Method and system for generating dental panoramic images with clear depictions of clinically relevant patient-specific preselected anatomical structures - Patent Application 20070123633

Abstract

The present teachings relate to a computer-implemented method for reconstructing a panoramic image of a patient's dental region, comprising the steps of: providing (102; 402) at least one initial panoramic image reconstructed by using a plurality of projection images of the patient and an initial layer including mapping points corresponding to the patient's dental region so as to be clearly depicted in the initial panoramic image; detecting (104, 404) one or more clinically relevant patient-specific preselected anatomical structures in the initial panoramic image to provide anatomical data (204, 206, 208, 210, 212) relating to spatial characteristics of the detected clinically relevant patient-specific preselected anatomical structures, respectively; calculating (106; 406) a new layer including the clinically relevant patient-specific preselected anatomical structures using the anatomical data and the initial panoramic image to provide a clearer depiction of the clinically relevant patient-specific preselected anatomical structures in the reconstructed new panoramic image compared to the depiction of the clinically relevant patient-specific preselected anatomical structures in the initial panoramic image; and reconstructing (108, 408) a new panoramic image using the plurality of projection images of the patient and the calculated new layer.

Description

[Technical field]

[0001] The entire content of priority application EP2215681.1 is incorporated herein by reference under the provisions of the PCT.

[0002] The present invention relates to a computer-implemented method and system for medical tomography, and more particularly to panoramic images of the dental region. [Background technology]

[0003] Dental panoramic imaging is commonly used for dental diagnosis and treatment of patients. A typical x-ray system for dental panoramic imaging includes an x-ray generator and an x-ray detector that rotates around the patient's head. The x-ray detector acquires multiple x-ray projections of the patient's dental region during the rotation.

[0004] The panoramic image is reconstructed from the multiple projection images by using reconstruction parameters describing a layer that contains mapping points corresponding to the areas of the patient's teeth that should be in focus, i.e., sharply depicted in the panoramic image. Typically, default reconstruction parameters describing a default layer are used to focus on the teeth areas, especially the patient's jaw arch. Areas of the teeth outside the default layer that contain anatomical parts that are not of interest to the dentist (e.g., the spine) are thus blurred and their influence on the panoramic image is reduced.

[0005] Because variations may occur in each individual panoramic image, particularly due to the position and orientation of the patient's anatomy during acquisition, it is possible that a panoramic image generated using default reconstruction parameters may not be perfectly focused on the region of interest, which can make diagnosis and treatment based on the panoramic image inaccurate and difficult.

[0006] Sometimes it is possible to manually adjust the focus of the panoramic image by adjusting the patient's position / orientation or by selecting one of the preset image programs that matches the patient's anatomy, e.g., jaw width or size, better than others. However, this may increase the manual effort required to adjust the patient and select the program, and may lead to multiple exposures.

[0007] In addition to manual approaches, certain automatic approaches for improving image quality are also known, in which the focus of the panoramic image is adjusted after the projection images have been acquired.

[0008] Autofocus methods are already known that attempt to automatically optimize the sharpness of the resulting panoramic image for a given set of projection images by adjusting reconstruction parameters. These methods use a mathematical metric, for example a measure of sharpness during image processing that can be calculated for the reconstructed panoramic image, and then optimize the value of this metric by adjusting reconstruction parameters, often in an iterative gradient descent type numerical optimization. Although this procedure results in a panoramic image with an optimal measure of sharpness in a mathematical sense, it is not necessarily optimal for diagnostic applications, since the in-focus structures do not necessarily correspond to the specific tooth structures relevant to the clinical situation of the dentist and the patient.

[0009] For example, in patients with metal brackets, these metric-optimized autofocus methods automatically calculate a large increase in the sharpness measurement when the focus is optimized on the high-contrast metal contour of the metal bracket. As a result, the sharpness improvement of the metal bracket dominates the possible sharpness improvement on the diagnostically more relevant tooth structure closer to the metal bracket, which is less noticeable in terms of radiopaque contrast. In these cases, the autofocus method may degrade image quality from a diagnostic standpoint.

[0010] Reference is made to EP3685752A1 which discloses a method for generating a panoramic image of the oral cavity. Summary of the Invention

[0011] The inventors have recognized that there is a need for improved techniques that enable automatic, optimized reconstruction of panoramic images in which clinically relevant, patient-specific, pre-selected anatomical structures can be more clearly displayed without being obscured by unrelated structures.

[0012] The inventors are currently not aware of any technique that enables targeted adaptive improvement of the sharpness of only desired or preselected anatomical structures in a panoramic image by corresponding adjustment of reconstruction parameters, i.e., the sharp layer.

[0013] It is an object of the present invention to provide a method and system for generating panoramic images having clearer depictions of clinically relevant, patient-specific, pre-selected anatomical structures that are desirable for a dentist to see.

[0014] This object is achieved by a method according to claim 1 and a system according to claim 7. The subject matter of the dependent claims relates to further developments and preferred embodiments.

[0015] The present invention provides a computer-implemented method for reconstructing a panoramic image of a dental region of a patient, the method comprising providing at least one initial panoramic image reconstructed to be clearly depicted in the initial panoramic image using a plurality of projection images of the patient and an initial layer including mapping points corresponding to the dental region of the patient, the method further comprising detecting one or more clinically relevant patient-specific preselected anatomical structures in the initial panoramic image to respectively provide anatomical data related to spatial characteristics of the detected clinically relevant patient-specific preselected anatomical structures, calculating a new layer including the clinically relevant patient-specific preselected anatomical structures using the anatomical data and the initial panoramic image to provide a clearer depiction of the clinically relevant patient-specific preselected anatomical structures in the reconstructed new panoramic image compared to the depiction of the clinically relevant patient-specific preselected anatomical structures in the initial panoramic image, and reconstructing the new panoramic image using the plurality of projection images of the patient and the calculated new layer.

[0016] The main advantageous effect of the present invention is that the new panoramic images can achieve higher clinical value with respect to anatomically or pathologically relevant structures.

[0017] Another major advantageous effect of the present invention is that a panoramic image can be automatically generated by focusing on one or more of the clinically relevant patient-specific preselected anatomical structures to be observed by the dentist. Thus, the method enables automatic adjustment of the sharp layer in the panoramic image by optimally determining new reconstruction parameters to be used to reconstruct a new panoramic image. The sharp layer in this context refers to a layer that includes mapping points corresponding to the patient that should be sharply depicted in the panoramic image. More specifically, the method enables adapting the depth of the sharp layer for multiple regions, preferably each region, of the reconstructed panoramic image such that a panoramic image having one or more regions including the clinically relevant patient-specific preselected anatomical structures of interest can be sharply depicted. By focusing on the clinically relevant patient-specific preselected anatomical structures, less calculations are required compared to conventional autofocus methods, and therefore faster calculations can be achieved.

[0018] For comparison, known autofocus methods are more likely to focus on the bracket boundaries of dental braces because they give high sharpness metric values, however the method of the present invention can ignore the bracket boundaries and focus on more important clinically relevant patient-specific pre-selected anatomical structures such as the teeth behind the bracket boundaries. More specifically, the initial panoramic image serves as a reference. Based on the anatomical data, new reconstruction parameters can be calculated. Thus, the initial panoramic image is used as a reference to locate preselected anatomical structures and to determine the depth of the preselected anatomical structures to obtain new layers. The technique of the present invention allows for the automatic calculation of new reconstruction parameters that allow clinically relevant patient-specific preselected anatomical structures to be included in the clear layers of the panoramic image.

[0019] For comparison, known autofocus methods optimize image quality metrics such as local sharpness without considering the relevance of depicted preselected anatomical structures. However, in the method of the present invention, an automatic pre-determination of relevant anatomical or pathological structures of the patient's anatomy prioritizes focus optimization of these preselected anatomical structures over optimization of nearby salient but diagnostically irrelevant structures. Since only one depth layer can be sharply depicted in the reconstructed panoramic image, this avoids a compromise in diagnostic quality for image regions where the parameter optimization must choose between relevant and irrelevant features to focus on.

[0020] As used herein, an anatomical structure refers to a portion or class of specific portions of a patient's oral anatomy. For example, an anatomical structure may relate to intraoral structures such as a group of teeth or a specific tooth, dentition, gums, nerve channels, tooth extraction sites, jawbone, or extraoral structures such as condyles, sinuses, paranasal sinuses, and bones. The anatomical structure may be an artificial structure, such as a dental replacement, e.g., a crown, brace, veneer, bridge. The anatomical structure may be an implant, or other natural or artificial structure attached to the patient's jaw. The anatomical structure may be associated with a medical condition or condition of the patient. As a non-exhaustive example, the medical condition or condition may be a tooth or bone fracture, bone loss, caries, radiolucency, demineralization, infection, impaction, cyst, cancer, or any other identifiable condition of the oral anatomical structure or any part thereof.

[0021] In one embodiment, the clinically relevant patient-specific pre-selected anatomical structures are detected through a segmentation operation. Alternatively, a localization operation can be performed. In one embodiment, the clinically relevant patient-specific pre-selected anatomical structures to be detected can be pre-selected by a user through input, for example, by using a display, a keyboard, or a mouse. The user input can comprise one or more of the following: a list of anatomical structures associated with a class or instance of a portion of the patient's oral anatomical structure to be detected, such as a tooth, a tooth with a specific tooth number, or a radiolucent lesion. This allows the dentist to improve the definition of the pre-selected anatomical structures according to the diagnostic and therapeutic procedures associated with the patient. The dentist can perform the pre-selection according to the medical indications leading to the x-ray acquisition. Alternatively, the dentist can perform the pre-selection after previewing the initial panoramic image and forming a diagnostic interest to focus on a specific anatomical structure.

[0022] In one embodiment, one or more of the calculation, detection, and reconstruction steps are performed by a corresponding trained artificial neural network. A single artificial neural network can also be used. Alternatively, any of the calculation, detection, and reconstruction steps can be performed without the use of an artificial neural network. The artificial neural network can comprise a convolutional neural network ("CNN") with multiple processing layers arranged in a serial and / or parallel configuration.

[0023] In one embodiment, the artificial neural network assigned to the detection step can be trained in a supervised manner using multiple input data pairs and output data pairs, each input data pair and output data pair comprising a panoramic image and associated anatomical data of an anatomical structure that can be preselected.

[0024] In one embodiment, the artificial neural network assigned to the computation step can be trained in a supervised manner using multiple input data pairs and output data pairs, each input data pair and output data pair comprising a panoramic image with associated anatomical data and an associated sharp layer.

[0025] In one embodiment, the artificial neural network assigned to the combined detection and calculation steps can be trained in a supervised manner using a plurality of input data pairs and output data pairs, each of which comprises a panoramic image and an associated sharp layer.

[0026] In one embodiment, the artificial neural network assigned to the combined steps of detection, calculation, and reconstruction may be trained in a supervised manner using multiple input and output data pairs, each comprising a projection image and an associated reconstructed panoramic image.

[0027] In one embodiment, an artificial neural network may be trained for the segmentation and / or localization operations for the detection step.

[0028] In the following description, the invention is explained in more detail using exemplary embodiments and by referring to the drawings. [Brief description of the drawings]

[0029] [Figure 1] FIG. 1 shows a flow chart according to an embodiment of the present invention. [Diagram 2] FIG. 2 shows a panoramic image with annotation of anatomical data. [Figure 3A] FIG. 3A shows the dependence of the panoramic image on the orientation of the patient position. [Figure 3B] FIG. 3B shows the dependence of the panoramic image on the orientation of the patient position. [Figure 3C]FIG. 3C shows the dependence of the panoramic image on the orientation of the patient position. [Figure 3D] FIG. 3D shows the dependence of the panoramic image on the orientation of the patient position. [Figure 4] FIG. 4 shows a flow chart according to a further embodiment of the invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0030] Enhanced delineation of clinically relevant, patient-specific, pre-selected anatomical structures The flow chart (100) shown in FIG. 1 may for example be implemented as a program executable by a computing means connected to an X-ray system for dental applications, such as a CBCT system.

[0031] In step (102), one or more initial panoramic images are provided. Each of the initial panoramic images is reconstructed from a number of projection images acquired during panoramic imaging performed by an X-ray system. Each of the initial panoramic images is reconstructed using corresponding initial reconstruction parameters, which are typically different from the parameters of the other initial panoramic images. The initial reconstruction parameters, i.e., the spatial relationship of the initial layers, for example, to the trajectory of the X-ray generator (304) and its corresponding X-ray detector (306), and to a bite block (not shown) and / or head fixation (not shown) for positioning the patient, are known in advance. The initial reconstruction parameters can be derived from the acquisition geometry or from defined layers in 3d space, or can be specified through default values.

[0032] In step (104), one or more clinically relevant patient-specific preselected anatomical structures are detected in at least one of the initial panoramic images, preferably by an artificial neural network, to provide anatomical data. The clinically relevant patient-specific preselected anatomical structures are detected through a segmentation operation. Alternatively, a localization operation can be performed.

[0033] The provided anatomical data describes the spatial relationships of pre-selected anatomical structures in the panoramic image. As some non-exhaustive examples, the anatomical data may comprise standardized tooth numbers of teeth, shape information of one or more tooth contours, the location of the pulp chambers of teeth, the location of the tips of teeth, or any other anatomical information derivable from the initial panoramic image. The anatomical data may comprise information about the jaw segments or other parts of the oral anatomy or surroundings, such as the mandible, maxilla, gums, tooth roots, nasal sinuses, paranasal sinuses, condyles, etc.

[0034] If the anatomical data are depicted sufficiently sharply in at least one of the initial panoramic images, their rough location in 3D space, i.e. their focal depth, can be derived by backprojection using the layer that shows a sharp depiction. The detection step can be performed by utilizing two or more initial panoramic images simultaneously.

[0035] In step (106), new reconstruction parameters, i.e. new (adjusted) layers, are calculated using the provided anatomical data and one or more initial panoramic images. The provided anatomical data includes spatial information regarding the detected clinically relevant patient-specific tooth structures. New reconstruction parameters are calculated, for example, to provide an appropriate depth of focus that provides a sharper depiction of the clinically relevant patient-specific anatomical features compared to the depiction of said structures in the initial panoramic image. The calculation is preferably performed by an artificial neural network.

[0036] When calculating new reconstruction parameters, one or more further constraints may be taken into account, such as, for example, preferring smooth layers in 3D space or weighting clinically relevant patient-specific pre-selected anatomical structures in conflicting image regions.

[0037] Calculating new reconstruction parameters that provide adequate depth of focus and provide sharper depiction of clinically relevant patient-specific anatomical features can be accomplished in several ways.

[0038] In one embodiment, new reconstruction parameters are calculated by varying the reconstruction parameters such that the sharpness of edges, contours and / or contrast of clinically relevant patient-specific pre-selected anatomical structures is optimal in the new panoramic image.

[0039] In another embodiment, new reconstruction parameters are calculated by detecting regions of clinically relevant, patient-specific, pre-selected anatomical structures within the panoramic image and modifying the reconstruction parameters such that the local definition of edges, contours and / or contrast of any structures within these regions is optimized within the new panoramic image.

[0040] In another embodiment, new reconstruction parameters are calculated by backprojecting the anatomical data to their positions in 3D space. These positions of the detected clinically relevant patient-specific preselected anatomical structures in 3D space can be used to calculate the positions of the sharp layers. Then, fine-tuning of the new reconstruction parameters can be calculated by optimizing the local sharpness of the edges or by optimizing the contrast by varying the reconstruction parameters in a small range.

[0041] The contrast or sharpness of structures other than the clinically relevant patient-specific preselected anatomical structures may be reduced when new reconstruction parameters are applied during reconstruction of the panoramic image.

[0042] In step 108, a panoramic image is reconstructed using the new reconstruction parameters, the panoramic image depicting clinically relevant patient-specific anatomical features in the sharpened new layer.

[0043] The reconstruction of a panoramic image can be done by weighted combination of pixels of the projection images to form a new reconstructed panoramic image. The reconstruction parameters dictate the exact mathematical manner in which the projection images are combined by specifying a) which pixels in the projection images should be combined with pixels in the reconstructed image, and b) how they are weighted during the combination.

[0044] In one embodiment, the clinically relevant patient-specific pre-selected anatomical structures to be detected may be input by a user, for example by using an input means (not shown), such as a display, keyboard, or mouse, allowing the dentist to adapt the pre-selection of anatomical structures to be polished according to the diagnostic and therapeutic procedures associated with the patient.

[0045] Trained Artificial Neural Networks In one embodiment, one or more of the calculation step (104), the detection step (106), and the reconstruction step (108) are performed by a trained artificial neural network, or they may be performed without the use of an artificial neural network.

[0046] The artificial neural network for the detection step can be trained in a supervised manner using multiple data pairs, each data pair including a previous panoramic image and associated anatomical data.

[0047] The artificial neural network for the computational step can be trained in a supervised manner using multiple data pairs, each data pair including a previous panoramic image provided with associated anatomical data and an associated sharp layer.

[0048] The artificial neural network may be further trained for the segmentation and localization operations applied in the detection step.

[0049] FIG. 2 shows an example of a panoramic image (202) that can be used in training and has annotations indicating pre-selected anatomical structures (204; 206; 208; 210; 212) such as teeth and condyles.

[0050] The panoramic image shows anatomical data superimposed on the initial panoramic image. A number of contours or boundaries around different teeth as well as other preselected anatomical structures such as the condyles are shown. Such boundaries or contours can be obtained by a segmentation operation. In FIG. 2, for example, the condyle boundaries (210) are shown. Furthermore, the localizations (212) of the condyles are shown as dots or points within the areas where the respective condyles are detected by the artificial neural network. FIG. 2 shows the boundaries (206, 208) associated with the detected teeth in the panoramic image. FIG. 2 also shows the tags (204) associated with the detected teeth in the panoramic image.

[0051] Patient misalignment 3A to 3D show the dependence of a panoramic image on the patient's position and orientation. FIG. 3A shows a first pose (308) of the patient's head (302) relative to an X-ray generator (304) and its corresponding X-ray detector (306). The first pose (308) may be considered an optimal pose. When the head (302) is scanned from the first pose (308), a panoramic image is generated showing a corresponding first view (310) of the intraoral cavity. The first view (310) may be considered a correctly registered view of the panoramic image. As can be seen, several anatomical structures such as teeth (320) and tooth roots (318) are visible in the panoramic image.

[0052] In some cases, it may happen that the head (302) is not optimally positioned or aligned. For example, FIG. 3C shows a case where the head (302) is tilted backward in a direction (316) relative to the placement of the X-ray source (304) and the X-ray detector (306). In such a case, distortions related to the patient's posture are introduced into the resulting panoramic image. This can be seen in the corresponding panoramic image shown in FIG. 3D, which is the corresponding second view (314). As can be seen, unlike the first view (310), the second view (314) does not display the dentition with a slightly upwardly curving curve, but rather somewhat resembles an S-shaped curve. Such posture-related distortions can make it difficult to compare different panoramic images.

[0053] Therefore, the method also provides a technique for adjusting the second view (210) so that the second view (314) is identical or similar to the first view (310). This is achieved by using anatomical data to calculate at least some of the new reconstruction parameters, whereby the new reconstruction parameters are adjusted to align the initial panoramic image with respect to a reference template that defines an optimal view. This adjustment is preferably performed by an artificial neural network.

[0054] 4 shows a flowchart (400) according to a further embodiment of the present invention. The flowchart (400) may be implemented as a program executable by a computing means connected to, for example, an X-ray system for dental applications, such as a CBCT system. The flowchart (400) and the flowchart (100) may be executed by the same computing means or by different computing means.

[0055] In step (402), one or more initial panoramic images are provided, each of which is reconstructed from a plurality of projection images acquired by an x-ray system, each of which is reconstructed using initial reconstruction parameters that may or may not be identical to the parameters of the other initial panoramic images.

[0056] In step (404), one or more clinically relevant patient-specific anatomical structures are detected, preferably via an artificial neural network, in at least one of the initial panoramic images to provide anatomical data.

[0057] In step (406), a number of new reconstruction parameters are calculated using the provided anatomical data and the one or more initial panoramic images. The new reconstruction parameters are calculated for aligning a projection of the panoramic image with respect to a reference template. The calculations are preferably performed by an artificial neural network.

[0058] In step 408, a panoramic image is reconstructed using the new reconstruction parameters. The panoramic image thus provides a registered view. When combined with flow chart 100, the present technique can simultaneously provide a panoramic image with clinically relevant patient-specific preselected anatomical structures in a clear layer and in a registered view.

Claims

1. 1. A computer-implemented method for reconstructing a panoramic image of a dental region of a patient, comprising: providing at least one initial panoramic image reconstructed by using a plurality of projection images of the patient and an initial layer including mapping points corresponding to dental regions of the patient so as to be clearly depicted in the initial panoramic image (102; 402); detecting (104, 404) one or more clinically relevant, patient-specific, pre-selected anatomical structures in said initial panoramic image to provide anatomical data (204, 206, 208, 210, 212) relating to spatial characteristics of the detected clinically relevant, patient-specific, pre-selected anatomical structures, respectively; using said anatomical data and said initial panoramic image to calculate (106; 406) a new layer including said clinically relevant patient-specific preselected anatomical structures to provide a clearer depiction of said clinically relevant patient-specific preselected anatomical structures in a new reconstructed panoramic image compared to the depiction of said clinically relevant patient-specific preselected anatomical structures in said initial panoramic image; and reconstructing (108, 408) the new panoramic image using the plurality of projection images of the patient and the calculated new layers.

2. The method of claim 1 , wherein the clinically relevant, patient-specific, pre-selected anatomical structures to be detected are input by a user.

3. 3. The method according to claim 1 or 2, characterized in that one or more of the calculation, detection and reconstruction steps are performed by a trained artificial neural network.

4. 4. The method of claim 3, wherein the artificial neural network assigned to the detection step is trained in a supervised manner using a plurality of input data pairs and output data pairs, each comprising a panoramic image and associated anatomical data.

5. 5. The method according to claim 3 or 4, characterized in that the artificial neural network assigned to the calculation step is trained in a supervised manner using a plurality of input data pairs and output data pairs, each data pair comprising a panoramic image with associated anatomical data and an associated sharp layer.

6. 4. The method of claim 3, wherein the artificial neural network assigned to the combination of the detecting and calculating steps is trained in a supervised manner using a plurality of input and output data pairs, each of the plurality of input and output data pairs comprising a panoramic image and an associated sharp layer.

7. 4. The method of claim 3, wherein the artificial neural network assigned to the combination of the detection, calculation and reconstruction steps is trained in a supervised manner using a plurality of input and output data pairs, each of the plurality of input and output data pairs comprising a plurality of projection images and an associated reconstructed panoramic image.

8. A computer readable storage medium storing a program including instructions which, when executed by a computerized X-ray system, cause the computerized X-ray system to carry out the steps of the method according to any one of claims 1 to 7.

9. A computerized X-ray system for reconstructing panoramic images, comprising means adapted to carry out the steps of the method according to any one of claims 1 to 7.

Images