Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
DETAILED ACTION
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102 of this title, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries set forth in Graham v. John Deere Co., 383 U.S. 1, 148 USPQ 459 (1966), that are applied for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1-4, 6-7, 9-20 are rejected under 35 U.S.C. 103 as being unpatentable over Cramer et al. (US Pub 2022/0262007 A1) in view of Parpara et al. (Us Pub 2019/0102880 A1).
As to claim 1, Cramer discloses a computer-implemented method for training one or more machine learning models to automatically validate (¶0045, “may be useful in planning and fabrication of dental appliances, including elastic polymeric positioning appliances” ¶0120, “Fabrication machine 822 fabricates dental appliances based on intermediate and final data set information acquired from data processing system 800.”), the method comprising:
receiving, by one or more computer processors, a first digital 3D oral care representation of a (¶0062, ““graph-based segmentation elements” can refer to classes of features or structures within the 3D models that can be used to identify, label, and segment the 3D models into individual dental components, including individual teeth, interproximal spaces between teeth, and/or gingiva.” ¶0098, “automatically receive three-dimensional (3D) meshes of various patients' dental arches.”);
receiving, by the one or more computer processors, a second digital 3D oral care representation of a (¶0023, “receiving, in the computing device, a ground truth input comprising a manual segmentation of the 3D mesh; and training a machine learning model of the computing device produce a segmentation output that attempts to achieve the ground truth input.” ¶0087, “a technician can evaluate 2D or 3D scan data, or a 3D model of patients' dentitions, and manually identify “graph-based segmentation elements” in the data.” “The technician's evaluation can then be input into the machine learning model as the “ground truth”, which is the desired output of the machine learning model for a given input.”);
using, by the one or more computer processors, one or more machine learning models that have been partially trained to assign one or more labels to the first digital representation, wherein at least one of the one or more labels specifies whether one or more aspects of the first digital representation is correctly formed (¶0101, ““graph-based segmentation elements” can refer to classes of features or structures within the 3D models that can be used to identify, label, and segment the 3D models into individual dental components, including individual teeth, interproximal spaces between teeth, and/or gingiva. Multiple training cases comprising a “ground truth” can be input into the machine learning model to further train the model. In some examples, the weights of the machine learning model can be adjusted while training to minimize the error between the ground truth and the output of the model.” ¶0064, “all features linked to a segmentation element label of “teeth” can be identified as teeth, all features linked to a segmentation element label of “interproximal spaces”, can be identified as interproximal spaces, and all features linked to a segmentation element label of “gingiva” can be identified as gingiva.” ¶0070, “the machine learning engine(s) 164 can be used to automatically segment the 3D model while the machine learning model is being trained.” ¶0091, “The distinct individual objects can be, for example, individually numbered or labeled in the output. The output from the segmentation engine may be used to automatically and accurately label the individual teeth of the 3D model”);
determining, by the one or more computer processors, whether the one or more aspects of the first digital representation is substantially similar to the one or more aspects of the second representation based at least in part on a comparison between the first digital representation and the second digital representation (¶0025, “Training the machine learning model may include adjusting weights of the machine learning model to minimize an error between the ground truth input and the segmentation output.” ¶0123, “segmentation by graph-based representation of the 3D mesh may be used to monitor, track and/or grade a margin line, e.g., of a tooth crown. An actual margin line may be determined using graph-based representation of the 3D mesh to segment the patient's dentition and to determine the margin line; this margin line may be compared to a target or desired margin line that may be set as a ground truth for the graph-based representation of the 3D mesh.”); and
automatically training, by the one or more computer processors, the machine learning model based on the results of the comparison (¶0096, “Multiple training cases comprising computed features can be input into the machine learning model to further train the model. In some examples, the inputs are used to train a submanifold convolutional neural network to recognize “graph-based segmentation elements”.” ¶0121, “comparing the results of a second segmentation technique for the same 3D data with a graph-based representation of the 3D mesh segmentation technique. For example, the same dataset may be segmented by both graph-based representation of the 3D mesh as described herein and a second segmentation technique (e.g., sparse voxel representation) and the resulting segmentations may be compared over sub-regions of the 3D data. The results of the comparison may be thresholded (e.g., within +/−1%, 2%, 3%, 4%, 5%, 10%, 15%, 20%, 25%, etc.), indicating regions of high and low confidence for the segmentation. Regions of low correlation may be re-segmented or segmented using a third technique, and/or may be flagged for intervention by a technician.” ¶0123, “segmentation by graph-based representation of the 3D mesh may be used to monitor, track and/or grade a margin line, e.g., of a tooth crown. An actual margin line may be determined using graph-based representation of the 3D mesh to segment the patient's dentition and to determine the margin line; this margin line may be compared to a target or desired margin line that may be set as a ground truth for the graph-based representation of the 3D mesh.”).
Even though Cramer discloses a dental appliance, Cramer does not explicitly disclose “a fixture model”.
Parpara discloses a fixture model (Parpara, abstract, “inspecting a customized orthodontic aligner for manufacturing defects, the customized orthodontic aligner customized for a specific arch of a specific patient and a specific stage of orthodontic treatment”” obtaining images of the customized orthodontic aligner; identifying an identifier of the customized orthodontic aligner, determining a digital file based on the identifier, the digital file associated with the customized orthodontic aligner including a digital model of a mold used during manufacture of the customized orthodontic aligner, determining a intended property for the customized orthodontic aligner by digitally manipulating the digital model of the mold, determining an actual property of the customized orthodontic aligner from the images, determining whether there is a manufacturing defect in the customized orthodontic aligner by comparing the intended property with the actual property, outputting an output associated with the determination of whether there is a manufacturing defect.” ¶0037, “determining a intended property for the dental appliance by digitally manipulating the digital model of the intermediate component used during manufacture of the dental appliance” ¶0097, “generate a digital model for the aligners, perform analysis that compares the digital model of an aligner with the image of the aligner to detect one or more quality issues (e.g., deformation, cutline variation, etc.), and classify aligners based on results of the analysis.” ¶0104, “The comparison may include computing a projection of the approximated outer surface (e.g., digital model) of the aligner into the same plane as the shape of the aligner and a region between contours of the approximated outer surface and the shape of the aligner in the image is identified.”).
Cramer and Parpara are considered to be analogous art because all pertain to dental appliance. It would have been obvious before the effective filing date of the claimed invention to have modified Cramer with the features of “a fixture model” as taught by Parpara. The suggestion/motivation would have been in order to determine whether there is a manufacturing defect in the customized orthodontic aligner by comparing the intended property with the actual property (Parpara, abstract.)
As to claim 2, claim 1 is incorporated and the combination of Cramer and Parpara discloses at least one of the one or more machine learning models is trained to detect one or more flaws in the first digital representation including excess material in the gums, one or more cracks, one or more chips, an undercut base, one or more kinks in the associated trimline, excess block out, excess interproximal webbing, one or more missing teeth, one or more erroneously present hardware representations (Cramer, ¶0048, “The apparatuses and/or methods described herein may be used to segment a patient's teeth from a three-dimensional model, such as a 3D mesh model or a 3D point cloud, and this segmentation information may be used to simulate, modify and/or choose between various orthodontic treatment plans.” ¶0050, “identify and/or number individual teeth and/or dental features of virtual representations of teeth, such as teeth represented in a three-dimensional dental mesh model resulting from a digital scan.” Parpara, ¶0099, “If an additional image is a side view of the plastic shell, bubbles that are present on a surface of the plastic shell may be detected and/or an inaccurate cutline may be detected.“ ¶0168, “a) digital files of a first set of plastic aligners with labels indicating whether or not each of the first set of plastic aligners experienced one or more defects or b) digital files of a second set of plastic aligners with labels indicating whether or not each of the second set of plastic aligners include one or more probable defects. Actual defects for aligners may be reported by manufacturing technicians, by an automated manufacturing system and/or by patients. Such historical data on actual defects on physical aligners may then be added as labels or metadata to the associated digital files of the aligners and/or images of the aligners. ¶0209, “A defect region for the aligner may be determined to be between the first line 1300 and the second line 1302. One or more measurements may be obtained from the defect region, as described above. For example, a thickness of the defect region, an area of the defect region, etc.” ¶0211, “deforming a digital model contour to more closely match the contour of the image of the aligner to detect other manufacturing defects (e.g., cutline variations, debris, webbing, trimmed attachments, and missing attachments, etc.),”).
As to claim 3, claim 1 is incorporated and the combination of Cramer and Parpara discloses generating, by the one or more computer processors, one or more suggestions of how to correct the first digital representation (Parpara, ¶0008, “identify one or more high risk areas for the one or more defects at one or more locations of the plastic shell.” ¶0055, “the proposed digital modifications to the digital model of the intermediate component include at least one of added virtual filler material, revisions to a cutline, and modifications to one or more attachments of the intermediate component.” ¶0219, “generating a modified projection by deforming an approximated outer surface contour to more closely match the contour of the image of the aligner to detect cutline variations”); and
outputting the generated suggestions, wherein the suggestions are only generated and outputted when it is determined that one or more aspects of the first digital representation are not correctly formed (Parpara, ¶0054, “the output includes a determination that a defect is present, and the output includes proposed digital modifications to the digital model of the intermediate component used during manufacture of the dental appliance to limit future defects.” ¶0055, “the proposed digital modifications to the digital model of the intermediate component include at least one of added virtual filler material, revisions to a cutline, and modifications to one or more attachments of the intermediate component.”).
As to claim 4, claim 1 is incorporated and the combination of Cramer and Parpara discloses one or more two dimensional (2D) representations are generated based on at least in part the first representation (Cramer, ¶0054, “The results may include 3D virtual representations of the dental arch, 2D images or renditions of the dental arch, etc.” ¶0072, “the 3D model can be rendered into one or more 2D image(s) from a plurality of viewing angles.”).
As to claim 6, claim 1 is incorporated and the combination of Cramer and Parpara discloses the one or more machine learning models have been trained to classify one or more 3D oral care representations (Cramer, ¶0055, ““Graph-based segmentation elements”, as used herein, can refer to classes of features or structures within a graph-based representation of the 3D mesh that can be used to identify, label, and segment the 3D mesh into individual dental components, including individual teeth, interproximal spaces between teeth, and/or gingiva.” ¶0064, “This style of network can be implemented with a graph-based representation that results in a network that accurately finds a 3D mask of each tooth and classifies each tooth. Alternatively, each instance proposal could be classified with a separate approach. In this implementation, the 3D model can be segmented using an instance segmentation combined with an instance classification.” ¶0091.).
As to claim 7, claim 1 is incorporated and the combination of Cramer and Parpara discloses at least one of the one or more machine learning models is a neural network (Cramer, ¶0007, “using machine learning neural networks.” “convolutional neural networks” ¶0011, ¶0043, ¶0065).
As to claim 9, claim 1 is incorporated and the combination of Cramer and Parpara discloses automatically generating, by the one or more computer processors, output that specifies whether the first digital representation can be used to generate an apparatus using thermoforming (Parpara, ¶0004, “The aligners may then be formed over the molds using thermoforming equipment.” ¶0018, “simulating a process of thermoforming a film over a digital model of the mold by enlarging the digital model of the mold into an enlarged digital model” ¶0117, “a sheet of material is pressure formed or thermoformed over the mold. The sheet may be, for example, a sheet of plastic (e.g., an elastic thermoplastic, a sheet of polymeric material, etc.). To thermoform the shell over the mold, the sheet of material may be heated to a temperature at which the sheet becomes pliable.”).
As to claim 10, claim 9 is incorporated and the combination of Cramer and Parpara discloses when it is determined that the first digital representation cannot be used to generate the apparatus using thermoforming, performing, by the one or more computer processors, the method of claim 1 (Parpara, ¶0004. ¶0142-0143, “processing logic may classify the plastic shell as defective and specify one or more remedies (e.g., add filler material, smooth cutline, modifications to one or more attachments on the mold or attachment cavities of the dental appliance, remanufacture, etc.) to attempt to remove the one or more defects” ¶0146, “The first image may be applied to a trained machine learning model that is trained to identify high risk areas for defects, or to a rules engine that includes rules specifying that certain features at locations indicate high risk areas for defects.” ¶0171, “these simulations may be run numerous times on multiple digital files of plastic aligners and labels may be included with the digital files indicating whether or not the digital files include one or more probable points of failure.”).
As to claim 11, claim 10 is incorporated and the combination of Cramer and Parpara discloses the apparatus is an indirect bonding tray or an orthodontic aligner (Cramer, ¶0008, “Automatic tooth segmentation may provide the basis for implementation of automated orthodontic treatment plans, design and/or manufacture of orthodontic aligners (including series of polymeric orthodontic aligners that provide forces to correct malocclusions in patients' teeth).”).
As to claim 12, claim 1 is incorporated and the combination of Cramer and Parpara discloses the determining comprises computing a loss value that quantifies one or more differences between the first representation and the second representation (Cramer, ¶0025, “Training the machine learning model may include adjusting weights of the machine learning model to minimize an error between the ground truth input and the segmentation output.”).
As to claim 13, claim 1 is incorporated and the combination of Cramer and Parpara discloses the first representation is a predicted representation (Parpara, ¶0101, “the model may be a predictive model that performs a numerical simulation on the digital file of the plastic aligner by applying one or more forces on the plastic aligner to simulate a removal process of the plastic aligner from a dental arch of a patient.”).
As to claim 14, claim 13 is incorporated and the combination of Cramer and Parpara discloses the predicted representation is generated by one or more machine learning models (Parpara, ¶0101, “the digital file of the plastic shell may be applied to a model (e.g., predictive model, machine learning model, etc.)” ¶0168, “digital files of aligners with associated probable defects (as provided by an output of a numerical simulation) may be used together to generate a robust machine learning model that can predict probable defects of new aligners from digital files of those aligners.”).
As to claim 15, claim 1 is incorporated and the combination of Cramer and Parpara discloses the second representation is a ground truth representation (Cramer, ¶0023, “a ground truth input comprising a manual segmentation of the 3D mesh; and training a machine learning model of the computing device produce a segmentation output that attempts to achieve the ground truth input.”).
As to claim 16, the combination of Cramer and Parpara discloses a system comprising: one or more computer processors; non-transitory computer-readable storage having stored thereon one or more machine learning models and instructions that when executed by the one or more processors cause the one or more processors to: receive a first digital 3D oral care representation of a fixture model; receive a second digital 3D oral care representation of a fixture model; determine whether the one or more aspects of the first digital representation is substantially similar to the one or more aspects of the second representation based at least in part on a comparison between the first digital representation and the second digital representation; automatically generate output that specifies whether the one or more aspects of the first digital representation is substantially similar to one or more aspects of the second digital representation; and automatically train the machine learning model based on the results of the comparison (See claim 1 for detailed analysis.).
As to claim 17, claim 16 is incorporated and the combination of Cramer and Parpara discloses the one or more machine learning models have been trained to classify one or more 3D oral care representations (See claim 6 for detailed analysis.).
As to claim 18, claim 16 is incorporated and the combination of Cramer and Parpara discloses at least one of the one or more machine learning models is trained to detect one or more flaws in the first digital representation including excess material in the gums, one or more cracks, one or more chips, an undercut base, one or more kinks in the associated trimline, excess block out, excess interproximal webbing, one or more missing teeth, one or more erroneously present hardware representations (Parpara, ¶0099, “a shape of the aligner, a size of the aligner, one or more features of the aligners, areas of higher risk for defects, one or more defects (e.g., deformation of the aligners, crack, etc.)”.
As to claim 19, claim 16 is incorporated and the combination of Cramer and Parpara discloses the instructions when executed by the one or more processors further cause the one or more processors to: generate one or more suggestions of how to correct the first digital representation; and output the generated suggestions, wherein the suggestions are only generated and outputted when it is determined that the one or more aspects of the first digital representation are not correctly formed (Parpara, ¶0101, “The machine learning model may be applied to a first image of the aligner and/or a digital file of the aligner and may generate an output indicating one or more high risk areas for defects at locations on the aligner.” ¶0143, “classify the plastic shell as defective and specify one or more remedies (e.g., add filler material, smooth cutline, modifications to one or more attachments on the mold or attachment cavities of the dental appliance, remanufacture, etc.) to attempt to remove the one or more defects.”).
As to claim 20, claim 16 is incorporated and the combination of Cramer and Parpara discloses the instructions when executed by the one or more processors further cause the one or more processors to generate output that specifies whether the first digital representation can be used to generate an apparatus using thermoforming (¶0143, “If it is determined at block 110 that there are one or more defects included in the plastic shell, then at block 112 processing logic may perform quality control for the plastic shell. For example, processing logic may classify the plastic shell as defective and specify one or more remedies (e.g., add filler material, smooth cutline, modifications to one or more attachments on the mold or attachment cavities of the dental appliance, remanufacture, etc.) to attempt to remove the one or more defects.” ¶0146, “a defect may be determined to be included in the plastic shell.”).
Claim 5 is rejected under 35 U.S.C. 103 as being unpatentable over Cramer et al. (US Pub 2022/0262007 A1) in view of Parpara et al. (Us Pub 2019/0102880 A1) and Farkash et al. (US Pub 2023/0309800 A1).
As to claim 5, claim 4 is incorporated and the combination of Cramer and Parpara discloses the one or more machine learning models are trained to (Parpara, ¶0229, “The digital model of the shell may result from the simulated trimming and the digital model may represent an outer surface of the aligner. The digital model of the aligner may be three-dimensional and various two-dimensional views (e.g., top view, side view) or three-dimensional views may be obtained using the digital model of the aligner.” ¶0231, “).
The combination of Cramer and Parpara does not disclose to classify the one or more 2D representations.
Farkash teaches the one or more machine learning models are trained to classify the one or more 2D representations (Farkash, ¶0263, “Tissue classifier 1152 may include one or more machine learning models that operate on 3D data or may include one or more machine learning models that operate on 2D data. If tissue classifier 1152 includes a machine learning model that operates on 2D data, then for each 3D model with labeled dental classes, a set of images (e.g., height maps) may be generated.” “If a machine learning model is being trained to perform image-level classification/prediction as opposed to pixel-level classification/segmentation, then a single value or label may be associated with a generated image as opposed to a map having pixel-level values.”).
Cramer, Parpara and Farkash are considered to be analogous art because all pertain to dental appliance. It would have been obvious before the effective filing date of the claimed invention to have modified Cramer with the features of “the one or more machine learning models are trained to classify the one or more 2D representations” as taught by Parpara. The claim would have been obvious because the technique for improving a particular class of devices was part of the ordinary capabilities of a person of ordinary skill in the art, in view of the teaching of the technique for improvement in other situations.
Claim 8 is rejected under 35 U.S.C. 103 as being unpatentable over Cramer et al. (US Pub 2022/0262007 A1) in view of Parpara et al. (Us Pub 2019/0102880 A1) and Saillet et al. (US Pub 2022/0414401 A1).
As to claim 8, claim 1 is incorporated and the combination of Cramer and Parpara discloses at least one of the one or more machine learning models is iteratively trained and (Parpara, ¶0168, “A machine learning model may refer to a model artifact that is created by a training engine using training data (e.g., training input and corresponding target outputs). Training may be performed using a set of training data including at least one of a) digital files of a first set of plastic aligners with labels indicating whether or not each of the first set of plastic aligners experienced one or more defects or b) digital files of a second set of plastic aligners with labels indicating whether or not each of the second set of plastic aligners include one or more probable defects.” ¶0171, “a strain or stress threshold may be used during the numerical simulation to determine when a point on the digital design of the aligner will likely fail. In this way, the numerical simulation may operate as a predictive model that predicts probable defects on the digital file of the aligner by identifying one or more high risk areas for the defects.” “these simulations may be run numerous times on multiple digital files of plastic aligners and labels may be included with the digital files indicating whether or not the digital files include one or more probable points of failure. The digital files including the labels indicating whether the digital file includes the one or more probable defect may be used as input to train the machine learning model.”).
The combination of Cramer and Parpara does not disclose “is considered fully trained when the machine learning model accuracy achieves a predefined threshold”.
However, this feature is part of the ordinary capabilities of a person of ordinary skill in the art.
Saillet teaches the one or more machine learning models is considered fully trained when the machine learning model accuracy achieves a predefined threshold (Saillet, ¶0011, “The training dataset may be divided into one training set (from which the model may learn its logic) and a test set (with which an accuracy of the model may be tested). This process may often require numerous iterations, where a model will be trained for a portion of time, and then tested for the analyst to test progress, and then feed more training data based on the test (e.g., where the model is feed training data that corresponds to predictions for which the model failed an accuracy threshold). While this disclosure primarily relates to the training data, these iterations also often includes changing various model parameters relating that are configurable by the data scientist training the model. In this way, a trained operator may validate a model as being fully trained at accurately predicting the training data, at which point can be deployed to the production environment.”).
Cramer, Parpara and Saillet are considered to be analogous art because all pertain to dental appliance. It would have been obvious before the effective filing date of the claimed invention to have modified Cramer with the features of “the one or more machine learning models is considered fully trained when the machine learning model accuracy achieves a predefined threshold” as taught by Saillet. The claim would have been obvious because the technique for improving a particular class of devices was part of the ordinary capabilities of a person of ordinary skill in the art, in view of the teaching of the technique for improvement in other situations.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Roschin et al. (US Pub 2023/0320824 A1) discloses raw dental features, principal component analysis (PCA) features, and/or other features may be extracted and compared to those of other teeth, such as those obtained through automated machine learning systems. A classifier can identify and/or output probability that the 3D tooth model requires trimming.
Kearney et al. (US Pub 2021/0358123 A1) discloses a machine learning model is trained to predict pixel spacing, distance, and volumetric measurements.
Claessen et al. (US Pub 2021/0082184 A1) discloses automated 3D root shape prediction.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to YU CHEN whose telephone number is (571)270-7951. The examiner can normally be reached on M-F 8-5 PST Mid-day flex.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Xiao Wu can be reached on 571-272-7761. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/YU CHEN/Primary Examiner, Art Unit 2613
Prosecution Insights