System and method for detecting faulty dental implant components

Abstract

A faulty dental implant detection system is provided. The system comprises: a processing unit configured to: receive a dental implant image of a dental implant; determine one or more characteristics of the dental implant based on the dental implant image; determine a surface characteristic of the dental implant based on the dental implant image; input the one or more characteristics and the surface characteristic into a pre-trained convolutional neural network (CNN); and obtain a dental implant classification from the CNN indicative of whether the dental implant is a faulty dental implant.

Description

[0001] SYSTEM AND METHOD FOR DETECTING FAULTY DENTAL IMPLANT COMPONENTS

[0002] Field of the invention

[0003] The present invention relates generally to the field of object classification. More particularly, the present invention relates to methods and systems for classifying and detecting faulty dental implant components, such as dental abutment screws.

[0004] Background to the invention

[0005] A dental abutment screw is often used in combination with an abutment and an implant system located within a patient’s gum for implanting a replacement crown within a patient’s mouth. Figure 1 shows a crown implant system having an implant system 4 located within a patient’s gum. As shown in Figure 1 , an abutment 3 may be located within the implant system 4, and an abutment screw 2 may be used to attach a replacement crown 1 to the patient’s gum by screwing the abutment screw 2 into the implant system 4.

[0006] Implant surgery may take a variety of forms. For example, a single stage implant surgery may comprise surgical placement of an implant system (such as the implant system 4), which is left exposed to the oral cavity of a patient following insertion. This single stage implant surgery is often used in non-submerged implant systems. Another example is a two-stage implant surgery, which may comprise an initial surgical placement of the implant system, buried beneath the mucosa of the gum, and then subsequent exposure some time later. This two- stage implant surgery is often used in submerged implant systems. Faulty dental implant components, such as faulty dental abutment screws described above, are currently being manufactured and supplied to dental labs and dentists worldwide. Such faulty or counterfeit dental implant components may not be safe for use. In particular, they may not meet appropriate standards as set out by, for example, International Organisation for Standardization (ISO), United Kingdom Medicines and Healthcare Products Regulatory Agency (MHRA) or US Food and Drug Administration (FDA) standards. For example, a silver dental abutment screw may be used instead of a gold abutment screw, which is not suitable for use in a patient’s mouth. Accordingly, these faulty dental implant components may cause negative side effects for the patients.

[0007] It may be difficult for dental professions to determine whether a dental abutment screw is counterfeit or faulty. Even with a trained eye, it is almost impossible to determine a counterfeit or faulty screw. This poses a threat to general public health significantly with impact on dental or oral health.

[0008] The present disclosure has been devised to mitigate at least some of the above-mentioned problems.

[0009] Summary of the invention

[0010] In accordance with a first aspect of the present disclosure, there is provided a faulty dental implant detection system, the system comprising: a processing unit configured to: receive a dental implant image of a dental implant; determine one or more characteristics of the dental implant based on the dental implant image; determine a surface characteristic of the dental implant based on the dental implant image; input the one or more characteristics and the surface characteristic into a pre-trained convolutional neural network (CNN); and obtain a dental implant classification from the CNN indicative of whether the dental implant is a faulty dental implant

[0011] The “one or more characteristics” may be understood as any characteristic associated with the dental implant. For example, the one or more characteristics may comprise objects or components of the dental implant, holes located on the dental implant, and / or dimensions of the holes.

[0012] The ‘surface characteristic’ may be understood as any characteristic associated with a surface of the dental implant. For example, the surface characteristic may be a spatial frequency. It will be appreciated that the surface characteristic may be any suitable surface characteristic including, but not limited to, a surface material, such as an anodization surface finishing material. For example, a non-faulty dental abutment screw may comprise low spatial frequency and / or a surface finishing material of titanium. Alternatively, the surface characteristic may be a surface colour of the dental implant. It will be appreciated that any suitable surface characteristic or any combination of surface characteristics may be used to distinguish between a non-faulty dental implant and a faulty dental implant.

[0013] The present invention may remedy the problem of detecting faulty dental implant components by leveraging a Convolutional Neural Network (CNN) to quickly identify faulty components, thereby improving patient safety and care. In particular, the present invention may utilise dimensional analysis (i.e., determine one or more characteristics of the dental implant) and material analysis (i.e., determine a surface characteristic of the dental implant) to obtain a dental implant classification indicative of whether the dental implant is a faulty dental implant. Advantageously, the classification may be provided based on the dental implant image alone, thereby improving an ease of detection of a faulty dental implant. In some embodiments, the faulty dental implant detection system further comprises: an image capture device arranged to: capture the dental implant image of the dental implant; and provide the dental implant image to the processing unit. In this way, the system may provide a means for a user to capture an image of a dental implant, and obtain a dental implant classification. Advantageously, the dental implant classification may be determined using a dental implant image alone. The present invention may therefore provide a fast and reliable means for a user, such as a dental professional, to determine whether a dental implant is appropriate for use with a patient.

[0014] In some embodiments, the one or more characteristics are determined using an edge detection algorithm, a contour analysis algorithm, a template-matching algorithm, or any combination thereof. Advantageously, the one or more characteristics may be detected using the dental implant image alone.

[0015] The one or more characteristics may comprise: one or more screw head dimensions; one or more screw shaft dimensions; and one or more screw thread dimensions. In particular, the dimensions may be dimensions of a dental abutment screw, which may be manufactured according to strict specifications that can be identified through image detection. Advantageously, the system may easily determine the one or more characteristics.

[0016] In some embodiments, the processing unit is further configured to: transmit, to a display, instructions for displaying a warning alert indicative that the dental implant is faulty; or transmit, to the display, instructions for displaying a verification alert indicative that the dental implant is not faulty. In this way, the system may clearly indicate whether a dental implant is faulty or not faulty, thereby further improving the technical advantage of improving patient safety and care. In some embodiments, the processing unit is further configured to: determine, based on the one or more characteristics and / or the surface characteristic, a dental implant type. For example, the system may determine a type of abutment screw. The type of abutment screw may be representative of any relevant information, such as a manufacturer of the abutment screw. Advantageously, the dental implant type may be determined prior to classifying the dental implant, therebyfurther improving the speed at which a dental implant can be classified.

[0017] In some embodiments, the processing unit is configured to determine the dental implant type by: accessing a dental implant database comprising manufacturing data associated with a plurality of dental implants; comparing the one or more characteristics and / or the surface characteristic to the manufacturing data; and determining a dental implant that matches the one or more characteristics and / or the surface characteristic. In this way, the present invention may communicate with such a third party database, compare manufacturing data stored on said database, and use the manufacturing data to determine whether the dental implant having the dimensions and / or surface characteristic matches a dental implant made by the manufacturer. Advantageously, a non-faulty dental implant may be determined prior to use of the CNN.

[0018] In some embodiments, the processing unit is further configured to: determine, based on the dental implant type, data associated with the dental implant type; and transmit, to a display, instructions for displaying the data associated with the dental implant type. In this way, useful information, such as a serial number, batch code, or any other relevant information, may be provided to the user.

[0019] In some embodiments, the processing unit is further configured to: upload, to a training database, a second dental implant image for use in a test dataset. The second dental implant image may be the same as the dental implant image received by the system, or may be a dental implant image of a different dental implant. In this way, the CNN, or a separate ANN, may be improved by using a test dataset that comprises the additional second dental implant image, and thus a greater number of training images.

[0020] In accordance with a second aspect of the present disclosure, there is provided a faulty dental implant detection method, the method comprising: receiving a dental implant image of a dental implant; determining one or more characteristics of the dental implant based on the dental implant image; determining a surface characteristic of the dental implant based on the dental implant image; inputting the one or more characteristics and the surface characteristic into a pre-trained convolutional neural network (CNN); and obtaining a dental implant classification from the CNN indicative of whether the dental implant is a faulty dental implant.

[0021] In some embodiments, the method further comprises: capturing, using an image capture device, the dental implant image of the dental implant; and providing the dental implant image to the processing unit.

[0022] In some embodiments, the method further comprises: determining the one or more characteristics using: an edge detection algorithm; a contour analysis algorithm; a templatematching algorithm; or any combination thereof.

[0023] In some embodiments, the one or more characteristics comprise: one or more screw head dimensions; one or more screw shaft dimensions; and one or more screw thread dimensions.

[0024] In some embodiments, the surface characteristic is a spatial frequency.

[0025] In some embodiments, the method further comprises: transmitting, to a display, instructions for displaying a warning alert indicative that the dental implant is faulty; and transmitting, to the display, instructions for displaying a verification alert indicative that the dental implant is not faulty.

[0026] In some embodiments, the method further comprises: determining, based on the one or more characteristics and / or the surface characteristic, a dental implant type.

[0027] In some embodiments, the dental implant type is determined by: accessing a dental implant database comprising manufacturing data associated with a plurality of dental implants; comparing the one or more characteristics and / or the surface characteristic to the manufacturing data; and determining a dental implant that matches the one or more characteristics and / or the surface characteristic.

[0028] In some embodiments, the method further comprises: determining, based on the dental implant type, data associated with the dental implant type; and transmitting, to the display, instructions for displaying the data associated with the dental implant type.

[0029] In some embodiments, the method further comprises: uploading, to a training database, the dental implant image for use in a test dataset.

[0030] It will be appreciated that any features described herein as being suitable for incorporation into one or more aspects or embodiments of the present disclosure are intended to be generalizable across any and all aspects and embodiments of the present disclosure. Other aspects of the present disclosure can be understood by those skilled in the art in light of the description, the claims, and the drawings of the present disclosure. The foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the claims. Brief Description of the Drawings

[0031] The invention will now be described by way of example only with reference to the following

[0032] Figures in which:

[0033] Figure 1 shows a crown implant system;

[0034] Figure 2 shows a faulty dental implant detection system in accordance with a first aspect of the present disclosure;

[0035] Figure 3 shows an abutment screw;

[0036] Figure 4 shows a smartphone comprising the faulty dental implant system of Figure 2; and

[0037] Figure 5 shows a flow diagram of a faulty dental implant detection method in accordance with a second aspect of the present disclosure, using the faulty dental implant detection system of Figure 2.

[0038] Detailed Description

[0039] Figure 1 shows a crown implant system having an implant system 4 located within a patient’s gum. As shown in Figure 1 , an abutment 3 may be located within the implant system 4, and an abutment screw 2 may be used to attach a replacement crown 1 to the patient’s gum by screwing the abutment screw 2 into the implant system 4. Figure 2 shows a faulty dental implant detection system 100 in accordance with a first aspect of the present disclosure. The faulty dental implant detection system 100 comprises an image capture device 102 and a processing unit 104.

[0040] The image capture device 102 is arranged to capture a dental implant image of a dental implant; and provide the dental implant image to the processing unit 104. The image capture device 102 captures the dental implant image prior to insertion of the dental implant within a patient’s gum. The image capture device 102 may be, for example, a multispectral imaging device. It will be understood that the type of image required for use with the present invention mat depend on the training data set used to train a convolutional neural network (CNN). The image capture device 102 may store the dental implant image in any suitable format.

[0041] An example dental implant is shown in the dental implant image of Figure 3. In particular, Figure 3 shows an abutment screw. The abutment screw is characterised by various features, including: a head height a; a neck length b; a head angle g; shank length c; a thread length d; a thread spacing e; a thread screw angle h; and an overall screw length f. The abutment screw is also characterised by various surface characteristics, including a spatial frequency of the surface. The dental implant image may therefore provide one or more of these characterising features.

[0042] The processing unit 104 is configured to execute a plurality of functions, and is configured to execute one or more steps of the faulty dental implant detection method 300 described below. The processing unit 104 comprises, or is in communication, with a pre-trained convolutional neural network (CNN). Further discussion regarding training of the CNN is provided in the “Training Process” section below. The dental implant detection system 100 may be part of a mobile device, such as a smartphone 200, as shown in Figure 4, comprising a display 202, and an imaging device (not shown), such as a CMOS camera. In particular, the image capture device 102 may be camera of the smartphone, and the processing unit 104 may be a processing unit of the smartphone 200. A user may use the smartphone 200 to execute the functions of the dental implant system 100, for example by executing the steps of the faulty dental implant detection method 300 described below. The smartphone 200 may comprise an application for providing a suitable interface on the display 202 for the user to execute the steps of the method. In this way, a dental professional (such as a dentist) may receive a dental implant, such as an abutment screw, and may capture an image of the dental implant with the camera of the smartphone 200. The dental professional may then interact with the application to execute the faulty dental implant detection method, and the application may provide an indication as to whether the dental implant is faulty or non-faulty on the display 202. The application may also provide data associated with the dental implant. The present invention may therefore provide a means for the dental professional to determine whether a dental implant is suitable for use with a patient quickly.

[0043] The smartphone 200 may utilise CMOS technology imaging principles, and smart camera and / or machine vision. The processor of the smartphone 200 may implement any combination of the following processes: electronic identifier, image arithmetic, colour analysis, binary & grayscale morphology, and edge detection. Image sensors in CMOS technology (for example, CMOS image sensor CIS) are implemented using an array of smart photo sensors called active pixel sensor (APS). CMOS photo sensors affixed to standard smartphones are used to determine its applicability in Image Analysis. Smartphones and mobile devices like tablets are equipped with active and passive features that are valuable for data collection and remote access to data acquisition and hardware. Two common types of digital photo sensors are charged-coupled device (CCD) and charged metal oxide semiconductor (CMOS). The CCD sensor technology converts photon light energy into electrical voltages and transmits signals as analogue signals, while CMOS technology has a built in analogue to digital (AD / DA) conversion embedded in individual photo electrical cells. CMOS technology is standard features in digital smartphones.

[0044] The faulty dental implant detection system 100 is in communication with a dental implant database 150, or a plurality of dental implant databases (not shown). The dental implant database 150 is a third party database maintained by one or more dental implant manufacturers. The dental implant database 150 comprises manufacturing data associated with a plurality of dental implants. In particular, the manufacturing data is any data associated with the manufacturing process of the dental implants. The manufacturing data may include: one or more characteristics; surface characteristics; a date of manufacture; a location of manufacture; and / or any other relevant manufacturing data. The manufacturing data may be used to determine whether a dental implant is faulty by matching one or more characteristics and / or surface characteristics of the dental implant matches manufacturing data stored in the dental implant database 150.

[0045] Figure 5 shows a flow diagram of a faulty dental implant detection method 300 in accordance with a second aspect of the present disclosure, using the faulty dental implant detection system 100 of Figure 1.

[0046] The following method 300 uses a first abutment screw and a second abutment screw as an example. The first abutment screw is non-faulty. The second abutment screw is faulty.

[0047] In a first step 302 of the method 300, the processing unit 102 receives a dental implant image of a dental implant. For example, the image capture device 102 captures the dental implant image of the dental implant. In the present example, the image capture device 102 captures a first dental implant image of the first abutment screw. The image capture device also captures a second dental implant image of the second abutment screw.

[0048] The image capture device 102 provides the dental implant image to the processing unit 104. The image capture device 102 may provide the dental implant image to the processing unit 104 in any suitable file type.

[0049] In an optional step 303, the processing unit 104 determines one or more discrete objects of the dental implant by image segmentation. In particular, the processor 104 characterises the dental implant image by individual objects or component of the dental implant such as the head; the neck; the shank; the threading; and the overall abutment screw.

[0050] In step 304, the processing unit 104 determines one or more characteristics of the dental implant based on the dental implant image. In particular, the processing unit 104 preferably determines one or more characteristics of each of the discrete objects determined in step 303. The one or more characteristics comprise: one or more screw head dimensions; one or more screw shaft dimensions; and one or more screw thread dimensions. The one or more screw head dimensions comprise: a head height; a neck length; and a head angle. The one or more screw shaft dimensions comprise: a shank length. The one or more screw thread dimensions comprise: a thread length; a thread spacing; and a thread angle. The one or more characteristics may further include shape statistics of each component, and / or shape boundary descriptions of each component.

[0051] The one or more characteristics preferably include a length and width distance of each component of the dental abutment screw. These measurements are to be distinguished over the determination of endpoint locations. Length and width distances may be preferable over absolute length and width endpoint locations because distance measurements may be less sensitive to object scaling, rotation, and / or translation. That is because length and width distance measurements are the same irrespective of where the object appears in the image and what the orientation of the object is in the image. Additionally, scaling invariance may be addressed by normalising the length and distance measurements to provide a unit distance and / or using a length-to-width distance ratio. The length-to-width distance ratio addresses scaling distortions because the ratio provides a single measure that guarantees that the length and width distances are correct relative to one another. Advantageously, the system may be robust to variance in image orientation and / or quality.

[0052] The one or more characteristics may comprise additional characteristics, such as a number of holes and / or a diameter of holes present in the components.

[0053] The processing unit 104 determines the one or more characteristics using an edge detection algorithm, a contour analysis algorithm, a template matching algorithm, or any combination thereof. The skilled person may utilize any further means for determining dimensions or characteristics of an object in an image as is known in the art.

[0054] The processing unit 104 utilises the template-matching algorithm to match a template dental implant image with the dental implant image. The template dental implant image may be an image of a dental implant stored on the dental implant database 150, or on a different datastore of, for example, the mobile device 200. The template dental implant image may have one or more characteristics associated with it. The processing unit 104 accesses the database of template dental implant images, compares the dental implant image with the template dental implant image, determines that the dental implant matches the template dental implant image, and provides the one or more characteristics associated with the template dental implant image. The template-matching algorithm may be particularly useful with dental implants, such as screws, having a distinct shape or pattern that may be used to identify it.

[0055] The processing unit 104 may utilise the edge detection algorithm to determine an image location of the edges of the dental implant features, such as edges of the screw threads. Once the edge locations are determined, the processing unit 104 determines a distance between the edge locations to determine a pixel separation between said edge locations. The processing unit 104 uses a scale reference to convert the pixel separation of the edges to corresponding physical dimensions of the dental implant. The scale reference may be a ruler or a calibration target present in the dental implant image. For example, the calibration target may be a standard object having known characteristics or dimensions.

[0056] The processing unit 104 utilises a contour analysis algorithm to identify contours or boundaries of the dental implant in the dental implant image. In particular, the processing unit 104 detects changes in pixel intensity along the dental implant boundary, which is indicative of a curve that represents the shape of the dental implant. Unlike the edge detection algorithm, the contour analysis algorithm focuses on the overall shape of the dental implant rather than the individual edges of the dental implant features. This can be particularly useful for measuring the dimensions of complex dental implants, such as screws, which may have irregular or curved shapes. By analysing the shape and size of the contours, it is possible to extract important dimensions of the object, such as its perimeter, area, and aspect ratio. A calibration target may also be used for the contour analysis algorithm.

[0057] The processing unit 104 may first execute an algorithm such as the template-matching algorithm to determine the one or more characteristics. If no match can be determined, the processing unit 104 may then proceed to one or more other algorithms. In the present example, the processing unit 104 determines that the one or more characteristics of the first abutment screw of the first dental implant image are: a head height of 1 mm; a neck length of 1.16 mm; a head angle of 55.223 degrees; a shank length of 2.350 mm; a thread length of 2.749 mm; a thread spacing of 0.393 mm; and a thread angle of 59.036 degrees. The processing unit 104 determines that the one or more characteristics of the second abutment screw of the second dental implant image are: a head height of 0.9 mm; a neck length of 1.261 mm; a head angle of 52.404 degrees; a shank length of 2.298 mm; a thread length of 2.841 mm; a thread spacing of 0.399 mm; and a thread angle of 59.638 degrees.

[0058] In step 306, the processing unit 104 determines a surface characteristic of the dental implant based on the dental implant image. In particular, the processing unit 104 determines a surface smoothness of the dental implant. It will be appreciated that any surface characteristic or combination of surface characteristics can be determined in this step.

[0059] To determine the surface smoothness of the dental implant, the processing unit 104 determines a spatial frequency of the dental implant. The spatial frequency of the dental implant is determined by determining a variation in brightness of the dental implant across the surface, said variation being indicative of a coarseness of the dental implant. The processor 104 may apply a Fourier transform to the dental implant image determine the presence of high frequencies. A coarse texture is indicative of a faulty dental implant and a smooth texture is indicative of a non-faulty dental implant.

[0060] In some embodiments, the surface characteristic is determined by a second pre-trained CNN (not shown). The second pre-trained CNN may be different to the CNN used to classify the dental implant. Alternatively, the second pre-trained CNN is the same as the CNN used to classify the dental implant. The processor 104 inputs the dental implant image to the second CNN. The second CNN, which has been trained using a labelled training set with surface characteristics labelled, can classify the surface characteristic of the dental implant. For example, the labelled training set may be labelled with spatial frequency values such that the second CNN is trained to classify spatial frequency values. Advantageously, a classification accuracy when using a surface characteristic to classify the dental implant may be increased when compared with using characteristics such as dimensions alone.

[0061] In the present example, the processing unit 104 determines a first spatial frequency for the first abutment screw of the first dental implant image and a second spatial frequency for the second abutment screw of the second dental implant image. ***if possible, it would be good to include an example spatial frequency of an original screw and a spatial frequency of a fake screw***

[0062] It will be understood that steps 304 and 306 may occur in parallel.

[0063] At step 308, the processing unit 104 inputs the one or more characteristics and the surface characteristic into a pre-trained convolutional neural network (CNN). Further discussion regarding how the pre-trained CNN is trained is provided in the “Training Process” section below.

[0064] In the present example, the processing unit 104 inputs the head height of 1 mm; the neck length of 1.16 mm; the head angle of 55.223 degrees; the shank length of 2.350 mm; the thread length of 2.749 mm; the thread spacing of 0.393 mm; the thread angle of 59.036 degrees; and the first spatial frequency into the CNN for the first abutment screw. Similarly, the processor 104 inputs the head height of 0.9 mm; the neck length of 1.261 mm; the head angle of 52.404 degrees; the shank length of 2.298 mm; the thread length of 2.841 mm; the thread spacing of 0.399 mm; the thread angle of 59.638 degrees; and the second spatial frequency into the CNN for the second abutment screw.

[0065] At step 310, the processing unit 104 obtains a dental implant classification from the CNN indicative of whether the dental implant is a faulty dental implant.

[0066] The CNN may use a classification tolerance for each of the one or more characteristics and the surface characteristics. For example, the CNN may use a length measurement tolerance of up to 20%. In this way, the classification may be robust to aberrations in the dental implant image, for example due to scaling. In particular, if a length (i.e., major axis length) of the component appears longer in the dental implant image, but within the classification tolerance, the component may still be correctly classified.

[0067] In the present example, the processing unit 104 obtains a first dental implant classification of the first abutment screw, the first dental implant classification indicating that the first abutment screw is not faulty. The processing unit 104 also obtains a second dental implant classification of the second abutment screw, the second dental implant classification indicating that the second abutment screw is faulty.

[0068] The following steps 312 to 318 are preferable but optional.

[0069] At step 312, the processing unit 104 transmits, to a display, instructions for displaying a warning alert indicative that the dental implant is faulty; or transmits, to the display, instructions for displaying a verification alert indicative that the dental implant is not faulty. For example, the processing unit 104 transmits, to the display 202 of the smartphone 200 the warning alert or the verification alert. In the present example, since the first dental implant classification is not faulty, the processing unit 104 transmits to the display 202, instructions for displaying a verification alert. Furthermore, since the second dental implant classification is faulty, the processing unit 104 transmits to the display 202, instructions for displaying a warning alert.

[0070] The following steps 314 to 322 may be executed prior to step 308, such that the system 100 compares the one or more characteristics and / or the surface characteristics prior to input to the CNN.

[0071] At step 314, the processing unit 104 accesses the dental implant database 150 comprising the manufacturing data associated with the plurality of dental implants.

[0072] At step 316, the processing unit 104 compares the one or more characteristics and / or the surface characteristic determined in steps 304 and 306 to the manufacturing data stored in the dental implant database 150.

[0073] At step 318, the processing unit 104 determines a dental implant type by determining that a dental implant matches the one or more characteristics and / or the surface characteristic of the first dental implant. In particular, the processing unit 104 determines that manufacturing data associated with said dental implant matches the one or more characteristics and / or the surface characteristic of the first dental implant. Since the second dental implant is a faulty dental implant, the one or more characteristics and / or the surface characteristics of the second dental implant will not match any manufacturing data and the method can proceed to step 308.

[0074] At step 320, the processing unit 104 determines, based on the dental implant type, data associated with the dental implant type. In the present example, the processing unit 104 determines, based on the dental implant type of the first dental implant, a serial number, a batch code, a CE conformity certificate, a date of manufacture, a manufacturer, abutment specifications, a compatibility to other implant systems, and prescriptions.

[0075] At step 322, the processing unit 104 transmits, to a display, instructions for displaying the data associated with the dental implant type. For example, the processing unit 104 transmits the display instructions to the display 202 of the smartphone 200. In the present example, the processing unit 104 transmits, to the display 202, instructions for displaying the serial number, the batch code, the CE conformity certificate, the date of manufacture, the manufacturer, the abutment specifications, the compatibility to other implant systems, and the prescriptions.

[0076] Accordingly, the user (such as a dental professional) may be presented with additional information that is relevant to the dental implant, and which may provide further guidance in choosing whether to use a particular dental implant with a patient.

[0077] Executing steps 314 to 322 prior to step 308, may allow the system 100 to compare the one or more characteristics and / or the surface characteristics prior to input to the CNN, such that a non-faulty dental implant can be detected prior to input to the CNN.

[0078] Training Process

[0079] Machine learning models can be trained based on sets of data. Image recognition algorithms can utilize machine learning models to categorise a given input, such as an input image. An example machine learning model is an artificial neural network. A specific example of an artificial neural network is a convolutional neural network (CNN). Artificial neural networks comprise an input layer, one or more hidden layers, and an output layer. A CNN may take an image as an input, and apply various convolution and pooling layers to the pixels of the input image. Then, a classification portion of the CNN can provide a probability for each type of classification from a selection of classifications. The classification with the highest probability may be the output classification of the CNN. In the present invention, the plurality of classifications include at least a faulty classification and a non-faulty classification, although additional classifications may be envisaged.

[0080] The CNN of the present invention is trained using a training data set comprising a plurality of dental implant images. The training data set may be acquired from any source, for example by capturing images of a plurality of dental implants, both faulty and non-faulty, using an image capturing device. Alternatively, the training data set may be acquired from an online repository. The training data set may be input to the CNN to obtain a classification for each dental implant image by forward propagation. A loss value may be determined based on the classification provided by the CNN for each dental implant image. The loss value may be used to perform back propagation to update weights of the different layers. The process may be repeated until the loss value meets a threshold.

[0081] The second pre-trained CNN used to determine the surface characteristic is trained using a dataset of dental implant images with labelled surface characteristics. The second CNN is trained using this labelled dataset. For example, the dental implant images, or objects or components within the dental implant images, may be labelled with a spatial frequency value indicative of a surface roughness of said objects or components. The second CNN learns to map the input image to a corresponding surface characteristic in order to classify the surface characteristic. Once the second pre-trained CNN is trained, it can be used to classify new dental implant images into one of a plurality of surface finishing characteristic classifications.

[0082] The description provided herein may be directed to specific implementations. It should be understood that the discussion provided herein is provided for the purpose of enabling a person with ordinary skill in the art to make and use any subject matter defined herein by the subject matter of the claims.

[0083] It should be intended that the subject matter of the claims not be limited to the implementations and illustrations provided herein, but include modified forms of those implementations including portions of implementations and combinations of elements of different implementations in accordance with the claims. It should be appreciated that in the development of any such implementation, as in any engineering or design project, numerous implementation-specific decisions should be made to achieve a developers’ specific goals, such as compliance with system-related and business related constraints, which may vary from one implementation to another. Moreover, it should be appreciated that such a development effort may be complex and time consuming, but would nevertheless be a routine undertaking of design, fabrication, and manufacture for those of ordinary skill having benefit of this disclosure.

[0084] Reference has been made in detail to various implementations, examples of which are illustrated in the accompanying drawings and figures. In the detailed description, numerous specific details are set forth to provide a thorough understanding of the disclosure provided herein. However, the disclosure provided herein may be practiced without these specific details. In some other instances, well-known methods, procedures, components, circuits, and networks have not been described in detail so as not to unnecessarily obscure details of the embodiments.

[0085] It should also be understood that, although the terms first, second, etc. may be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another. For example, a first element could be termed a second element, and, similarly, a second element could be termed a first element. The first element and the second element are both elements, respectively, but they are not to be considered the same element.

[0086] The terminology used in the description of the disclosure provided herein is for the purpose of describing particular implementations and is not intended to limit the disclosure provided herein. As used in the description of the disclosure provided herein and appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. The term “and / or” as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items. The terms “includes,” “including,” “comprises,” and / or “comprising,” when used in this specification, specify a presence of stated features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof.

[0087] While the foregoing is directed to implementations of various techniques described herein, other and further implementations may be devised in accordance with the disclosure herein, which may be determined by the claims that follow. Although the subject matter has been described in language specific to structural features and / or methodological acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are disclosed as example forms of implementing the claims.

Claims

CLAIMS1. A faulty dental implant detection system, the system comprising: a processing unit configured to: receive a dental implant image of a dental implant; determine one or more characteristics of the dental implant based on the dental implant image; determine a surface characteristic of the dental implant based on the dental implant image; input the one or more characteristics and the surface characteristic into a pre-trained convolutional neural network (CNN); and obtain a dental implant classification from the CNN indicative of whether the dental implant is a faulty dental implant.

2. The faulty dental implant detection system of claim 1 , further comprising: an image capture device arranged to: capture the dental implant image of the dental implant; and provide the dental implant image to the processing unit.

3. The faulty dental implant detection system of claim 1 or claim 2, wherein the one or more characteristics are determined using an edge detection algorithm, a contour analysis algorithm, a template-matching algorithm, or any combination thereof.

4. The faulty dental implant detection system of any preceding claim, wherein the one or more characteristics comprise: one or more screw head dimensions; one or more screw shaft dimensions; andone or more screw thread dimensions.

5. The faulty dental implant detection system of any preceding claim, wherein the surface characteristic is a spatial frequency.

6. The faulty dental implant detection system of any preceding claim, wherein the processing unit is further configured to: transmit, to a display, instructions for displaying a warning alert indicative that the dental implant is faulty; or transmit, to the display, instructions for displaying a verification alert indicative that the dental implant is not faulty.

7. The faulty dental implant detection system of any preceding claim, wherein the processing unit is further configured to: determine, based on the one or more characteristics and / or the surface characteristic, a dental implant type.

8. The faulty dental implant detection system of claim 7, wherein the processing unit is configured to determine the dental implant type by: accessing a dental implant database comprising manufacturing data associated with a plurality of dental implants; comparing the one or more characteristics and / or the surface characteristic to the manufacturing data; and determining a dental implant that matches the one or more characteristics and / or the surface characteristic.

9. The faulty dental implant detection system of claim 7 or claim 8, wherein the processing unit is further configured to: determine, based on the dental implant type, data associated with the dental implant type; and transmit, to a display, instructions for displaying the data associated with the dental implant type.

10. The faulty dental implant detection system of any preceding claim, wherein the processing unit is further configured to: upload, to a training database, a second dental implant image for use in a test dataset.

11. A faulty dental implant detection method, the method comprising: receiving a dental implant image of a dental implant; determining one or more characteristics of the dental implant based on the dental implant image; determining a surface characteristic of the dental implant based on the dental implant image; inputting the one or more characteristics and the surface characteristic into a pretrained convolutional neural network (CNN); and obtaining a dental implant classification from the CNN indicative of whether the dental implant is a faulty dental implant.

12. The faulty dental implant detection method of claim 11 , further comprising: capturing, using an image capture device, the dental implant image of the dental implant; and providing the dental implant image to the processing unit.

13. The faulty dental implant detection method of claim 11 or claim 12, further comprising determining the one or more characteristics using: an edge detection algorithm; a contour analysis algorithm; a template-matching algorithm; or any combination thereof.

14. The faulty dental implant detection method of any of claims 11 to 13, wherein the one or more characteristics comprise: one or more screw head dimensions; one or more screw shaft dimensions; and one or more screw thread dimensions.

15. The faulty dental implant detection method of any of claims 11 to 14, wherein the surface characteristic is a spatial frequency.

16. The faulty dental implant detection method of any of claims 11 to 15, further comprising: transmitting, to a display, instructions for displaying a warning alert indicative that the dental implant is faulty; or transmitting, to the display, instructions for displaying a verification alert indicative that the dental implant is not faulty.

17. The faulty dental implant detection method of any of claims 11 to 16, further comprising: determining, based on the one or more characteristics and / or the surface characteristic, a dental implant type.

18. The faulty dental implant detection method of claim 17, wherein the dental implant type is determined by: accessing a dental implant database comprising manufacturing data associated with a plurality of dental implants; comparing the one or more characteristics and / or the surface characteristic to the manufacturing data; and determining a dental implant that matches the one or more characteristics and / or the surface characteristic.

19. The faulty dental implant detection method of claim 18, further comprising: determining, based on the dental implant type, data associated with the dental implant type; and transmitting, to a display, instructions for displaying the data associated with the dental implant type.

20. The faulty dental implant detection method of any of claims 11 to 19, further comprising: uploading, to a training database, the dental implant image for use in a test dataset.

Images