Computer-implemented method for training a machine learning model to identify artefacts in a digital impression of a dental feature

A machine learning model trained on digital dental impressions automatically identifies and corrects artefacts, addressing the inefficiencies of manual correction and pre-scanning cleaning, enhancing precision and reducing patient discomfort.

WO2026234546A1PCT designated stage Publication Date: 2026-11-12ZAAMIGO AG
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
PCT/IB2026/054390
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2025-05-05
Filing Date
2026-05-04
Publication Date
2026-11-12

AI Technical Summary

Technical Problem

Digital dental impressions often contain artefacts from unwanted external elements, requiring manual correction by operators, which increases time and discomfort for patients, and existing digital solutions fail to effectively reduce these artefacts without invasive pre-scanning cleaning.

Method used

A machine learning model trained on datasets of digital impressions to identify and correct artefacts, allowing for automated artefact removal and correction, reducing the need for manual intervention and pre-scanning cleaning.

Benefits of technology

Automated artefact identification and correction in digital dental impressions enhance precision and reduce patient discomfort by minimizing manual cleaning and scanning time.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

Computer-implemented method for training a machine learning model to identify artefacts in a digital impression of a dental feature, the method comprising the following steps: - providing a plurality of datasets, each dataset comprising a plurality of digital impressions of the same dental feature taken at different points in time; - for each dataset, determining an estimated three-dimensional structure of the corresponding dental feature, from which any artefacts are absent, on the basis of the corresponding digital impressions; - training the machine learning model in a supervised way, on the basis of each dataset associated with the corresponding estimated three-dimensional structure as ground truth. The resulting trained machine learning model is able to identify, and possibly also remove and correct, any artefacts in any new digital impression provided, either in real time or as a post processing step.
Need to check novelty before this filing date? Find Prior Art

Description

[0001] COMPUTER- IMPLEMENTED METHOD FOR TRAINING A MACHINE LEARNING MODEL TO IDENTIFY ARTEFACTS IN A DIGITAL IMPRESSION OF A DENTAL FEATURE

[0002] The present invention relates to the technical field of dental technology, and refers to a computer-implemented method for training a machine learning model to identify artefacts in a digital impression of a dental feature . In the dental industry, the necessity of obtaining an estimate of the three-dimensional structure of a dental feature of a patient is well-known, for example in order to manufacture dental appliances (such as dental braces, clear aligners, retainers, implants, dentures and other prosthetics) especially tailored for the patient, or in order to help a dentist to diagnose dental diseases and devise corresponding treatment plans . Examples of dental features are one or more teeth, one or more parts of a tooth, one or more parts of a dental arch, and an entire dental arch.

[0003] Traditionally, the dental industry has relied heavily on physical dental impressions for obtaining an estimate of the three-dimensional structure of the dental features of a patient . These dental impressions are usually obtained by placing a semi-solid impression material (such as alginate) on a tray shaped so as to fit over the dental arch of the patient, and then placing the tray in the mouth over the arch: as soon as the impression material solidifies, it bears the physical negative imprint of the dental arch, from which a positive reproduction (such as a cast) can be formed obtaining the estimate of the three-dimensional structure of the dental features of the dental arch.This solution, however, is relatively invasive, since the tray should be kept in the mouth for several minutes in order to allow the impression material to completely solidify, creating discomfort for the patient and the risk of gag reflex . In addition, the retrieval of the impression material without deformation is a complicated operation which is difficult to perform properly and requires specific training. Finally, each physical dental impression requires a fresh amount of impression material, the preparation of which is time consuming. As an alternative, digital dental impressions are becoming more and more common in the dental industry. This technology usually involves the use of an intraoral scanner which is inserted into the mouth of the patient and captures a set of images that are used to reconstruct a digital impression of the observed dental feature : the digital impression contains three-dimensional structures that together form the estimated three-dimensional structure of the observed dental feature .

[0004] Actually, the digital dental impressions could be obtained by means of many different solutions, including active wavefront sampling, "quasi" confocal microscopy and active or passive stereo cameras .

[0005] These "digital" solutions allow to obtain the impressions in a less invasive and more rapid way. However, the digital impressions usually contain also the so-called artefacts, i . e . three-dimensional structures created by unwanted external elements which lie on the surface of the observed dental feature, that are erroneously considered to be part of the estimated three-dimensional structure of the dental feature . Theseartefacts could be created for example by temporary elements such as saliva, food residues, dental plaque or blood which lie on the surface of the observed dental feature, or by external elements such as the cheeks, the tongue or the dentist' s glove .

[0006] A digital impression containing one or more artefacts could lead for example to the manufacture of dental appliances which are not perfectly tailored for the patient . Therefore, nowadays the operator usually has to identify and manually remove and correct each artefact from the obtained digital impression: this solution inevitably increases the total time needed to obtain the final digital impression, and relies on the operator' s expertise in distinguishing the artefacts from the actual structure of the dental feature .

[0007] Further, the amount of saliva inside the mouth usually increases while the different images of the set are being captured (in particular for the lower dental arch) , leading to distorted digital impressions when the same point of the dental features appears in images captured at different moments .

[0008] Therefore, in order to reduce the number of artefacts and distortions beforehand, the operator is requested to carefully clean the teeth immediately before the impression is taken, and to continuously remove saliva and blood during the scan process, further increasing the total time needed to obtain the final digital impression and creating discomfort for the patient .

[0009] An obj ect of the present invention is to overcome the above-mentioned drawbacks and in particular to provide a machine learning model specifically trained to identify artefacts in a digital impression, thereforereducing or even completely avoiding the need for a manual intervention on the artefacts by the operator, for a prior cleaning of the teeth and for a continuous removal of saliva and blood during the scan process . Another obj ect of the present invention is to provide a computer-implemented method for estimating the three-dimensional structure of a dental feature, using the above-mentioned machine learning model .

[0010] Another obj ect of the present invention is to provide a device for estimating the three-dimensional structure of a dental feature, making use of the above-mentioned computer-implemented method.

[0011] Another obj ect of the present invention is to provide a method for producing a dental appliance suited for a patient, by first estimating the three-dimensional structure of a dental feature making use of the above-mentioned computer-implemented method.

[0012] These and other results are achieved according to the present invention providing a computer-implemented method for training a machine learning model according to claim 1, a computer-implemented method for estimating the three-dimensional structure of a dental feature according to claim 6, a device for estimating the three-dimensional structure of a dental feature according to claim 9, a method for producing a dental appliance according to claim 10, a computer program according to claim 11 and a computer program according to claim 12. Further preferred characteristics are the obj ect of the dependent claims .

[0013] The present invention will be now described, in an illustrative and non-limiting way.

[0014] In a first embodiment, the invention refers to acomputer-implemented method for training a machine learning model to identify artefacts in a digital impression of a dental feature .

[0015] First of all, this method comprises a data supplying step of providing a plurality of datasets, wherein each dataset comprises a plurality of digital impressions of the same dental feature taken at different points in time . The digital impressions of a single dataset can be taken for example at regular or irregular time intervals, and the time period between the taking of two consecutive impressions can be in the order of magnitude of minutes, hours, days, weeks, months or years . In this way, it is reasonable to expect that the artefacts included will be different for each digital impression, or at least that the same or similar artefacts will be located in different positions . However, the time period between the taking of two consecutive impressions should not be too long, in order to avoid tooth migrations and changes in the dimensions of the gums .

[0016] In any case, it is advantageous that, for each dataset, at least one digital impression includes at least one artefact .

[0017] The digital impressions could be obtained with any suitable technique, including active wavefront sampling, "quasi" confocal microscopy, active or passive stereo cameras and monocular depth estimation.

[0018] Afterwards, on the basis of the digital impressions of each single dataset, an estimated three-dimensional structure of the corresponding dental feature is determined (data preparation step) , where any artefacts are absent . This estimated three-dimensional structure is the best estimation of the actual three-dimensionalstructure of the dental feature, without any artefacts . The estimated three-dimensional structure of each dental feature can be determined in different ways .

[0019] In some embodiments the estimated three-dimensional structure of each dental feature is determined as comprising only the three-dimensional structures that are common to each digital impression; in this way, only the three-dimensional structures that are present in each one of the digital impressions are determined to be part of the estimated three-dimensional structure of the dental feature, while any three-dimensional structure that is absent from at least one of the digital impression is determined to be an artefact . This particular solution maximizes the probability of identifying all the artefacts from the estimated three-dimensional structure of the dental feature .

[0020] As an alternative, the estimated three-dimensional structure of each dental feature can be determined as comprising only the three-dimensional structures that are common to the majority of the digital impressions ; in this way, only the three-dimensional structures that are present in at least half of the digital impressions are determined to be part of the estimated three-dimensional structure of the dental feature . This particular solution reduces the impact on the estimated three-dimensional structure of the digital impressions erroneously lacking a portion of the actual dental feature .

[0021] As a further alternative, the estimated three-dimensional structure of each dental feature can be determined as the average of the three-dimensional structures in the digital impressions . In this way, allthe impressions are equally taken into consideration . In all the previous cases, the digital impressions of each dataset can be compared by rigidly aligning them to each other, or by aligning corresponding portions thereof, in order to determine the common three-dimensional structures .

[0022] Preferably, each dataset comprises at least three digital impressions of the same dental feature taken at different points in time : in this way, a more accurate estimated three-dimensional structure of the dental feature can be determined.

[0023] Once the estimated three-dimensional structure of the corresponding dental feature is determined for each dataset, the machine learning model is trained in a supervised way, on the basis of each dataset associated with the corresponding estimated three-dimensional structure, which acts as ground truth . In this way, the machine learning model is trained to associate to each digital impression of a dental feature a corresponding estimated three-dimensional structure of the dental feature without artefacts .

[0024] The resulting trained machine learning model is able to identify, and possibly also remove (i . e . delete them) and correct (i . e . recover the underlying three-dimensional structure of the dental feature) , any artefacts in any new digital impression provided, reducing or avoiding the need for a manual intervention on the artefacts by the operator, for a prior cleaning of the teeth and for a continuous removal of saliva and blood during the scan process . The trained machine learning model could also identify and possibly remove and correct any artefacts from a digital impression inreal time while the digital impression itself is taken with an intra-oral scanner, in order to provide the operator with a live estimated three-dimensional structure of the observed dental feature from which any artefacts are absent . As an alternative, the trained machine learning model could be used in a post processing step to identify and possibly remove and correct any artefacts from a complete digital impression previously taken .

[0025] The trained machine learning model could also be limited to only identify the artefacts, pointing them out and relying on manual intervention or another software to remove and correct them.

[0026] The expression "identify the artefacts" also comprises the case where the trained machine learning model makes the scanner' s software "ignore" the artefacts while the three-dimensional structure of the dental feature is being reconstructed, allowing to directly predict the real three-dimensional structure of the dental feature even if it is "hidden" behind any artefacts, without being influenced by the artefacts .

[0027] For example, the trained machine learning model, in use, may compare the new digital impression provided with the training digital impressions of the training datasets, determine which training digital impression is most similar to the new digital impression, and identify any artefacts in the new digital impression by identifying all the three-dimensional structures of the new digital impression that are included in the most similar training digital impression and are not included in the corresponding estimated three-dimensional structure, marking them as artefacts, and possibly removing andcorrecting them.

[0028] The present invention also relates to a computer-implemented method for estimating the three-dimensional structure of a dental feature, in which a digital impression of the dental feature is provided, any artefact is identified in the digital impression using the trained machine learning model, and the identified artefacts are removed and corrected.

[0029] The present invention also relates to a device for estimating the three-dimensional structure of a dental feature, comprising a scanner configured to take a digital impression of the dental feature and a central control unit programmed to identify any artefacts in the digital impression using the trained machine learning model and to remove and correct the identified artefacts . The present invention also relates to a method for producing a dental appliance suited for a patient, for example dental braces, clear aligners, retainers, implants, dentures or other prosthetics . In particular, this method comprises the steps of estimating the three-dimensional structure of a dental feature inside the mouth of the patient, by using the trained machine learning model to identify any artefacts in a digital impression of the dental feature and by removing and correcting the identified artefacts, and then creating digital data associated to the estimated three-dimensional structure, transferring the digital data to a manufacturing system, and manufacturing the dental appliance by means of the manufacturing system on the basis of the digital data . In this way, the dental appliance can be manufactured to be perfectly tailored for the patient .The manufacturing system could be a CAD / CAM system or a 3D printing system.

[0030] The present invention has been described in an illustrative and non-limiting way according to relative favourite embodiments, but it is clear that the person skilled in the art could perform many variations and modifications, all of which are within the scope of the invention .

Claims

CLAIMS1 ) Computer-implemented method for training a machine learning model to identify artefacts in a digital impression of a dental feature, the method comprising the following steps :providing a plurality of datasets, each dataset comprising a plurality of digital impressions of the same dental feature taken at different points in time;for each dataset, determining an estimated three-dimensional structure of the corresponding dental feature, from which any artefacts are absent, on the basis of the corresponding digital impressions ;- training the machine learning model in a supervised way, on the basis of each dataset associated with the corresponding estimated three-dimensional structure as ground truth.2 ) Method according to claim 1, wherein, for each dataset, the estimated three-dimensional structure of the corresponding dental feature is determined as comprising only the three-dimensional structures that are common to each digital impression.3) Method according to claim 1, wherein, for each dataset, the estimated three-dimensional structure of the corresponding dental feature is determined as comprising only the three-dimensional structures that are common to the majority of the digital impressions .4 ) Method according to claim 1, wherein, for each dataset, the estimated three-dimensional structure of the corresponding dental feature is determined as the average of the three-dimensional structures in the digital impressions .5) Method according to any of the preceding claims,wherein each dataset comprises at least three digital impressions of the same dental feature taken at different points in time .6) Computer-implemented method for estimating the three-dimensional structure of a dental feature, comprising the following steps :- providing a digital impression of the dental feature; - identifying any artefacts in the digital impression by means of a machine learning model trained according to any of claims 1 to 5;preferably removing and correcting the identified artefacts .7 ) Method according to claim 6, wherein the artefacts are identified and preferably removed and corrected in real time while the digital impression is being taken.8 ) Method according to claim 6, wherein the artefacts are identified and preferably removed and corrected as a post processing step performed on a digital impression already taken.9) Device for estimating the three-dimensional structure of a dental feature, comprising a scanner configured to take a digital impression of the dental feature and a central control unit programmed to carry out a method according to any of claims 6 to 8 .10) Method for producing a dental appliance suited for a patient, comprising the following steps :- estimating the three-dimensional structure of a dental feature inside the mouth of the patient, by means of a method according to any of claims 6 to 8 ;creating digital data associated to the estimated three-dimensional structure;transferring the digital data to a manufacturingsystem;- manufacturing the dental appliance by means of the manufacturing system on the basis of the digital data .11 ) Computer program comprising instructions which, when the program is executed by a computer, cause the computer to carry out the method of any of claims 1 to 5 .12 ) Computer program comprising instructions which, when the program is executed by a computer, cause the computer to carry out the method of any of claims 6 to 8 .