Artificial intelligence-based cavity diagnosis

Abstract

Providing devices for diagnosing the condition of teeth. [Solution] The device captures image data representing the patient's teeth based on data obtained from a hardware device that scans the teeth. The device inputs the image data into a first supervised machine learning model and receives multiple biomarkers from the first supervised machine learning model as output, each biomarker corresponding to a different location on the tooth. The device inputs the multiple biomarkers into a second supervised machine learning model and receives a diagnosis of the tooth condition from the second supervised machine learning model as output.

Description

Background Art

[0001] (Cross - reference to related applications) This application claims the benefit of U.S. Provisional Patent Application No. 63 / 219,216, filed on December 17, 2021, the content of which is incorporated herein by reference in its entirety.

[0002] (Background) The risk of dental caries (cavities) is widespread throughout society, and anyone is at risk of dental caries, especially by a diet rich in carbohydrates and ultra - processed foods. Dental caries are a major cause of disability, pain, and medical costs. Effective treatments for preventing and improving dental caries exist, especially when it is found at an early stage and is still asymptomatic. However, compliance with early diagnosis is not optimal when the standard treatment is a dental examination by a dentist, leading to unnecessary progression of dental caries. Dental examinations by dentists and additionally by X - ray as standard treatment are effective for early diagnosis of dental caries, but due to the need for either a highly skilled professional (i.e., a dentist) with requirements that vary greatly by state or a highly skilled radiographer, as well as the exposure to X - rays that can have long - term health effects, and in addition, the relatively high cost, scaling this is difficult due to the need for expert evaluation through some forms of telemedicine that reduce access.

Summary of the Invention

Means for Solving the Problems

[0003] (Abstract) Therefore, autonomous artificial intelligence (AI) technology for dental screening that a) does not require X-rays and b) does not require expertise to evaluate images would be highly beneficial in increasing access, reducing costs, and thereby avoiding permanent damage and costs. Specifically, deep penetration optical coherence tomography (DPOCT) is a non-radioactive optical technique based on an interferogram of low-coherence light to penetrate up to 4-5 mm into tooth elements, including molars, allowing all surfaces of each element, including any cavities on all surfaces, to be imaged. In particular, it allows imaging of so-called interdental cavities between two elements, where 60% of cavities occur and which are inaccessible by visible examination. Autonomous AI is a technology that enables real-time point-of-care diagnosis of cavities from multi-plane DPOCT images.

[0004] Since DPOCT cannot penetrate the entire tooth, especially the larger molars, imaging of each accessible surface may be required. This can be achieved through multi-plane DPOCT using a small probe or multiple probes covering at least the facial, lingual, and occlusal surfaces of the molar arch, illuminating each plane with low-coherence light using a so-called sweep light source or a Michelson interferometry approach in domain OCT interference fringes that allows for high-speed scanning of the entire element, and collecting the low-coherence light reflected to the DPOCT device through three or more parallel fiber optics that calculate the interference image. It is obtained by taking one or more B scans (2-sided scans) from these three planes coaxial with the probe, and as the probe is advanced back and forth manually or servo-driven, a multi-plane 3D PDOCT volume is collected. In one example, the probe is applied to the other three rows of molars to compute multi-sided images of all molar elements in the patient.

[0005] The autonomous AI takes a multi-faceted DPOCT volume as input and outputs either the likelihood of cavities in the volume or a diagnostically binary or multi-stage presence or severity of cavities in the volume. Other diagnostically relevant outputs may also be made. The AI ​​can be constructed as a hybrid partially independent biomarker multi-detector AI with a fusion stage, or as a multi-image-based convolutional neural network using deep learning, shallow learning, recurrent networks, or any combination of such AI designs, as are well known to those skilled in the art. An exemplary AI of this technique is disclosed in further detail in U.S. Patent No. 10,115,194, jointly owned, issued October 30, 2018, which is incorporated herein by reference in its entirety. Extension and transfer learning approaches, including unsupervised transfer learning, may also be used. DPOCT training images labeled with the presence or location of cavities or their likelihood or probability can be used to train such AI systems, or, in the case of a multi-detector-based approach, labeled image samples can be used to train detectors, or mathematical descriptions of biomarkers can be used to design such detectors. DPOCT is merely illustrative, and whenever DPOCT or any other imaging technique is mentioned, any other imaging technique that captures the functionality described herein with respect to DPOCT (e.g., any other waveform generation technique using low-coherence light, sound, vibration, or interferometry) is equally applicable. Furthermore, any low-coherence tomography (LCT) technique is also equally applicable where DPOCT is mentioned herein, and includes ultraviolet, infrared, dental penetrant or deep-penetration optical coherence tomography, microwave, visible-light LCT, ultrasound imaging techniques and similar. The present invention provides, for example, the following: (Item 1) A method for diagnosing the condition of teeth, wherein the method is Based on data obtained from a hardware device that scans the patient's teeth, image data representing the teeth is captured. The aforementioned image data is input into the first supervised machine learning model, The output is to receive multiple biomarkers from the first supervised machine learning model, each biomarker corresponding to a different position on the tooth. The above-mentioned multiple biomarkers are input into a second supervised machine learning model, The output is to receive a diagnosis of the tooth condition from the second supervised machine learning model. A method that includes [a certain feature]. (Item 2) The method according to item 1, wherein the data obtained from the hardware device for scanning the tooth comprises deep penetration optical coherence tomography (DPOCT) data, and the image data comprises an intensity map that reflects one or more optical properties of the tooth based on the DPOCT data. (Item 3) The method according to item 2, wherein the first supervised machine learning model is trained to detect changes in intensity between regions in the intensity map and to output each classification for each different location of the tooth based on the changes in intensity, and each classification forms a biomarker. (Item 4) The method according to item 3, wherein the first supervised machine learning model is further trained to take the tooth color image as input, to output a biomarker based on both the image and the intensity map, and the output of the first supervised machine learning model removes color data from the color image. (Item 5) The method according to item 1, further comprising accessing the historical biomarker of the tooth, wherein the historical biomarker is input into the second supervised machine learning model together with the plurality of biomarkers, and the second supervised machine learning model outputs the diagnosis based on both the historical biomarker and the plurality of biomarkers. (Item 6) The method according to item 5, wherein inputting the historical biomarker together with the plurality of biomarkers into the second supervised machine learning model comprises calculating the intensity difference of each of the plurality of biomarkers with respect to the historical biomarker, and inputting each intensity difference into the second supervised machine learning model. (Item 7) The method according to item 1, wherein the output includes receiving a diagnosis of the tooth condition from the second supervised machine learning model, receiving multiple diagnoses, and each of the multiple diagnoses corresponds to one different of the multiple biomarkers. (Item 8) A non-temporary computer-readable medium comprising a memory having instructions encoded therein for diagnosing the condition of teeth, wherein, when executed, the instructions cause one or more processors to perform actions, and the instructions Based on data obtained from a hardware device that scans the patient's teeth, image data representing the teeth is captured. The aforementioned image data is input into the first supervised machine learning model, The output is to receive multiple biomarkers from the first supervised machine learning model, where each biomarker corresponds to a different position on the tooth. The above-mentioned multiple biomarkers are input into a second supervised machine learning model, The output is to receive a diagnosis of the tooth condition from the second supervised machine learning model. A non-temporary, computer-readable medium equipped with instructions to perform a certain action. (Item 9) A non-transient computer-readable medium according to item 8, wherein the data obtained from the hardware device for scanning the tooth comprises deep penetration optical coherence tomography (DPOCT) data, and the image data comprises an intensity map that reflects one or more optical properties of the tooth based on the DPOCT data. (Item 10) The non-transient computer-readable medium described in item 9, wherein the first supervised machine learning model is trained to detect changes in intensity between regions in the intensity map and to output each classification for each different location of the tooth based on the changes in intensity, and each classification forms a biomarker. (Item 11) The non-temporary computer-readable medium described in item 10, wherein the first supervised machine learning model is further trained to take the tooth color image as input, to output a biomarker based on both the image and the intensity map, and the output of the first supervised machine learning model excludes color data from the color image. (Item 12) The method further comprises accessing the tooth history biomarker, the history biomarker being input into the second supervised machine learning model together with the plurality of biomarkers, and the second supervised machine learning model outputting the diagnosis based on both the history biomarker and the plurality of biomarkers, in a non-temporary computer-readable medium as described in item 8. (Item 13) A non-temporary computer-readable medium as described in item 12, wherein the command to input the historical biomarker along with the plurality of biomarkers into the second supervised machine learning model comprises a command to calculate the intensity difference of each biomarker among the plurality of biomarkers with respect to the historical biomarker, and inputting the respective intensity differences into the second supervised machine learning model. (Item 14) A non-temporary computer-readable medium as described in item 8, wherein the command to receive the diagnosis of the tooth condition from the second supervised machine learning model as output comprises a command to receive multiple diagnoses, and each of the multiple diagnoses corresponds to one different of the multiple biomarkers. (Item 15) A system, wherein the system is A memory containing encoded instructions for diagnosing the condition of teeth, 1 or more processors The processors, comprising the above 1 or more processors, perform an operation when executing the instruction, and the operation is Based on data obtained from a hardware device that scans the patient's teeth, image data representing the teeth is captured. The aforementioned image data is input into the first supervised machine learning model, The output is to receive multiple biomarkers from the first supervised machine learning model, where each biomarker corresponds to a different position on the tooth. The above-mentioned multiple biomarkers are input into a second supervised machine learning model, The output is to receive a diagnosis of the tooth condition from the second supervised machine learning model. A system that includes these features. (Item 16) The system according to item 15, wherein the data obtained from the hardware device for scanning the tooth comprises deep penetration optical coherence tomography (DPOCT) data, and the image data comprises an intensity map that reflects one or more optical properties of the tooth based on the DPOCT data. (Item 17) The system according to item 16, wherein the first supervised machine learning model is trained to detect changes in intensity between regions in the intensity map and to output a classification for each different location of the tooth based on the changes in intensity, and each classification forms a biomarker. (Item 18) The system according to item 17, wherein the first supervised machine learning model is further trained to take the tooth color image as input, to output a biomarker based on both the image and the intensity map, and the output of the first supervised machine learning model removes color data from the color image. (Item 19) The operation further comprises accessing the historical biomarker of the tooth, the historical biomarker is input into the second supervised machine learning model together with the plurality of biomarkers, and the second supervised machine learning model outputs the diagnosis based on both the historical biomarker and the plurality of biomarkers. The system according to item 15. (Item 20) Inputting a historical biomarker together with the plurality of biomarkers into the second supervised machine learning model comprises calculating the intensity difference of each biomarker of the plurality of biomarkers with respect to the historical biomarker and inputting each intensity difference into the second supervised machine learning model. The system according to item 19.

Brief Description of Drawings

[0006] [Figure 1] FIG. 1 illustrates one embodiment of a system environment for implementing a dental caries determination tool.

[0007] [Figure 2] FIG. 2 illustrates one embodiment of an exemplary module and database used by a treatment decision tool in the use of artificial intelligence that directly outputs treatment.

[0008] [Figure 3] FIG. 3 shows one embodiment of further details of a submodule used by the dental caries determination module of a treatment decision tool.

[0009] [Figure 4] FIG. 4 shows one embodiment of a DPOCT intensity map.

[0010] [Figure 5] FIG. 5 illustrates an exemplary process for determining a diagnosis of a dental condition using a supervised machine learning approach.

[0011] The figures illustrate various embodiments of the present invention for illustrative purposes only. Those skilled in the art will readily recognize from the following discussion that alternative embodiments of the structures and methods illustrated herein may be used without departing from the principles of the invention described herein. [Modes for carrying out the invention]

[0012] (Detailed explanation) (a) Overview Figure 1 illustrates one embodiment of a system environment for implementing a caries determination tool. As depicted in Figure 1, the environment 100 includes an imaging device 110, a network 120, a caries determination tool 130, and patient data 140. The elements of environment 100 are illustrative, and fewer or more elements may be incorporated into environment 100 to achieve the functionality disclosed herein.

[0013] The imaging device 110 may be any device for capturing images of the patient's teeth. The imaging device 110 may include a camera sensor, a DPOCT sensor, and any other long-wave sensors capable of capturing anything indicating dental problems and caries. The imaging device 110 may acquire credentials and other information about the patient. This information may be entered by a technician through a user interface or directly by the patient. In other words, the imaging device 110 may include a keyboard or touchscreen interface for entering data about the patient, and in addition, may include one or more imaging components for capturing images. The imaging components may be any known imaging components (e.g., a slider slid over and around the teeth by a technician). The imaging device 110 may capture images and prompts for patient information based on running software controlled by or otherwise distributed by the caries determination tool 130. Because the imaging device 110 uses non-X-ray technology, it can be operated by persons who are not qualified to operate X-ray equipment (e.g., technicians with minimal training and high school education), thus enabling scalability for caries detection.

[0014] The imaging device 110 may use multi-planar OCT technology in addition to the DPOCT sensor. For example, since DPOCT cannot completely penetrate some teeth, multi-planar OCT may be used to capture images showing caries that DPOCT may not be able to capture. The imaging device 110 may use multi-planar OCT to measure captured images (e.g., 2D images) simultaneously or sequentially and convert these images into 3D or other multi-volume measurements of any parameter representation of caries (or defects) at any given point on the tooth. The AI ​​disclosed herein may process multiple 3D volumes or other multi-volumes captured together in this technique. For example, if multiple teeth or multiple sections of teeth are scanned together, the 3D or other multi-volume images may be captures of those multiple teeth and / or multiple sections of teeth, and may be processed considering each other if the information in adjacent volumes is bound to each other (e.g., by occlusion).

[0015] Network 120 can be any data network capable of transmitting data communication between the client device 110 and the treatment decision tool 130. Network 120 can be, for example, the Internet, a local area network, a wide area network, or any other network.

[0016] The caries determination tool 130 receives and processes images from the imaging device 110. The caries determination tool 130 may also receive other patient information from the imaging device 110 and may retrieve information about the patient from the patient data 140. Based on some or all of the received information, the caries determination tool 130 may output the patient's condition. Further details of the operation of the caries determination tool 130 are discussed below with reference to Figure 2. The operation of the treatment determination tool 130 can be instantiated on the imaging device 110, either as a whole or as a part (for example, through an application installed on the imaging device 110 or accessed by the imaging device 110 through a browser).

[0017] Patient data 140 is a database that stores records of data for one or more patients. Patient data 140 may include hospital records, personal patient records, physician notes, and similar. Patient data 140 may be located in the same location as the imaging instrument 110 and / or the caries determination tool 130. In an exemplary embodiment, patient data 140 may include profile information about the patient (e.g., age, sex, race, etc.), previous images of the patient's teeth captured by the imaging instrument 110 (if any), and previous conditions detected by the caries determination tool 130 (if any). Patient data 140 may be integrated with a dental practice where the dentist may indicate a decision (e.g., filling a cavity) or a decision that no treatment was necessary (e.g., a false positive) in an examination prompted by a condition identified by the caries determination tool 130. Patient data 140 may be used as a baseline for detecting the progression of caries when images of the patient's teeth are captured over time. Patient data 140 may be used to retrain the machine learning model used by the caries determination tool 130 (for example, relabeling the training data to indicate that a false positive is indicated by a dentist, indicating that it does not correspond to a caries condition). In some embodiments, the use of patient data 140 is optional (for example, if patient data 140 may contain noise that degrades the accuracy or other performance of the AI ​​model).

[0018] Figure 2 illustrates one embodiment of exemplary modules and databases used by a caries determination tool in the use of artificial intelligence to output the condition of teeth. As depicted in Figure 2, the caries determination tool 130 includes an image capture module 231, a caries determination module 232 and an image preprocessing module 233, as well as a machine learning module 241. The modules and databases depicted in Figure 2 are illustrative only, and the caries determination tool 130 may include more or fewer modules and / or databases and still achieve the functionality described herein. Furthermore, the modules and / or databases of the treatment determination tool 130 may be instantiated as a whole or as part on a client device 110 and / or one or more servers.

[0019] The image capture module 231 performs activities related to causing the imaging instrument 110 to capture information and transmit that information to the caries determination tool 130 and / or the patient data database 140. This activity may include instructing the imaging instrument 110 to capture images of the patient's teeth (e.g., via a locally installed application or via remote commands) and transmit them to the image capture module 231 for processing. In response to receiving the images, the image capture module 231 optionally performs pre-processing on the images (e.g., using the image pre-processing module 233). Pre-processing may include optimizing the images, removing artifacts, ensuring that the images conform to one or more protocols (e.g., that the images show certain teeth they are expected to show), and similar.

[0020] In one embodiment, the image capture module 231 prompts the technician to perform certain activities. For example, the technician may be prompted by the image capture module 231 to capture an image of a tooth, a gum region, slide the instrument to capture an image from a certain angle, or perform any other operation of the image capture device. Images captured between actions corresponding to these prompts are validated through a pre-processing protocol that verifies images conforming to the protocol, and in response to them not conforming to the protocol, the image capture module 231 may prompt the technician to recapture such images.

[0021] The caries determination module 232 determines one or more conditions of the patient based on the captured image. In one embodiment, the caries determination module 232 determines a condition that is a binary decision: whether or not the patient may have a caries. That is, the output of the caries determination module 232 is either that the patient may have a caries and should see a dentist, or that the patient does not have a caries and does not need to see a dentist. In other embodiments, the caries determination module 232 may output a diagnosis (e.g., the upper right incisor has a caries through the enamel). In addition or alternatively, the caries determination module 232 may output a treatment (e.g., a filling is needed on the upper right incisor). In addition or alternatively, the caries determination module 232 may output the urgency of the risk (e.g., a binary "urgent" or "not urgent", a scale from 1 to 10 increasing in proportion to urgency, or any other method of representation) based on how likely it is that the caries will become severe within a threshold period (e.g., a filling is needed now, but root canal treatment will be needed in one month). In addition or alternatively, the caries determination module 232 may output any risks associated with the caries (e.g., developing periodontal disease if the caries is left untreated).

[0022] The caries determination module 232 can make these decisions by using machine learning. In one embodiment, one or more longwave images (e.g., DPOCT images) are obtained by the image capture module 231 and input into the machine learning model by the caries determination module 232. The machine learning model is trained with training data that labels the images into one or more states (e.g., training images are labeled to indicate whether a dental examination is needed, whether a specific diagnosis exists, urgency, whether a certain treatment is needed, risk, and similar). The machine learning model may be a neural network (e.g., a convolutional neural network), a deep learning network, or any other type of machine learning model. The machine learning model may be trained with other auxiliary data, such as patient profile information paired with patient images, and may accept given patient profile data in addition to longwave images as input. Optical images showing the tooth surface, gum surface, and similar may be used as input to the machine learning model or as parts thereof (e.g., tooth boundary polygons).

[0023] A single-model approach can be accurate but may tolerate bias. For example, if a captured image contains gum pigment, the gum pigment can form the basis for training a model to determine a certain state. For this purpose, two machine learning models may be used. A first machine learning model may accept an image as input and output, for each region of a tooth, the likelihood that the tooth contains a certain property (e.g., caries, discoloration, previous fillings or sealants, or other dental treatments, etc.) or some other representation of the image (e.g., intensity at each location in the image). A second machine learning model may take a likelihood (or other representation) vector for a given tooth or set of teeth as input and be trained to output a state based on that vector. This eliminates the risk that irrelevant factors will bias the state determination. In some embodiments, the likelihood vector may be supplemented with auxiliary information, such as a visible light image of the periodontal tissue.

[0024] In one embodiment, patient data 140 may be input to a model in association with long-wave images. For example, a baseline image from a previous image may be input, or image processing may be used to identify the difference between the baseline image and the new image. The difference may be input into the machine learning model(s) to output disease status. The machine learning model may be trained using historical data of the same tooth at different time points, and differences in the intensity of biomarkers over time are labeled to indicate the severity of the condition (e.g., worsening or improvement of cavities, worsening of discoloration, etc.).

[0025] Figure 3 shows one embodiment of further details of the submodules used by the caries determination module of the treatment decision tool. As depicted in Figure 3, the submodules of the caries determination module 232 may include the DPOCT abstraction module 315, the intensity map module 325, the biomarker determination module 335, and the caries diagnosis module 345. Fewer or more modules may be used to achieve the functionality described herein.

[0026] As described above, in one embodiment, the caries determination module 232 inputs an image into a machine learning model. However, in addition or intentionally, one or more abstractions of the DPOCT image may be input into the machine learning model. Abstractions as described herein may refer to representations of an image. For example, a DPOCT image may contain pixels of varying intensities. An abstraction may include the difference in intensity between adjacent pixels in each region of the image, and the intensity difference may be input into the machine learning model. In one embodiment, the DPOCT abstraction module 315 may determine regions where there is at least a threshold intensity difference (e.g., mean intensity difference or intermediate intensity difference) between those regions and adjacent pixels, and may use a map of those regions as input into the machine learning model. Ground truth training data may exist for any form of abstraction that labels the abstractions. In embodiments where a single machine learning model is used, the training data may include labels for caries diagnoses corresponding to the abstractions. In embodiments where an image is input into a first machine learning model, the abstractions may be labeled with ground truth for any of the aforementioned output classifications.

[0027] The intensity map 325 can generate one or more intensity maps for an image (e.g., a DPOCT image). An intensity map can indicate the intensity at each pixel of an image for a given property (e.g., reflectance, refractive index, brightness). These intensity maps can be used for any purpose (e.g., to abstract intensity differences, for direct input into machine learning models, and similar). Generally, healthy dental materials do not absorb DPOCT scans very well. However, caries or other materials (e.g., fillings) absorb relatively more and may have higher intensity. Therefore, ground truth data labeling the intensity maps with classifications (e.g., caries, discoloration, previous fillings) can enable machine learning models to output classifications based on the intensity maps.

[0028] In one embodiment, the intensity map module 325 may determine the intensity map using the difference in intensity between the intensity of the most recent image and the intensity of a previous image from the patient data 140 (for example, the difference in intensity of the current image compared to a previous image of the same tooth).

[0029] The caries determination module 232 can input full-color images, DPOCT images, and any of the mentioned abstractions, either individually or together, into either a machine learning model trained to output a diagnosis directly, or a machine learning model trained to output a likelihood for a tooth region containing a certain property. If the machine learning model is trained to take two or more of these inputs together, the ground truth data for each type of input is utilized to optimally determine the relevant classification.

[0030] In one embodiment, the biomarker determination module 335 applies heuristics to an intensity map and / or an abstraction showing intensity differences, the heuristics identify parameters for the size of a region and / or intensity changes of a region that are classified as biomarkers. These biomarkers can be input into a single machine learning model that outputs a diagnosis based on the biomarkers. In an embodiment where two machine learning models are used, the first machine learning model can be trained to detect intensity changes from the intensity map and output each classification for each location on the tooth based on the intensity changes. From the classifications and optionally other heuristics (e.g., size of location), the biomarker determination module 335 can determine the likelihood that each region with a given intensity change forms a particular type of biomarker (e.g., cavity, previous filling, etc.). The biomarker determination module 335 can compare these likelihoods to previous likelihoods from patient data 140 (e.g., from images taken in a previous scan during an early examination of the same tooth), and this comparison can be used as input to a second machine learning model. The biomarker determination module 335 may, in addition or alternatively, annotate biomarkers using the intensity differences of biomarkers from early scans of patients 140 for input to a second machine learning model.

[0031] The caries diagnosis module 345 generates diagnoses for one or more areas of a tooth. In one embodiment, the caries diagnosis module 345 determines a diagnosis for each biomarker of the tooth. That is, a machine learning model (e.g., a single model or a second machine learning model in a two-model approach) may output the probability that the tooth (or biomarker) has one or more conditions (e.g., cavity, filling, discoloration, etc.) for either the tooth as a whole or for different biomarkers of the tooth. The caries diagnosis module 345 may apply a threshold such that a probability higher than a minimum threshold leads to the caries diagnosis module 345 deciding that the corresponding condition is applied as a diagnosis. Thus, a tooth with three biomarkers (one for cavity, one for filling, and one for discoloration) may have each of those biomarkers, which are diagnosed individually in connection with its location on the tooth. Alternatively, a tooth may be superficially labeled to have each of its three conditions without labeling different areas of the tooth accordingly. Finally, the caries diagnosis module 345 can output a diagnostic abstraction. For example, heuristics can be applied to indicate whether a diagnosed condition should be classified as “see dentist” or “no problem detected.” For instance, a classification indicating discoloration and previous fillings in a tooth might lead to “no problem detected” based on those heuristics, while any finding of a cavity or pre-existing cavity material might lead to a classification of “see dentist.”

[0032] Figure 4 shows one embodiment of the DPOCT intensity map. As shown in Figure 4, image 400 shows a tooth color image and a corresponding intensity map. These intensity maps can be used as training data for experts to point to labels (e.g., enamel, dentin, secondary caries, cracks, and composite fillings). Labels can be for any type of tooth, such as healthy molars versus reconstructed molars. Patient data 140 can be used to determine the type of tooth (e.g., patient data 140 determines that a molar has been reconstructed, and then the machine learning model fits the reconstructed molar to the ground truth data). Different machine learning models can be trained for different types of teeth, and the caries determination module 232 can select a machine learning model that fits the patient's tooth type so that it is determined from patient data 140 to diagnose the condition for a given tooth. This can improve efficiency and reduce noise in the output of the machine learning model by using a better target set of ground truth data.

[0033] Figure 5 illustrates an exemplary process for determining a diagnosis of a tooth condition using a supervised machine learning approach. As depicted in Figure 5, process 500 begins with the caries determination tool 130 capturing an image data display of the patient's teeth 502 (e.g., using an image capture module 231) based on data obtained from a hardware device that scans the teeth. In one embodiment, the data obtained from the hardware device that scans the teeth may include deep penetration optical coherence tomography (DPOCT) data. The image data may include, or may be, intensity maps that reflect one or more optical properties of the teeth based on the DPOCT data.

[0034] The caries determination module 130 inputs image data into a first supervised machine learning model (for example, from machine learning model 241) 504. In one embodiment, the first supervised machine learning model may be trained to detect changes in intensity between regions in an intensity map and to output a classification for each different location of the tooth based on the changes in intensity, each of which forms a biomarker. In one embodiment, the first supervised machine learning model may also take in a tooth color image as input and output a biomarker based on both the image and the intensity map.

[0035] The caries determination module 130 receives multiple biomarkers from a first supervised machine learning model as output 506, each biomarker corresponding to a different location on a tooth. If the input includes a color image, the output of the first machine learning model may exclude color data from the color image (e.g., to eliminate bias in the use of the second machine learning model). The caries determination module 130 inputs the multiple biomarkers into the second supervised machine learning model 508 and receives a diagnosis of the tooth condition from the second supervised machine learning model as output 510. In one embodiment, a history of tooth biomarkers (e.g., from patient data 140 in which previous images have been taken and decisions made from the first and / or second models) is retrieved and input into the second machine learning model. In one embodiment, intensity information is input into the model, and the intensity difference of the current intensity relative to the previous intensity may be input into the second machine learning model, as changes in intensity over time may be useful for some classification. In one embodiment, the diagnosis may include a classification of each biomarker for a given tooth.

[0036] The disclosure herein refers to human patients, but patients may not be human. Dogs, horses, and the like may have caries that can be treated with the systems and methods disclosed herein. For example, training data may be fed into the teeth of animals to train one or more animal-specific machine learning models for making condition diagnoses. In one embodiment, the machine learning models may be trained using cross-species training data. The ground truth in any of these training sets may be based on one or more of the following: outcomes from treatment, pain, and clinician labeling.

[0037] (b) Summary The foregoing description of embodiments of the present invention has been presented for illustrative purposes only. It is not intended to be exhaustive or to limit the invention to the specific shapes disclosed. Those skilled in the art will understand that many modifications and variations are possible in consideration of the above disclosure.

[0038] Several parts of this description describe embodiments of the present invention with respect to algorithms and symbolic representations of operations on information. These descriptions and representations of algorithms are commonly used by those skilled in the art in data processing techniques to efficiently communicate the nature of the work to others skilled in the art. These operations are described functionally, computationally, or logically, but are understood to be implemented by computer programs or equivalent electronic circuits, microcode, or similar. Furthermore, without loss of generality, it has also proven convenient at times to refer to these arrangements of operations as modules. The operations described and their associated modules may be embodied in software, firmware, hardware, or any combination thereof.

[0039] Any of the steps, operations, or processes described herein may be performed or implemented using one or more hardware or software modules, either alone or in combination with other devices. In some embodiments, the software module is implemented using a computer program product comprising a computer-readable medium containing computer program code, which can be executed by a computer processor to perform any or all of the steps, operations, or processes described herein.

[0040] Embodiments of the present invention may also relate to apparatus for performing the operations described herein. Such apparatus may be specifically constructed for a particular purpose and / or may comprise a general-purpose computing device that is selectively invoked or reconfigured by a computer program stored in a computer. Such a computer program may be stored in a non-temporary, tangible, computer-readable storage medium or any kind of medium suitable for storing electronic instructions, which may be coupled to a computer system bus. Furthermore, any computing system referenced herein may comprise a single processor or may be an architecture employing multiple processor designs for increased computing power.

[0041] Embodiments of the present invention may also relate to products produced by the computational processes described herein. Such products may contain information derived from the computational processes, which is stored on a non-temporary, tangible, computer-readable storage medium and may include any embodiment of a computer program product or any combination of other data described herein.

[0042] Finally, the language used herein has been chosen in principle for readability and teaching purposes, and may not be chosen to describe or limit the subject matter of the invention. Therefore, the scope of the invention is intended to be limited not by this detailed description, but rather by any claims to be issued in an application based on this specification. In addition, the disclosure of embodiments of the invention is illustrative but not intended to be restrictive of the scope of the invention, as stated in the following claims.

Claims

1. A method for diagnosing the condition of teeth, wherein the method is performed on a computer having one or more processors, and the method is The aforementioned one or more processors capture deep penetration optical coherence tomography (DPOCT) image data from a tooth scan, The one or more processors capture the tooth color image, The process involves one or more processors inputting the DPOCT image data and the color image into a first supervised machine learning model, the first supervised machine learning model being trained to detect changes in intensity between regions in the image data and to output classifications for each different position of the tooth based on the changes in intensity, and each classification forming a biomarker. The one or more processors receive multiple biomarkers as output from the first supervised machine learning model, each biomarker corresponding to a different location on the tooth. The one or more processors input the multiple biomarkers into a second supervised machine learning model, The one or more processors receive a diagnosis of the condition of the teeth as output from the second supervised machine learning model. Methods that include...

2. The method according to claim 1, wherein the DPOCT data is obtained from a hardware device that scans the teeth.

3. The method according to claim 1, wherein each of the classifications is obtained by removing color data from the color image.

4. The method according to claim 1, further comprising the one or more processors accessing the historical biomarkers of the tooth, the historical biomarkers of the tooth being input together with the plurality of biomarkers into the second supervised machine learning model.

5. The method according to claim 4, wherein the second supervised machine learning model outputs the diagnosis based on both the historical biomarker and the plurality of biomarkers.

6. The method according to claim 5, wherein inputting the history biomarker together with the plurality of biomarkers into the second supervised machine learning model includes one or more processors calculating the intensity difference of each biomarker of the plurality of biomarkers with respect to the history biomarker and inputting each intensity difference into the second supervised machine learning model.

7. The method according to claim 1, wherein receiving the diagnosis of the tooth condition as output from the second supervised machine learning model comprises one or more processors receiving a plurality of diagnoses, each of the plurality of diagnoses corresponding to one different of the plurality of biomarkers.

8. A non-temporary computer-readable medium comprising a memory having instructions encoded thereon for diagnosing the condition of teeth, wherein, when executed, the instructions cause one or more processors to perform an action, This involves capturing deep penetration optical coherence tomography (DPOCT) image data from a dental scan, The aforementioned tooth color image is captured, The DPOCT image data and the color image are input into a first supervised machine learning model, the first supervised machine learning model is trained to detect changes in intensity between regions in the image data and to output classifications for each different position of the tooth based on the changes in intensity, and each classification forms a biomarker. The process involves receiving multiple biomarkers as output from the first supervised machine learning model, where each biomarker corresponds to a different location on the tooth. The above-mentioned multiple biomarkers are input into a second supervised machine learning model, The output from the second supervised machine learning model mentioned above is a diagnosis of the condition of the teeth. A non-temporary, computer-readable medium containing instructions for performing certain actions.

9. The DPOCT data is obtained from a hardware device for scanning teeth, in a non-temporary computer-readable medium according to claim 8.

10. The non-temporary computer-readable medium according to claim 8, wherein each of the classifications excludes color data from the color image.

11. The non-temporary computer-readable medium according to claim 8, wherein the instruction further comprises an instruction for accessing the historical biomarkers of the tooth, the historical biomarkers of the tooth, together with the plurality of biomarkers, are input into the second supervised machine learning model.

12. The non-temporary computer-readable medium according to claim 11, wherein the second supervised machine learning model outputs the diagnosis based on both the historical biomarker and the plurality of biomarkers.

13. The non-temporary computer-readable medium according to claim 12, wherein the instruction for inputting the chronological biomarker together with the plurality of biomarkers into the second supervised machine learning model includes an instruction for calculating the intensity difference of each biomarker of the plurality of biomarkers with respect to the chronological biomarker, and inputting each intensity difference into the second supervised machine learning model.

14. The non-temporary computer-readable medium according to claim 8, wherein the instruction for receiving the diagnosis of the tooth condition as output from the second supervised machine learning model comprises an instruction for receiving a plurality of diagnoses, each of the plurality of diagnoses corresponds to a different one of the plurality of biomarkers.

15. A system, wherein the system is A memory having encoded instructions on it for diagnosing the condition of teeth, One or more processors and Equipped with, When the one or more processors execute the instruction, they perform an operation, and the operation is: This involves capturing deep penetration optical coherence tomography (DPOCT) image data from a dental scan, The aforementioned tooth color image is captured, The DPOCT image data and the color image are input into a first supervised machine learning model, the first supervised machine learning model is trained to detect changes in intensity between regions in the image data and to output classifications for each different position of the tooth based on the changes in intensity, and each classification forms a biomarker. The process involves receiving multiple biomarkers as output from the first supervised machine learning model, where each biomarker corresponds to a different location on the tooth. The above-mentioned multiple biomarkers are input into a second supervised machine learning model, The output from the second supervised machine learning model mentioned above is a diagnosis of the condition of the teeth. A system that includes this.

16. The system according to claim 15, wherein the DPOCT data is obtained from a hardware device that scans the teeth.

17. The system according to claim 15, wherein each of the classifications excludes color data from the color image.

18. The system according to claim 15, wherein the operation further comprises accessing the historical biomarkers of the tooth, the historical biomarkers of the tooth, together with the plurality of biomarkers, are input into the second supervised machine learning model.

19. The system according to claim 18, wherein the second supervised machine learning model outputs the diagnosis based on both the historical biomarker and the plurality of biomarkers.

20. The system according to claim 19, wherein inputting the chronological biomarker together with the plurality of biomarkers into the second supervised machine learning model includes calculating the intensity difference of each biomarker of the plurality of biomarkers with respect to the chronological biomarker, and inputting each intensity difference into the second supervised machine learning model.

Images