Materials detection for scanners having penetration capability

Abstract

A method for imaging an intraoral sample using optical coherence tomography (OCT) or ultrasound to acquire image data content from multiple scans of the sample using a hand-held, depth-penetrating scanner and reconstruct a 3D surface according to the acquired measurements. For coordinate locations on the reconstructed 3D surface, the method repeats a sequence of identifying scan data related to material at the coordinate location; computing material content according to a consensus of the identified scan data; and labeling the coordinate location according to the computed consensus for the material content. The method records the label assignments and displays, stores, or transmits the labeled, reconstructed image content.

Description

[0001] MATERIALS DETECTION FOR SCANNERS HAVING PENETRATION CAPABILITY

[0002] TECHNICAL FIELD

[0003] The disclosure relates generally to hand-held imaging and more particularly to scanning apparatus and methods using penetrating signals in order to identify material composition of intraoral features.

[0004] BACKGROUND

[0005] Non-invasive imaging apparatus that use a low-energy penetrating signal can obtain beneficial information about features both on and below the surface of an object, which can be highly useful for various types of intraoral imaging applications. One such low-energy imaging apparatus is the ultrasound system, that measures a depth profile with sound energy that penetrates through soft tissues and materials and provides useful information on structures and features that lie behind the surface of an examined object.

[0006] Optical coherence tomography (OCT) is a non-invasive imaging technique that employs scanning of low levels of light energy and interferometric principles in order to obtain high resolution, cross-sectional tomographic images that characterize the depth structure of a sample. Particularly suitable for in vivo imaging of human tissue, OCT has shown its usefulness in a range of biomedical research and medical imaging applications, such as in ophthalmology, dermatology, oncology, and other fields, as well as in ear-nose- throat (ENT) and dental imaging.

[0007] OCT has been described as a type of "optical ultrasound", imaging reflected energy from within living tissue to obtain cross-sectional data. In an OCT imaging system, light from a wide-bandwidth source, such as a super luminescent diode (SLD) or other light source, is directed along two different optical paths: a reference arm or path of known optical path length and a sample arm or path that illuminates the tissue or other subject under study. Reflected and back-scattered light from the reference and sample arms is then recombined in the OCT apparatus and interference effects are used to determine characteristics of the surface and near-surface underlying structure of the sample. Interference data can be acquired by rapidly scanning the illumination across the sample. At each of several thousand points along the sample surface, the OCT apparatus obtains an interference profde which can be used to reconstruct an A-scan with an axial depth into the material that is largely a factor of light source coherence. For most tissue imaging applications, OCT uses broadband illumination sources and can provide image content at depths of up to a few millimeters (mm). Visual observation of a tooth is constrained by scatter from multiple sources within the tooth sample. OCT, however, effectively ignores scattered light, allowing the OCT measurement to reveal sub-surface content without the need for ionizing radiation.

[0008] There are significant limitations to the various techniques and approaches that have been applied to the problems of intraoral imaging. Constraints on camera and scanner size and form factor and the confined space requirements of the intraoral imaging environment make it challenging to accurately characterize intraoral surfaces. It can be difficult to focus with accuracy on individual surface features, to provide image content of broad areas of patient dentition at suitable resolution and focus and to provide sufficient illumination for diagnostic purposes. In addition, it can be challenging to accurately combine or “stitch together” content from successive manual scans over adjacent areas.

[0009] One area of particular interest for properly characterizing intraoral features and structure relates to materials detection and identification. Conventional 2D reflective imaging can’t provide detailed information on the material composition of intraoral features. In conventional image capture and presentation, only the contour of tooth, gum, and other dental surfaces can be accurately mapped. Conventional OCT imaging characterizes surface and depth features, but any information on type of materials that form the viewed features is indirect, such as by observation, prior knowledge, inference, or guesswork, without reporting or confirmation by the imaging apparatus.

[0010] Conventional ultrasonic imaging provides a mapping of ultrasonic echo signals onto an image plane, wherein relative differences in sound echo intensity are represented by corresponding differences in pixel brightness. The resulting images can help to distinguish some features and coarse structure of an imaged material, but provide limited insight into the type of material that is imaged or its physical properties. For OCT imaging, as with ultrasound imaging, data about density and depth resolution from the reflected signal that is returned from the object scanned provide some level of useful information on internal structures and features for the practitioner. It would be useful, however, to be able to identify various types of materials that form the detected structures and features. Materials identification, not available with existing systems, can provide a number of benefits, including the following:

[0011] (i) identification of areas having amalgams or composite treatment, allowing automated generation of dental chart data;

[0012] (ii) sensitive detection of demineralization that can indicate early caries;

[0013] (iii) detection of periodontal pocketing in areas of poor visibility;

[0014] (iv) detection of gingival lesions or inflamed tissue;

[0015] (v) detection of cracks within the depth profile;

[0016] (vi) detect the gap formation between the tooth and restoration material.

[0017] Improvement in the quality and quantity of information available using OCT or ultrasound imaging for intraoral applications can be of considerable benefit to the practitioner, particularly where the improvement can be achieved without adding components or accessories to the existing OCT or ultrasound imaging apparatus.

[0018] SUMMARY

[0019] An object of the present disclosure is to advance the art of intraoral imaging for apparatus that provide image content using a non-ionizing penetrating energy.

[0020] Another object of this application is to address, in whole or in part, at least the foregoing and other deficiencies in the related art.

[0021] It is a related object of this application to provide, in whole or in part, at least the advantages described herein.

[0022] These objects are given only by way of illustrative example, and such objects may be exemplary of one or more embodiments of the application. Other desirable objectives and advantages inherently achieved by the disclosed methods may occur or become apparent to those skilled in the art. The invention is defined by the appended claims. According to one aspect of the disclosure, there is provided a method for imaging an intraoral sample using optical coherence tomography (OCT), the method comprising:

[0023] (a) acquiring a plurality of OCT measurements from multiple scans of the sample using a hand-held scanner and detecting one or more interfaces between materials in the acquired OCT measurements;

[0024] (b) reconstructing a 3D surface according to the acquired OCT measurements and detected interfaces and associating OCT measurement data indicative of material composition with 3D coordinate locations of the reconstructed 3D surface;

[0025] (c) for each of a plurality of the reconstructed 3D surface coordinate locations, executing a sequence of:

[0026] (i) combining the associated OCT measurement data indicative of material composition, relative to the 3D surface location;

[0027] (ii) assigning a label identifying an intraoral material according to the combination of OCT measurement data corresponding to the coordinate location;

[0028] (f) forming labeled reconstructed surface image content that associates the label assignments for each of the plurality of 3D surface coordinates to the reconstructed 3D surface; and

[0029] (g) displaying, storing, or transmitting the labeled reconstructed surface image content.

[0030] BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The foregoing and other objects, features, and advantages of the invention will be apparent from the following more particular description of the embodiments of the disclosure, as illustrated in the accompanying drawings.

[0032] The elements of the drawings are not necessarily to scale relative to each other.

[0033] FIG. 1 is a schematic diagram showing an exemplary swept-source OCT (SS-OCT) apparatus according to an embodiment of the present disclosure. FIG. 2A shows a schematic representation of scanning operation for obtaining a B-scan.

[0034] FIG. 2B shows an OCT scanning pattern for C-scan acquisition.

[0035] FIG. 3 is a logic flow diagram that shows basic steps in the process of processing and reporting materials information.

[0036] FIGs. 4A, 4B, and 4C show aspects of the Applicant’s solution for addressing the need to identify materials in a single 2D OCT scan, for providing training data to classifier logic.

[0037] FIG. 5 shows a 3D surface contour, formed by stitching together multiple C- scans.

[0038] FIG. 6 is a logic flow diagram that outlines a process for classifier training that can be used according to an embodiment of the present disclosure.

[0039] FIG. 7A is a schematic diagram that shows, by way of example, a feature vector generated from a single A-scan according to an embodiment.

[0040] FIG. 7B shows examples of generated feature vectors for 3 different materials.

[0041] FIG. 8 is a diagram that shows stitching of visual images with overlapping content and classifier results.

[0042] FIG. 9 is a logic flow diagram that shows a sequence for processing OCT scan data according to an embodiment of the present disclosure.

[0043] FIG. 10 is a diagram that shows OCT results displayed for corresponding teeth in a dental chart.

[0044] FIG. 11 is a diagram that shows a 3D surface reconstruction with a user selected slice.

[0045] FIG. 12 is a diagram that shows an example display representing a portion of a 3D surface having regions formed of different materials.

[0046] FIGs. 13 A and 13B show example displays for mesh and point cloud surface contours of intraoral features, respectively.

[0047] FIG. 14 is a diagram showing an arrangement for a user interface according to an embodiment of the present disclosure. DESCRIPTION OF EXEMPLARY EMBODIMENTS

[0048] The following is a detailed description of exemplary embodiments, reference being made to the drawings in which the same reference numerals identify the same elements of structure in each of the several figures.

[0049] Where they are used in the context of the present disclosure, the terms “first”, “second”, and so on, do not necessarily denote any ordinal, sequential, or priority relation, but are simply used to more clearly distinguish one step, element, or set of elements from another, unless specified otherwise.

[0050] In conventional optics terminology, an optical system forms an image of an object. With respect to an intraoral feature that is scanned as a sample for obtaining image content, the terms “object”, “subject”, and “sample” can be used interchangeably, unless otherwise specifically distinguished.

[0051] The term “set”, as used herein, refers to a non-empty set, as the concept of a collection of elements or members of a set is widely understood in elementary mathematics. The term “subset”, unless otherwise explicitly stated, is used herein to refer to a non-empty proper subset, that is, to a subset of the larger set, having one or more members. For a set S, a subset may comprise the complete set S. A “proper subset” of set S, however, is strictly contained in set S and excludes at least one member of set S.

[0052] The general term “depth scanning signal” refers to an electromagnetic or acoustical signal that, by reflection from within an object, characterizes both the surface of the object and a portion of the adjacent sub-surface within the object. Reflection is in the reverse direction of signal incidence. Two types of imaging systems provide the depthscanning signal for intraoral imaging applications: ultrasound imaging apparatus and OCT imaging apparatus.

[0053] The general terms "depth scanner" or “depth-penetrating scanner” relate to an apparatus that acquires image content from reflected energy of a depth-scanning signal directed toward an object. In the context of the present disclosure, a depth scanner or depth-penetrating scanner can be either an ultrasonic scanner that uses reflected acoustical, ultrasound energy from within an intraoral feature or an optical OCT scanner using OCT. The OCT imaging scanner is energizable to project a scanned beam of light, such as broadband near-IR (BNIR) light, that can be directed to the intraoral surface through a sample arm and acquired, as reflected and scattered light returned in the sample arm, for measuring interference with light from a reference arm used in OCT imaging of a surface.

[0054] The term "raster scanner" relates to the combination of hardware components that sequentially scan a depth-scanning signal toward uniformly spaced locations along a sample, as described in more detail subsequently.

[0055] Hand-held scanners used in intraoral imaging are compact devices that transmit the depth-scanning signal toward the subject and acquire reflected energy returned from the subject surface. Some portion of the signal processing can be performed within the probe of the hand-held scanner; however, these devices typically process and store scanned data at processing devices that are in signal communication with the hand-held probe.

[0056] In the context of the present disclosure, the phrase “imaging range” relates to the effective distance (generally considered in the z-axis or A-scan direction) over which depth-scanner measurement is available. When using OCT, the light beam is considered to be within focus over the imaging range. Image depth relates to imaging range, but has additional factors related to signal penetration through the sample tooth or other tissue.

[0057] By way of example, the simplified schematic diagram of FIG. 1 shows the components of one type of OCT apparatus, here, a conventional swept-source OCT (SS- OCT) apparatus 100 using an interferometer with a light source provided by a programmable filter 10 that is part of a tuned laser 50. For intraoral OCT, for example, laser 50 can be tunable over a range of frequencies (expressed in terms of wave-numbers k) corresponding to wavelengths between about 400 and 1600 nm. According to an embodiment of the present disclosure, a tunable range of about 60nm bandwidth centered about 1300nm is used for intraoral OCT, with a swept rate around 200 Hertz.

[0058] In the FIG. 1 device, the variable tuned laser 50 output goes through a coupler 38 and to a sample arm 40 and a reference arm 42. The sample arm 40 signal goes through a circulator 44 and is directed for imaging of a sample S from a handpiece or probe 46. The sampled signal is directed back through circulator 44 and to a detector 60 through a coupler 58. The reference arm 42 signal is directed by a reference 34, which can be a mirror or a light guide, through coupler 58 to detector 60. The detector 60 may use a pair of balanced photodetectors configured to cancel common mode noise.

[0059] A control logic processor (control processing unit CPU) 70 is in signal communication with tuned laser 50 and its programmable filter 10 and with detector 60. Processor 70 can control the scanning function of probe 46 and store any needed calibration data for obtaining a linear response to scan signals. Processor 70 obtains and processes the output from detector 60. CPU 70 is also in signal communication with a display 72 for command entry and OCT results display.

[0060] It should be noted that the swept-source architecture of FIG. 1 is one example configuration only; there are a number of ways in which the interferometer components could be arranged for providing swept-source OCT imaging.

[0061] By way of further background, FIGs. 2A and 2B give an overview of the OCT scanning pattern as executed by probe 46. At each of several locations or points in the scanning pattern, the OCT device performs an A-scan. A linear succession of A-scans then forms a B-scan, corresponding to the x-axis direction as shown. Successive B-scan rows, side-by-side, then form a C-scan which provides the 3D OCT image content for the sample S. Sample S is formed of one or more materials. For intraoral imaging, the sample materials can include teeth, filling materials such as amalgam, full dentition, soft tissues, or other materials of interest to a practitioner.

[0062] The term “point” as used herein is not abstract or dimensionless, but instead relates to a location on the surface of the sample which, in practice, extends over some finite dimensional area (length x width), depending on scanning probe resolution. Coordinates (x, y, z) provide the spatial position of this scanned location; these coordinates (x, y, z) necessarily occupy some volume. For example, a scan can have a coordinate location (x, y, z) that actually extends over an area 100 x 100 microns wide or wider, depending on the resolution available from the scanning head of the intraoral OCT probe. Given these practical considerations, it can be appreciated that the terms “coordinate location” and “coordinate point” are considered synonymous in subsequent description.

[0063] FIG. 2A schematically shows the information acquired during each A-scan, a depth signal extending along the direction of the scanning light beam. In a unidirectional raster scanning arrangement, the scan signal for obtaining each B-scan image has two linear sections in the example shown, with a scan portion 92, during which the scanning mirror is driven to direct the sampling beam from a beginning to an ending position for data acquisition, and a retro-scan 93, during which the scanning mirror is restored to its beginning x position, without data acquisition. An interference signal 88, shown with DC signal content removed, is acquired over the time interval for each point location 82, wherein the signal is a function of the time interval required for the sweep, with the signal that is acquired indicative of the spectral interference fringes generated by combining the light from reference and feedback sample arms of the interferometer (FIG. 1). The Fast Fourier Transform (FFT) generates a transform T for each A-scan. One transform signal corresponding to an A-scan is shown by way of example in FIG. 2 A.

[0064] According to an alternate embodiment of the present disclosure, the scanner can operate with a bidirectional raster scan pattern, wherein the signal for two B-scan images is sampled during both a forward scan portion and a retro-scan portion.

[0065] In Fourier domain OCT, the A scan corresponds to one line of spectrum acquisition which generates a line of depth (z-axis) resolved OCT signal. The B scan data generates a 2-D OCT image along the corresponding scanned line.

[0066] Raster scanning is used to obtain multiple B-scan data by incrementing the raster scanner 90 acquisition in the C-scan (y-axis) direction. The scan patterns in both a unidirectional raster scan pattern and the bidirectional raster scan pattern are represented schematically in FIG. 2B, which shows how 3-D OCT image content is generated using the A-, B-, and C-scan data.

[0067] The wavelength or frequency sweep sequence that is used at each A-scan point 82 can be modified from the ascending or descending wavelength sequence that is typically used. Arbitrary wavelength sequencing can alternately be used. In the case of arbitrary wavelength sequencing, which may be useful for some particular implementations of OCT, only a portion of the available wavelengths are provided as a result of each sweep. In arbitrary wavelength sequencing, each wavelength can be randomly selected, in arbitrary sequential order, to be used in the OCT system during a single sweep. A-scan point locations 82 can be uniformly spaced from each other with respect to the x axis, providing a substantially equal x-axis distance between adjacent points 82 along any B-scan image. Similarly, the distance between lines of scan points 82 for each B scan can be uniform with respect to the y axis. X-axis spacing may differ from y-axis spacing; alternately, spacing along these orthogonal axes of the scanned surface may be equal.

[0068] Embodiments of the present disclosure are directed to the task of identifying material types and their interfaces in intraoral depth scanning, using either an OCT or ultrasound depth scanner. The logic flow diagram of FIG. 3 shows basic steps in the process of processing and reporting materials information. An acquisition step S300 acquires the scan data from the subject, obtained from the hand-held intraoral ultrasound or OCT scanner. The reflected energy detected by the hand-held scanner is conditioned by materials on or adjacent to the surface of the subject and thus provides information content that can be processed and used to characterize surface features and their material make-up. Depth information can thus be obtained on rigid surface features and interfaces. A reconstruction step S400 then stitches together the acquired information from multiple scans in order to visually model or represent the surface and near-surface features of the scanned region and to identify the structure and extent of rigid surfaces. The reconstruction process generates a common 3D coordinate system, used in stitching for relative positioning of each scan in the scanned surface data.

[0069] Unlike 3D reconstruction that forms a volume image of its object from a series of 2D images at predetermined angles in computed tomography (CT) imaging, reconstruction for manually scanned OCT data forms a contour image, in 3D space, of the scanned surface, incorporating the near-surface content available from the OCT scan data. In the context of the present disclosure, the contour image that is formed has coordinates in 3D space. Extending in three dimensions, the contour image can thus can be considered a type of “volume” image for analysis and display, whereby point locations of the image that is formed act as “voxels” with 3D coordinates (x, y, z), but with the constraint that the resulting stitched image content provides a type of volume image that can only represent surface and near-surface features of the scanned subject. Compared against a CT volume image that is generated using ionizing radiation, the OCT surface contour provides an “outer shell” of limited thickness. Only content and features of the subject within the depth range of the OCT scan can be represented. In the context of the present disclosure, unless otherwise specified, the phrases “surface contour”, “3D surface image”, and “volume image” can be used equivalently to describe the surface reconstruction formed by stitching using the OCT-scanned content.

[0070] The 3D reconstruction formed from OCT data is formed in 3D space. This reconstruction has image content that represents the surface contour of a tooth or other intraoral feature and some portion beneath the surface, as limited to the depth achievable by the OCT scan; typically, this depth for near-surface content is no more than a few millimeters. The stitching process allows reconstruction of the surface contour, with some added depth data, by identifying overlapping scan content for identical locations in multiple scanned images, using overlapping locations as a guide to correlating spatial position of the depth scanning data from adjacent scans. The resulting stitched 3D reconstruction of the surface contour and its near-surface components is typically represented as a point cloud or mesh.

[0071] Because the surface reconstruction occupies a volume, various utilities can be used for viewing the surface reconstruction data at different angles, including pan, zoom, and rotate capabilities, for example. Aspects of tasks such as image segmentation, well- known to those skilled in handling volume image content in medical and dental fields, can also be used.

[0072] Both acquisition step S300 and reconstruction step S400 use procedures that are familiar to those skilled in ultrasound or OCT imaging.

[0073] Continuing with the FIG. 3 process, in a materials identification step S500, the scan data can be processed to identify material content of regions of the reconstructed surface. The setup and operation used for this classification process is described in more detail subsequently. The common (x, y, z) coordinate system that is assigned as part of the stitching process allows scanned surface data to be combined into a reconstructed 3D surface contour or volume. The identified materials are associated with features at coordinate locations or positions in the reconstructed 3D surface contour. A representation combining the reconstructed 3D surface and associated materials is generated in a representation step S600. The representation having material information can then be displayed or otherwise distributed in a reporting step S700.

[0074] It should be observed that materials identification, shown in step S500 in FIG. 3, can execute on individual B-scans for OCT or can be performed on the reconstructed 3D surface. The image content that is analyzed for material composition can be from either 2D or 3D image content obtained from scan data. As is described in more detail subsequently, 3D image content from stitched scan data can be particularly effective to provide materials characterization and presentation of this data.

[0075] FIGs. 4A, 4B, and 4C show aspects of the Applicant’s solution for addressing the need to identify materials in a single 2D OCT scan, for providing training data to classifier logic. FIG. 4A shows an example of a single intraoral B-scan from an OCT scanner, defining a number of regions: air (background); a liquid bubble 230; enamel cuspids 232, and an unlabeled region that lies beyond the depth range. FIG. 4B shows manually drawn labeling for the three regions in the FIG. 4A scan, useful for training classifier logic, as described in more detail subsequently. FIG. 4C shows manual highlighting of transitions between identifiable materials. For example, color-coding can be used to indicate transitions between different liquid or solid materials such as at the enamel interface with fluid, for example. A “label” can be any appropriate text annotation, coloring or other visual highlighting, or other assignment of a distinguishing attribute to an identified material from the OCT scan data.

[0076] FIG. 5 is a mesh surface representation of a reconstructed 3D surface, formed by stitching together multiple C-scans as described, for example, in commonly assigned U.S. Patent No. 11,006,835 to Inglese et al., entitled “Surface Mapping Using an Intraoral Scanner with Penetrating Capabilities”, incorporated herein in its entirety. The composite, reconstructed 3D surface contour can be represented in the form of a surface mesh or point cloud, for example. The surface contour that can be reconstructed within 3D space can be very sizable relative to the scanner field of view (FOV) and typically spans several times the scanner FOV. Using data from the depth-penetrating scanner, regions of the contour image can be identified and labeled, either manually for training classifier logic or automatically as performed by the image processing system. In the FIG. 5 example, regions within the surface contour image space that are labeled include gum tissue 240, enamel 242, composite filling 244, amalgam 246, and unlabeled 248. While the example of FIG. 5 shows only outer surfaces (typically those externally visible, wherein the outer interface is between air and a material), embodiments of the present disclosure can also extend to inner or “hidden” surfaces that lie to at least some depth within the intraoral structure.

[0077] The ability to generate a reconstructed 3D surface contour, represented as a point cloud or mesh that assigns reference 3D coordinates (x, y, z) to each surface location, is the result of stitching logic that integrates the information from multiple samples scanned over the same surface region. In manual scanning with a hand-held scanner, the same coordinate location, idealized as a “point” or “point location” along the surface, considered as a coordinate point (x, y, z), or more generally, as a coordinate location (x, y, z), can be sampled multiple times, with each scan sampling the coordinate location from a slightly different direction. Embodiments of the present disclosure take advantage of multiple samples for most of the scanned coordinate locations (x, y, z) by using the accumulated information of multiple scans in order to more accurately characterize the material content. Thus, for a given point or coordinate location on the reconstructed surface, at coordinates (x, y, z) in the 3D surface contour image space, there can be any number of samples obtained, each sample indicative of the material content, as learned by classifier logic. With respect to the data obtained, each scan sampling that includes coordinate location (x, y, z) can provide a “vote” indicative of the most likely material that was encountered at that point. Summing or otherwise accumulating the votes for each scanned coordinate location (x, y, z) provides the classifier with a suitable characterization of the material content at that position, wherein the characterization has a high likelihood of accuracy. Multiple adjacent or neighboring coordinates (xi, yi, zi), (x2, y2, Z2), (xn, yn, zn) can be suitably grouped together according to the classification provided, forming regions that represent different natural and fabricated intraoral features of the scanned patient. The results provided for adjacent coordinates can then be used to add materials data to the reconstructed surface, point cloud or mesh, and to more positively identify different regions, as shown in the example of FIG. 5.

[0078] Efficient identification of materials content for a region, as shown in FIG. 5, simplifies classification of scanned material within that region. Labeling of surfaces of the 3D surface contour can be applied for all of the scan data within the corresponding reconstructed volume. With this capability, a significant amount of scan data can be incorporated into a training set for the classifier, as described in more detail subsequently. Classifier Setup and Training

[0079] Embodiments of the present disclosure can employ trained classifier logic in order to evaluate the scan data and select the most likely materials arrangement for the given scan content. A classifier can use custom-designed programmed or trained logic or can use a commercially available machine learning utility that employs known machinelearning architectures such as decision trees, naive Bayes classifier, K-Nearest neighbors, Support Vector Machines, and Artificial Neural networks, for example. For OCT operation, as described previously with respect to FIG. 1, classifier execution can be performed on processor 70 or on an external processor or network of processors in signal communication with processor 70 and the overall OCT or ultrasound imaging apparatus.

[0080] Training is the process of generating model values that minimize the likelihood of prediction error, based on a training set that establishes a basis for materials determination. Classifier logic can be trained for material detection using supervised learning. Supervised learning techniques employ a set of labeled training data that serves as “ground truth” data and operate to improve results accuracy by generating and processing feature vectors for the input data. The training set itself can be labeled by human operators or can be machine-labeled in order to guide the training logic.

[0081] The logic flow diagram of FIG. 6 outlines a process for classifier training that can be used according to an embodiment of the present disclosure. In a training set generation step S200, a set of exemplary OCT scan data is provided with labeled data for generating a training set. As described previously with respect to FIG. 5, labeling of a reconstructed 3D surface contour can then be applied to individual scans that form part of the corresponding volume. In this way, a significant number of training set elements can be generated from a single labeled surface, accelerating the training cycle and helping to improve the overall accuracy of the trained classifier.

[0082] Each element of the training set has a known set of materials with known features, arranged as feature vectors, arrangements of multi-parameter data, designed to facilitate learning for classifier logic. A training step S210 executes, training the classifier logic to generate suitable parameters into feature vectors according to the training set. A recording step S220 then stores the generated classifier parameters for subsequent classification tasks.

[0083] FIG. 7A is a schematic diagram that shows, by way of example, how a feature vector can be generated from a single A-scan according to an embodiment. At left is a “raw data” OCT slice (B-scan) highlighting a detected interface. A rectangular window, at the position shown, indicates the scan data used for the feature vector. A collection of triangular filters, 15 in the example shown, is used for computation of local average and local gradient values. Using this type of low-pass filtering helps to suppress noise content and allows for increased processing speed. The generated numerical values (30 in this example) can be concatenated to build the feature vector represented at the right in FIG. 7A. An additional value provides depth information along the (vertical) scan direction. Additional information, such as color information, normal direction, confidence value, local signal-to-noise value, higher-order derivatives, linear / nonlinear transform value, and images with different illumination setup such as visual light, UV light or IR light, for example, can also be incorporated within the feature vector.

[0084] The rectangular window in the FIG. 7A example is placed relative to the location of the detected interface between two materials. The window length is chosen to contain some signal content from before the interface, and some from after the interface, in order to obtain physical quantities such as the variation of intensity at the interface, the intensity of backscattered light after the interface, and the slope at which the signal drops off beyond the interface. These characteristics are related to the material reflectivity at the interface and to material absorption and scattering. These may include physical characteristics that are not measured directly; however, they affect the intensity profile. The rectangular window width can be the width of a single A-scan or can have a width corresponding to multiple adjacent or neighboring A-scans, in order to reduce noise in the computed feature vector.

[0085] FIG. 7B shows examples of generated feature vectors for 3 different materials (Enamel, Gums, Amalgam). The material classifier can learn to identify the distinctive “signature” of the feature vector for each of a set of intraoral materials and learn to distinguish materials from each other according to their respective feature vectors. By way of further example, FIG. 8 shows, at the left, a crude stitching of two visual images 254a and 254b having partially overlapping content. Each image 254a, 254b represents a top view of a scanned region having OCT scanned content. Regions of amalgam 246 and composite 244 are shown. The reconstructed 3D surface contour of the OCT stitched content is represented in a labeled surface 250, part of a training set considered to present the “ground truth” for classifier training and manually labeled as part of the training set setup, with at least amalgam 246 and composite 244 regions identified. Trained classifier results from the full training set are shown in the automated labeling provided on a reconstructed surface 252. A high degree of accuracy thus allows the classifier logic to identify materials within the reconstructed OCT contour image content and to provide this information to the practitioner. It can be appreciated that unlabeled data

[0086] 248 from manually labeled surface 250 is replaced with a best match on surface 252, according to execution of the learned logic.

[0087] It can be appreciated that the reconstruction 252 is generated by stitching together data from multiple scans, that is, from multiple manual passes of a scanning probe across the scanned region. Each scan over any specific (x,y,z) location typically differs in probe direction, head angle, scan duration, starting point and ending point. For each scan that provides OCT data on location (x,y,z), a decision or vote is provided indicative of candidate material characteristics, that is, according to the feature vector generated for coordinate location or point (x,y,z) during that scan. Where votes conflict or data is sufficiently ambiguous, classifier logic cannot reliably identify the material at a specific point, and can leave the point location unlabeled as to material content. A “no vote” label

[0088] 249 is assigned by the classifier when confidence for a known candidate material at a particular location is too low for probability thresholds.

[0089] Processing for Materials Identification

[0090] According to an embodiment of the present disclosure, as noted previously, machine learning, executed by a suitable computer or other logic processor or by a network of logic processors, can be used to identify the type of material being scanned by the intraoral OCT scanning device or, similarly, by an ultrasound probe. Using machine learning, for example, classifier logic can be trained to identify the signal “signature” from returned signal content reflected by the subject for each of a set of materials {A, B, C, D, etc. }. This processing can be repeated for each scan, generating a probabilistic metric or “vote” relating to the materials content of multiple coordinate locations on the reconstructed surface.

[0091] The logic flow diagram of FIG. 9 shows a sequence for processing OCT scan data according to an embodiment of the present disclosure. An acquisition step S900 acquires the OCT scan data obtained from a scan session with an intraoral scanner. According to an alternate embodiment of the present disclosure, the OCT scan data can be supplemented by images obtained from a camera and using visible light, combined with OCT scan data. A reconstruction step S910 assembles a 3D surface associated with a common coordinate system, typically visualized as a point cloud or mesh, according to the acquired scan data. As described previously, the combined OCT data in the common coordinate system form a reconstructed 3D surface or surface contour. An iterative sequence S920 can then execute the classification task for each of a set of 3D coordinates (x, y, z) of the reconstructed surface. This can include coordinate points that are just beneath the 3D surface, within the depth range of the OCT signal. 3D coordinates can also correspond to interfaces within the scanned surface, transitions where two different materials face each other. Color camera images can be used to help provide reference information for the surface image detection and generation processing described herein.

[0092] Continuing with the flow of iterative sequence S920 in FIG. 9, the process repeated for each of multiple point location coordinates (x, y, z) includes a computation step S922 that identifies, collects, and analyzes one or more feature vectors, set up for classifier logic as described previously, that correspond to OCT measurements obtained through the particular (xm, ym, zm) coordinate. Because each coordinate location (xm, ym, zm) on the scanned surface can be scanned multiple times, each time from a different direction and from a different angle of scanner inclination, there can be any number of feature vectors that intersect at (xm, ym, zm), having data related to that coordinate location. These feature vectors that relate to OCT scan measurements taken through a particular coordinate location are considered to intersect with the individual (x, y, z) coordinate. In a calculation step S924, classifier logic then analyzes the feature vector for each intersecting OCT scan and accumulates the vote data thus provided. Using this data, step S924 then determines the most likely candidate material associated with the scan measurement. According to an embodiment of the present disclosure, for each intersecting scan measurement through coordinate (xm, ym, zm), this analysis effectively provides a “vote” that indicates the likely probability that the coordinate location (xm, ym, zm) is formed of a particular material from the set {A, B, C, D, etc.} for which the classifier is trained. The cumulative count of votes obtained from the one or more scan measurements that intersect through location coordinate (xm, ym, zm) can then be used to assign a label to coordinate (xm, ym, zm) that gives the best possible approximation of material content at that location.

[0093] In the FIG. 9 process, an appropriate assignment step S926, S928 can then assign an appropriate label related to a material or can assign a status of “unknown” to the location coordinate (x, y, z) under consideration. A confidence value may also be assigned to the selected label. A recording step S929 records the label assignments. A reporting step S930 then records, transmits, or displays the information related to the surface.

[0094] Embodiments of the present disclosure can be particularly useful for detecting transitions between naturally occurring or fabricated materials and for identifying the materials on each side of an interface. Naturally occurring materials can include enamel, dentin, pulp, gums, tongue, bone, and blood and other fluids. Fabricated or artificial materials can include composite, amalgam, metals, plastics, ceramics, zirconium, crown material, or appliance formed from such materials or using some combination of materials. Interface transitions can also be particularly useful for highlighting structural variations and flaws, gum condition, caries-infected areas, cracking, periodontal cavities, cancerous areas, and microleakage, for example.

[0095] Detection of an interface between two materials can define a surface according to changes in gradient, which correspond to transitions between different types of material. For example, a crack, the end or edges of a composite filling, or a dentin-enamel junction can be a type of internal interface. Of particular interest are transitions related to natural materials (enamel, dentin, pulp, gums, blood, other fluids), artificial materials (composite, amalgam, metal, transparent plastic or ceramic, zirconium, crown or any other restoration material or appliance, structural variations (healthy vs. unhealthy gums), caries areas vs healthy enamel, cracks, periodontal cavities, and microleakage. For convenience of graphical representation, the figures provided herein primarily depict interfaces wherein one material is air. However, it must be noted that the methods of the present disclosure can similarly be used for classification of other interfaces, such as interfaces below the surface wherein the first material is not air, without loss of generality.

[0096] Reporting results

[0097] The depth imaging apparatus can report results related to materials identification in a number of ways, as described following.

[0098] Populating a Dental Chart

[0099] According to an embodiment of the present disclosure, results from OCT scanning and identification of materials can be used to populate a dental chart or other mapping of teeth and other intraoral features. FIG. 10 is a diagram that shows OCT results displayed for corresponding teeth in a dental chart 110. In the example of FIG. 10, teeth 14, 15, and 16 can be highlighted, using color or other visual treatment, to indicate detection of crown material. Teeth 11 and 21 can show highlighting that is indicative of filling materials, for example.

[0100] In order to generate dental chart 110 from OCT scan data, the imaging system processor can stitch together multiple scans in order to generate reconstructed 3D surface contour images for upper and lower jaws. Each reconstructed 3D surface can then be automatically segmented to detect tooth location. Segmentation is a standard processing technique that allows identification of individual anatomy features, such as distinguishing two adjacent teeth from each other, for example. Each tooth crown can be assigned a number, using an appropriate numbering standard.

[0101] The processing logic can also analyze tooth shape, such as cuspids, preparation, bridge, emergence profile, attached orthodontic braces, or other feature. Tooth identification logic can use the identified materials results to exclude gum tissue and other materials not typically needed for the dental chart.

[0102] For each segmented crown, the corresponding anatomical region (occlusion, lingual, buccal, mesial and distal sides) can be identified. Then, for each region, the system can look up the corresponding material categories and extract the one with highest probability. This category can then be reported in the dental chart automatically, along with the associated color, texture, label or text. The generated dental chart can be automatically fdled with labels for each tooth and each anatomical region. The operator can then check and validate the automatically generated information and correct errors as needed.

[0103] According to an embodiment of the present disclosure, the processor can respond to operator instructions entered on the dental chart, such as by displaying the surface contour for a selected tooth.

[0104] Displayed as Labeled Slice

[0105] According to an alternate embodiment of the present disclosure, the depth imaging apparatus can display a labeled 2D portion or “slice” of the obtained data obtained from the 3D surface reconstruction. The diagram of FIG. 11 shows a representative graphical user interface (GUI) on display 72 with a reconstructed 3D surface representation on the left, having a selected slice direction as indicated by the dashed line; on the right is shown the viewer-selected slice, either from a single OCT B-scan or C-scan volume or from the accumulation of information from multiple OCT scans stitched together, annotated with materials information. The GUI can allow the viewer to indicate a position for slice direction and orientation, for example. The user-selected portion could be identified in any suitable form, such as using a polygon representing a 2D slice in a 3D coordinate system, for example, or using a static slice, defined relative to the current field of view of the scanner, as shown in FIG. 11.

[0106] Annotation can be highlighted or outlined, marked with an arrow, encircled within a polygon, or displayed with a color, a texture, or showing text or any other label. In the FIG. 11 example, the detected material is a potential tooth decay (a special unhealthy region, as evidenced by material properties identified) that requires further inspection by the dentist. In this example, the detected material transitions would be air-to-caries region and caries region-to-tooth, which can then be highlighted or enhanced to indicate tooth decay. Represented as Regions of 3D Surface on 2D Projection

[0107] According to an alternate embodiment, the reconstructed surface contour obtained from stitching OCT scans can be shown. This 3D image content can appear, for example, as a point cloud or mesh, with color or other indication showing the detected material type over any portion of the surface. Visual presentation of the subject surfaces can be displayed using color, including false color or grayscale, or other image treatments to present or highlight materials content.

[0108] The diagram of FIG. 12 shows an example display representing a portion of a 3D surface having regions formed of different materials. The display shows a 2D surface projection having separated shaded regions indicating gum tissue 124, enamel 122, and composite 120.

[0109] FIG. 13 A shows a 3D surface mesh. FIG. 13B shows an example point cloud display of intraoral features enhanced to indicate different materials. The reconstructed 3D surface contour can be partitioned into small cubes. Each cube is assigned the majority vote or other combined results of all material contributions, and the region of the 3D surface mesh or point cloud within each cube is rendered with the corresponding color for that material.

[0110] According to an alternate embodiment of the present disclosure, a listing of materials identified in an intraoral OCT scan can be provided as a text listing. This listing can include other details, such as measured lengths, surfaces, volumes, thickness, or other dimensional data. For example, the thickness of a sealant or the depth and width of a detected caries region may be of particular interest. In addition to measured data, other information can be provided, such as a severity rating, for example: deep, superficial, or potential. A confidence rating or other probability can also be indicated.

[0111] A text report may contain computer-generated text phrases or sentences such as “tooth #23 is a crown” or “a periodontal pocket of about 3 millimeters needs confirmation on the buccal side of the first molar”.

[0112] The OCT signal may be automatically adjusted to reveal the detected conditions. For instance, signal brightness and contrast may be enhanced to highlight healthy and unhealthy regions. According to an alternate embodiment of the present disclosure, as shown in the example of FIG. 14, the operator interface can provide the capability to display particular materials or to remove particular materials from the display, that is, to make one or more specified materials substantially transparent, that is, less than about 20 percent opaque. Thus, for example, the display can show only those materials that are of interest for a particular procedure, as selected using the GUI.

[0113] For instance, by instructing the system to remove the composite material from visibility, a dentist can inspect the cavity drilled in the enamel below the composite. Disabling visibility of a material triggers the removal of the corresponding surface from the 3D contour, thus revealing the next interface beneath it on the 3D contour, which can be the enamel or dentin, for example.

[0114] In an embodiment, the next interface may be corrected to account for refraction (shift of light direction) and path length (optical index for OCT or sound velocity for ultrasound) from the disabled materials. An A-scan corresponds to a signal which propagated through a non-uniform medium and therefore did not follow a straight line. Correct measurements can require optical path corrections.

[0115] According to an alternate embodiment of the present disclosure, the GUI viewer can select a point on the displayed surface, such as with a mouse click or other cursor, to obtain materials information for the selected location.

[0116] Selective representation by material type can allow the viewer to reveal interfaces from fluids (blood, water, saliva) and to reconstruct only hard surfaces, for example. This helps visualization of the outline of the margin line for restoration. This feature can also allow better visibility of periodontal pockets or of portions of the root of a tooth or the metallic surface of an implant through a thin layer of gum tissues. In displaying internal surfaces, inner interfaces may require correction to compensate for signal refraction.

[0117] According to the embodiment as shown in the example user interface display of FIG. 14, the user may select, from a pull-down menu 140, the materials of interest for displaying an image 144, such as viewing all caries regions, viewing only composite filling surfaces, etc. Selective transparency of one or more materials, as well as enhancement of other materials, can be employed to successively show internal interfaces in a sequence useful to the viewer.

[0118] For 2D display, the viewer can have the option to select a particular slice of the reconstructed 3D surface contour. This can include views taken along the axis of the scan signal or orthogonal or oblique to that axis, for example. For 3D display, view manipulation utilities can allow the viewer to rotate the view of the reconstructed surface.

[0119] The invention has been described in detail with particular reference to a presently understood exemplary embodiments, but it will be understood that variations and modifications can be affected within the spirit and scope of the disclosure.

[0120] For example, control logic processor 70 can be any of a number of types of logic processing device, including a computer or computer workstation, a dedicated host processor, a microprocessor, logic array, or other device that executes stored program logic instructions. The interferometer that is used for one or more channels, described in the example configurations given hereinabove as a type of Mach-Zehnder interferometer, can alternately be another appropriate type, such as a Michelson interferometer, for example, with appropriate component re-arrangement.

[0121] The presently disclosed exemplary embodiments are therefore considered in all respects to be illustrative and not restrictive. The scope of the disclosure is indicated by the appended claims, and all changes that come within the meaning and range of equivalents thereof are intended to be embraced therein.

[0122] Consistent with at least one exemplary embodiment, exemplary methods / apparatus can use a computer program with stored instructions that perform on image data that is accessed from an electronic memory. As can be appreciated by those skilled in the image processing arts, a computer program of an exemplary embodiment herein can be utilized by a suitable, general-purpose computer system, such as a personal computer or workstation. However, many other types of computer systems can be used to execute the computer program of described exemplary embodiments, including an arrangement of one or networked processors, for example.

[0123] A computer program for performing methods of certain exemplary embodiments described herein may be stored in a computer readable storage medium. This medium may comprise, for example; magnetic storage media such as a magnetic disk such as a hard drive or removable device or magnetic tape; optical storage media such as an optical disc, optical tape, or machine-readable optical encoding; solid state electronic storage devices such as random-access memory (RAM), or read only memory (ROM); or any other physical device or medium employed to store a computer program. Computer programs for performing exemplary methods of described embodiments may also be stored on computer readable storage medium that is connected to the image processor by way of the internet or other network or communication medium. Those skilled in the art will further readily recognize that the equivalent of such a computer program product may also be constructed in hardware.

[0124] It should be noted that the term “memory”, equivalent to “computer- accessible memory” in the context of the application, can refer to any type of temporary or more enduring data storage workspace used for storing and operating upon image data and accessible to a computer system, including a database, for example. The memory could be non-volatile, using, for example, a long-term storage medium such as magnetic or optical storage. Alternately, the memory could be of a more volatile nature, using an electronic circuit, such as random-access memory (RAM) that is used as a temporary buffer or workspace by a microprocessor or other control logic processor device. Display data, for example, is typically stored in a temporary storage buffer that can be directly associated with a display device and is periodically refreshed as needed in order to provide displayed data. This temporary storage buffer can also be considered to be a memory, as the term is used in the application. Memory is also used as the data workspace for executing and storing intermediate and final results of calculations and other processing. Computer- accessible memory can be volatile, non-volatile, or a hybrid combination of volatile and non-volatile types.

[0125] It will be understood that computer program products for exemplary embodiments herein may make use of various image manipulation algorithms and / or processes that are well known. It will be further understood that exemplary computer program product embodiments herein may embody algorithms and / or processes not specifically shown or described herein that are useful for implementation. Such algorithms and processes may include conventional utilities that are within the ordinary skill of the image processing arts. Additional aspects of such algorithms and systems, and hardware and / or software for producing and otherwise processing the images or co-operating with the computer program product of the application, are not specifically shown or described herein and may be selected from such algorithms, systems, hardware, components and elements known in the art.

[0126] Exemplary embodiments according to the application can include various features described herein (individually or in combination).

[0127] While the invention has been illustrated with respect to one or more implementations, alterations and / or modifications can be made to the illustrated examples without departing from the spirit and scope of the appended claims. In addition, while a particular feature of the invention can have been disclosed with respect to only one of several implementations / exemplary embodiments, such feature can be combined with one or more other features of the other implementations / exemplary embodiments as can be desired and advantageous for any given or particular function.

[0128] The term “a” or “at least one of’ is used to mean one or more of the listed items can be selected. The term “about” indicates that the value listed can be somewhat altered, as long as the alteration does not result in nonconformance of the process or structure to the illustrated exemplary embodiment. Finally, “exemplary” indicates the description is used as an example, rather than implying that it is an ideal.

[0129] Other embodiments of the invention will be apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the invention being indicated by the following claims.

Claims

CLAIMSWHAT IS CLAIMED IS:

1. A method for imaging an intraoral sample using a depthpenetrating scanner, the method comprising the steps of:(a) acquiring a plurality of depth-penetrating scanner measurements from multiple scans of the sample using a hand-held, depth-penetrating scanner and detecting one or more interfaces between materials in the acquired depthpenetrating scanner measurements;(b) reconstructing a 3D surface according to the acquired depth-penetrating scanner measurements and detected interfaces and associating depthpenetrating scanner measurement data indicative of material composition with 3D coordinate locations of the reconstructed 3D surface;(c) for each of a plurality of the reconstructed 3D surface coordinate locations, executing a sequence of:(i) combining the associated depth-penetrating scanner measurement data indicative of material composition, relative to the 3D surface location;(ii) assigning a label identifying an intraoral material according to the combination of depth-penetrating scanner measurement data corresponding to the coordinate location;(f) forming labeled reconstructed surface image content that associates the label assignments for each of the plurality of 3D surface coordinates to the reconstructed 3D surface; and(g) displaying, storing, or transmitting the labeled reconstructed surface image content.

2. The method of claim 1 further comprising forming a feature vector corresponding to each depth-penetrating scanner measurement associated with a 3D coordinate location.

3. The method of claim 2 wherein the feature vector corresponds to an interface between materials.

4. The method of claim 1 wherein reconstructing the 3D surface further comprises using color data from a camera or color sensor.

5. The method of claim 2 further comprising using a camera image to generate the feature vector.

6. The method of claim 2 wherein associating depth-penetrating scannermeasurement data indicative of material composition with 3D coordinate locations of the reconstructed 3D surface comprises applying machine logic.

7. The method of claim 1 wherein reconstructing the 3D surface comprises forming a point cloud or mesh.

8. The method of claim 1 wherein the labeled reconstructed surface image content distinguishes intraoral tissue of a patient from fabricated materials.

9. The method of claim 1 wherein the labeled reconstructed surface image content distinguishes a first fabricated material from a second fabricated material.

10. The method of claim 1 wherein the labeled reconstructed surface image distinguishes healthy from infected tissue.

11. The method of claim 1 wherein the labeled reconstructed surface image distinguishes bone from enamel.

12. The method of claim 1 further comprising applying image segmentation to the labeled, reconstructed surface image content.

13. The method of claim 1, wherein the depth-penetrating scanner comprises an ultrasonic scanner.

14. The method of claim 1, wherein the depth-penetrating scanner comprises an optical OCT scanner.

15. A processor programmed to execute programmed instructions to:(a) acquire a plurality of depth-penetrating scanner measurements from multiple scans of the sample using a hand-held, depth-penetrating scanner and detect one or more interfaces between materials in the acquired depthpenetrating scanner measurements;(b) reconstruct a 3D surface according to the acquired depth-penetrating scanner measurements and detect interfaces and associating depth-penetrating scanner measurement data indicative of material composition with 3D coordinate locations of the reconstructed 3D surface;(c) for each of a plurality of the reconstructed 3D surface coordinate locations, executing a sequence of:(i) combining the associated depth-penetrating scanner measurement data indicative of material composition, relative to the 3D surface location;(ii) assigning a label identifying an intraoral material according to the combination of depth-penetrating scanner measurement data corresponding to the coordinate location;(f) form labeled reconstructed surface image content that associates the label assignments for each of the plurality of 3D surface coordinates to the reconstructed 3D surface; and(g) display, store, or transmit the labeled reconstructed surface image content.

16. The processor of claim 15 wherein reconstructing the 3D surface comprises forming a point cloud or mesh.

17. The processor of claim 15 wherein label assignments have an identifying color.

18. The processor of claim 15 wherein the processor is further programmed to distinguish between a first and a second dental filling material according to the depth-penetrating scanner measurement data.

19. The processor of claim 15 further configured to reconstruct the 3D surface according to acquired data from a camera.

20. The processor of claim 15 wherein the processor is configured to distinguish between two or more metals.

21. The processor of claim 15 wherein the processor is configured to distinguish between healthy and diseased tissue.

23. The processor of claim 15, wherein the depth-penetrating scanner comprises an ultrasonic scanner.

25. The processor of claim 15, wherein the depth-penetrating scanner comprises an optical OCT scanner.

26. A method for identifying material content for an intraoral feature, the method comprising the steps of: at each of a plurality of coordinate locations along a surface of the intraoral feature: a) acquiring signal data from two or more depth-penetrating scanner scans taken through the coordinate location with a depth-penetrating scanner; b) analyzing the acquired signal data from each of the two or more depthpenetrating scanner scans to identify a material corresponding to the coordinate location; c) combining the analyzed signal data and assigning a label for the material at the corresponding coordinate location; and d) reporting the assigned label corresponding to the intraoral feature.

27. The method of claim 26 wherein reporting the identified material content comprises displaying the coordinate location in a color.

28. The method of claim 26 wherein reporting the identified material content comprises assigning a label to the coordinate location.

29. The method of claim 26 wherein the depth-penetrating scanner comprises an ultrasound scanner.

30. The method of claim 26, wherein the depth-penetrating scanner comprises an optical OCT scanner.

31. The method of claim 26 further comprising obtaining one or more camera images of the intraoral feature.

32. The method of claim 26 wherein identifying the material corresponding to the coordinate location comprises distinguishing normal from diseased tissue.

33. The method of claim 26 wherein identifying the material corresponding to the coordinate location comprises identifying a metal.

34. The method of claim 26 wherein identifying the material corresponding to the coordinate location comprises identifying an amalgam.

35. The method of claim 26 further comprising identifying an interface between different materials, wherein the interface lies beneath the outer surface of a tooth or gum feature.

36. The method of claim 26 wherein reporting the assigned label comprises transmitting data to a remote networked computer.

37. An OCT imaging apparatus comprising: a hand-held scanning probe configured to direct scanning light toward an intraoral sample and to direct reflected light from the sample along a detection path; a detector configured to generate an output signal according to interference of combined light from the detection path with light from a reference path;a processor that is configured to identify one or more materials at corresponding locations within the intraoral sample according to the detector output signal, wherein the processor is further configured to(a) acquire a plurality of OCT measurements from multiple scans of the sample using a hand-held scanner and detect one or more interfaces between materials in the acquired OCT measurements;(b) reconstruct a 3D surface according to the acquired OCT measurements and detect interfaces and associating OCT measurement data indicative of material composition with 3D coordinate locations of the reconstructed 3D surface;(c) for each of a plurality of the reconstructed 3D surface coordinate locations, executing a sequence of(i) combining the associated OCT measurement data indicative of material composition, relative to the 3D surface location;(ii) assigning a label identifying an intraoral material according to the combination of OCT measurement data corresponding to the coordinate location;(f) form labeled reconstructed surface image content that associates the label assignments for each of the plurality of 3D surface coordinates to the reconstructed 3D surface; and a display, in signal communication with the detector and configured to render an image showing the labeled reconstructed surface image content.

38. The apparatus of claim 37 wherein the processor uses trained classifier logic.

39. The apparatus of claim 37 wherein the labeled reconstructed surface image content further represents locations wherein the material is undetermined.

40. The apparatus of claim 37 wherein the processor generates a feature vector according to each OCT measurement.

41. The apparatus of claim 37 wherein the probe further comprises a camera.

42. A method for imaging an intraoral sample using a depthpenetrating scanner, the method comprising the steps of:(a) acquiring depth-penetrating scanner measurements from multiple scans of the sample using a hand-held, depth-penetrating scanner, wherein each depthpenetrating scanner measurement identifies an interface material and corresponding candidate material data at a coordinate location;(b) reconstructing a 3D surface according to the acquired depth-penetrating scanner measurements, wherein reconstructing comprises combining candidate material data from multiple measurements at related coordinate locations;(c) assigning a label to a plurality of coordinate locations according to the combination of corresponding candidate material data; and(e) displaying, storing, or transmitting the reconstructed 3D surface.

43. The method of claim 42 wherein the label has a designated color.

44. The method of claim 42 wherein one or more label assignments indicate an unknown material.

45. The method of claim 42 wherein the related coordinate locations include adjacent coordinate locations.

46. The method of claim 42, wherein the depth-penetrating scanner comprises an ultrasonic scanner.

47. The method of claim 42, wherein the depth-penetrating scanner comprises an optical OCT scanner.

48. A method for imaging an intraoral sample using depthpenetrating scanner, the method comprising:(a) acquiring a plurality of depth-penetrating scanner measurements from multiple scans of the sample using a hand-held, depth-penetrating scanner;(b) reconstructing a 3D surface according to the acquired depth-penetrating scanner measurements;(c) labeling one or more regions of the 3D surface according to materials identified from the acquired depth-penetrating scanner measurements; and(d) in response to an operator instruction entry, rendering results of the materials identification on a display.

49. The method of claim 48 further comprising: associating a label assignment to a tooth; and automatically updating a dental chart according to the label assignment.

50. The method of claim 48 wherein rendering results of the materials identification comprises displaying a 2D surface with one or more labeled regions.

51. The method of claim 48 wherein rendering results of the materials identification comprises displaying a labeled 2D slice extracted from the reconstructed 3D surface.

52. The method of claim 48 wherein rendering results of the materials identification comprises assigning a color corresponding to the identified material.

53. The method of claim 48 wherein rendering results of the materials identification comprises displaying text annotation indicating the identified materials.

54. The method of claim 48 wherein rendering results of the materials identification comprises highlighting an interface between two materials using color.

55. The method of claim 48 wherein rendering results of the materials identification comprises rendering one of the labeled regions of the reconstructed surface with reduced opacity over another labeled region.

56. The method of claim 48 wherein the display renders a 2D image.

57. The method of claim 48 wherein the display renders a 3D image.

58. The method of claim 48 wherein the operator instruction entry specifies a coordinate location on the reconstructed surface.

59. The method of claim 48 wherein the rendering results comprise text annotation that indicates a corresponding identified material.

60. The method of claim 48 wherein rendering results of the materials identification comprises displaying a point cloud or mesh of the surface.

61. The method of claim 48 wherein rendering results of the materials identification comprises displaying one or more regions of a single identified material.

62. The method of claim 48 wherein rendering results of the materials identification comprises displaying a portion of the surface reconstruction that contains a selected tooth.

63. The method of claim 48 wherein the operator instruction entry is entered using a display.

64. The method of claim 48, wherein the depth-penetrating scanner comprises an ultrasonic scanner.

65. The method of claim 48, wherein the depth-penetrating scanner comprises an optical OCT scanner.

66. A method for imaging an intraoral sample using a depthpenetrating scanner, the method comprising:(a) acquiring a plurality of depth-penetrating scanner measurements from multiple scans of the sample using a hand-held, depth-penetrating scanner;(b) reconstructing a 3D surface according to the acquired depth-penetrating scanner measurements;(c) labeling one or more regions of the 3D surface according to materials identified from the acquired depth-penetrating scanner measurements; and(d) automatically generating and displaying a dental chart that is annotated according to the reconstructed 3D surface and to the identified material at each of one or more teeth.

67. A method for imaging an intraoral sample using a depthpenetrating scanner, the method comprising:(a) acquiring a plurality of depth-penetrating scanner measurements from multiple scans of the sample using a hand-held, depth-penetrating scanner;(b) reconstructing a 3D surface according to the acquired depth-penetrating scanner measurements;(c) labeling one or more regions of the 3D surface according to materials identified from the acquired depth-penetrating scanner measurements; and a) displaying two or more of the labeled regions in different colors.

68. The method of claim 67, wherein the depth-penetrating scanner comprises an ultrasonic scanner.

69. The method of claim 67, wherein the depth-penetrating scanner comprises an optical OCT scanner.

70. A method for training a machine learning classifier to identify a surface material from depth-penetrating scanner scanned data, the method comprising: a) generating, from depth-penetrating scanner scan measurements obtained via a depth-penetrating scanner, a reconstruction of an intraoral surface; b) obtaining a set having a plurality of coordinate locations along the reconstructed surface; c) labeling each member of the set according to the surface material at the corresponding coordinate location; d) for each member of the obtained set, identifying one or more depthpenetrating scanner scan measurements taken at said member coordinate location; and e) submitting the one or more scan measurements to the classifier for generating machine learning logic.

71. The method of claim 70 wherein the intraoral material comprises intraoral gum tissue.

72. The method of claim 70 wherein the intraoral material is taken from the group consisting of intraoral composite dental filling materials and amalgam dental filling materials.

73. The method of claim 70 wherein the intraoral material comprises tooth bone or enamel.

74. The method of claim 70 wherein assigning the label is performed automatically.

75. The method of claim 70 wherein assigning the label is performed manually.

76. The method of claim 70 wherein the intraoral material includes infected tissue.

77. The method of claim 70, wherein the depth-penetrating scanner comprises an ultrasonic scanner.

78. The method of claim 70, wherein the depth-penetrating scanner comprises an optical OCT scanner.

Images