysis, Flexible Discriminant Analysis), Ensemble Algorithms (e.g., Augmentation, Bootstrap Aggregation, AdaBoost, Hybrid, Gradient Boosting Machines, Gradient Boosting Regression Trees, Random Forests), or suitable combinations thereof. For example, the reverse algorithm can be or includes CNN, RNN, GAN, GNN, capsule network, autoencoder, ViT, etc. II. Dental Instruments and Associated Methods

[0136] Figure 15AThe illustration shows a representative example of a tooth repositioning appliance 1500 configured according to an embodiment of the present technology. Appliance 1500 can be manufactured using any of the systems, methods, and apparatuses described herein. Appliance 1500 (also referred to herein as an "orthodontic appliance") can be worn by a patient to achieve incremental repositioning of individual teeth 1502 in the jaw. Appliance 1500 may include a housing (e.g., a continuous polymer housing or a segmented housing) having a tooth receiving cavity that receives and resiliently repositions the teeth. Appliance 1500 or portions thereof can be fabricated indirectly using a physical model of the teeth. For example, an appliance (e.g., a polymer appliance) can be formed using a physical model of the teeth and suitable polymer material sheets. In some embodiments, for example, a physical appliance is fabricated directly from a digital model of the appliance using additive manufacturing techniques.

[0137] The appliance 1500 can be adapted to all teeth present in the maxilla or mandible, or to fewer than all teeth. The appliance 1500 can be specifically designed to receive a patient's teeth (e.g., the morphology of the tooth receiving cavity matches the morphology of the patient's teeth) and can be fabricated based on a positive or negative mold of the patient's teeth generated by impression, scanning, etc. Alternatively, the appliance 1500 can be a general-purpose appliance configured to receive teeth, but not necessarily shaped to match the morphology of the patient's teeth. In some cases, only certain teeth received by the appliance 1500 are repositioned by the appliance 1500, while other teeth can provide a base or anchoring area for holding the appliance 1500 in place when the appliance 1500 applies force to one or more teeth that are the repositioning targets. In some cases, some, most, or even all teeth can be repositioned at some point during treatment. The moved teeth can also be used as a base or anchor for holding the appliance when the patient wears it. In a preferred embodiment, no thread or other component is provided for holding the appliance 1500 in a suitable position on the tooth. However, in some cases, it may be desirable or necessary to provide various attachments 1504 or other anchoring elements on the tooth 1502, having corresponding receiving portions 1506 or apertures in the appliance 1500, such that the appliance 1500 can apply selected forces on the tooth. Representative examples of appliances, including those used in the Invisalign® system, are described in numerous patents and patent applications assigned to Align Technology, Inc., including, for example, U.S. Patent Nos. 6,450,807 and 5,975,893, and on the company’s website accessible via the World Wide Web (e.g., see URL “invisalign.com”). Examples of tooth-mounted attachments suitable for use with orthodontic appliances are also described in patents and patent applications assigned to Align Technology, Inc., including, for example, U.S. Patent Nos. 6,309,215 and 6,830,450.

[0138] Figure 15BThe illustration depicts a tooth repositioning system 1510 comprising multiple appliances 1512, 1514, 1516 according to an embodiment of the present technology. Any appliance described herein may be designed and / or provided as part of a set of multiple appliances for use in a tooth repositioning system. Each appliance may be configured such that the tooth receiving cavity has a geometry corresponding to an intermediate or final tooth arrangement intended for use with that appliance. By placing a series of incremental position adjustment appliances on the patient's teeth, the patient's teeth can be progressively repositioned from an initial tooth arrangement to a target tooth arrangement. For example, the tooth repositioning system 1510 may include a first appliance 1512 corresponding to the initial tooth arrangement, one or more intermediate appliances 1514 corresponding to one or more intermediate arrangements, and a final appliance 1516 corresponding to the target arrangement. The target tooth arrangement may be a planned final tooth arrangement selected for the patient's teeth at the end of all planned orthodontic treatment. Alternatively, the target tooth arrangement can be one of several intermediate arrangements used for the patient's teeth during orthodontic treatment. These intermediate arrangements can include a variety of different treatment scenarios, including but not limited to cases where surgery is recommended, interproximal enamel reduction (IPR) is appropriate, progress checks are scheduled, anchor placement is optimal, palatal expansion is desired, and cases involving restorative dentistry (e.g., inlays, onlays, crowns, bridges, implants, veneers, etc.). Thus, it should be understood that the target tooth arrangement can be any resulting arrangement planned for the patient's teeth following one or more incremental repositioning stages. Similarly, the initial tooth arrangement can be any initial arrangement column of the patient's teeth followed by one or more incremental repositioning stages.

[0139] Figure 15CThe illustration depicts a method 1520 for orthodontic treatment using multiple appliances according to an embodiment of the present technology. Method 1520 can be practiced using any of the appliances or groups of appliances described herein. In box 1522, a first orthodontic appliance is applied to the patient's teeth to reposition the teeth from a first dental arrangement to a second dental arrangement. In box 1524, a second orthodontic appliance is applied to the patient's teeth to reposition the teeth from a second dental arrangement to a third dental arrangement. Method 1520 can be repeated as needed using any suitable number and combination of sequential appliances and combinations of sequential appliances to incrementally reposition the patient's teeth from an initial arrangement to a target arrangement. Appliances can be manufactured all at once, in groups, or in batches (e.g., at the beginning of a treatment phase), or appliances can be manufactured one at a time, and the patient can wear each appliance until pressure on the teeth from each appliance is no longer felt or until the maximum amount of expressed tooth movement for that given phase has been achieved. Multiple different appliances (e.g., a set) can be designed or even manufactured before the patient wears any of the appliances in the multiple appliances. After an appropriate period of appliance wear, the patient can replace the current appliance with the next appliance in the series until no more appliances are available. Appliances are typically not attached to the teeth, and the patient can place and change appliances at any time during the procedure (e.g., the patient can remove the appliance). The final appliance or several appliances in the series may have one or more geometries selected for overcorrecting the tooth arrangement. For example, one or more appliances may have geometries that will (if fully realized) move individual teeth beyond the tooth arrangement that has been selected as “final.” This overcorrection may be desirable to counteract potential regression after the repositioning method has been terminated (e.g., allowing individual teeth to move back toward their pre-correction positions). Overcorrection can also be beneficial to accelerate the correction rate (e.g., an appliance with a geometry positioned beyond a desired intermediate or final position can move individual teeth toward that position at a greater rate). In this case, the use of the appliance may be terminated before the teeth reach the position defined by the appliance. Furthermore, overcorrection may be intentionally applied to compensate for any inaccuracies or limitations of the appliance.

[0140] Figure 16 A method 1600 for designing orthodontic appliances according to an embodiment of the present technology is illustrated. Method 1600 can be applied to any embodiment of the orthodontic appliances described herein. Some or all of the steps in method 1600 can be performed by any suitable data processing system or device (e.g., one or more processors configured with suitable instructions).

[0141] In box 1602, a movement path for moving one or more teeth from an initial arrangement to a target arrangement is determined. The initial arrangement can be determined, for example, using wax bite method, direct contact scanning, X-ray imaging, tomography, ultrasound imaging, and other techniques for obtaining information about the location and structure of teeth, jaws, gingiva, and other orthodontic-related tissues, based on a mold or scan of the patient's teeth or oral tissues. A digital dataset representing the initial (e.g., pre-treatment) arrangement of the patient's teeth and other tissues can be derived from the obtained data. Optionally, the initial digital dataset is processed to segment the tissue components to each other. For example, a data structure representing the individual crowns digitally can be generated. Advantageously, a digital model of the entire tooth can be generated, including the measured or inferred hidden surfaces and root structures, as well as the surrounding bone and soft tissue.

[0142] The target tooth arrangement (e.g., the expectation and anticipated end result of orthodontic treatment) can be received from the clinician in the form of a prescription, calculated based on fundamental orthodontic principles, and / or inferred computationally from the clinical prescription. By specifying the expected final position of the teeth and a digital representation of the teeth themselves, the final position and surface geometry of each tooth can be specified to form a complete model of the tooth arrangement at the expected end of treatment.

[0143] Given that each tooth has an initial position and a target position, a movement path can be defined for the movement of each tooth. In some embodiments, the movement path is configured to move the tooth from its initial position to its desired target position in the fastest way with the least amount of round trips. The tooth path can optionally be segmented, and segments can be computed such that the movement of each tooth within a segment remains within threshold limits for linear translation and rotational translation. Thus, the endpoints of each path segment can constitute a clinically feasible repositioning, and the aggregation of the segment endpoints can constitute a clinically feasible sequence of tooth positions such that moving from one point to the next in the sequence does not result in tooth collision.

[0144] In box 1604, a force system for generating movement of one or more teeth along the movement path is defined. The force system may include one or more forces and / or one or more torques. Different force systems can produce different types of tooth movement, such as tilting, translation, rotation, extrusion, intrusion, root movement, etc. Biomechanical principles, modeling techniques, force calculation / measurement techniques, etc. (including knowledge and approaches commonly used in orthodontics) can be used to determine the appropriate force system to be applied to the teeth to achieve tooth movement. Sources may be considered when determining the force system to be applied, including literature, force systems determined through experimental or virtual modeling, computer-based modeling, clinical experience, minimization of unwanted forces, etc.

[0145] The determination of the force system can be performed in a variety of ways. For example, in some embodiments, the force system is determined on a patient-by-patient basis, for example, using patient-specific data. Alternatively or in combination, the force system can be determined based on a generalized model of tooth movement (e.g., based on experimental, modeling, clinical data, etc.), making it not necessarily necessary to use patient-specific data. In some embodiments, determining the force system includes calculating specific force values ​​to be applied to one or more teeth to produce specific movement. Alternatively, the determination of the force system can be performed at a high level without calculating specific force values ​​for the teeth. For example, block 1604 may include determining a specific type of force (e.g., compressive force, invasive force, translational force, rotational force, tilting force, torsional force, etc.) to be applied without calculating the specific amplitude and / or direction of the force.

[0146] Determining the force system can include constraints on permissible forces (such as permissible direction and amplitude) and the desired movement to be induced by the applied forces. For example, different patients may expect different movement strategies when fabricating a palatal expander. For instance, the amount of force required to separate the palate can depend on the patient's age, as very young patients may not have a fully formed suture. Therefore, in adolescent patients and others with incompletely closed palatal sutures, palatal expansion can be accomplished with a lower force amplitude. Slower palatal movement can also help bone growth to fill the expanding suture. For other patients, a faster expansion may be desired, which can be achieved by applying a larger force. The structure and materials of the appliance can be selected as needed; for example, by selecting a palatal expander capable of applying a larger force to open the palatal suture and / or cause rapid expansion of the palate. Subsequent appliance stages can be designed to apply varying amounts of force, such as initially applying a larger force to break the suture, followed by a smaller force to maintain suture separation or gradually expand the palate and / or dental arch.

[0147] The force determination system may also include modeling the patient's facial structures, such as the skeletal structure of the jaw and palate. For example, scan data of the palate and dental arch (such as X-ray data or 3D optical scan data) can be used to determine parameters of the skeletal and muscular systems of the patient's oral cavity in order to determine the force sufficient to provide the desired expansion of the palate and / or dental arch. In some embodiments, the thickness and / or density of the palatal suture may be measured or entered by a medical professional. In other embodiments, the medical professional may select appropriate treatment based on the patient's physiological characteristics. For example, the characteristics of the palate may also be estimated based on factors such as the patient's age; for instance, a young adolescent patient may require less force to expand the suture than an older patient because the suture has not yet fully formed.

[0148] In box 1606, the design for an orthodontic appliance configured to generate a force system is defined. This design may include appliance geometry, material composition, and / or material properties, and may be determined in various ways, such as using a treatment or force application simulation environment. The simulation environment may include, for example, a computer modeling system, a biomechanical system, or a device. Optionally, a digital model of the appliance and / or teeth, such as a finite element model, may be generated. The finite element model can be created using computer program application software available from various vendors. To create the solid geometry model, computer-aided engineering (CAE) or computer-aided design (CAD) programs, such as AutoCAD® software products available from Autodesk, Inc., San Rafael, California, can be used. To create and analyze the finite element model, program products from several vendors can be used, including, but not limited to, the finite element analysis package from ANSYS, Inc., Fort Cannons, Pennsylvania, and the SIMULIA (Abaqus) software product from Dassault Systèmes, Waltham, Massachusetts.

[0149] Optionally, one or more designs can be selected for testing or force modeling. As noted above, the desired tooth movement and the required or desired force system to induce the desired tooth movement can be identified. Using a simulation environment, candidate designs can be analyzed or modeled to determine the actual force system generated by using the candidate apparatus. Optionally, one or more modifications can be made to the candidate apparatus, and force modeling can be further analyzed based on the description, for example, to iteratively determine the apparatus design that produces the desired force system.

[0150] In block 1608, instructions for fabricating orthodontic appliances incorporating the design are generated. The instructions may be configured to control a fabrication system or apparatus to produce orthodontic appliances having the specified design. In some embodiments, the instructions are configured to fabricate the orthodontic appliance using direct fabrication (e.g., stereolithography, selective laser sintering, fused deposition modeling, 3D printing, continuous direct fabrication, multimaterial direct fabrication, etc.) according to various methods presented herein. In alternative embodiments, the instructions may be configured to fabricate the appliance indirectly (e.g., by thermoforming).

[0151] While the steps described above illustrate a method 1500 for designing orthodontic appliances according to some embodiments, those skilled in the art will recognize some variations based on the teachings described herein. Some steps may include sub-steps. Some steps may be repeated frequently as needed. One or more steps of method 1600 may be performed using any suitable fabrication system or apparatus, such as those described herein. Some steps may be optional; for example, the process in box 1604 may be omitted, allowing the orthodontic appliance to be designed based on desired tooth movement and / or determined tooth movement paths rather than on a force system. Moreover, the order of steps may vary as desired.

[0152] Figure 17 The illustration depicts a method 1700 for digitally planning orthodontic treatment and / or designing or fabricating an appliance according to an embodiment. Method 1700 can be applied to any treatment procedure described herein and can be performed by any suitable data processing system.

[0153] In box 1702, a digital representation of the patient's teeth is received. The digital representation may include surface topography data of the patient's oral cavity (including teeth, gingival tissue, etc.). The surface topography data can be generated by directly scanning the oral cavity, a physical model (positive or negative mold) of the oral cavity, or an impression of the oral cavity using a suitable scanning device (e.g., a handheld scanner, a desktop scanner, etc.).

[0154] In box 1704, one or more treatment phases are generated based on a digital representation of the teeth. A treatment phase can be an incremental repositioning phase in the orthodontic treatment process, designed to move one or more of the patient's teeth from an initial tooth arrangement to a target arrangement. For example, a treatment phase can be generated by determining the initial tooth arrangement indicated by the digital representation, determining the target tooth arrangement, and determining the movement path for one or more teeth in the initial arrangement required to achieve the target tooth arrangement. The movement path can be optimized based on minimizing the total distance of movement, preventing collisions between teeth, avoiding more difficult tooth movements, or any other suitable criteria.

[0155] In box 1706, at least one orthodontic appliance is fabricated based on the generated treatment phase. For example, a set of appliances may be fabricated, each appliance shaped according to a tooth arrangement specified by one of the treatment phases, such that the appliances can be sequentially worn by the patient to incrementally reposition the teeth from the initial arrangement to the target arrangement. The appliance set may include one or more orthodontic appliances described herein. Fabricating the appliance may include: creating a digital model of the appliance for use as input to a computer-controlled fabrication system. Depending on the desired outcome, the appliance may be formed using direct fabrication methods, indirect fabrication methods, or a combination thereof.

[0156] In some cases, the phased arrangement or treatment phases may not be necessary for the design and / or fabrication of the apparatus. For example... Figure 17 As illustrated by the dashed lines, the design and / or fabrication of orthodontic appliances and possible specific orthodontic treatments may include using a representation of the patient's teeth (e.g., including receiving a digital representation of the patient's teeth (box 1702)), and then designing / or fabricating orthodontic appliances based on the representation of the patient's teeth in the arrangement represented by the received representation.

[0157] As noted herein, the techniques described herein can be used to directly fabricate dental appliances, such as orthodontic appliances and / or a series of appliances having tooth receiving cavities, which are configured to move a person’s teeth from an initial arrangement toward a target arrangement according to a treatment plan. Orthodontic appliances may include mandibular repositioning elements, such as those described in the following documents, which are incorporated herein by reference in their entirety: U.S. Patent No. 10,912,629, filed November 30, 2015, entitled “Dental Appliances with Repositioning Jaw Elements”; U.S. Patent No. 10,537,406, filed September 19, 2014, entitled “Dental Appliances with Repositioning Jaw Elements”; and U.S. Patent No. 9,844,424, filed February 21, 2014, entitled “Dental Appliances with Repositioning Jaw Elements”.

[0158] The techniques used in this article can also be used to manufacture attachment placement devices, such as appliances used to position prefabricated attachments on a person's teeth according to one or more aspects of a treatment plan. Examples of attachment placement devices (also referred to as “attachment placement templates” or “attachment fabrication templates”) can be found, in their entirety, in the following: U.S. Patent Application No. 17 / 249,218, filed February 24, 2021, entitled “Flexible 3D Printed Orthodontic Device”; U.S. Patent Application No. 16 / 366,686, filed March 27, 2019, entitled “Dental Attachment Placement Structure”; U.S. Patent Application No. 15 / 674,662, filed August 11, 2017, entitled “Devices and Systems for Creation of Attachments”; and U.S. Patent Application No. 14 / Dental Attachment Placement, filed June 14, 2017, entitled “Dental Attachment Placement”. U.S. Patent No. 11,103,330 entitled “Dental Attachment Placement Structure”, filed December 9, 2015; U.S. Application No. 14 / 963,527 entitled “Dental Attachment Placement Structure”, filed November 12, 2015; U.S. Application No. 14 / 939,246 entitled “Dental Attachment Placement Structure”, filed November 12, 2015; U.S. Application No. 14 / 939,252 entitled “Dental Attachment Formation Structures”, filed November 12, 2015; and U.S. Patent No. 9,700,385 entitled “Attachment Structure”, filed August 22, 2014.

[0159] The techniques described herein can be used to manufacture incremental palatal expanders and / or a series of incremental palatal expanders used to expand a person's palate from an initial position toward a target position according to one or more aspects of a treatment plan. Examples of incremental palatal expanders can be found at least in the following documents, which are incorporated herein by reference in their entirety: U.S. Application No. 16 / 380,801, entitled "Releasable Palatal Expanders," filed April 10, 2019; U.S. Application No. 16 / 022,552, entitled "Devices, Systems, and Methods for Dental Arch Expansion," filed June 28, 2018; U.S. Patent No. 11,045,283, entitled "Palatal Expander with Skeletal Anchorage Devices," filed June 8, 2018; and U.S. Patent No. 11,045,283, entitled "Palatal Expanders and Methods of Expanding," filed December 4, 2017. U.S. Patent Application No. 15 / 831,159 entitled "aPalate (a palatal expander and method of expanding the palate)"; U.S. Patent No. 10,993,783 entitled "Methods and Apparatuses for Customizing a Rapid Palatal Expander" filed on December 4, 2017; and U.S. Patent No. 7,192,273 entitled "System and Method for Palatal Expansion" filed on August 7, 2003. Example

[0160] The following examples are included to further describe some aspects of this technology, and these examples should not be used to limit the scope of this technology.

[0161] Example 1: A method that includes: Receive at least one image, the at least one image representing the target geometry of an object to be manufactured using an additive manufacturing process; At least one modified image is generated by inputting the at least one image into a machine learning algorithm, wherein the machine learning algorithm is trained to determine one or more modifications to the at least one image, and wherein the one or more modifications are configured to: compensate for predicted deviations from the target geometry of the object when the object is manufactured via the additive manufacturing process based on the at least one image; and Instructions for manufacturing the object using the additive manufacturing process are generated based on the at least one modified image.

[0162] Example 2: According to the method described in Example 1, the machine learning algorithm is trained based on initial image data of multiple additive manufacturing objects and corresponding modified image data.

[0163] Example 3: The method described in Example 1 or 2, wherein the machine learning algorithm includes a convolutional neural network (CNN).

[0164] Example 4: The method according to any one of Examples 1 to 3, wherein the one or more modifications are configured to compensate for the predicted deviation from the target geometry of the object due to the additive manufacturing process, post-processing operation or a combination thereof.

[0165] Example 5: The method according to any one of Examples 1 to 4, wherein the one or more modifications are configured to compensate for the predicted deviation from the target geometry of the object due to over-curing of the material used to make the object, over-construction of the material used to make the object, retention of the material on the surface of the object, material loss from the object, deformation of the object, or a combination thereof.

[0166] Example 6: The method according to any one of Examples 1 to 5, wherein the one or more modifications include removing material from a portion of the object represented in the at least one image.

[0167] Example 7: The method according to any one of Examples 1 to 6, wherein the one or more modifications include adding material to a portion of the object represented in the at least one image.

[0168] Example 8: The method according to any one of Examples 1 to 7, wherein the additive manufacturing process includes one or more of the following: stereolithography, digital light processing, selective laser sintering, material jetting, or material extrusion.

[0169] Example 9: The method according to any one of Examples 1 to 8, wherein the additive manufacturing process includes applying energy to a precursor material to form a plurality of object layers.

[0170] Example 10: The method according to Example 9, wherein the instructions are configured to control the application of the energy to the precursor material.

[0171] Example 11: The method according to Example 9 or 10, wherein the instructions are configured to form at least one object layer corresponding to the at least one modified image.

[0172] Example 12: The method according to any one of Examples 1 to 11, wherein the at least one image corresponds to at least one 2D cross section of a 3D digital representation of the object.

[0173] Example 13: The method according to any one of Examples 1 to 12 further includes: determining a predicted geometry of the object after it has been fabricated using the additive manufacturing process, based on the at least one modified image.

[0174] Example 14: The method according to Example 13 further includes: identifying the deviation between the target geometry and the predicted geometry.

[0175] Example 15: The method according to Example 14 further includes: outputting an indication of the identified deviation.

[0176] Example 16: The method according to any one of Examples 1 to 15 further includes: evaluating whether the at least one modified image satisfies one or more quality control parameters.

[0177] Example 17: The method according to Example 16, wherein the evaluation includes: detecting artifacts, detecting broken features, detecting insufficiently supported features, detecting features smaller than the minimum feature size, or a combination thereof.

[0178] Example 18: The method according to Example 16 or 17 further includes: adjusting the at least one modified image in response to the at least one modified image not meeting the evaluation of the one or more quality control parameters.

[0179] Example 19: The method according to any one of Examples 1 to 18, wherein the at least one modified image represents a modified geometry of the object that is different from the target geometry.

[0180] Example 20: The method according to any one of Examples 1 to 19 further includes: fabricating the object using the additive manufacturing process based on the instructions.

[0181] Example 21: A system comprising: One or more processors; and A memory operatively coupled to one or more processors and storing instructions that, when executed by the one or more processors, cause the system to perform operations including: Receive at least one image, said at least one image representing the target geometry of an object to be manufactured using an additive manufacturing process. At least one modified image is generated by inputting the at least one image into a machine learning algorithm, wherein the machine learning algorithm is trained to determine one or more modifications to the at least one image, and wherein the one or more modifications are configured to: compensate for predicted deviations from the target geometry of the object when the object is fabricated via the additive manufacturing process based on the at least one image, and Instructions for manufacturing the object using the additive manufacturing process are generated based on the at least one modified image.

[0182] Example 22: According to the system of Example 21, the machine learning algorithm is trained based on initial image data of a plurality of additive manufacturing objects and corresponding modified image data.

[0183] Example 23: The system according to Example 21 or 22, wherein the machine learning algorithm includes a convolutional neural network (CNN).

[0184] Example 24: A system according to any one of Examples 21 to 23, wherein the one or more modifications are configured to compensate for predicted deviations from the target geometry of the object due to the additive manufacturing process, post-processing operations, or a combination thereof.

[0185] Example 25: A system according to any one of Examples 21 to 24, wherein the one or more modifications are configured to compensate for the predicted deviation from the target geometry of the object due to over-curing of the material used to make the object, over-construction of the material used to make the object, retention of material on the surface of the object, material loss from the object, deformation of the object, or a combination thereof.

[0186] Example 26: A system according to any one of Examples 21 to 25, wherein the one or more modifications include removing material from a portion of the object represented in the at least one image.

[0187] Example 27: A system according to any one of Examples 21 to 26, wherein the one or more modifications include adding material to a portion of the object represented in the at least one image.

[0188] Example 28: The system according to any one of Examples 21 to 27, wherein the additive manufacturing process includes one or more of the following: stereolithography, digital light processing, selective laser sintering, material jetting, or material extrusion.

[0189] Example 29: A system according to any one of Examples 21 to 28, wherein the additive manufacturing process includes applying energy to a precursor material to form a plurality of object layers.

[0190] Example 30: The system according to Example 29, wherein the instructions are configured to control the application of energy to the precursor material.

[0191] Example 31: The system according to Example 29 or 30, wherein the instructions are configured to cause the formation of at least one object layer corresponding to the at least one modified image.

[0192] Example 32: A system according to any one of Examples 21 to 31, wherein the at least one image corresponds to at least one 2D cross section of a 3D digital representation of the object.

[0193] Example 33: The system according to any one of Examples 21 to 32 further includes: determining a predicted geometry of the object after it has been fabricated using the additive manufacturing process, based on the at least one modified image.

[0194] Example 34: The system according to Example 33 further includes: identifying the deviation between the target geometry and the predicted geometry.

[0195] Example 35: The system according to Example 34 further includes: outputting an indication of the identified deviation.

[0196] Example 36: The system according to any one of Examples 21 to 35 further includes: evaluating whether the at least one modified image satisfies one or more quality control parameters.

[0197] Example 37: The system according to Example 36, wherein the evaluation includes: detecting artifacts, detecting broken features, detecting insufficiently supported features, detecting features smaller than the minimum feature size, or a combination thereof.

[0198] Example 38: The system according to Example 36 or 37 further includes: adjusting the at least one modified image in response to the evaluation that the at least one modified image does not meet the evaluation of the one or more quality control parameters.

[0199] Example 39: A system according to any one of Examples 21 to 38, wherein the at least one modified image represents a modified geometry of the object that is different from the target geometry.

[0200] Example 40: The system according to any one of Examples 21 to 39 further includes: fabricating the object using the additive manufacturing process based on the instructions.

[0201] Example 41: A method comprising: Receive at least one image, the at least one image representing the target geometry of an object to be manufactured using an additive manufacturing process; The predicted geometry of the object after it has been fabricated using the additive manufacturing process is determined based on the at least one image. Identify the deviation between the target geometry and the predicted geometry; Modify at least one image based on the identified deviations; and Instructions for manufacturing the object using the additive manufacturing process are generated based on the at least one modified image.

[0202] Example 42: The method described in Example 41, wherein the prediction is determined using a machine learning algorithm.

[0203] Example 43: The method described in Example 42, wherein the machine learning algorithm includes a convolutional neural network (CNN).

[0204] Example 44: The method described in Example 43, wherein the CNN is trained using images of other objects, which are fabricated using the additive manufacturing process.

[0205] Example 45: The method according to Example 43 or 44, wherein the CNN includes a kernel, and wherein the kernel includes a function representing a physical or chemical phenomenon associated with the additive manufacturing process.

[0206] Example 46: The method according to Example 45, wherein the physical or chemical phenomenon includes light scattering.

[0207] Example 47: The method according to any one of Examples 42 to 46, wherein the machine learning algorithm is configured to predict deviations from the target geometry of the object due to the additive manufacturing process, post-processing operations, or a combination thereof.

[0208] Example 48: The method according to any one of Examples 42 to 47, wherein the machine learning algorithm is configured to predict deviations from the target geometry of the object due to over-curing of the material used to make the object, over-construction of the material used to make the object, retention of the material on the surface of the object, material loss from the object, deformation of the object, or a combination thereof.

[0209] Example 49: The method according to any one of Examples 41 to 48, wherein the additive manufacturing process includes one or more of the following: stereolithography, digital light processing, selective laser sintering, material jetting, or material extrusion.

[0210] Example 50: The method according to any one of Examples 41 to 49, wherein the additive manufacturing process includes applying energy to a precursor material to form a plurality of object layers.

[0211] Example 51: The method according to Example 50, wherein the instructions are configured to control the application of the energy to the precursor material.

[0212] Example 52: The method according to Example 50 or 51, wherein the instructions are configured to cause the formation of at least one object layer corresponding to the at least one modified image.

[0213] Example 53: The method according to any one of Examples 41 to 52, wherein the predicted geometry represents the geometry of the object after it has been made using the additive manufacturing process and after undergoing post-processing operations.

[0214] Example 54: According to the method of Example 53, the post-processing operation includes one or more of the following: centrifuging the object, post-curing the object, or washing the object.

[0215] Example 55: The method according to any one of Examples 41 to 54, wherein the at least one image corresponds to at least one 2D cross section of a 3D digital representation of the object.

[0216] Example 56: The method according to any one of Examples 41 to 55, wherein the modification is performed using an optimization algorithm.

[0217] Example 57: The method described in Example 56, wherein the optimization algorithm is a reverse optimization algorithm.

[0218] Example 58: The method according to any one of Examples 41 to 57 further includes: generating at least one second image, the at least one second image representing the predicted geometry of the object.

[0219] Example 59: The method according to Example 58, wherein the deviation is identified based on the at least one image and the at least one second image.

[0220] Example 60: According to the method of Example 59, the deviation is identified by comparing the at least one image with the at least one second image.

[0221] Example 61: The method according to any one of Examples 41 to 60, wherein the object includes a dental appliance.

[0222] Example 62: The method according to Example 61, wherein the dental appliance is an orthodontic appliance, palatal expander, retainer, attachment placement device or oral protector.

[0223] Example 63: A system comprising: One or more processors; and A memory operatively coupled to one or more processors and storing instructions that, when executed by the one or more processors, cause the system to perform operations including: Receive at least one image, said at least one image representing the target geometry of an object to be manufactured using an additive manufacturing process. The predicted geometry of the object after it has been fabricated using the additive manufacturing process is determined based on the at least one first image. Identify the deviation between the target geometry and the predicted geometry. Modify at least one image based on the identified deviations, and Instructions for manufacturing the object using the additive manufacturing process are generated based on the at least one modified image.

[0224] Example 64: The system according to Example 63, wherein the at least one second image is generated using a machine learning algorithm.

[0225] Example 65: The system according to Example 64, wherein the machine learning algorithm includes a convolutional neural network (CNN).

[0226] Example 66: The system according to Example 65, wherein the CNN is trained using images of other objects, which are fabricated using the additive manufacturing process.

[0227] Example 67: The system according to Example 65 or 66, wherein the CNN includes a kernel, and wherein the kernel includes a function representing a physical or chemical phenomenon associated with the additive manufacturing process.

[0228] Example 68: The system according to Example 67, wherein the physical or chemical phenomenon includes light scattering.

[0229] Example 69: A system according to any one of Examples 64 to 68, wherein the machine learning algorithm is configured to predict deviations from the target geometry of the object due to the additive manufacturing process, post-processing operations, or a combination thereof.

[0230] Example 70: A system according to any one of Examples 64 to 69, wherein the machine learning algorithm is configured to predict deviations from the target geometry of the object due to over-curing of the material used to make the object, over-construction of the material used to make the object, retention of the material on the surface of the object, material loss from the object, deformation of the object, or a combination thereof.

[0231] Example 71: The system according to any one of Examples 63 to 70, wherein the additive manufacturing process includes one or more of the following: stereolithography, digital light processing, selective laser sintering, material jetting, or material extrusion.

[0232] Example 72: A system according to any one of Examples 63 to 71, wherein the additive manufacturing process includes applying energy to a precursor material to form a plurality of object layers.

[0233] Example 73: The system according to Example 72, wherein the instructions are configured to control the application of energy to the precursor material.

[0234] Example 74: The system according to Example 72 or 73, wherein the instructions are configured to cause the formation of at least one object layer corresponding to the at least one modified image.

[0235] Example 75: The system according to any one of Examples 72 to 74 further includes an additive manufacturing system, wherein the additive manufacturing system comprises: An energy source, configured to output the energy; The source of the precursor material; and A controller configured to cause the energy source to apply energy to the precursor material according to the instructions.

[0236] Example 76: A system according to any one of Examples 63 to 75, wherein the predicted geometry represents the geometry of the object after it has been fabricated using the additive manufacturing process and after undergoing post-processing operations.

[0237] Example 77: According to the system of Example 76, the post-processing operation includes one or more of the following: centrifuging the object, post-curing the object, or washing the object.

[0238] Example 78: A system according to any one of Examples 63 to 77, wherein the at least one image corresponds to at least one 2D cross-section of a 3D digital representation of the object.

[0239] Example 79: A system according to any one of Examples 63 to 78, wherein the modification is performed using an optimization algorithm.

[0240] Example 80: The system according to Example 79, wherein the optimization algorithm is a reverse optimization algorithm.

[0241] Example 81: The system according to any one of Examples 63 to 80, wherein the operation further comprises: generating at least one second image, the at least one second image representing the predicted geometry of the object.

[0242] Example 82: The system according to Example 81, wherein the deviation is identified based on the at least one image and the at least one second image.

[0243] Example 83: The system according to Example 82, wherein the deviation is identified by comparing the at least one image with the at least one second image.

[0244] Example 84: A system according to any one of Examples 63 to 83, wherein the object includes a dental appliance.

[0245] Example 85: The system according to Example 84, wherein the dental appliance is an orthodontic appliance, palatal expander, retainer, attachment placement device, or oral protector.

[0246] Example 86: A non-transitory computer-readable medium comprising instructions that, when executed by one or more processors of a computing system, cause the computing system to perform operations including: Receive at least one image, the at least one image representing the target geometry of an object to be manufactured using an additive manufacturing process; Based on the at least one image, generate a predicted geometry of the object after it has been fabricated using the additive manufacturing process; Identify the deviation between the target geometry and the predicted geometry; Modify at least one image based on the identified deviations; and Instructions for manufacturing the object using the additive manufacturing process are generated based on the at least one modified image. in conclusion

[0247] Although many embodiments of the systems, apparatus, and methods for manufacturing dental appliances have been described above, the technology is applicable to other applications and / or other means, such as manufacturing other types of objects. Moreover, other embodiments besides those described herein are also within the scope of this technology. Additionally, several other embodiments of the technology may have configurations, components, or programs different from those described herein. Therefore, those skilled in the art will accordingly understand that the technology may have other embodiments with additional elements, or the technology may have embodiments without the foregoing reference. Figures 1 to 17 Other embodiments of some of the features shown and described.

[0248] The various processes described herein can be implemented, partially or entirely, using program code comprising instructions executable by one or more processors of a computing system to implement specific logical functions or steps within the process. The program code can be stored on any type of computer-readable medium, such as storage devices including disks or hard disk drives. Computer-readable media including code or portions thereof can include any suitable medium known in the art, such as non-transitory computer-readable storage media. Computer-readable media can include volatile and non-volatile, removable and non-removable media implemented using any method or technology for storing and / or transmitting information, including, but not limited to, random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies; compact disc read-only memory (CD-ROM), digital video disc (DVD) or other optical storage devices; magnetic tape cassettes, magnetic tape, disk storage or other magnetic storage devices; solid-state drives (SSDs) or other solid-state storage devices; or any other medium that can be used to store desired information and is accessible by system devices.

[0249] The description of embodiments of the present technology is not intended to be exhaustive or to limit the technology to the precise forms disclosed above. Where the context permits, singular or plural terms may also include plural or singular terms, respectively. Although specific embodiments and examples of the present technology have been described above for illustrative purposes, various equivalent modifications are possible within the scope of the present technology, as will be recognized by those skilled in the art. For example, while the steps are presented in a given order, alternative embodiments may perform the steps in a different order. Various embodiments described herein may also be combined to provide other embodiments.

[0250] As used herein, the terms “generally,” “basically,” “about,” and similar terms are used as approximate terms rather than terms of degree and are intended to explain the inherent variations in measurements or calculations that would be recognized by one of ordinary skill in the art.

[0251] Furthermore, unless the word “or” is explicitly limited to referring only to a single item that excludes other items in a list of two or more items, its use in such a list is to be interpreted as including (a) any single item in the list, (b) all items in the list, or (c) any combination of items in the list. As used herein, the phrase “and / or” in “A and / or B” refers to A alone, B alone, and both A and B. Additionally, the term “including” is used throughout to mean at least one or more of the stated features, such that no further number of the same features and / or additional features of the same type are excluded.

[0252] In the event of any conflict between this disclosure and any material incorporated herein by reference, this disclosure shall prevail.

[0253] It should also be understood that specific embodiments have been described herein for illustrative purposes, but various modifications may be made without departing from the present invention. Furthermore, while advantages associated with certain embodiments of the present invention have been described in the context of those embodiments, other embodiments may also exhibit such advantages, and not all embodiments must exhibit such advantages to fall within the scope of the present invention. Therefore, this disclosure and related technologies may cover other embodiments not expressly shown or described herein.

Claims

1. A method comprising: Receive at least one image, the at least one image representing the target geometry of an object to be manufactured using an additive manufacturing process; At least one modified image is generated by inputting the at least one image into a machine learning algorithm, wherein the machine learning algorithm is trained to determine one or more modifications to the at least one image, and wherein the one or more modifications are configured to compensate for predicted deviations from the target geometry of the object when the object is manufactured via the additive manufacturing process based on the at least one image; as well as Instructions for manufacturing the object using the additive manufacturing process are generated based on the at least one modified image.

2. The method of claim 1, wherein the machine learning algorithm is trained based on initial image data of a plurality of additive manufacturing objects and corresponding modified image data.

3. The method according to claim 1 or 2, wherein the machine learning algorithm comprises a convolutional neural network (CNN).

4. The method according to any one of claims 1 to 3, wherein the one or more modifications are configured to compensate for predicted deviations from the target geometry of the object due to the additive manufacturing process, post-processing operations, or a combination thereof.

5. The method according to any one of claims 1 to 4, wherein the one or more modifications are configured to compensate for the predicted deviation from the target geometry of the object due to over-curing of the material used to make the object, over-construction of the material used to make the object, retention of material on the surface of the object, material loss from the object, deformation of the object, or a combination thereof.

6. The method according to any one of claims 1 to 5, wherein the one or more modifications include removing material from a portion of the object represented in the at least one image.

7. The method according to any one of claims 1 to 6, wherein the one or more modifications include adding material to a portion of the object represented in the at least one image.

8. The method according to any one of claims 1 to 7, wherein the additive manufacturing process comprises one or more of the following: stereolithography, digital light processing, selective laser sintering, material jetting, or material extrusion.

9. The method according to any one of claims 1 to 8, wherein the additive manufacturing process includes applying energy to a precursor material to form a plurality of object layers.

10. The method of claim 9, wherein the instructions are configured to control the application of the energy to the precursor material.

11. The method of claim 9 or 10, wherein the instructions are configured to form at least one object layer corresponding to the at least one modified image.

12. The method according to any one of claims 1 to 11, wherein the at least one image corresponds to at least one 2D cross-section of a 3D digital representation of the object.

13. The method according to any one of claims 1 to 12, further comprising: The predicted geometry of the object after it has been fabricated using the additive manufacturing process is determined based on the at least one modified image.

14. The method of claim 13, further comprising: Identify the deviation between the target geometry and the predicted geometry.

15. The method of claim 14, further comprising: Output an indication of the identified deviations.

16. The method according to any one of claims 1 to 15, further comprising: Evaluate whether the at least one modified image meets one or more quality control parameters.

17. The method of claim 16, wherein the evaluation includes: Detect artifacts, detect broken features, detect insufficiently supported features, detect features smaller than the minimum feature size, or combinations thereof.

18. The method according to claim 16 or 17, further comprising: In response to the fact that the at least one modified image does not meet the evaluation of the one or more quality control parameters, the at least one modified image is adjusted.

19. The method according to any one of claims 1 to 18, wherein the at least one modified image represents a modified geometry of the object that is different from the target geometry.

20. The method according to any one of claims 1 to 19, further comprising: The object is fabricated using the additive manufacturing process based on the instructions.

21. A system comprising: One or more processors; as well as A memory operatively coupled to one or more processors and storing instructions that, when executed by the one or more processors, cause the system to perform operations including: Receive at least one image, said at least one image representing the target geometry of an object to be manufactured using an additive manufacturing process. At least one modified image is generated by inputting the at least one image into a machine learning algorithm, wherein the machine learning algorithm is trained to determine one or more modifications to the at least one image, and wherein the one or more modifications are configured to: compensate for predicted deviations from the target geometry of the object when the object is fabricated via the additive manufacturing process based on the at least one image, and Instructions for manufacturing the object using the additive manufacturing process are generated based on the at least one modified image.

22. The system of claim 21, wherein the machine learning algorithm is trained based on initial image data of a plurality of additive manufacturing objects and corresponding modified image data.

23. The system of claim 21 or 22, wherein the machine learning algorithm comprises a convolutional neural network (CNN).

24. The system according to any one of claims 21 to 23, wherein the one or more modifications are configured to compensate for predicted deviations from the target geometry of the object due to the additive manufacturing process, post-processing operations, or a combination thereof.

25. The system according to any one of claims 21 to 24, wherein the one or more modifications are configured to compensate for predicted deviations from the target geometry of the object due to over-curing of the material used to make the object, over-construction of the material used to make the object, retention of material on the surface of the object, material loss from the object, deformation of the object, or a combination thereof.

26. The system according to any one of claims 21 to 25, wherein the one or more modifications include removing material from a portion of the object represented in the at least one image.

27. The system according to any one of claims 21 to 26, wherein the one or more modifications include adding material to a portion of the object represented in the at least one image.

28. The system according to any one of claims 21 to 27, wherein the additive manufacturing process comprises one or more of the following: stereolithography, digital light processing, selective laser sintering, material jetting, or material extrusion.

29. The system according to any one of claims 21 to 28, wherein the additive manufacturing process includes applying energy to a precursor material to form a plurality of object layers.

30. The system of claim 29, wherein the instructions are configured to control the application of the energy to the precursor material.

31. The system of claim 29 or 30, wherein the instructions are configured to cause the formation of at least one object layer corresponding to the at least one modified image.

32. The system according to any one of claims 21 to 31, wherein the at least one image corresponds to at least one 2D cross-section of a 3D digital representation of the object.

33. The system according to any one of claims 21 to 32, further comprising: The predicted geometry of the object after it has been fabricated using the additive manufacturing process is determined based on the at least one modified image.

34. The system of claim 33, further comprising: Identify the deviation between the target geometry and the predicted geometry.

35. The system according to claim 34, further comprising: Output an indication of the identified deviations.

36. The system according to any one of claims 21 to 35, further comprising: Evaluate whether the at least one modified image meets one or more quality control parameters.

37. The system of claim 36, wherein the evaluation includes: Detect artifacts, detect broken features, detect insufficiently supported features, detect features smaller than the minimum feature size, or combinations thereof.

38. The system according to claim 36 or 37, further comprising: In response to the fact that the at least one modified image does not meet the evaluation of the one or more quality control parameters, the at least one modified image is adjusted.

39. The system according to any one of claims 21 to 38, wherein the at least one modified image represents a modified geometry of the object that is different from the target geometry.

40. The system according to any one of claims 21 to 39, further comprising: The object is fabricated using the additive manufacturing process based on the instructions.

41. A method comprising: Receive at least one image, the at least one image representing the target geometry of an object to be manufactured using an additive manufacturing process; The predicted geometry of the object after it has been fabricated using the additive manufacturing process is determined based on the at least one image. Identify the deviation between the target geometry and the predicted geometry; Modify at least one image based on the identified deviations; as well as Instructions for manufacturing the object using the additive manufacturing process are generated based on the at least one modified image.

42. The method of claim 41, wherein the prediction is determined using a machine learning algorithm.

43. The method of claim 42, wherein the machine learning algorithm comprises a convolutional neural network (CNN).

44. The method of claim 43, wherein the CNN is trained using images of other objects, which are fabricated using the additive manufacturing process.

45. The method of claim 43 or 44, wherein the CNN includes a kernel, and wherein the kernel includes a function representing a physical or chemical phenomenon associated with the additive manufacturing process.

46. ​​The method of claim 45, wherein the physical or chemical phenomenon includes light scattering.

47. The method of any one of claims 42 to 46, wherein the machine learning algorithm is configured to predict deviations from the target geometry of the object due to the additive manufacturing process, post-processing operations, or a combination thereof.

48. The method of any one of claims 42 to 47, wherein the machine learning algorithm is configured to predict deviations from the target geometry of the object due to over-curing of the material used to make the object, over-construction of the material used to make the object, retention of material on the surface of the object, material loss from the object, deformation of the object, or a combination thereof.

49. The method according to any one of claims 41 to 48, wherein the additive manufacturing process comprises one or more of the following: stereolithography, digital light processing, selective laser sintering, material jetting, or material extrusion.

50. The method according to any one of claims 41 to 49, wherein the additive manufacturing process includes applying energy to a precursor material to form a plurality of object layers.

51. The method of claim 50, wherein the instructions are configured to control the application of the energy to the precursor material.

52. The method of claim 50 or 51, wherein the instructions are configured to cause the formation of at least one object layer corresponding to the at least one modified image.

53. The method according to any one of claims 41 to 52, wherein the predicted geometry represents the geometry of the object after it has been fabricated using the additive manufacturing process and after undergoing post-processing operations.

54. The method of claim 53, wherein the post-processing operation includes one or more of the following: centrifuging the object, post-curing the object, or washing the object.

55. The method according to any one of claims 41 to 54, wherein the at least one image corresponds to at least one 2D cross-section of a 3D digital representation of the object.

56. The method according to any one of claims 41 to 55, wherein the modification is performed using an optimization algorithm.

57. The method according to claim 56, wherein the optimization algorithm is a reverse optimization algorithm.

58. The method according to any one of claims 41 to 57, further comprising: Generate at least one second image, the at least one second image representing the predicted geometry of the object.

59. The method of claim 58, wherein the deviation is identified based on the at least one image and the at least one second image.

60. The method of claim 59, wherein the deviation is identified by comparing the at least one image with the at least one second image.

61. The method according to any one of claims 41 to 60, wherein the object comprises a dental appliance.

62. The method of claim 61, wherein the dental appliance is an orthodontic appliance, a palatal expander, a retainer, an attachment placement device, or an oral protector.

63. A system comprising: One or more processors; as well as A memory operatively coupled to one or more processors and storing instructions that, when executed by the one or more processors, cause the system to perform operations including: Receive at least one image, said at least one image representing the target geometry of an object to be manufactured using an additive manufacturing process. The predicted geometry of the object after it has been fabricated using the additive manufacturing process is determined based on the at least one first image. Identify the deviation between the target geometry and the predicted geometry. Modify at least one image based on the identified deviations, and Instructions for manufacturing the object using the additive manufacturing process are generated based on the at least one modified image.

64. The system of claim 63, wherein the at least one second image is generated using a machine learning algorithm.

65. The system of claim 64, wherein the machine learning algorithm comprises a convolutional neural network (CNN).

66. The system of claim 65, wherein the CNN is trained using images of other objects, which are fabricated using the additive manufacturing process.

67. The system of claim 65 or 66, wherein the CNN includes a kernel, and wherein the kernel includes a function representing a physical or chemical phenomenon associated with the additive manufacturing process.

68. The system of claim 67, wherein the physical or chemical phenomenon includes light scattering.

69. The system according to any one of claims 64 to 68, wherein the machine learning algorithm is configured to predict deviations from the target geometry of the object due to the additive manufacturing process, post-processing operations, or a combination thereof.

70. The system of any one of claims 64 to 69, wherein the machine learning algorithm is configured to predict deviations from the target geometry of the object due to over-curing of the material used to make the object, over-construction of the material used to make the object, retention of material on the surface of the object, material loss from the object, deformation of the object, or a combination thereof.

71. The system according to any one of claims 63 to 70, wherein the additive manufacturing process comprises one or more of the following: stereolithography, digital light processing, selective laser sintering, material jetting, or material extrusion.

72. The system according to any one of claims 63 to 71, wherein the additive manufacturing process includes applying energy to a precursor material to form a plurality of object layers.

73. The system of claim 72, wherein the instructions are configured to control the application of the energy to the precursor material.

74. The system according to claim 72 or 73, wherein the instructions are configured to cause the formation of at least one object layer corresponding to the at least one modified image.

75. The system according to any one of claims 72 to 74, further comprising an additive manufacturing system, wherein the additive manufacturing system includes: An energy source, configured to output the energy; The source of the precursor material; as well as A controller configured to cause the energy source to apply energy to the precursor material according to the instructions.

76. The system according to any one of claims 63 to 75, wherein the predicted geometry represents the geometry of the object after it has been fabricated using the additive manufacturing process and after undergoing post-processing operations.

77. The system of claim 76, wherein the post-processing operation includes one or more of the following: centrifuging the object, post-curing the object, or washing the object.

78. The system according to any one of claims 63 to 77, wherein the at least one image corresponds to at least one 2D cross-section of a 3D digital representation of the object.

79. The system according to any one of claims 63 to 78, wherein the modification is performed using an optimization algorithm.

80. The system according to claim 79, wherein the optimization algorithm is a reverse optimization algorithm.

81. The system according to any one of claims 63 to 80, wherein the operation further comprises: Generate at least one second image, the at least one second image representing the predicted geometry of the object.

82. The system of claim 81, wherein the deviation is identified based on the at least one image and the at least one second image.

83. The system of claim 82, wherein the deviation is identified by comparing the at least one image with the at least one second image.

84. The system according to any one of claims 63 to 83, wherein the object comprises a dental appliance.

85. The system of claim 84, wherein the dental appliance is an orthodontic appliance, a palatal expander, a retainer, an attachment placement device, or an oral protector.

86. A non-transitory computer-readable medium comprising instructions that, when executed by one or more processors of a computing system, cause the computing system to perform operations including: Receive at least one image, the at least one image representing the target geometry of an object to be manufactured using an additive manufacturing process; Based on the at least one image, generate a predicted geometry of the object after it has been fabricated using the additive manufacturing process; Identify the deviation between the target geometry and the predicted geometry; Modify at least one image based on the identified deviations; as well as Instructions for manufacturing the object using the additive manufacturing process are generated based on the at least one modified image.

Images