Method, system, and computer program for performing periodontal measurement from ultrasonic images

Abstract

According to some embodiments of the present invention, a method for performing periodontal measurements from ultrasonic images is provided. After obtaining at least one ultrasonic image of a jaw portion including periodontal tissue of at least one tooth, from the at least one obtained ultrasonic image, a plurality of images are generated, each indicating the presence of at least one anatomical periodontal feature in the at least one obtained ultrasonic image, and the presence of at least one anatomical periodontal feature in the at least one obtained ultrasonic image is confirmed from the corresponding generated images. When the presence of an anatomical periodontal feature in the at least one obtained ultrasonic image required for performing at least one periodontal measurement is confirmed, at least one periodontal measurement is performed from the anatomical periodontal feature whose presence has been confirmed.

Description

Technical Field

[0001] The present invention relates to the field of oral measurement methods and devices for the health management industry. In particular, the present invention relates to methods, systems, and computer programs for performing periodontal measurements from images acquired by an ultrasonic sensor, including but not limited to periodontal pocket depth measurement.

Background Art

[0002] One important measurement indicating the health of the gums is the depth of the periodontal pocket located around each tooth. The space between the gum and the tooth in front of it becomes deeper due to the presence of gum disease. To monitor and evaluate the health of the gums, it is necessary to measure, record, and monitor the pocket depth over time. In a healthy mouth, the pocket may be any depth between 1 millimeter and 3 millimeters (if the depth is shallower than 3 mm, the pocket is judged to be healthy, and if the depth is deeper than 4 mm, it is judged to be unhealthy).

[0003] Using a periodontal probe with graduations, the depth of the periodontal pocket can be measured, that is, the distance from the top of the periodontal pocket to the bottom of the periodontal pocket where the tissue is connected to the tooth root via the ligament can be measured.

[0004] FIG. 1 is a cross-sectional view schematically showing a part of a tooth and jaw (only the structure located in the left part of the tooth is shown in detail for clarity).

[0005] As shown, the tooth 100 includes enamel 105 and cementum 110 that contacts along a line 115 shown as the CEJ (cementum-enamel junction). Cementum is a calcified tissue that forms the outer covering of the tooth root. Cementum indicates the tooth fixation point of the alveolar-periodontal ligament's band teeth fibers.

[0006] The jaw includes gum 120, alveolar bone 125, and periodontal ligament (or band teeth) 130.

[0007] The gum 120 is a thick, highly vascularized mucous membrane that covers the bone and generally attaches to the tooth root by forming a small fold called the gingival sulcus or periodontal pocket. Thus, the gum 120 includes a gingival portion 120-1 attached to the alveolar bone and a free gingival portion 120-2. The periodontal ligament (or dentogingival ligament) consists of fibers that enable the attachment of the tooth to the alveolar bone.

[0008] The measurement of the periodontal pocket aims to determine the height between the end of the free gingival portion and the bottom of the periodontal pocket, for example, the height between the end of the free gingival portion indicated by reference numeral 135 and the bottom of the periodontal pocket indicated by reference numeral 140. As shown in the figure, such height can be measured by inserting this probe into the bottom of the cavity with a graduated periodontal lobe 140. When performed by a skilled practitioner, the obtained measurement value is accurate.

[0009] However, although generally efficient, measuring the periodontal pocket using a graduated periodontal probe requires several measurements (for example, six measurements) for each tooth and the need to write down or record them, so this operation is very uncomfortable for the patient and time-consuming for the practitioner.

[0010] To address such drawbacks, ultrasonic imaging has been adapted for intraoral use in several implementation methods and has been found to be particularly useful for operations such as measuring the depth of the periodontal pocket. Conditions such as gingivitis can be detected, for example, by detecting the acoustic response of the tissue.

[0011] Since it does not emit ionizing radiation, ultrasonic imaging is essentially safer than ionizing methods and can also be repeated for screening if necessary. Ultrasonic imaging can be used as an alternative to or in complement to various types of radiography (cone beam computed tomography or CBCT, panoramic X-ray, intraoral X-ray), magnetic resonance imaging (MRI), or nuclear medicine.

[0012] Ultrasonic imaging methods can use high-frequency sound waves, typically in the range of 1 MHz to 100 MHz. High-frequency waves have greater attenuation over a given distance than low-frequency waves, so high-frequency waves are primarily suitable for imaging superficial structures such as, for example, dermatology and dental imaging. For example, high-frequency sound waves may preferably be in the range of 10 MHz to 50 MHz for periodontal pocket examinations. Conversely, low-frequency waves are suitable for imaging the deepest structures within the body.

[0013] An ultrasonic imaging device generally includes one or several transducers that function as an ultrasonic beam transmitter and / or an ultrasonic beam receiver and receive echoes from the emitted signals. The ultrasonic imaging device may also include various processing components and display components used to generate and present an image from the acquired signals. The ultrasonic beam transmitter generates an ultrasonic signal from an electrical signal, and conversely, the ultrasonic receiver generates an electrical pulse from a mechanical ultrasonic signal.

[0014] An object in the path of the transmitted ultrasonic signal returns a portion of the ultrasonic energy to the transducer, and the transducer generates an electrical signal indicative of the detected structure. The electrical signal generated from the received ultrasonic signal can be delayed by a selected time specific to each transducer, whereby the ultrasonic energy scattered from the selected region is coherently added, and the ultrasonic energy from other regions has a negligible effect. Further, the transmission of the ultrasonic signal may be delayed to enable adaptive focusing. Electronic adaptive focusing makes it possible to increase the resolution depending on the depth of the imaged organ.

[0015] The array processing technique used to generate and process the signals received in this manner is called "beamforming".

[0016] Special challenges in intraoral ultrasound imaging generally relate to interpreting images that are typically gray-level images and from which it is difficult to recognize different parts of the patient's mouth, and thus performing specific measurements such as determining the height of the periodontal pocket. Also, for reliable measurements, the ultrasound probe used to acquire the ultrasound image should be correctly oriented and positioned.

[0017] Accordingly, there is a need for a method, system, and computer program that enable automatic acquisition of specific features of a patient's mouth, in particular, that enable automatic acquisition of the depth of the periodontal pocket and / or the distance between the epithelial attachment and the cementoenamel junction (CEJ).

Prior Art Documents

Patent Documents

[0018]

Patent Document 1

Patent Document 2

Non-Patent Documents

[0019]

Non-Patent Document 1

[0020] The present invention has been devised to address one or more of the above concerns. MEANS FOR SOLVING THE PROBLEMS

[0021] In this context, a method, a system, and a computer program are provided that enable automatic acquisition of specific features of a patient's mouth from ultrasonic images of the patient's mouth without the need for these ultrasonic images to be analyzed by a practitioner.

[0022] According to one aspect of the present invention, there is provided a method for performing periodontal measurements from an ultrasonic image, the method comprising: obtaining at least one ultrasonic image of a jaw portion including periodontal tissue of at least one tooth; generating a plurality of images from the at least one obtained ultrasonic image, each of the generated images indicating the presence of at least one anatomical periodontal feature in the at least one obtained ultrasonic image; verifying the presence of at least one anatomical periodontal feature in the at least one obtained ultrasonic image from the corresponding generated image; performing at least one periodontal measurement from the anatomical periodontal feature whose presence has been verified when the presence of the anatomical periodontal feature in the at least one obtained ultrasonic image required for performing at least one periodontal measurement is verified; and displaying, storing, or sending the result of the at least one periodontal measurement.

[0023] The method of the present invention enables several periodontal measurements to be automatically provided from ultrasonic images of the periodontal tissue of teeth without requiring techniques related to the interpretation of ultrasonic images, thereby enabling the practitioner to focus on handling the sensor probe and also on the patient's mouth.

[0024] According to some embodiments of the present invention, the at least one periodontal measurement includes at least one of a clinical attachment level (CAL) measurement, a pocket depth (PD) measurement, and a gingival margin (GM) measurement.

[0025] According to some embodiments of the present invention, the method further includes obtaining at least one other ultrasonic image of a jaw portion including periodontal tissue of at least one tooth, and repeating generating and verifying using the at least one other ultrasonic image when it is determined that the at least one obtained ultrasonic image does not include the anatomical periodontal feature required for performing at least one periodontal measurement.

[0026] According to some embodiments of the present invention, the method further includes displaying a message indicating that at least one periodontal measurement cannot be performed from the acquired at least one ultrasonic image when it is determined that the acquired at least one ultrasonic image does not include anatomical periodontal features.

[0027] According to some embodiments of the present invention, the confirmation includes determining, within the generated image, the number of pixels whose value is greater than or equal to a first threshold value, and comparing the determined number of pixels with a second threshold value.

[0028] According to some embodiments of the present invention, the method further includes identifying a structure within at least one of the generated images of the generated images, the structure being identified as a function of the positions of pixels whose value is greater than or equal to a third threshold value, and performing at least one periodontal measurement based on the identified structure.

[0029] According to some embodiments of the present invention, the method further includes identifying a target landmark within at least one of the generated images of the generated images, the target landmark being identified as a function of the number of predetermined pixels having the highest value, and performing at least one measurement based on the identified landmark.

[0030] According to some embodiments of the present invention, the method further includes reducing at least one of the acquired images, and the generation is based on the reduced at least one of the acquired images.

[0031] According to some embodiments of the present invention, the generation is performed using a deep neural network.

[0032] According to some embodiments of the present invention, the deep neural network is a U-net type convolutional neural network.

[0033] According to some embodiments of the present invention, the method further includes reducing the coding bit depth of the weights of the deep neural network after training the deep neural network.

[0034] According to some embodiments of the present invention, at least one acquired ultrasonic image belongs to a plane including at least one point on the surface of a tooth belonging to the tooth axis of the tooth and the periodontal tissue of the tooth.

[0035] According to some embodiments of the present invention, at least one anatomical periodontal feature includes enamel, cementum, gingiva, alveolar bone, cementum-enamel junction, and / or epithelial attachment.

[0036] According to another aspect of the present invention, there is provided an apparatus including a processing device configured to execute each step of the above-described method. Another aspect of the present disclosure has the same effect as one of the above-described aspects.

[0037] According to some embodiments of the present invention, the processing device includes a specific integrated circuit in which a deep neural network is incorporated, and the deep neural network is each used to generate an image indicating the presence of at least one anatomical periodontal feature in at least one acquired ultrasonic image.

[0038] At least a part of the method according to the present invention can be implemented by a computer. Therefore, the present invention may be in any form of an embodiment that is all hardware, an embodiment that is all software (including firmware, resident software, microcode, etc.), or an embodiment that combines software and hardware aspects, and all of these can generally be referred to as "circuits", "modules", or "systems" in this specification. Further, the present invention may be in the form of a computer program product implemented on any tangible expression medium in which computer-usable program code is implemented in the medium.

[0039] Since the present invention can be implemented by software, it can be implemented as computer-readable code on any suitable carrier medium for providing to a programmable device. The tangible carrier medium may include a storage medium such as a floppy disk, a CD-ROM, a hard disk drive, a magnetic tape device, or a solid-state memory device. The transient carrier medium may include signals such as electrical signals, electronic signals, optical signals, acoustic signals, magnetic signals, or electromagnetic signals, for example, signals such as microwaves or RF signals. Next, embodiments of the present invention will be described with reference to the following drawings for illustrative purposes only.

Brief Description of the Drawings

[0040]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Modes for Carrying Out the Invention

[0041] The following is a detailed description of specific embodiments of the present invention with reference to the drawings, and in each figure, the same reference numerals indicate the same elements of the structure.

[0042] In the drawings and the following description, like elements are designated by like reference numerals, and descriptions of like elements and their configurations or interactions that have already been described are omitted. When terms such as "first," "second," etc. are used, these terms do not necessarily imply any ranking or priority relationship, and may simply be used to more clearly distinguish one element from another without specific mention.

[0043] In the context of the present disclosure, the terms "viewer," "operator," and "user" are considered to be equivalent to and to mean a practitioner, technician, or other person who acquires and views ultrasonic images such as intraoral images on a display monitor. "Operator command," "user command," or "viewer command" is obtained by the clear instruction input by the viewer, such as clicking a button on an ultrasonic probe or system hardware, clicking a button using a computer mouse or touch screen, or keyboard input.

[0044] In the context of the present disclosure, the phrase "in signal communication" means that two or more devices and / or components can communicate with each other via a signal moving on some kind of signal path. The signal communication may be wired or wireless. The signal may be a communication, power, data, or energy signal. The signal path may include a physical, electrical, magnetic, electromagnetic, optical, wired, and / or wireless connection between a first device and / or component and a second device and / or component. The signal path may also include additional devices and / or components between the first device and / or component and the second device and / or component.

[0045] The term "object" means the gingiva of the patient being imaged and other soft tissues in the oral cavity (in some cases, the tooth surface), and can also be considered equivalent to the "subject" of the corresponding imaging system in optical terms.

[0046] According to some embodiments of the present invention, the depth of the periodontal pocket is automatically measured from an ultrasonic image. If the device that generates the image is not accurately positioned (i.e., if the depth of the periodontal pocket cannot be reliably measured from the acquired image), the measurement is not performed and an error message is displayed. The measurement is based on the detection of specific materials or tissues within the ultrasonic image and the identification of specific landmarks (or specific anatomical points) such as the epithelial attachment and / or the cementoenamel junction. Such detection and identification are performed in real time, thereby enabling, for example, processing at least 10 images per second. Since the measurement of the depth of the periodontal pocket is automatically performed, there is no need to display the ultrasonic image from which the measurement value is obtained.

[0047] According to some specific embodiments, at least some of the following structures and landmarks are detected and identified to enable reliable measurement of the depth of the periodontal pocket. - Enamel, - Cementum, - Gingiva, - Alveolar bone, - Cementoenamel junction, and / or - Epithelial attachment.

[0048] According to some embodiments, machine learning is used for the detection of anatomical structures and landmarks that are important for performing the desired measurement. For this purpose, the system can determine (classify) whether the structures and landmarks are present in the image, and if so, the system can accurately determine their location (segmentation or landmark localization). Next, from these structures and landmarks, the system performs a measurement, for example, a periodontal measurement.

[0049] FIG. 2 shows an example of detected structures and identified landmarks in an ultrasonic image that enable reliable measurement of the depth of the periodontal pocket. For illustrative purposes, the ultrasonic image used shows the tooth and jaw portion indicated by reference numeral 200 in FIG. 1.

[0050] After processing the ultrasonic image 200, the alveolar bone, the gingival portion attached to the alveolar bone, the free gingival portion, cementum, and enamel, each indicated by reference numerals 205, 210, 215, 220, and 225, are detected and their positions are determined. Similarly, the cementum-enamel junction and the epithelial attachment, each indicated by reference numerals 230 and 235, are identified and their positions are determined.

[0051] FIG. 3 shows an example of steps of a method for performing a periodontal measurement from an ultrasonic image to provide, for example, a measurement of the depth of a periodontal pocket, according to some embodiments of the present invention.

[0052] As shown, the first step is directed to obtaining (step 300) one or several images, for example, a 2D (two-dimensional) image resulting from data obtained from a 2D ultrasonic probe shown as an ultrasonic image. The obtained image should be taken from a perspective that enables detection of the structure to be explored and the landmark of interest to be explored. Thus, the obtained image preferably belongs to a plane that includes at least one point on the tooth axis of the tooth and the surface of the tooth belonging to the periodontal tissue of the tooth.

[0053] Next, the obtained ultrasonic image is reduced (step 305) to a resolution that enables fast and accurate detection of the structure and / or landmark, for example, a resolution of 128×128 pixels. The reduced image is then processed to analyze and extract features that correspond to the structure to be explored or enable identification of the landmark. According to some embodiments of the present invention, this step involves a deep neural network such as a convolutional neural network that generates several result images, for example, six images (one for each of the anatomical landmarks and structures to be explored, namely, enamel, cementum, gingiva, alveolar bone, cementum-enamel junction, and epithelial attachment).

[0054] Next, each of the acquired images is analyzed, for example, using an appropriate threshold value, to determine whether this image is important for performing a periodontal measurement (step 315). Such an analysis may consist of comparing the value of each pixel of the generated image under consideration with a first threshold value and comparing the number of pixels whose value is greater than or equal to the first threshold value with a second threshold value. The values of the first threshold value and the second threshold value may be determined in advance or may be determined dynamically. For illustrative purposes, they may be automatically determined during the training of the deep neural network used to analyze and extract features and generate the image.

[0055] The process that enables determination of whether the generated image is important for performing a periodontal measurement may be applied to a single input ultrasonic image, or, for example, to several input ultrasonic images (preferably adjacent images) in order to take into account small movements of the probe (this process may improve the importance of the data of the input ultrasonic image). It should be noted that according to some embodiments, one of several generated images corresponding to the same structure or the same target landmark is selected based on the number of pixels whose value is greater than or equal to the first threshold value (for example, the generated image with the highest number of pixels whose value is greater than or equal to the first threshold value is selected).

[0056] If the number of pixels whose value is greater than or equal to the first threshold value is lower than the second threshold value, it means that there is no structure or target landmark to be searched for, or that the probe is placed in the wrong position. Therefore, it is concluded that the input ultrasonic image cannot be used for performing a periodontal measurement, and this image is rejected (step 325). According to some embodiments, an indication that the input image is rejected is displayed, indicating to the practitioner that this image cannot be used for performing a periodontal measurement (step 330). From such an indication, the practitioner can understand that the probe is not correctly placed, and thus can move the probe until the desired measurement is obtained.

[0057] As shown, next, the algorithm loops to step 300 where a new ultrasound image is processed.

[0058] Note that by rejecting images that appear to contain no important data, the loss of processing time due to processing meaningless data is avoided, thus enabling high frame rate performance.

[0059] Conversely, if the number of pixels whose value is greater than or equal to a first threshold is greater than or equal to a second threshold, then preferably, for each of the generated images corresponding to the input ultrasound image under consideration, it is concluded that the input ultrasound image contains important information from which a periodontal measurement can be performed.

[0060] Accordingly, for example, by converting the corresponding generated image into a binary image, the structure to be searched for is identified (step 335). For illustrative purposes, the structure to be searched for may correspond to an area in the generated image where the pixel value is greater than or equal to a third threshold. The value of the third threshold may be predetermined or may be determined dynamically. The third threshold may be equal to the first threshold. From the coordinates in the generated image at these points (the same in the reduced image), the position of the structure to be searched for in the ultrasound image can be calculated (based on the parameters of the reduction operation).

[0061] Next or in parallel, fiducial marks (e.g., CEJ and epithelial attachment) are extracted from two corresponding generated images (step 340). For this purpose, the number of pixels (K pixels) determined to have the highest value in these two images is selected and used to calculate two reference points (one for each generated image). Such reference points are, for example, the centroids of the selected points in each of the two generated images. For illustrative purposes, K may be selected from the range of 2 to 50. For example, K is set to 10. These two centroids correspond to the positions of the CEJ and epithelial attachment.

[0062] Based on the coordinates of these points in the generated image (the same applies to the reduced image), the positions of the CEJ and epithelial attachment in the original image can be calculated (based on the parameters of the reduction operation).

[0063] Next, using the positions of the structures to be searched for and the markers to be searched for in the input image, periodontal measurements can be performed (step 345). For illustrative purposes, at least some of the following measurements can be performed. - The distance corresponding to the distance between the CEJ and the epithelial attachment, shown as the clinical attachment level (CAL), - The pocket depth (PD) corresponding to the distance between the top of the free gingiva that preferably projects perpendicularly to the tooth axis on the tooth surface and the epithelial attachment, and / or - The gingival margin (GM) corresponding to the distance between the top of the free gingiva that preferably projects perpendicularly to the tooth axis on the tooth surface and the CEJ.

[0064] The calculated distances are displayed (step 350). According to some embodiments, the ultrasonic images used to measure these distances are not displayed. Regardless of the direction of the probe, immediately after a minimum amount of periodontal tissue structures are detected in the image, the algorithm provides the measurement values.

[0065] As shown, the algorithm then loops back to step 300 and a new ultrasonic image is processed.

[0066] The illustrated example is based on, for example, 2D ultrasonic images taken at a frame rate of 10 images per second, but a 3D approach using a 3D volume obtained by directly imaging a thick layer of the jaw with a 3D probe (e.g., using a multi-layer oscillator) or a 3D volume obtained by reconstructing a 3D volume from several 2D images as the input to a deep neural network should also be noted.

[0067] As disclosed above, and according to some specific embodiments, a deep neural network is used to process the reduced ultrasonic image, thereby detecting the structure to be searched for and identifying the landmark of the object to be searched for. Such a deep neural network may be derived from a convolutional neural network known as U-Net. The U-Net network includes a contracting path and an expanding path. The contracting path is a typical convolutional network consisting of repeated applications of convolution, followed by a rectified linear unit (ReLU) and a max-pooling operation after each convolution. During contraction, the spatial information is reduced and the feature information is increased. The expanding path combines the feature information and the spatial information by concatenation with high-resolution features from a series of up-convolutions and the contracting path. The U-net type artificial neural network is described, for example, in the following paper: "U-net: Convolutional networks for biomedical image segmentation" (Ronneberger, O., Fischer, P. & Brox, T., Medical Image Computing and Computer-Assisted Intervention - MICCAI 2015, Lecture Notes in Computer Science, 9351, 234-241 (Springer International Publishing, 2015)).

[0068] Figure 4 shows an example of a deep neural network that can be used to process an ultrasonic image, preferably a reduced ultrasonic image, to detect structures such as alveolar bone, gingival portions attached to the alveolar bone, free gingival portions, cementum, and enamel, and to identify landmarks such as the cementoenamel junction and epithelial attachment.

[0069] According to the figure example, the input of the deep neural network is a single image with 128x128 pixels, where each pixel may be coded in 1 byte. The output also includes six images with 128x128 pixels, where each pixel may also be coded in 1 byte. Each box shows a multi-channel feature map. The x size and y size of the feature map are indicated by the first two numbers of the label associated with the box, and the number of channels is indicated by the third number (in parentheses). The cross-hatched boxes show replicated feature maps, and the arrows represent the operations shown in Figure 4.

[0070] As shown in the figure, the structure of the deep neural network includes a contracting path (left side) and an expanding path (right side). The contracting path aims to apply double 3x3 convolutions several times and then downsample by a subsequent rectified linear unit (ReLU) and a 2x2 max pooling operation with a stride of 2. In each downsampling step, the number of feature channels doubles. Conversely, the expanding path aims to upsample the feature map and then apply a 2x2 convolution (an "up-convolution") that halves the number of feature channels, a concatenation with the corresponding cropped feature map from the contracting path, and two 3x3 convolutions followed by a ReLU for each. In the last layer, 1x1 convolutions are used to map each of the 32-component feature vectors to six classifications, each representing enamel, cementum, gingiva, alveolar bone, cementum-enamel junction, or epithelial attachment, thereby enabling the detection and localization of six anatomical structures and landmarks by a single pathway.

[0071] According to some embodiments, the deep neural network shown in FIG. 4 is trained with real-world data in which the correct structures and landmarks have been determined by experts. For illustration purposes, a first database containing 950 images (where all anatomical structures and landmarks are present and identified), and a second database containing 400 images taken in cases of poor probe placement that do not contain anatomical structures and / or landmarks, were created. These two databases were split into a training set, a test set, and a validation set and used for training.

[0072] After training the deep neural network, the coding bit depth of the parameters of the deep neural network, in particular the weights of the neurons, is preferably changed to maintain the accuracy and precision at a good level while reducing the size of the deep neural network. For example, the coding bit depth may be to code the neuron weights to 1 byte, which enables the deep neural network to be executed on dedicated hardware such as an Edge TPU (an ASIC designed for a specific purpose to execute an AI engine in an embedded system).

[0073] Regarding the deep neural network shown in FIG. 4, although its effectiveness has been proven, it should be noted that some parameters such as the size of the feature map and / or the number of channels may be changed.

[0074] FIG. 5 is a schematic block diagram showing an arithmetic unit for implementing one or more embodiments of the present invention, particularly for executing the steps described with reference to FIG. 3 or some of these steps.

[0075] The arithmetic unit 500 includes a communication bus that may be connected to all or some of the following elements. - A central processing unit 505 represented as a CPU, such as a microprocessor, - A random access memory 510 represented as a RAM for storing executable code of the method according to some embodiments of the present invention, and a register adapted to store variables and parameters necessary for implementing a method for performing periodontal measurement from an ultrasonic image according to some embodiments of the present invention, the memory capacity of which can be expanded, for example, by any RAM connected to an expansion port. - A read-only memory 515 represented as a ROM for storing a computer program for implementing some embodiments of the present invention. - A user interface and / or input / output interface 520 for receiving input from a user, providing information to the user, and / or receiving or sending data from / to internal sensors and / or external devices, in particular, for receiving data from a sensor such as an ultrasonic sensor 525 which may be incorporated within the arithmetic unit 500 or may be an external device connected to the arithmetic unit 500 via a wired or wireless link. - An AI engine 530 configured to analyze an ultrasonic image or a reduced ultrasonic image and generate an image representing a structure to be searched and / or a target landmark to be searched. It should be noted that the AI engine may be incorporated within the arithmetic unit 500, or may be a remote processing device, or may be distributed, for example, across one or more servers on the cloud.

[0076] Optionally, the communication bus of the arithmetic unit 500 may be connected to a solid state disk 535 (or a hard disk drive) represented as an SSD which is used as a mass storage device and / or, for example, a display device 540 for displaying periodontal measurements.

[0077] The communication bus of the computing device 500 may also typically be connected to a network interface 545 that is connected to a communication network capable of transmitting or receiving digital data to receive data from or send data to remote devices, particularly dental information systems and / or storage devices 535. The network interface 545 may be a single network interface or may consist of a set of different network interfaces (e.g., wired and wireless interfaces, or different types of wired or wireless interfaces). Data packets are written to the network interface for transmission or read from the network interface for reception under the control of a software application being executed by the CPU 505.

[0078] The executable code may be stored in any of the read-only memory 515, the solid-state device 535, or a removable digital medium such as a memory card, for example. According to a variant, the executable code of the program can be received via the network interface 534 by the communication network and thereby stored in one of the storage means of the computing device 500, such as the solid-state device 535, before execution.

[0079] The central processing unit 505 is adapted to control and manage the execution of a program or instructions of a plurality of programs or a part of the software code according to some embodiments of the present invention, and these instructions are stored in one of the above storage means. After power-on, the CPU 505 can execute instructions from the main RAM memory 510 related to the software application after these instructions are loaded, for example, from the ROM 515 or from the solid-state device 535. Such a software application, when executed by the CPU 505, causes the steps of the present disclosure to be executed.

[0080] Any of the steps of the present disclosure can be implemented as software by execution of a set of instructions or programs by a programmable computing machine such as a PC (“personal computer”), a DSP (“digital signal processor”), or a microcontroller, or else can be implemented as hardware by a machine or dedicated components such as an FPGA (“field programmable gate array”) or an ASIC (“application specific integrated circuit”).

[0081] The present disclosure has been described above with respect to several specific embodiments, but the invention is not limited to these specific embodiments, and modifications within the scope of the invention will be apparent to those skilled in the art.

[0082] Many further modifications and variations will be suggested to those skilled in the art by reference to the above exemplary embodiments, which are shown by way of example only and are not intended to limit the scope of the invention as determined by the appended claims. In particular, different features of different embodiments may be exchanged where appropriate.

[0083] In the claims, the term “comprising” does not exclude other elements or steps, and the indefinite articles “a” or “an” do not exclude a plurality. The fact that different features are recited in mutually different independent claims does not mean that these features cannot be combined to advantage.

Claims

1. A computer program for a programmable device, which, when loaded into and executed on the programmable device, includes a set of instructions for implementing each of the steps for performing periodontal measurement from an ultrasonic image, the steps being: Acquire at least one ultrasound image of the jaw region including the periodontal tissue of at least one tooth (300), (310) generating a plurality of images from the acquired at least one ultrasound image, wherein each of the generated images shows the presence of at least one anatomical periodontal feature in the acquired at least one ultrasound image. The presence of at least one anatomical periodontal feature in the acquired at least one ultrasound image is confirmed from the corresponding generated image, If the presence of an anatomical periodontal feature in the acquired at least one ultrasound image is confirmed (320) as required to perform at least one periodontal measurement, then the at least one periodontal measurement is performed from the confirmed anatomical periodontal feature (345), Displaying, saving, or sending the results of at least one of the periodontal measurements, A computer program that includes [this].

2. A computer program according to claim 1, A computer program wherein the at least one periodontal measurement includes at least one of clinical attachment level (CAL) measurement, pocket depth (PD) measurement, and gingival margin (GM) measurement.

3. A computer program according to claim 1, A computer program that, if it is determined that the acquired at least one ultrasound image does not contain the anatomical periodontal features required to perform the at least one periodontal measurement, acquires at least one other ultrasound image of the jaw region including the periodontal tissue of at least one tooth, and further includes repeating the generation and verification using the at least one other ultrasound image.

4. A computer program according to claim 1, A computer program that includes determining the number of pixels in the generated image whose value is greater than or equal to a first threshold, and comparing the determined number of pixels with a second threshold.

5. A computer program according to claim 4, A computer program that identifies (335) a structure in at least one of the generated images, wherein the structure is identified as a function of the position of a pixel whose value is greater than or equal to a third threshold, and performs at least one periodontal measurement based on the identified structure.

6. A computer program according to claim 1, A computer program further comprising identifying a marker of interest in at least one of the generated images (340), wherein the marker of interest is identified as a function of a predetermined number of pixels having the highest value, and performing the at least one measurement based on the identified marker.

7. A computer program according to claim 1, A computer program further comprising reducing the size of the acquired at least one image (305), wherein the generation is based on the reduced acquired at least one image.

8. A computer program according to claim 1, A computer program whose generation is performed using a deep neural network.

9. A computer program according to claim 8, The aforementioned deep neural network is a computer program that is a U-net type convolutional neural network.

10. A computer program according to claim 8, A computer program that, after training the deep neural network, further includes reducing the coding bit depth of the weights of the deep neural network.

11. A computer program according to claim 1, A computer program wherein the acquired at least one ultrasound image falls within a plane that includes the tooth axis of the tooth and at least one point on the tooth surface belonging to the periodontal tissue of the tooth.

12. A computer program according to claim 1, A computer program in which the at least one anatomical feature includes enamel, cementum, gum tissue, alveolar bone, cementum-enamel junction, and / or epithelial attachments.

13. An apparatus including a processing unit configured to perform each of the steps of a computer program described in any one of claims 1 to 12.

14. The apparatus according to claim 13, The apparatus comprises a specific integrated circuit incorporating a deep neural network, the deep neural network being used to generate images, each of which shows the presence of at least one anatomical periodontal feature in the acquired at least one ultrasound image.

Images