DETERMINATION OF STRUCTURAL CHARACTERISTICS OF AN OBJECT

Abstract

The present invention relates generally to a system and method for measuring the structural characteristics of an object. The object is subjected to an energy application processes and provides an objective, quantitative measurement of structural characteristics of an object. The system may include a device, for example, a percussion instrument, capable of being reproducibly placed against the object undergoing such measurement for reproducible positioning. The invention provides for a system and methods for analyzing measured characteristics utilizing machine learning to create a system for predicting pathologies from measurements.

Claims

1-24. (canceled)

25. A computerized system for evaluating the structural characteristics of physical objects, the system comprising: a device having an energy application tool capable of applying energy to an object to generate a measurement and an interface to the computer for storing the device measurements; a first program logic module for applying noise filtering and/or frequency domain transformations to the device measurements to produce feature vectors; a second program logic module for applying unsupervised clustering to the feature vectors; and a third program logic module for accepting a device measurement and uses membership in a specific cluster to predict structural characteristics.

26. The system of claim 25 further comprising a fourth program logic module that accepts annotations with structural characteristics to associate with specific device measurements as a means of informing the third program logic module of which clusters are associated with which structural characteristics.

27. The system of claim 25 wherein said measurement comprises energy reflected from the object as a result of energy application, or deceleration information of the energy application tool.

28. A system for detecting previously seen patterns/classes of defects in objects, comprising: at least one device having an energy application tool capable of applying energy to an object to generate a measurement and an interface to the computer for storing the device measurements; and at least one quantitative percussion device for capturing energy return data connected to one or more computers for storing and displaying energy return data, annotations, prediction results or combinations thereof; wherein said at least one computer utilize deep learning to mathematically identify patterns in energy return data sets annotated as being similar/related in one or more ways and to predict the status of new object samples.

29. The system of claim 28 wherein said at least one computer for storing is different from said at least one computer used for displaying.

30. The system of claim 28 wherein said at least one computer for storing and displaying is different from said at least one computer used for deep learning.

31. The system of claim 28 wherein said object includes anatomical and non-anatomical objects.

32. The system of claim 31 wherein said anatomical objects are teeth, teeth structures and implants.

33. The system of claim 31 wherein said non-anatomical objects includes physical structures.

34. The system of claim 28 wherein said patterns/classes being detected are cracks. damage, defects, tissue decay or combinations thereof.

35. The system of claim 33 wherein the patterns/classes being detected are physical characteristics including size, material type, shapes or geometry.

36. The system of claim 28 wherein said energy return data is captured by said device at multiple anatomical locations on a tooth.

37. The system of claim 28 wherein said objects are teeth and said classes of teeth are on a continuous scale of damage score, mobility score, or combinations thereof.

38. The system of claim 28 further comprising simulated energy return data, wherein said simulated energy return data is incorporated into the training process to strengthen the identification of specific patterns.

39. The system of claim 38 wherein said simulated energy return data comprises data from random energy return data pattern generators, varied Finite Element Models, or combinations thereof.

40. The system of claim 28 wherein said at least one computer collect repeated measurements on the same tooth by repeated application of energy on the object using the energy application tool.

41. The system of claim 40 wherein said energy application tool is a tapping rod for tapping said object, and said repeated measurements comprises varying the number of taps, varying the force level of the taps or combinations thereof on the same object.

42. The system of claim 28 further comprising incorporating a physical mount and software automation to generate and collect a large number of energy return data or to generate or collect energy return data at higher throughput.

43. The system of claim 28 further comprising a force measurement device incorporated into the system to calibrate/adjust/normalize energy return data used for training.

44. The system of claim 44 wherein said force measurement device comprises a load cell.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0152] FIG. 1 illustrates the connection of an endpoint device with computing device and a network;

[0153] FIG. 2 illustrates the operation of an endpoint device by an operator;

[0154] FIG. 3 illustrates a block diagram of a measurement device of the present invention;

[0155] FIG. 3a illustrates a cross-sectional view of a handpiece showing the internals of the present invention;

[0156] FIG. 3b illustrates a cross-sectional view of a tapping rod within the handpiece of FIG. 3a;

[0157] FIG. 4 illustrates an energy return graph of an undamaged tooth;

[0158] FIG. 5 illustrates an energy return graph of a tooth with damage;

[0159] FIG. 6 illustrates the operation of an endpoint device by an operator for entering annotations;

[0160] FIG. 7 illustrates the operation of an endpoint device by an operator for retrieving predictions;

[0161] FIG. 8 illustrates connectivity of an endpoint device with a server;

[0162] FIG. 9 shows an example of a finite element model (FEA) used as a physical simulation model;

[0163] FIG. 10 illustrates examples of physiological models of teeth and surrounding anatomy;

[0164] FIG. 11 illustrates an example of server architecture of the present invention;

[0165] FIG. 12 shows the process by which the training phase is performed when at least one trigger fires;

[0166] FIG. 13 shows the process by which unsupervised clustering is performed and a transformation function is generated;

[0167] FIG. 14 shows the process by which the production interface performs predictions;

[0168] FIG. 15 shows an example of a screen that displays predicted results;

[0169] FIGS. 16, 16a, 16b and 16c illustrate general Energy Return Graphs (ERGs) from teeth having different levels of pathology;

[0170] FIG. 17 illustrates an embodiment of a sleeve portion for the measurement devices of the present invention;

[0171] FIGS. 18, 18a, 18b and 18c illustrate block diagrams of measurement devices of the present invention;

[0172] FIGS. 19, 19(a), 19(b), 19(c), 19(d), 19(e) and 19(f) illustrate examples of Energy Return Graphs of objects or teeth with defects; and

[0173] FIG. 20 illustrates in chart form of using Same General Machine Learning Process for Multiple Applications.

DETAILED DESCRIPTION OF THE INVENTION

[0174] The detailed description set forth below is intended as a description of some of the exemplified systems, devices and methods provided in accordance with aspects of the present invention and is not intended to represent the only forms in which the present invention may be prepared or utilized. It is to be understood, rather, that the same or equivalent functions and components may be accomplished by different embodiments that are also intended to be encompassed within the spirit and scope of the invention.

[0175] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood to one of ordinary skill in the art to which this invention belongs. Although any systems, methods, devices and materials similar or equivalent to those described herein may be used in the practice or testing of the invention, some of the exemplified systems, methods, devices and materials are now described.

[0176] All publications mentioned herein are incorporated herein by reference for the purpose of describing and disclosing, for example, the designs and methodologies that are described in the publications which might be used in connection with the presently described invention. The publications listed or discussed above, below and throughout the text are provided solely for their disclosure prior to the filing date of the present application. Nothing herein is to be construed as an admission that the inventors are not entitled to antedate such disclosure by virtue of prior invention.

[0177] In general, the system may include the following aspects and/or a sub-combination thereof, which are discussed additionally above: an aspect including a first program logic module (“Acquisition”) that includes the functions for collecting device measurements of physical objects (a “Collection” aspect), which may be anatomical or non-anatomical; and accepting, from a user, expert annotations on device measurements regarding the structural characteristics of the measured physical objects (an “Annotation” aspect); another aspect including a program logic module (“Training”) executing a training cycle for training a machine learning algorithm on a ground truth dataset including stored device measurements and/or expert annotations to create a transformation function; and a further aspect including a program logic module (“Prediction”) exposing a production interface for performing predictions on the device measurements utilizing the transformation function. In general, structural characteristics may include, but are not limited to, the presence or absence of a physical characteristic, trait or property, and/or an indication of a degree, level or severity of such (e.g. a discrete, metered or arbitrary indication such as low, medium, high or the like).

[0178] In the aspect termed as Collection, which may form part of the Acquisition aspect, the present invention relates to a system for compiling test results from a multitude of objects which may or may not include test results of an object tested over a period of time. In some embodiments, each of the test results may be generated using an instrument or device designed and/or adapted for collecting test results or data, such as an endpoint device.

[0179] FIG. 1 illustrates an embodiment of the architecture of an endpoint device, for example, for a dental measurement. Endpoint devices may generally be located in, for example, dentists' offices or other locations where measurements may be taken on object (e.g. teeth, implants, etc.) of patients. An endpoint device may generally include or be connected to a personal computing device such as a PC workstation, a laptop, a tablet, or some other general computing device that may connect to a larger network such as the Internet or a private network. At least one device may be attached to the endpoint device, either via a wired data transmission technology such as for example USB or FireWire or via a wireless data transmission technology such as Bluetooth.

[0180] FIG. 2 illustrates an embodiment of the operation of the endpoint device by a user, such as technician. First, the endpoint device presents a login screen to the technician, in order to establish his/her identity, and this identity is checked via a call to the central server. The endpoint device then presents a user interface for creating and/or searching for a record corresponding to the patient being examined. The endpoint device then presents a screen showing the physical objects, for example teeth, to be annotated. The physical objects may be laid out in a graphical form, such as in the shape of a jaw, in order to more easily navigate to one specific object based on its anatomical location relative to other objects. The endpoint device prompts the technician to choose one of these physical objects, for example a tooth, and the connected device may then be placed over the corresponding tooth on the patient. The connected device captures measurement data, and the endpoint device saves this measurement data to the endpoint device's storage and/or the central server. This process may then be repeated until all teeth whose data is to be captured have been examined.

[0181] The device suitable for use in testing the object may include may include housing having a longitudinal axis, with an open end and an energy application tool, for example, a tapping rod, or impact rod mounted inside the housing for axial movement along the longitudinal axis of the housing, as shown in FIG. 3 which shows an embodiment of the device discussed above. In some embodiments, the system may include a handpiece 104, in the form of a percussion instrument. The handpiece 104 may have a cylindrical housing 132 with an open end 132a and a closed end 132b. The open end 132a is tapered as exemplified here, though other configurations are also contemplated. An energy application tool 120, for example, a tapping rod 120, may be mounted inside the housing 132 for axial movement, as noted above. The handpiece also includes a drive mechanism 160, mounted inside the housing 132 for driving the tapping rod 120 axially within the housing 132 between a retracted position 128 and an extended position 129 during operation. In the extended position 129, the free end of the tapping rod 120 extends or protrudes from the open end 132a of the housing 132, as shown. The drive mechanism 160 may include an electromagnetic coil 156, to be discussed further below. The tapping rod 120 may have a permanent magnetic ensemble 157 mounted at the end away from the free end. The electromagnetic coil 156 of the drive mechanism 160 may be situated behind the other end of the tapping rod 120, resulting in a relatively small outside diameter for the handpiece 104.

[0182] The mounting mechanism for the energy application tool 120, for example, tapping rod 120 may be formed by bearings 1003 and 1004, as shown in FIG. 3b, for receiving or supporting the tapping rod 120 in a largely friction-free manner. The magnetic or propulsion coil 156 may be situated in the housing 132 adjacent to the permanent magnet 157 and is axially behind the permanent magnet 157. The magnetic coil 156 and the permanent magnet 157 form a drive for the forward and return motion of the tapping rod 120. The drive coil 156 may be an integral component of the housing 130 and may be connected to a supply hose or line 1000.

[0183] The two bearings 1003 and 1004 may be substantially frictionless and may include, as shown in FIG. 3b, a plurality of radially inwardly extending ridges separated by axial openings 1400. The axial openings 1400 of the bearing 1003 allow the movement of air between a chamber 1500 which is separated by the bearing 1003 from a chamber 1600, which chambers are formed between an inner wall surface of the housing 132 and the tapping rod 120. Air movement between these chambers 1500 and 1600 may thus compensate for movement of the tapping rod 120.

[0184] Referring again to FIG. 3, a sleeve 108 is positioned towards the end 132a and extending beyond it. The sleeve 108 envelops the end of the housing 132a and is flattened at its end 116 for ease of positioning against a surface of an object 112 during operation. The sleeve aids in the positioning of the handpiece 104 on the object to stabilize the handpiece during operation. The sleeve 108 may also include a tab 110, as shown in FIG. 17, protruding from a portion of its end 116, so that when the open end 116 of the sleeve 108 is in contact with a surface of the object 112 undergoing the measurement, the tab 110 may be resting on a portion of the top of the object 112, as shown in the FIG. 3b. The tab 110 and the sleeve 108 both assist in the stabilizing and repeatable positioning of the handpiece 104 with respect to the object 112 and the tab 110 may be placed substantially at the same distance from the top of the object 112 every time. As noted above, the object may include an anatomical structure or a physical structure.

[0185] FIG. 18 depicts embodiments of other devices that are applicable for the present invention. The system may include a handpiece 100 having a housing 102 which houses the energy application tool and sensing mechanism, as illustrated in the block diagram of FIG. 18. In general, a handpiece may refer to a handheld device, but may also include, without limitation, any other appropriate form for the desired application, such as mounted devices or tool/mechanically/robotically articulated devices. The handpiece 100 may also be referred to, for example, as a device or instrument interchangeably herein. In some embodiments, the energy application tool 110, as illustrated, may be mounted within the housing 102 for axial movement in the direction A toward an object, and such axial movement may be accomplished via a drive mechanism 140. Drive mechanism 140 may generally be a linear motor or actuator, such as an electromagnetic mechanism which may affect the axial position of the energy application tool 110, such as by producing a magnetic field which interacts with at least a portion of the energy application tool 110 to control its position, velocity and/or acceleration through magnetic interaction. For example, an electromagnetic coil disposed at least partially about the energy application 110 may be energized to propel the energy application tool 110 forward toward the object to be measured, as illustrated with the electromagnetic coil 140. The electromagnetic coil may also, for example, be alternatively energized to propel the energy application tool 110 backward to prepare for a subsequent impact. Other elements, such as rebound magnetic elements, may also be included, such as to aid in repositioning of the energy application tool 110 after propelling via the electromagnetic coil. The drive mechanism 140 and/or other portions of the instrument may generally be powered by a power source, as shown with power source 146, which may be a battery, capacitor, solar cell, transducer, connection to an external power source and/or any appropriate combination. An external connection to a power source, either to power the handpiece 100 or to charge the internal power source, such as the power source 146, may be provided, such as a power interface 147 in FIG. 18, which may include, for example, a power contact for direct conductive charging, or the power interface 147 may utilize wireless charging, such as inductive charging.

[0186] In some other embodiments, the energy application tool 110 may be utilized to move substantially in a direction A which may be perpendicular or substantially perpendicular to the longitudinal axis of the housing 102, as illustrated in the block diagram of a handpiece 100 in FIG. 18a. As illustrated, the energy application tool 110 may, for example, be substantially L-shaped to accommodate the interaction with the drive mechanism 140 and protrude in direction A, substantially perpendicular to the axis of the housing 102. As illustrated in an example, the drive mechanism 140 may act on the energy application tool 110 to cause it to rock on a pivot 110a, causing it to move in direction A at its tip. The drive mechanism 140 may utilize, for example, an alternating magnetic element which may act on the energy application tool 110 to cause it to move alternatingly in two directions, such as up and down. In another example, the bend portion of the L-shaped energy application tool 110, such as shown with bend 110b, may include a flexing and/or deformable construction such that a linear force applied by the drive mechanism 140 may push the energy application tool 110 in the direction A at the tip by conveying the forward motion around bend 110b. For example, the bend 110b may include a braided, segmented, spring-like and/or otherwise bendable section that may also convey motion and/or force around a bend. In general, the shape of the L-shaped energy application tool 110 may generally include other angles besides 90 degrees, such as between approximately +/−45 degrees from the rearward portion 110d. In some embodiments, the energy application tool 110 may also include multiple portions which may be separable, such as portions 110c and 110d, such that, for example, the portion 110c may be removed and disposed between uses or patients, such as to aid in preventing cross-contamination. In general, the separable portions may include an interface to couple them for use in a measurement such that they substantially act as a unitary energy application tool 110, as described below.

[0187] In some embodiments, the L-shaped energy application tool 110 may rock on a pivot 110a, such as, for example, with an external force applied from a drive mechanism 140, as shown in FIGS. 18b and 15c. For example, the drive mechanism 140 may apply alternating forces to the energy application tool 110 to cause it to rock about the pivot 110a, such as with a force applied D from portion 140d applied to the rearward portion 110d to cause rocking in direction A′ away from a target object, as shown in FIG. 18b, or with a force applied E from portion 140c applied to the rearward portion 110d to cause rocking in a direction A″ toward the target object such that the energy application tool 110 is driven in direction A, as shown in FIG. 18c. The forces D and E may be applied by any appropriate method, such as, for example, by applying a magnetic force on the energy application tool 110, which may contain a magnetic or metallic element which may respond to the application of force from the drive mechanism 140. In general, the shape and arc of the rocking motions A′ and A″ may be designed such that the energy application tool 110 impacts the target object in a direction substantially perpendicular to the target object surface, as shown with the rocking A″ into a substantially vertical orientation of the bent portion 110c around bend 110b in FIG. 18c. To reset the device 100 for a subsequent measurement, the portion 140d may apply a return force D, as shown in FIG. 18b, to cause rocking A′ to return the energy application tool 110 to a withdrawn or resting state. In general, the interior of the device 100 may be adapted to allow for the rocking motions A′ and A″ without interfering with the energy application tool 110.

[0188] Other examples of endpoint devices may include, for example and without limitation, those described in U.S. Pat. Nos. 6,120,466, 7,008,385, 6,997,887, 9,358,089 9869606, U.S. Pat. No. 10,488,312, PCT/US17/69164, PCT Patent Application Ser. No. PCT/US 20/40386, U.S. patent publication No. 20190331573, PCT/US2018/068083 and/or PCT publication WO2019133946, which are incorporated by reference in their entireties.

[0189] Any of the devices described above, for example, a handpiece 100, delivers a free-floating or substantially free-floating energy application tool, such as the energy application tool 110 or 120, for example, a tapping probe to the object, for example a tooth and/or implant with a consistent kinetic energy just prior to each percussion of the object. From the resulting data, the energy returned to the energy application tool normalized by the kinetic energy of the energy application tool prior to impact vs time may be determined and analyzed. The response, such as a percussion response is plotted as Percent Energy Return (ER) on the vertical axis and Time (micro seconds or μs) on the horizontal axis. Each ER value is measured and plotted, at time increments of, for example, 4 μs along the horizontal axis. The vertical axis may autoscale to the highest ER value and the horizontal axis may range from 0 to 0.5 ms (milliseconds), as shown in the percussion response curves of FIGS. 4 and 5. The numerous ERGs of FIGS. 4 and 5 may be representative of tests on numerous defect free objects or numerous objects with defects, respectively.

[0190] The same percussion response as shown in FIGS. 4 and 5, from a tooth or an implant may be analyzed by two different methods. The first method analyzes the percussion response for the tooth/implant mobility using Loss Coefficient (LC) characteristics whereas the second method analyzes the same percussion response for the tooth/implant internal/proximal mobility using Normal Fit Error (NFE) characteristics. The results of the percussion response for a dental setting, including ER graph, LC and NFE values are displayed on the computer screen for the clinician's review. Thus each one of the curves in FIGS. 4 and 5 illustrate typical representations of computer screen displays of percussion responses for a tooth with no pathology and with structural pathology, respectively.

[0191] The device may be coupled to a computer that uses an additional path to analyze the percussion response to generate mobility of the object and its associated fixed structures for example, teeth and/or implants. The software assesses the characteristics of the teeth and implants by identifying the presence of any structural characteristics or pathology (e.g. crack) within the internal and/or proximal structure of tooth or implant. As the tooth structure breaks down over time due to the normal tooth/implant wear and parafunction, the level of structural pathology may increase over time resulting in development of additional mobility.

[0192] The clinical implication is the higher the mechanical interaction between the structures within the site the higher the internal/proximal mobility detected by the device described above. Additional site mobility is demonstrated in the Energy Return Curve or Energy Return Graph (ERG) by the shape of the curve. Specifically, the more structural pathology there is within the structure of the site (internal/proximal), the more the shape of the curve deviates from a uniform single peak. In addition, the software uses Levenberg-Marquardt algorithm to characterize the shape of the curve in terms of Normal Fit Error or NFE: higher NFE values are associated with greater additional mobility within the internal and/or proximal structure of tooth and/or implant. FIGS. 16, 16a, 16b and 16c illustrate generally characteristic ERG shapes that correspond to different levels of pathology. FIG. 16 shows a generally normal tooth with no structural pathology, FIG. 16a shows a generally mild level of pathology (as shown with the small additional peak), FIG. 16b shows a generally moderate level of pathology (as shown with the multiple small to medium peaks), and FIG. 16c shows a severe level of pathology (as shown with the large additional peaks).

[0193] Any of the devices described above, for example, a handpiece, delivers a free-floating energy application tool, for example, a tapping rod to the object, for example a tooth and/or implant with a consistent kinetic energy just prior to each percussion of the object. From the resulting data, the energy returned to the energy application tool be normalized by the kinetic energy of the energy application tool prior to impact vs time is determined and analyzed. The percussion response is plotted as Percent Energy Return (ER) on the vertical axis and Time micro seconds (or μs) on the horizontal axis. Each ER value is measured and plotted, at time increments of, for example, 4 μs along the horizontal axis. The vertical axis may auto-scale to the highest ER value and the horizontal axis may range from 0 to 0.5 ms (milliseconds) (get one curve form addition explanation paper). The energy return curve shown in FIG. 4 represents an object without any defects, as shown with a single peak without accessory peaks. In FIG. 5, an energy return curve for an object with defect is shown, as shown with the main peak and an additional peak that indicates damage.

[0194] In general, when a percussion device is used, the Normal Fit Error (NFE) may be determined as follows:

The Defect Severity Quotient (DSQ) is equal to NFE×1000.
The Damage (D) is given by D=27×ln(DSQ)−61, where ln is the natural logarithm. In some cases, the Damage (D) may also be simply equal to DSQ.

[0195] As noted above, the force may be determined by a sensor coupled to the energy application tool. The energy return data generated in a test is normalized before the fit to these data is performed to make the fitting process simpler. In other words, for an energy application tool that is a tapping rod, the equation to be fitted is for E.sub.r/E.sub.rmax instead of just the energy return, E.sub.r, where E.sub.r is the energy return, which characterizes the elastic energy of this force measurement. For example, E.sub.r is defined as E.sub.r=F.sup.2/2K, where F is the resultant percussion force and K is the stiffness of the energy application tool, for example, the tapping rod assembly. The normalized energy return, E.sub.rt is the energy return during impact divided by the kinetic energy of the taping rod just before impact with the sample. The energy return/impact energy variation with tie for a defect-free calibration sample could be expressed in the following form:

[00005]${\stackrel{\_}{E}}_r=\beta {\sin }^2\left(\gamma t\right)\exp \left[-\frac{{\left(t-\varphi \right)}^2}{\psi }\right]$

where t is time and β, γ, ϕ, and ψ are parameters that are determined via a nonlinear regression fit to measured data; or the energy return/impact energy variation with tie for a defect-free calibration sample could be expressed in the following alternate form, as noted before:

[00006]${\stackrel{\_}{E}}_r=\begin{array}{cc}\beta & \exp [-\frac{{\left(t-\varphi \right)}^2}{\psi }\end{array}].$

The NFE is equal to the cumulative error between the normalized measured data and a nonlinear regression fit of the equation above to the normalized measured data. Thus, the NFE represents that overall difference between the shape of an ideal energy return response for a defect-free sample and that for the measured data.

[0196] In general, the loss coefficient may be derived from damping characteristics of an object, for example, tooth and implant. After application of kinetic energy to the object, the relative extent to which the object deforms inelastically and dampens elastic energy may be characterized as its loss coefficient, η, given by:

η=D/2πU

where D is the total energy dissipated (or lost) per unit volume and U is the elastic energy per unit volume. The stability index (SI) is equal to Fp{circumflex over ( )}2/Fc{circumflex over ( )}2×100 where Fp is the maximum percussion force measured by the sensor in the percussion rod for the sample tested and Fc is the maximum percussion force for a stiff calibration sample (e.g. aluminum alloy or stainless-steel block). Other calibration materials may also be used. The Mobility is equal to 100−SI.

[0197] In the Annotation aspect of the invention, which may form part of the Acquisition aspect, a system may include an annotation user interface for collecting annotations, such as expert annotations. These annotations may associate data records for physical objects such as teeth with a label naming a specific defect indication, such as “loss of cement seal”, “oblique crack”, etc. For each such annotation, the interface may allow at least one device measurement to be captured for at least one, for example, tooth of a single patient and for at least one clinical finding of the physical condition of the teeth to be inputted for the teeth. In other embodiments, the interface for collecting expert annotations may use a pictorial diagram of the object and other surrounding objects, for example, a depiction of a patient's lower and/or upper jaw. In some embodiments, the tooth measurement and annotation may be captured at the same time. In other embodiments, the annotation may be captured independently from (for example, at a later point in time relative to) any target device measurement.

[0198] In some exemplary embodiments, a user may enter an annotation by selecting one or more matching categories from a single-level or multi-level ontology, e.g., crack in restoration (crown, filling, etc.), crack between restoration and tooth (microleakage, open margins, decay, etc.), crack between tooth and bone (infection, orthodontic origin, trauma origin, etc.), enamel cracks, etc.

[0199] In some embodiments, the device measurements and/or expert annotations may be stored using a distributed computing environment, such as a cloud. Storage on, for example, a cloud may allow multiple expert annotations to be collected simultaneously and decrease the time for accumulating an expert annotation dataset in order to improve prediction accuracy. In some embodiments, device measurements and/or expert annotations may be collected on multiple instances of the system and consolidated onto one or more of those instances. In some embodiments, the device measurements and/or expert annotations entries may be encrypted.

[0200] As mentioned before, machine learning techniques may include regression (e.g., logistic, linear), clustering (e.g., k-means), neural networks (e.g., deep learning), classifiers (e.g., support vector machine, decision tree, random forest), deep learning, etc. The base machine learning techniques utilized may themselves be standardized techniques, and not themselves unique; however, the present inventors have found certain unique adaptations to the types of data stored in a case file to make them useful to a machine learning algorithm. For example, percussive energy return graphs produced by a system and method as described herein above and below, for measuring and evaluating structural characteristics of an object, whether anatomical or non-anatomical, in a non-invasive manner and/or using a non-destructive method of measurement may be employed. Structural characteristics of an object may be identified based on measurements of the same or other objects previously made using and captured by the system using a device such as exemplified in FIGS. 3, 3a, 3b, 17, 18a-c and their corresponding descriptions, or those as described in U.S. Pat. Nos. 6,120,466, 7,008,385, 6,997,887, 9,358,089 9,869,606, U.S. Pat. No. 10,488,312, PCT/US17/69164, PCT Patent Application Ser. No. PCT/US 20/40386, U.S. patent publication No. 20190331573, PCT/US2018/068083 and/or PCT publication WO2019133946, which are incorporated by reference in their entireties, may be filtered and transformed into spectrograms for use in deep learning. Models may then be trained, versioned, and stored in a secure database running on a set of centralized cloud-based servers. Models may take the form of complex mathematical transformation functions that, when presented with a set of inputs for a new case file, may reveal one or more likely structural defect types as output. The primary users of this aspect of the invention are data scientists and software engineers that perform maintenance on the system.

[0201] In some exemplary embodiments, at least one machine learning algorithm may be configured to train a computer and construct at least one transformation function to predict the structural characteristics of an object, anatomical or anatomical, such as cracks, micro-cracks, fractures, microfractures; loss of cement seal; cement failure; bond failure; microleakage; lesions; decay; structural characteristic of the foundation or environment to which the object may be anchored or present in, structural integrity in general or structural stability in general, or for a tooth or tooth structure, the physical conditions, for example, crack in restoration (crown, filling, etc.), crack between restoration and tooth (microleakage, open margins, decay, etc.), crack between tooth and bone (infection, orthodontic origin, trauma origin, etc.), enamel cracks, etc., using a dataset composed of device measurements and the associated annotations, which in combination may also be known as ground truth, or a subset thereof. The device measurement may include the ERGs, the loss coefficients, the normal fit error or NFEs or a combination of loss coefficient or mobility and normal fit error (NFE) or damage or stability based on the shape of the ERG versus time curve. The feature vector used by the machine learning algorithms may be constructed for each object, for example, a tooth based on one or more elements present in the device measurement and in zero or more associated annotations, e.g., clinical findings, ontological classifications such as defect type and/or tooth geometry, etc., or laboratory findings, actual measurements, etc. The labels to be predicted by the machine learning algorithms may include specific physical conditions in specific regions of the physical object, for example, a tooth, or regions relative to some physical characteristic, for example, a tooth crack. The labels may also include measures such as a severity level of the condition and an estimate of remaining lifespan or usefulness.

[0202] An ERG of most defect-free object or tooth may generally have the shape of a single gaussian graph, as noted above, and as shown in FIG. 16. For an object or tooth with any defect, its ERG generally has more than one peak, as shown in FIGS. 16a-c. However, for a tooth or object with defects, its ERG generally has more than one peak and the heights of the peaks may vary.

[0203] In some embodiments, in NFE calculation, attempts to fit the ERGs of objects or teeth with defects that generally have more than one peak, into a best fit single gaussian with one peak and then report the difference as error may yield different error information. In general, the larger the error numbers, the farther the best fits are from the actual ERGs of the objects or teeth. For example, an ERG of a perfect, defect free object or teeth such as shown in FIG. 16, it has an NFE of zero (0); while the ERG for a tooth or object with defects and having two side-by-side, well-formed and separate peaks, as shown in FIG. 19(e) may have an NFE value that may vary, depending on whether the best fit is to fit over a single peak or to fit over both peaks, the ERG for a tooth or object with defects and having a split peak, as shown in FIG. 19(f), will generally have an NFE value close to 0.02 (or 20 if adjusted by multiplying by 1000) because the split peak produces a minimal error relative to a single gaussian. For the above examples, one may be able to infer from the NFE numbers that the defects of a tooth or object with an ERG similar to FIG. 19 (f) might be different from the defects of a tooth or object with an ERG similar to FIG. 19(e).

[0204] In other embodiments, instead of fitting an ERG having more than one peak into a single gaussian and calculating how different the single best-fit gaussian is to the ERG, as in NFE calculations, the ERG maybe fitted instead into multiple gaussians and summarized as a Gaussian Score, as shown in FIGS. 19(a) and (b). As noted above, teeth or objects may have defects, and thus most ERGs generated by the system and method as described herein above and below, for measuring and evaluating structural characteristics of an object, whether anatomical or non-anatomical, in a non-invasive manner and/or using a non-destructive method of measurement may have more than one peak. Rather than fitting potentially multiple-peaked ERGs to a single gaussian, one may choose to fit the multiple-peaked ERGs to multiple gaussians, as shown in FIGS. 19(a) and (b) with the dotted line curves under ERGs. Thus, a Gaussian Score represents a decomposition of an ERG of an object or tooth with defect or defects into a set of gaussians which fundamentally is a count of the number of gaussians or Gaussian Mixture Models (GMM) with some adjustments, including discounts to be discussed further. For example, as noted above, an ERG of a perfect, defect free object or teeth such as shown in FIG. 16, it has a GS of one (1); while the ERG for a tooth or object with defects and having two side-by-side, well-formed and separate peaks, as shown in FIG. 19(e) has a GS score of two (2), the ERG for a tooth or object with defects and having a split peak, as shown in FIG. 19(f), has a GS value of about 1.5, because there is uncertainty about whether this is a 1 gaussian or 2 gaussian decomposition. Similar to NFEs, one may infer from the GS numbers that the defects of a tooth or object with an ERG similar to FIG. 19(f) might be different from the defects of a tooth or object with an ERG similar to FIG. 19 (e).

[0205] In the previously discussed annotation step, in one aspect, a human may notate which ERGs may represent an object or tooth with cracks. AI may then predict the structural characteristics, for example, the structural characteristics like cracks, micro-cracks, fractures, microfractures; loss of cement seal; cement failure; bond failure; microleakage; lesions; decay; structural characteristic of the foundation or environment to which the object may be anchored or present in, structural integrity in general or structural stability in general, of the object or tooth. In another aspect, a human may identify a good Gaussian Mixture Model (GMM) for each ERG, and the GMM may be manifested as an annotation for that ERG. Though they may not be perfect, there are software libraries for computing GMMs and Deep Learning may be used to train algorithms to recognize GMMs from thousands of artificially-created ERGs which can be advantageous as reliance may not need to be solely on actual measurements on tooth or other objects for data collection. In a further or final optimization step, similar to what happens with NFE calculation, the error between the GMM and the original ERG may be minimized. Unlike in NFE, which is a measurement of residual error relative to a single gaussian, the deviations from a single gaussian for the GS are captured by additional gaussians and there may be nearly zero (0) residual error for GMMs. Thus, ERGs that cannot achieve a satisfactory near zero error GMM may be abandoned. Further, Deep Learning may be used to reduce the cases of significant residual error as much as possible. AI may then predict the structural characteristics as above, for example, the structural characteristics like cracks, micro-cracks, fractures, microfractures; loss of cement seal; cement failure; bond failure; microleakage; lesions; decay; structural characteristic of the foundation or environment to which the object may be anchored or present in, structural integrity in general or structural stability in general, of the object or tooth by predicting a GMM. The computer may count the number of Gaussians and may subtract discounts if some predicted Gaussians are redundant. Though they may not be perfect, there are software libraries for computing GMMs.

[0206] Discounts may be defined in the following manner. For example, for a perfect, defect-free tooth or object, as shown in FIG. 16, an NFE will be zero (0), while the GS will be one (1) since only one gaussian is present, according to above. For two side-by-side, well-formed and separate peaks, as shown in the first subgraph of FIG. 19(e), the NFE value may vary, depending on whether the best fit is to fit a single peak or fit over both peaks; while the GS will be two since there are two (2) gaussians present, as shown in the second subgraph (gaussians summed) and third subgraph (gaussians separate). For an ERG with a split peak, as shown in the first subgraph of FIG. 19(f), an NFE score of the ERG will be generally close to 0.02 (or 20 if adjusted by multiplying by 1000) because the split peak produces a minimal error relative to a single gaussian; while the GS will be about one and a half, (1.5) since there is uncertainty about whether this is a one gaussian or two gaussians, as shown in the second subgraph (gaussians summed) and third subgraph (gaussians separate). Thus, there are generally multiple possible well-fitting GMM decompositions for any ERG. Therefore, rather than directly reporting the total number of gaussians, a discount may be computed based on how many of the gaussians for which there is uncertainty about whether they ought to be combined with others. All combinations of a GMM's gaussians are considered to see whether the gaussians could be replaced by a single gaussian, resulting in a probability-weighted discount. For example, as shown in FIGS. 19, both of the graphs on the right-hand and left hand side have a same score of about 1.86 because the first two (2) gaussians can be combined and there is a small probability of about fourteen percent (14%) that the two peaks are actually one peak.

[0207] FIGS. 19(c) and 19(d) show two not too differently shape ERGs generated using an embodiment of the system of the present invention, having a device with a tapping rod for performing a percussion action on the same object at different but closely spaced times. In calculating NFEs for these two ERGs, FIG. 19 (c) has an NFE of 0.03 (or 30 after multiplying the result by 1000), while FIG. 19(d) has an NFE of 0.06 (or 60 after multiplying the result by 1000). In calculating Gaussian Scores, the variation is between 2.45-2.9. Although not always the case. in this particular example, the GS representation may be better.

[0208] Not all ERGs may be decomposed and analyzed to find a low residual error GMM. Adding additional deep learning models for common ERG patterns may be able to further improve the residual error. However, GMMs with an excessive number of gaussians may take a long time to analyze, thus further optimization may be employed to help to resolve this issue.

[0209] Thus, while an NFE measures deviation from the ERG of the object or tooth that is defect free by calculating the difference between its best-fit gaussian from the ERG, a Gaussian Score fundamentally is a count of the number of gaussians or Gaussian Mixture Models (GMM) with some adjustments for calculating the difference between the decomposition and the ERG. It may also be possible to employ both NFE and GS to gain more insight into some defects. GMMs may also be used and may be a more useful feature space for doing deep learning pattern detection for small cracks, or may be even different crack types than just raw energy/force return data.

[0210] FIG. 20 illustrates how the different aspects of the machine learning system described in this specification are applied to solve two different clinical problems of interest: detecting cracks and determining crack severity (GS). The main difference is that the annotations for crack detection are comprised of clinicians' crack indications, and the annotations for GS are GMMs which, during the post processing step, are mathematically transformed into Gaussian Scores as described earlier.

[0211] FIG. 6 illustrates an embodiment of the operation of the endpoint device by a user for entering annotations. Such a user may generally be an expert practitioner for entering annotations such that, for example, the records may be annotated properly. First, the endpoint device presents a login screen to the expert, in order to establish his/her identity, and this identity is checked via a call to the central server. The expert then sees a menu of device measurement records, possibly across multiple patients, from which to choose to work on, based on data stored on the central server. After the expert selects one of these records, the endpoint device presents a screen showing the physical objects, for example teeth, to be annotated. The physical objects may be laid out in a graphical form, such as in the shape of a jaw, in order to more easily navigate to one specific object based on its anatomical location relative to other objects. The endpoint device prompts the expert to select a single tooth and annotate the tooth's record with any discernable clinical findings and/or other data, if any, such as whether a crack exists or if a filling or crown is present. Annotations are stored on the endpoint device and/or the central server. This process may then be repeated until the expert has completed all annotations the expert is able to enter.

[0212] FIG. 8 shows an embodiment of the connectivity between an endpoint device and the central server. During operations such as logging in and saving of device measurement data and/or annotations, network connections may be established to the central server. These network connections may generally be established over a public network such as the Internet or a private network. The connections may further be encrypted, such as to prevent interception or tampering of records and/or to comply with government regulations. The central server may also use these connections to return previously saved device measurement records, annotations, and physical condition predictions to the endpoint device. Network connections may use a standard protocol/convention for transmitting requests and responses, such as Representational State Transfer over Hypertext Transfer Protocol (REST/HTTP) and JavaScript Object Notation (JSON).

[0213] FIG. 11 shows an embodiment of the architecture of the central server. The central server may physically reside on a single computer or across a set of multiple computers connected by a private network. One component of the central server is responsible for storing and managing user credentials for the users of the endpoint devices. This component may use standard technologies such as Lightweight Directory Access Protocol (LDAP) or another directory server. The other components delegate all user management and authentication of users during information exchanges over the network to this component. Another component is responsible for storing and managing data from the Acquisition aspect (e.g. the Collection and/or Annotation), such as the device measurement data, annotations, transformation functions, and physical simulation models, potentially in an encrypted form. This component may use standard technologies such as a SQL relational database and filesystem-level encryption built into operating systems such as Linux. Yet another component is responsible for running training cycles in the Training aspect, in which data from the data management component is retrieved on a periodic basis and fed into a machine learning algorithm. This component of the central server may typically be equipped with graphics processing units (GPUs) to accelerate machine learning tasks. A final component is responsible for exposing a production interface that accepts requests from endpoint devices for use in the Prediction aspect, such as for predicting physical conditions of objects, for example teeth, once provided with a set of device measurements. The components of the central server communicate with one another and/or directly with endpoint devices in order to service requests that are received via the network from endpoint devices.

[0214] In the Training aspect, as discussed additionally above, a system trains at least one machine learning model to produce at least one transformation function using the data collected in the Acquisition aspect or Collection aspect and the annotations collected in the Acquisition aspect or Annotation aspect. In some exemplary embodiments, a combination of one or more of the device measurement data, ancillary data, simulated data, and expert annotations (the combination being referred to as “ground truth data”), may be utilized in training a machine learning model on the patterns that exist in device measurements, ancillary data, and/or simulated data, and the resulting transformation function may be used to predict the true condition of an object, for example, an anatomical object such as a human tooth. Applied repeatedly when new and/or updated ground truth data and/or annotations are introduced, the process may be capable of generating better and more accurate predictions.

[0215] FIG. 12 shows an embodiment of the process by which the training phase is performed when at least one trigger fires. A trigger may be based on an event such as for example the introduction of new annotations, based on a pre-established timing interval, and/or based on a manual triggering by a user. When a trigger fires, the central server retrieves relevant ground truth data from the device measurement and annotation database component. A physical simulation model is applied to generate additional ground truth dataset entries to be included. The machine learning training algorithm is then performed, and this may take a long time to execute. At the end of the training, the resulting transformation function is then saved to the database component.

[0216] FIG. 13 shows an embodiment of the process by which unsupervised clustering is performed and a transformation function is generated. First, at least one device measurement, for example, waveform data from a probe device, may be passed through a filter to remove high and low frequencies. Then, transformations such as a Fast Fourier Transformation may be used to convert the data into a form more useful for a machine learning algorithm, such as a feature vector comprised of frequency components. The feature vectors are then assembled together within the feature space and clustered using a technique such as k-means clustering. Each cluster is analyzed to identify the most prominent clinical findings that are associated with the cluster's device measurements as recorded inside annotations. The generated transformation function is thus configured to accept a device measurement, perform the same filtering and transformation steps described above, place the generated feature vector in the feature space, and return the clinical findings determined to be most prominent in the cluster that is the closest match to the generated feature vector.

[0217] The present invention may also include a simulation model component, which may form part of the Training aspect. FIG. 9 shows an example of a finite element model (FEA) used as a physical simulation model. This analysis method may involve the use of numerical models to simulate actual testing using the device described herein. In general, modeling and simulation may be desirable for training the system and its predictive capabilities with simulated models that may embody test objects that have not been tested, are not readily available for actual physical testing, etc.

[0218] In an example of the modeling, a physiologically accurate 3D model of a mandibular second molar (site 18) was created using a solid modeling computer-aided design program using 3D x-ray computer tomography tooth data, but the same process may be applicable to other teeth as well as other solid objects. The models include both enamel and dentin together with a pulp chamber, the periodontal ligament (PDL) and surrounding bone, examples of which are shown in FIG. 10.

[0219] The solid models were then exported to a computer-aided engineering program for meshing the solids. A non-linear finite element solver may be appropriate for modeling nonlinear material behaviors such as those reported for the PDL as well as transient environmental conditions including percussion were used. It was necessary to include a percussion rod in the present simulation models to fully analyze a percussion event using comparisons with experimental data. The elastic modulus of the percussion rod, its mass and an initial velocity were inputted into the program. The resultant percussion force was measured by a piezoelectric sensor in the rod.

[0220] The percussion response from the measurement was plotted as an Energy Return (ER) versus time curve. Examples of energy return versus time curves are shown in FIGS. 4 and 5, for an object without defects and one with defects, respectively.

[0221] The FEA models may include a large number of elements, for example, about 500,000 to about 1,000,000 elements each (FIG. 9). A second-order isoparametric three-dimensional 4-node tetrahedron for the PDL, an 8-node, isoparametric, arbitrary hexahedral for the percussion probe, and a linear isoparametric three-dimensional tetrahedron for the rest of the model may be used. Boundary conditions may be defined to minimize or prevent free body motion so that the elements on the outer surfaces of the object, for example, the bone may be constrained. The models were run with a time increment of for example, 4 μs.

[0222] A direct integration method may be used to obtain the solution to the equations of motion for the models. Additionally, viscous damping may be included in the analysis using classical Rayleigh Damping (RD) which is convenient for an incremental approach to a numerical solution. The damping matrix D is defined as a linear combination of the mass and stiffness matrices of the system and damping coefficients are specified on an element-by-element basis. Rayleigh damping uses coefficients on the element matrices and is represented by the equation

[00007]$D=\stackrel{n}{\underset{i=1}{.Math.}}\left\{{\alpha }_i{M}_i+\left({\beta }_i+{\gamma }_i\frac{\Delta t}{\pi }\right)K_i\right\},$

where D is the global damping matrix, M.sub.i is the mass matrix multiplier for the i.sup.th element, K.sub.i is the stiffness matrix multiplier for the i.sup.th element, α.sub.i is the mass damping coefficient on the i.sup.th element, β.sub.i is the usual stiffness damping coefficient on the i.sup.th element, γ.sub.i is the numerical damping coefficient on the i.sup.th element, and Δt is the time increment. The same damping coefficients may be used throughout the PDL in a given model. The mechanical properties of, for example, the hard dental tissues (i.e. other than the PDL) are assumed to have linear-elastic and isotropic behavior following σ.sub.ij=C.sub.ijklε.sub.kl, where the nonzero components of C.sub.ijkl are a function of elastic modulus E and Poisson's ratio.

[0223] In the Prediction aspect, as discussed additionally above, a system applies transformation functions to predict one or more structural defect classifications. In some exemplary embodiments, the transformation function used for prediction is chosen from the transformation functions generated in the Training aspect. In some embodiments, a predefined, canonical transformation function may be included and used to perform predictions. In some exemplary embodiments, the transformation function used may be based on a standardized predictive model.

[0224] FIG. 7 illustrates an embodiment of the operation of the endpoint device by a user, such as an expert practitioner, for retrieving predictions. This may generally assume that another user, such as a technician, has already followed the steps as illustrated in FIG. 2. The endpoint device or associated computing device may present a login screen to the expert, in order to establish his/her identity, and this identity is checked via a call to the central server. The expert then sees a menu of device measurement records, possibly across multiple patients, from which to choose to work on, based on data stored on the central server. After the expert selects one of these records, a button is shown to allow the expert to request the predicted physical conditions of at least one physical object, such as a tooth, from the central server. The predicted physical conditions may then displayed.

[0225] FIG. 14 shows an embodiment of the process by which the production interface performs predictions. First, a set of device measurements for objects, for example teeth, are supplied to the interface by, for example, an endpoint device. Next, a transformation function is selected from the database and loaded into memory. Afterwards, the transformation function is applied to the device measurements and a result is rendered. Finally, the result is returned to the endpoint device.

[0226] FIG. 15 shows an example of a screen that displays predicted results. One area may show identifying information for the source of the objects being examined, such as for example the patient. Other areas of the screen may be used to depict for example the location of the object being examined relative to other objects, the predicted location of the damage relative to the object being examined, and/or any computed quantitative scores derived from the underlying device measurements. Predicted labels may be displayed in an easy-to-understand form, such as an indication such as “Cement Failure”, the restorative condition such as “crowned root canal”, the urgency such as “red”, and the location of the damage such as “vertical to root”. The certainty score of the prediction may be presented as for example a probability score such as “97%”. Additional information based on the predictions may also be presented under an area such as “clinical considerations”.

[0227] In an example, when an operator uses the system to examine a tooth, the endpoint device, such as a handpiece, collects the ERG for the tooth and sends it to the local computer and/or to the integrated computing components of the endpoint device, as applicable. The local computer or integrated computer calculates some parameters (such as Damage and Mobility) from the raw data. At the end of the exam, the local computer puts the raw data, calculated parameters, and meta data in the correct format and uploads it to a cloud server. The cloud server analysis the data it receives a, makes predictions about the structural integrity of the teeth using pre-trained machine learning algorithms. The cloud sever then sends the predictions and insight as a code to the local computer. The local computer reads the code, looks up the definition of the codes in a local database, and displays the prediction text in the local software. If the code does not exist in the local database, the system will look up the master table which is hosted in a cloud database and replicates the table locally. All subsequent predictions will be looked up in the local database until the cloud database changes.

[0228] Although the invention has been described with respect to specific aspects, embodiments and examples thereof, these are merely illustrative, and not restrictive of the invention. The description herein of illustrated embodiments of the invention, including the description in the Abstract and Summary, is not intended to be exhaustive or to limit the invention to the precise forms disclosed herein (and in particular, the inclusion of any particular embodiment, feature or function within the Abstract or Summary is not intended to limit the scope of the invention to such embodiment, feature or function). Rather, the description is intended to describe illustrative embodiments, features and functions in order to provide a person of ordinary skill in the art context to understand the invention without limiting the invention to any particularly described embodiment, feature or function, including any such embodiment feature or function described in the Abstract or Summary. While specific embodiments of, and examples for, the invention are described herein for illustrative purposes only, various equivalent modifications are possible within the spirit and scope of the invention, as those skilled in the relevant art will recognize and appreciate. As indicated, these modifications may be made to the invention in light of the foregoing description of illustrated embodiments of the invention and are to be included within the spirit and scope of the invention. Thus, while the invention has been described herein with reference to particular embodiments thereof, a latitude of modification, various changes and substitutions are intended in the foregoing disclosures, and it will be appreciated that in some instances some features of embodiments of the invention will be employed without a corresponding use of other features without departing from the scope and spirit of the invention as set forth. Therefore, many modifications may be made to adapt a particular situation or material to the essential scope and spirit of the invention.

[0229] In general, references to the “cloud” may include both internet connected computing services and/or resources, or those that may exist on smaller or private networks.

[0230] In general, “program logic modules” and software elements may generally be configured onto, run, stored, processed and/or executed on separately on different computer processors and/or memories, in combination with each other on the same computer processors and/or memories, and/or on any of the above in varied temporal arrangements, as applicable. Nothing should be implied or construed in this specification as requiring any program logic modules and/or software elements to be run on any one or combination of computing processors and/or memories, and any suitable combination or singular unit may be utilized.

[0231] References throughout this specification to “one embodiment”, “an embodiment”, or “a specific embodiment” or similar terminology mean that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment and may not necessarily be present in all embodiments. Thus, respective appearances of the phrases “in one embodiment”, “in an embodiment”, or “in a specific embodiment” or similar terminology in various places throughout this specification are not necessarily referring to the same embodiment. Furthermore, the particular features, structures, or characteristics of any particular embodiment may be combined in any suitable manner with one or more other embodiments. It is to be understood that other variations and modifications of the embodiments described and illustrated herein are possible in light of the teachings herein and are to be considered as part of the spirit and scope of the invention.

[0232] In the description herein, numerous specific details are provided, such as examples of components and/or methods, to provide a thorough understanding of embodiments of the invention. One skilled in the relevant art will recognize, however, that an embodiment may be able to be practiced without one or more of the specific details, or with other apparatus, systems, assemblies, methods, components, materials, parts, and/or the like. In other instances, well-known structures, components, systems, materials, or operations are not specifically shown or described in detail to avoid obscuring aspects of embodiments of the invention. While the invention may be illustrated by using a particular embodiment, this is not and does not limit the invention to any particular embodiment and a person of ordinary skill in the art will recognize that additional embodiments are readily understandable and are a part of this invention.

[0233] As used herein, the terms “comprises,” “comprising,” “includes,” “including,” “has,” “having,” or any other variation thereof, are intended to cover a non-exclusive inclusion. For example, a process, product, article, or apparatus that comprises a list of elements is not necessarily limited only those elements but may include other elements not expressly listed or inherent to such process, process, article, or apparatus.

[0234] Furthermore, the term “or” as used herein is generally intended to mean “and/or” unless otherwise indicated. For example, a condition A or B is satisfied by any one of the following: A is true (or present) and B is false (or not present), A is false (or not present) and B is true (or present), and both A and B are true (or present). As used herein, including the claims that follow, a term preceded by “a” or “an” (and “the” when antecedent basis is “a” or “an”) includes both singular and plural of such term, unless clearly indicated within the claim otherwise (i.e., that the reference “a” or “an” clearly indicates only the singular or only the plural). Also, as used in the description herein, the meaning of “in” includes “in” and “on” unless the context clearly dictates otherwise.