A dental implant three-dimensional modeling design method and system based on oral scanning data

Abstract

The application provides a kind of implant three-dimensional modeling design method and system based on oral scanning data, related to digital oral technology field, by obtaining equipment geometric measurement deviation information, by spectrum analysis, obtain the biological optical response information of specific area soft tissue of patient;When processing original point cloud data, the point cloud geometry is compared with the optical characteristics and the aforementioned information multidimensionally, and the systematic geometric deformation is accurately identified according to the preset rule, and the differential correction algorithm is called according to the deformation type for targeted correction;When reconstructing three-dimensional model, dynamically adjust parameters, and reconstruct important anatomical regions such as gingival papilla with high fidelity.The application effectively corrects the systematic distortion that the traditional method cannot handle, generates a three-dimensional digital model that fits the real oral anatomy of the patient, provides accurate basis for implant and prosthesis design, optimizes the treatment plan, and improves the accuracy and reliability of digital oral diagnosis and treatment.

Description

Technical Field

[0001] This application relates to the field of digital oral technology, and more specifically, to a method and system for three-dimensional modeling and design of implants based on oral scan data. Background Technology

[0002] In modern dental implant restoration procedures, acquiring three-dimensional point cloud data of the patient's oral cavity using an intraoral scanner and constructing a digital model based on this data is a crucial step in achieving precision treatment. However, existing technology suffers from a deep and subtle technical problem in this process. Specifically, during factory calibration or routine maintenance, the coating of the standard model used by the intraoral scanner may exhibit microscopic wear or aging. This invisible imperfection alters the local optical reflectivity of the standard model, resulting in a small but systematic measurement bias embedded within the scanner during data acquisition. The problem becomes more complex when this potentially biased scanner is used to scan a patient's oral cavity. Particularly for patients with gingival recession or unique physiological conditions, their soft tissues, such as the gingival papillae, may exhibit optical response characteristics that happen to resemble the areas with imperfections on the standard model due to their moist, translucent texture and unique microstructure. This coincidence can "trigger" and significantly amplify the scanner's built-in systematic bias, causing regular geometric distortions in the acquired raw point cloud data in these key soft tissue areas, such as stretching or twisting of the shape. However, existing 3D modeling software typically lacks a mechanism to identify such complex errors when processing data. It often incorrectly classifies these errors as random noise and processes them using a uniform smoothing algorithm. This "one-size-fits-all" approach to noise reduction not only fails to correct potential geometric deformations but also further smooths out or distorts already distorted fine anatomical structures such as gingival papillae and mucosal folds, resulting in the permanent loss of crucial physiological information. Ultimately, implants and restorations designed based on such distorted models cannot accurately reflect the patient's biological basis and may lead to long-term complications such as peri-implantitis and soft tissue recession. Summary of the Invention

[0003] This application provides a method and system for three-dimensional modeling and design of implants based on oral scan data, aiming to solve the technical problem that the existing technology cannot identify and correct the systematic geometric deformation caused by the combined effect of scanner bias and biological characteristics, resulting in insufficient accuracy of the three-dimensional model and loss of key soft tissue details.

[0004] In a first aspect, this application discloses a method for three-dimensional modeling and designing implants based on oral cavity scan data, including: It receives raw point cloud data of the oral soft tissue region acquired by an intraoral scanner, and obtains device measurement deviation information containing geometric measurement offset, which describes the intraoral scanner under specific lighting conditions, as well as bio-optical response information reflecting the optical response characteristics of the oral soft tissue. Geometric morphological feature analysis and optical reflection feature analysis are performed on each local region in the original point cloud data to obtain the feature analysis results; The feature analysis results are compared with the equipment measurement deviation information and the bio-optical response information. By using preset judgment rules, the systematic geometric deformation in the original point cloud data caused by the combined effect of scanner deviation and biological characteristics is identified. Based on the type of geometric deformation, a corresponding differential correction strategy is adopted to perform geometric correction and obtain corrected point cloud data; Based on the corrected point cloud data, the surface reconstruction algorithm parameters are adjusted to preserve the fine anatomical structure of the oral soft tissue, and a three-dimensional oral model is constructed; and based on the three-dimensional oral model, implant placement parameters are planned and prosthesis morphology parameters are designed.

[0005] Secondly, this application discloses a 3D modeling and design system for implants based on oral cavity scan data, the system comprising: The data acquisition module is used to receive raw point cloud data of the oral soft tissue region collected by the intraoral scanner, and to acquire device measurement deviation information containing geometric measurement offsets that describes the intraoral scanner under specific lighting conditions, as well as bio-optical response information reflecting the optical response characteristics of the oral soft tissue. The feature analysis module is used to perform geometric morphological feature analysis and optical reflection feature analysis on each local region in the original point cloud data to obtain the feature analysis results. The geometric deformation recognition module is used to compare the feature analysis results with the equipment measurement deviation information and the bio-optical response information. Through preset judgment rules, it identifies the systematic geometric deformation in the original point cloud data caused by the combined effect of scanner deviation and biological characteristics. The geometric correction module is used to perform geometric correction based on the type of geometric deformation using corresponding differentiated correction strategies, and obtain corrected point cloud data. The 3D model building module is used to construct a 3D oral cavity model by adjusting the surface reconstruction algorithm parameters based on the corrected point cloud data to preserve the fine anatomical structure of the oral soft tissue; and to plan the implantation parameters and design the prosthesis morphology parameters based on the 3D oral cavity model.

[0006] This technical solution provides a physical system capable of executing the aforementioned innovative methods, transforming abstract algorithmic processes into concrete functional modules. It offers clear architectural support for the practical application and productization of this technology, and has real industrialization value. Beneficial effects

[0007] This application provides a method and system for 3D modeling and designing implants based on oral scan data, which achieves significant advantages over existing technologies. Existing technologies, when processing intraoral scan data, cannot distinguish between regular geometric deformations and random noise caused by a combination of systematic equipment bias and soft tissue bio-optical characteristics. They typically employ uniform smoothing, leading to distortion and loss of fine structures in key soft tissues such as the gingival papilla. This application's technical solution addresses the root cause of the problem. First, it obtains pure equipment geometric measurement bias information through a precise calibration process, and simultaneously obtains bio-optical response information of specific soft tissue regions of the patient through spectral analysis. When processing raw point cloud data, this solution does not blindly denoise; instead, it compares the geometric and optical features of the point cloud with the pre-acquired equipment bias information and bio-optical information in multiple dimensions. Based on preset logical rules, it accurately identifies systematic geometric deformations with specific patterns. Once identified, it calls a matching differential correction algorithm for targeted correction based on the type of deformation (e.g., stretching or twisting). Finally, when reconstructing the 3D model, a strategy of dynamically adjusting parameters is employed to perform high-fidelity reconstruction of important anatomical regions such as the gingival papilla. Through this series of innovative steps, this application can effectively correct systematic distortions that traditional methods cannot handle, and ultimately generate a three-dimensional digital model that is highly faithful to the patient's actual oral anatomy. This provides an unprecedentedly accurate basis for the subsequent design of implants and prostheses, thereby enabling the planning of treatment plans that are more in line with biological principles and help maintain the long-term health of soft tissues, significantly improving the accuracy and reliability of digital oral diagnosis and treatment. Attached Figure Description

[0008] To illustrate this application more clearly, the accompanying drawings used in the embodiments will be briefly described below. Obviously, those skilled in the art can obtain other drawings based on these drawings without any creative effort.

[0009] Figure 1 The diagram above illustrates a flowchart of a method for designing three-dimensional implant models based on oral scan data. Figure 2 The diagram above illustrates a structural schematic of a 3D modeling and design system for implants based on oral scan data.

[0010] Figure reference numerals: 100, Implant 3D Modeling and Design System Based on Oral Scan Data; 10, Data Acquisition Module; 20, Feature Analysis Module; 30, Geometric Deformation Recognition Module; 40, Geometric Correction Module; 50, 3D Model Construction Module. Detailed Implementation

[0011] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0012] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0013] In the digital workflow of dental implant restoration, obtaining accurate three-dimensional models of soft tissue is the cornerstone of developing a successful treatment plan. However, a common but often overlooked problem is that when dentists use intraoral scanners to scan a patient's alveolar ridge, especially in areas with gingival recession, the resulting digital model on the computer screen often exhibits systematic distortions in its soft tissue morphology, particularly intricate structures such as gingival papillae and gingival margins, that do not match what is seen with the naked eye. For example, a gingival papilla that should be full and sharp may appear slightly flattened and stretched in a specific direction on the model, or a wavy gingival margin may become abnormally straight. This distortion is not random noise, but rather a geometrically regular deformation stemming from the complex interaction between the inherent minute measurement biases of the scanning equipment and the unique moist, translucent optical properties of receding gingival tissue. Traditional modeling software often misjudges this deformation as general noise and smooths it out. As a result, it not only fails to restore the true shape, but also further smooths out key anatomical details, making the final model unable to provide reliable biological basis for the design of implant depth, angle, and prosthesis edge.

[0014] like Figure 1As shown, an exemplary flowchart of a method for 3D modeling and designing implants based on oral cavity scan data is illustrated. This paper proposes a mechanism to fundamentally solve the problem of soft tissue scanning distortion. By actively acquiring two core pieces of information—device deviation and bio-optical response—and comparing them with measured features of point cloud data, it can accurately identify systematic geometric deformations that are treated as random noise in traditional methods. This lays the foundation for subsequent accurate correction and high-fidelity modeling, fundamentally improving the accuracy of oral cavity 3D models. This application provides a method for 3D modeling and designing implants based on oral cavity scan data, including: S10, receives raw point cloud data of the oral soft tissue region collected by the intraoral scanner, and obtains device measurement deviation information containing geometric measurement offset, which describes the intraoral scanner under specific lighting conditions, as well as bio-optical response information reflecting the optical response characteristics of the oral soft tissue. S20, Perform geometric morphological feature analysis and optical reflection feature analysis on each local region in the original point cloud data to obtain the feature analysis results; S30 compares the feature analysis results with the equipment measurement deviation information and the bio-optical response information, and identifies the systematic geometric deformation in the original point cloud data caused by the combined effect of scanner deviation and biological characteristics through preset judgment rules. S40. Based on the type of geometric deformation, a corresponding differential correction strategy is adopted to perform geometric correction and obtain corrected point cloud data. S50, based on the corrected point cloud data, adjusts the surface reconstruction algorithm parameters to preserve the fine anatomical structure of the oral soft tissue and constructs a three-dimensional oral model; and based on the three-dimensional oral model, plans the implantation parameters and designs the prosthesis morphology parameters.

[0015] To better understand the technical solution of this application, some technical concepts involved in this application will be explained first.

[0016] Equipment measurement deviation information refers to a characteristic data file that specifically describes the geometric measurement output deviations of a particular intraoral scanner under specific optical input conditions. This information is not a general error range, but rather precise to the point that "when the scanner light source illuminates a surface with specific reflectivity and scattering characteristics, its three-dimensional coordinate measurement results will systematically deviate by a specific distance in a certain direction." This information is typically generated before the scanner leaves the factory or during periodic maintenance by scanning a series of standard calibration bodies with known geometric and optical characteristics. It can be understood as a unique, intrinsic "error fingerprint" for each scanner, recording predictable systematic weaknesses in the device's optical system and geometric calculation chain.

[0017] Bio-optical response information refers to a database that stores the optical properties of human oral soft tissue under different types and physiological conditions. For example, the database records typical reflectance spectra, absorption coefficients, and scattering coefficients of different tissues such as healthy gingiva, hyperplastic gingiva, atrophic gingiva, and oral mucosa. Especially for atrophic gingiva, its thinner tissue, altered water content, and increased translucency lead to a significantly different response to scanning light compared to healthy tissue. This information provides a biological basis for the system to determine the type of tissue being scanned and the potential scanning abnormalities it may cause.

[0018] Systematic geometric distortion refers to non-random geometric morphological distortions with specific patterns that appear in raw point cloud data. It is not random noise caused by operator hand tremors or accidental patient movements, but rather a reproducible distortion resulting from the combined and continuous effects of both equipment measurement bias and bio-optical response. For example, when an inherent bias of the scanner is "triggered" by the optical characteristics of receding gingiva, a "stretching deformation" or "torsion deformation" may stably occur in the gingival papilla region.

[0019] In a specific implementation scenario of this application, the entire execution process of the method can be broken down in detail. Imagine a patient whose long-term tooth loss has led to alveolar ridge recession and thinning of the gingival tissue in the mandibular posterior region, requiring the dentist to perform implant restoration.

[0020] First, the dentist uses an intraoral scanner to scan the patient's edentulous area. The scanner's probe emits structured light, illuminating the patient's teeth and soft tissue surfaces. Its built-in camera captures the modulated fringe image, and the internal processor calculates millions of three-dimensional spatial coordinates in real time, forming raw point cloud data. This data is transmitted as a data stream to a connected computer workstation. Simultaneously, the workstation's specialized software loads two key databases from local storage: one uniquely corresponding to the scanner's model and serial number, containing device measurement deviation information; and the other a general database of oral bio-optical responses.

[0021] Next, the software begins processing the received raw point cloud data. It divides the entire point cloud into numerous small local regions, for example, one cubic millimeter as an analysis unit. For each local region, the software performs preliminary feature analysis. In a basic implementation, the software might only analyze the currently acquired single frame of point cloud data. It calculates the geometric features of the point cloud within that region, such as obtaining surface curvature by calculating the curvature of the local fitted plane of the point cloud, or calculating the average direction and dispersion of the normal vectors of all points within that region. Simultaneously, the software also analyzes optical reflection characteristics, recording the intensity or color information of the light reflected back from each point. In this way, the software obtains a preliminary geometric and optical profile of the entire scanned area.

[0022] Next, the crucial identification step begins. The software compares the feature analysis results of each local region obtained in the previous step with two pre-loaded databases. In a more basic implementation, the judgment rules might be relatively simple. For example, if the average reflectance of a region is approximately equal to the feature value of "receding gingiva" in the bio-optical response information database within a wide range, and the curvature value of that region is lower than the threshold of normal gingiva, the system might mark it as a suspected deformed area. While this method can initially screen out some abnormalities, it is prone to misjudgment due to its broad judgment criteria.

[0023] After identifying suspected deformed areas, the system enters the geometric correction stage. In a simple implementation, the system might indiscriminately apply a generic smoothing or morphological filtering algorithm to all marked deformed areas, attempting to "pull" the abnormal geometry back to a more normal state. This "one-size-fits-all" correction method, while potentially improving visual appeal to some extent, has limited effectiveness because it doesn't address the root cause and specific pattern of the deformation, and may even introduce new distortions.

[0024] Finally, after all the point cloud data has been corrected through the above processing steps, the software calls a surface reconstruction algorithm, such as the Poisson surface reconstruction algorithm, to transform the discrete point cloud data into a continuous, smooth triangular mesh surface, which is the final 3D model of the oral cavity. Once the model is obtained, the dentist can perform virtual implant placement planning on it, designing the ideal position, angle, and depth of the implant, and further designing the morphology of the superimposed restoration, such as the marginal contour of the crown and the contact point with adjacent teeth.

[0025] Through the aforementioned process, even in a relatively basic implementation, the technical solution of this application demonstrates its core innovative value. It is the first to incorporate two long-neglected error sources—equipment bias and biological characteristics—into the modeling process, establishing a complete logical closed loop of "problem identification - problem analysis - attempt to solve the problem." Compared to existing technologies that treat all anomalies as random noise and blur them, this application consciously distinguishes and attempts to locate systematic deformations. This in itself represents a significant step forward in improving model accuracy, making it possible to obtain more realistic soft tissue morphology, thus laying a far more solid foundation for subsequent clinical design than in the past.

[0026] In some embodiments, to further improve the accuracy of feature analysis and avoid misjudgments caused by minute dynamic changes during the scanning process, this application proposes a more optimized scheme. The steps of performing geometric morphological feature analysis and optical reflection feature analysis on each local region in the original point cloud data to obtain the feature analysis results can specifically include: During oral cavity scanning, multiple frames of raw point cloud data are acquired at continuous time points to obtain the geometric morphological features and optical reflection features of local areas at different time points. Identify the geometric features of stable reference regions in raw point cloud data; The geometric morphological features and optical reflection features of the local area collected at each time point are registered with the geometric features of the stable reference area to eliminate the influence of overall displacement and obtain the registered local area geometric morphological features and optical reflection features. Time series analysis is performed on the geometric morphology and optical reflection features of the registered local region to identify their change trajectories at continuous time points; and based on the amplitude and frequency of the change trajectory, the feature differences caused by minute irregular movements are distinguished from actual physiological structural changes or scanner errors, thereby obtaining feature analysis results.

[0027] The reason for this improvement is that, in actual scanning operations, even with a stable handheld device, the patient's subtle breathing, heartbeat, and involuntary movements of the tongue or cheek muscles can cause high-frequency, small-amplitude irregular movements on the soft tissue surface. If only single-frame data is analyzed, these dynamic artifacts may be mistaken for the true geometric features of the tissue or scanning errors. The multi-frame analysis method proposed in this application effectively solves this problem.

[0028] Specifically, when the scanner probe pauses or moves slowly in an area (such as the aforementioned receding gingival papilla), the system continuously acquires and records 30 frames of point cloud data within a very short time, such as 0.5 seconds. This is equivalent to briefly "recording" the area. Simultaneously, the system automatically searches for a geometrically stable reference object within the scanning range. Ideally, this would be the enamel surface of a nearby healthy natural tooth or the surface of a stabilized implant abutment. The morphology of these hard tissues remains absolutely unchanged during the scanning process. Then, using the position and orientation of the stable reference area in the first frame as a baseline, the system aligns all subsequent 29 frames of data to this baseline using an iterative nearest-closest point (ICP) algorithm or its variant. This registration process effectively eliminates global coordinate system drift caused by slight movements or rotations of the dentist's hand, ensuring that all frames are compared within the same coordinate system.

[0029] After registration, the system obtains a sequence of geometric and optical features of the same local region (such as the gingival papilla) at 30 consecutive time points. For a specific point on the gingival papilla, the system can track its three-dimensional coordinate changes over these 30 frames. If the point's trajectory is a high-frequency, irregular back-and-forth vibration with a small amplitude, the system will determine that this is likely caused by the patient's physiological tremors or data noise, and will suppress this influence through time-domain averaging or filtering when forming the final feature analysis results. Conversely, if the system observes that this point and a cluster of points around it exhibit a continuous, low-frequency "drift" trend in the same direction over the 30 frames, this is highly likely strong evidence of systematic geometric deformation. By analyzing the amplitude and frequency of the changing trajectory, the system can clearly distinguish between random disturbances and regular systematic deviations, thus obtaining a more reliable feature analysis result that has been dynamically validated, laying a solid data foundation for subsequent accurate identification.

[0030] In some embodiments, to make the identification process of systematic geometric deformation more accurate and reliable and avoid fuzzy judgments, this application further proposes that the step of comparing the feature analysis results with device measurement deviation information and bio-optical response information, and identifying the systematic geometric deformation in the original point cloud data caused by the combined effect of scanner deviation and biological characteristics through preset judgment rules, may specifically include: The local geometric features and optical reflection features of the original point cloud data at multiple time points are acquired to identify the geometric features of stable regions in the original point cloud data as a reference. The local geometric features include surface curvature variation features and surface normal direction consistency features; the optical reflection features include light intensity variation features and reflectivity variation features. By comparing the local geometric features and optical reflection features at each time point with the equipment measurement deviation information, bio-optical response information and reference benchmark, the type of geometric deformation that may exist at that time point is identified. The type of geometric deformation is a regular geometric distortion type produced by the combined effect of scanner deviation and biological characteristics, including at least one of stretching deformation and torsional deformation. By tracking the changing trends of geometric deformation patterns identified in the same local area at different time points and recognizing the regular evolution of geometric deformation patterns at continuous time points, and by pre-setting judgment rules, systematic geometric deformations can be identified.

[0031] Furthermore, a specific and feasible pre-defined judgment rule is: when the point cloud data of a local area simultaneously meets the following conditions, such as sparse point distribution, abnormal curvature change, poor consistency of normal direction, light intensity change pattern matching the abnormal scattering pattern of optical deviation data points, and regional optical characteristics consistent with the bio-optical response characteristics of the patient's atrophied gingiva, the area is judged to be a systematic geometric deformation area.

[0032] This process is equivalent to establishing a multi-dimensional chain of evidence, conducting a rigorous "courtroom-style" trial on the suspected area. Let's return to the aforementioned application scenario and analyze the gingival papilla area that appears to be stretched.

[0033] The system first extracts more specific geometric and optical indicators from the feature results obtained after time-series analysis. Geometrically, it calculates the surface curvature changes of the region. A healthy gingival papilla should be approximately conical or pyramidal in shape, with smooth and continuous curvature changes. Deformed areas may exhibit abnormally flat tops (curvature close to zero) or abrupt curvature changes at the interface with surrounding tissues. Simultaneously, the system calculates the consistency of surface normal directions. On a smooth surface, the normal vectors of adjacent points should point almost in the same direction, exhibiting high consistency. However, in deformed or stretched areas, the arrangement of points is disturbed, potentially leading to irregular deflections in the calculated normal vectors, resulting in poorer consistency. Optically, the system analyzes the light intensity variation patterns and reflectivity changes of the region.

[0034] Then, the system begins to perform cross-validation based on preset judgment rules: First, the system checks the point cloud density of the area and finds that the average distance between points is greater than that of the surrounding healthy gingival area, satisfying the condition of "sparse point distribution". This is because geometric stretching will "stretch" the originally dense points apart.

[0035] Second, the system examined its geometric features and found that the curvature of the top of the region was much smaller than the reference value of the curvature database of normal gingival papilla, and there was an unnatural jump in curvature at the edge, which met the conditions of "abnormal curvature change" and "poor consistency of normal direction".

[0036] Third, and most crucially, the system matches the optical reflection characteristics of the region—the specific light intensity distribution pattern formed after light is scattered and absorbed by the tissue surface—with the equipment's measurement deviation database. The system discovered that this light intensity pattern perfectly matches "Optical Trigger Mode A," recorded in the deviation database, which triggers the scanner to produce a geometric deviation of "0.1 mm stretch along the positive Y-axis." This establishes a direct correlation between equipment defects and geometric deformation.

[0037] Fourth, the system then compares the optical characteristics of this area with the bio-optical response database. The system confirms that the reflectance spectrum of this "optical triggering mode A" is completely consistent with the optical response characteristics of "moderately atrophied gingiva in a saliva-moistened state" recorded in the database. This explains why this discrepancy occurred here.

[0038] Finally, the system tracked the region's performance across multiple frames of data, finding that all the aforementioned features and matching relationships remained stable in each frame, exhibiting a continuous and regular evolution. At this point, the entire chain of evidence was completely closed. Based on this rigorous pre-defined judgment rule, the system ultimately made a high-confidence determination: the gingival papilla region does indeed exhibit systematic geometric deformation caused by the combined effects of scanner bias and biological characteristics, and the deformation type is "stretching deformation." This identification method, based on multiple pieces of evidence and rigorous logic, greatly eliminates randomness and uncertainty, ensuring the targeted and accurate nature of subsequent corrective actions.

[0039] In some embodiments, after accurately identifying the type of systematic geometric deformation, in order to perform targeted repairs rather than using a uniform and ineffective fuzzing process, this application further proposes that the step of using a corresponding differentiated correction strategy for geometric correction based on the type of geometric deformation may specifically include: A mapping relationship library between geometric deformation types and correction strategies is established. For tensile deformation, a non-uniform scaling correction algorithm is configured to restore the original geometric proportions. For torsional deformation, a local non-rigid deformation correction algorithm based on radial basis functions is configured to restore the natural anatomical shape. If the geometric deformation is determined to be a stretching deformation, the main stretching direction of the local area where the stretching deformation is located is identified, and a reverse non-uniform scaling correction algorithm is applied along the direction to correct the deformation, so that the coordinates of each point in the area are adjusted according to the preset shrinkage ratio to restore the original geometric proportions. If the geometric deformation is determined to be a tortuous deformation, multiple control points in the area where the tortuous deformation is located are selected, the target position of each control point is determined according to the geometric trend of the surrounding healthy tissue, and then a local non-rigid deformation correction algorithm based on radial basis functions is used for correction.

[0040] The core idea of ​​this strategy is to prepare different "special treatments" for different types of "diseases". The software will have a mapping library, which can be in the form of a simple lookup table or a configuration file, which clearly defines the correspondence between the deformation type and the correction algorithm.

[0041] Now, let's take a closer look at how these two differentiated correction strategies are implemented.

[0042] Correction for "stretching deformation". In the aforementioned scenario, the system has determined that the gingival papilla region has experienced "stretching deformation". The system first performs principal component analysis (PCA) on the point cloud of this region to determine the principal direction of stretching. The analysis results may show that the variance of the data points is largest in the Y-axis direction, indicating that the principal stretching direction is the Y-axis direction. Simultaneously, by comparing the dimensions with a standard gingival papilla model in a digital dental morphology database, the system calculates that this region is stretched by approximately 15% in the Y-axis direction, i.e., a stretching ratio of 1.15. At this point, the system invokes a non-uniform scaling correction algorithm. This algorithm fixes the geometric center of the region and then multiplies the Y-coordinate value of each point within the region by a shrinkage ratio, which is the reciprocal of the stretching ratio, i.e., 1 / 1.15. The X and Z coordinate values ​​of these points remain unchanged. Through this operation, the entire region is like a unidirectionally stretched spring being precisely compressed back to its original length; the original geometric proportions between its height, width, and thickness are restored, and the flattened gingival papilla shape is "erected" again at the data level.

[0043] To demonstrate the applicability of this application to different situations, we envision another scenario: the correction of "distorted deformation." Suppose that during the scanning of a single implant in a patient, due to the challenging angle of the scanning probe, the acquired gingival cuff point cloud data around the implant exhibits a certain degree of rotation and distortion, resulting in an unnatural shape. The system, through the aforementioned identification process, determines this area to be "distorted deformation." At this point, the mapping database instructs the system to invoke a local non-rigid deformation correction algorithm based on radial basis functions (RBF). The algorithm's execution process is as follows: First, the system automatically selects several control points within the distorted gingival cuff area, and simultaneously selects several reference points on the surrounding, morphologically stable, and healthy gingival surface. Then, based on the smooth geometric contours of the surrounding healthy gingiva, the system calculates the ideal three-dimensional positions of the control points within the distorted area using surface interpolation or extrapolation; these ideal positions are called target positions. Next, the RBF algorithm begins operation. It can be simply understood as a kind of spatial "deformation magic." It can construct a smooth mathematical function that causes all points within the distortion area to undergo smooth displacement. The final effect of this displacement is that each selected control point is precisely moved to its respective target position. Because RBF ensures the smoothness of the deformation, other points between the control points will also undergo natural, non-rigid movement. The entire distorted area is like a wrung-out cloth being pulled and automatically smoothed out by a few key points, restoring it to its proper natural state that conforms to the anatomical shape of the surrounding tissue.

[0044] By employing this differentiated correction strategy based on accurate identification, this application can specifically address different types of systematic geometric deformations. Its correction effect is far superior to that of traditional single smoothing algorithms, resulting in high-fidelity corrected point cloud data that is infinitely close to the real anatomical morphology in detail.

[0045] To achieve the aforementioned accurate identification and correction, a crucial prerequisite is to obtain two high-precision data files in advance: one about the "defects" of the scanning equipment itself, and the other about the "characteristics" of the biological tissue being scanned.

[0046] First, in order to obtain pure and accurate equipment measurement deviation information and avoid conflating optical noise such as light source aging and ambient light interference with actual geometric measurement errors, the steps for obtaining equipment measurement deviation information containing geometric measurement offsets generated by an intraoral scanner under specific lighting conditions can specifically include: When the reference light source illuminates multiple calibration points with known optical reflection characteristics set on the standard model, the reflection spectrum data of each calibration point is collected. The reflection spectrum data is compared with the preset pure spectrum reference of the calibration point to identify the spectral drift of the reference light source. Based on the spectral drift, adjust the driving parameters of the reference light source to compensate for the spectral drift of the reference light source; When the scanner's own light source illuminates the calibration point, reflectance spectral data is collected. Differential analysis is performed on the spectral data collected under the scanner's own light source and the spectral data collected under the compensated reference light source to obtain the differential analysis results. Based on the differential analysis results, the optical noise introduced by the decrease in the spectral purity of the scanner's own light source and the photosensitive reaction characteristics of the coating on the standard model body was removed; The optical signal after removing optical noise is compared with a preset digital reference model. Based on the comparison results, the pure geometric deviation data points are quantified and recorded as equipment measurement deviation information.

[0047] This process can be understood as an extremely precise "physical examination" of the scanner. In one specific embodiment, the process is carried out in a specialized calibration laboratory.

[0048] The standard model used for calibration is a geometric object made of a highly stable ceramic material (such as zirconium oxide), with dozens of tiny hemispherical pits precisely machined on its surface as calibration points. The three-dimensional coordinates and optical reflection characteristics of these calibration points are pre-calibrated using a higher-precision coordinate measuring machine and spectrometer.

[0049] At the start of calibration, a spectrally tunable, highly stable laser is used as a reference light source. The system controls the reference light source to emit a laser beam of a specific wavelength (e.g., 660 nm) and illuminate one of the calibration points. The scanner's photosensitive element acquires the reflectance spectrum data at this moment. The software compares this measured spectrum with a "pure spectral reference" for that calibration point under an ideal 660 nm light source, stored in a pre-existing database. If the peak wavelength of the measured spectrum deviates from 660 nm, for example, to 660.5 nm, the system identifies a 0.5 nm spectral drift in the reference light source. Subsequently, the system sends an adjustment command to the reference light source's drive controller, fine-tuning its drive current or temperature until its output spectrum perfectly matches the reference, completing the compensation for the reference light source.

[0050] Next, the reference light source is turned off, and the intraoral scanner's own LED light source is turned on, illuminating the same calibration point and acquiring reflectance spectral data. At this point, the software performs a crucial differential analysis: subtracting the previously acquired spectral data from the compensated reference light source from the newly acquired spectral data generated by the scanner's own light source. This difference exposes all optical noise introduced by non-ideal factors. For example, if the scanner's LED aging causes a decrease in spectral purity, or if the anti-reflective coating on the standard model surface produces a weak fluorescence effect under specific lighting conditions, these will appear as specific noise signals in the differential results.

[0051] Based on the differential analysis results, the software removes optical noise components from the raw optical signal acquired by the scanner's own light source using digital filtering and other methods, resulting in a "clean" optical signal. This signal theoretically only reflects the deviations introduced by the scanner's optical lens and geometric calculation algorithms.

[0052] Finally, the software calculates the three-dimensional coordinates of the calibration point based on this "pure" optical signal and compares them with the preset, real digital reference coordinates of the calibration point. The spatial vector difference between the two, for example (Δx=0.01mm, Δy=-0.02mm, Δz=0.005mm), is quantified and recorded. This process is repeated for all calibration points, eventually compiling a detailed "map" describing the pure geometric deviations of the scanner under different spatial positions and optical conditions—that is, the equipment measurement deviation information.

[0053] Similarly, to understand another key factor contributing to the bias—the individualized characteristics of the patient's soft tissue—the steps to obtain bio-optical response information reflecting the optical response characteristics of oral soft tissue can specifically include: The original point cloud data is divided into regions to obtain multiple sub-regions; Perform local pre-scanning on each sub-region and collect reflectance spectral data of each sub-region under multiple specific wavelength illuminations; Based on the reflectance spectral data of each sub-region, the tissue type and physiological state of each sub-region are classified to obtain the classification results; Based on the classification results of each sub-region, a set of optical response parameters matching that sub-region is extracted from the bio-optical property database as bio-optical response information.

[0054] This process is equivalent to performing a rapid "composition analysis" of the patient's oral soft tissues while undergoing a formal scan.

[0055] In one specific embodiment, when the doctor moves the scanner probe inside the patient's mouth, the software dynamically divides the real-time point cloud data into multiple sub-regions, such as dividing the gums, mucosa, etc. into different blocks based on the teeth.

[0056] As the scanning probe sweeps across a sub-region (such as the labial gingiva in the anterior teeth region), the scanner performs a rapid local pre-scan. This pre-scan is not for acquiring three-dimensional morphology, but rather for collecting spectral information. The scanner's light source system flashes three or more monochromatic lights with different center wavelengths at extremely high frequencies (imperceptible to the human eye), such as blue light (450nm), green light (530nm), and red light (660nm). For each point within this sub-region, the photosensitive element records its reflectance intensity under these three light illuminations.

[0057] The software generates a reflectance spectrum curve for this sub-region based on the collected multi-wavelength reflectance intensity data. Then, the software performs pattern matching between this measured spectrum curve and a built-in database of bio-optical properties. This database pre-stores a large amount of spectral data from pathologically confirmed oral tissue samples. For example: - Healthy gums have a higher reflectivity in the green light band (appearing pink), while they absorb more in the red light band (because they are rich in blood). - Due to connective tissue fibrosis and reduced blood vessels, atrophic gingiva will have reduced absorption in the red light band, and its overall reflectivity may be increased. Inflamed gums, due to vasodilation and congestion, absorb green and blue light significantly more readily and reflect red light more strongly.

[0058] By comparing the data with standard spectral curves in the database, the software can classify the tissue type and physiological state of the sub-region, obtaining classification results such as "healthy attached gingiva", "moderately atrophied free gingiva" or "mildly inflamed gingival papilla".

[0059] Once the classification is complete, the software extracts a complete set of optical response parameters from the database that perfectly matches the classification result, including but not limited to detailed absorption coefficients, scattering coefficients, and anisotropy factors. This set of parameters is the bio-optical response information tailored to this specific region, which will serve as an important biological basis for subsequent identification of systematic geometric deformations.

[0060] After obtaining high-fidelity corrected point cloud data, another crucial step is converting it into a 3D model without losing painstakingly recovered details. To this end, adjusting the surface reconstruction algorithm parameters based on the corrected point cloud data to preserve the fine anatomical structure of the oral soft tissue can specifically include: A 3D oral cavity model was reconstructed using the Poisson surface reconstruction algorithm based on the corrected point cloud data. During the reconstruction process, the reconstruction parameters were dynamically adjusted according to the anatomical importance of different regions in the corrected point cloud data. Anatomical importance was determined based on whether a region contained the fine anatomical structures required for implant design. Fine anatomical structures included gingival papillae and mucosal folds, while non-fine anatomical structures were flat soft tissue areas. For areas identified as fine anatomical structures such as gingival papillae or mucosal folds, a first grid sampling density and a first smoothness coefficient are set; For flat soft tissue regions, a second grid sampling density and a second smoothing intensity coefficient are set; The sampling density of the first grid is higher than that of the second grid; the first smoothing intensity coefficient is lower than that of the second smoothing intensity coefficient.

[0061] This process can be seen as a refined strategy of "differentiated treatment" during model reconstruction.

[0062] In one specific embodiment, when the software begins executing the Poisson surface reconstruction algorithm, it first automatically identifies which regions belong to anatomically important fine structures based on the geometric features of the point cloud data (such as high curvature regions, edge lines, etc.). For example, the software identifies gingival papillae by detecting the sharpness and conical features of local point cloud morphology; and identifies mucosal folds by detecting linear, regularly undulating fold morphology.

[0063] For regions identified as gingival papillae or mucosal folds, the software dynamically adjusts two core parameters of the Poisson reconstruction algorithm. It sets the "mesh sampling density" for that region to a higher value (e.g., setting the subdivision depth to 10 in the octree subdivision of Poisson reconstruction). This means the system uses more and smaller triangular meshes to construct the surface of this area, thus capturing more subtle morphological variations. Simultaneously, it sets the "smoothing intensity coefficient" to a lower value (e.g., setting the smoothing weight in the solver to 0.2). This reduces the smoothing effect on data points when fitting the surface, maximizing fidelity to the original position of the corrected point cloud and preserving its sharpness and detail.

[0064] Conversely, for areas identified as flat soft tissue, such as the buccal slope of the alveolar ridge or the hard palate region of the maxilla, the morphology is relatively simple and has less impact on implant design. Therefore, the software uses a different set of parameters: setting the "mesh sampling density" to a lower value (e.g., an octree subdivision depth of 7) to construct the surface with fewer, larger triangular meshes, which helps improve computational efficiency and reduce the file size of the final model. At the same time, setting the "smoothing intensity coefficient" to a higher value (e.g., a smoothing weight of 1.0) allows the algorithm to perform moderate smoothing to generate a visually smoother surface without too many micro-undulations.

[0065] By dynamically adjusting reconstruction parameters based on anatomical importance, the resulting 3D oral model exhibits extremely high detail fidelity in critical areas (such as gingival papillae), while maintaining overall model smoothness and data processing economy in non-critical areas, achieving the best balance between model accuracy and efficiency.

[0066] Ultimately, this high-precision 3D model will serve the final goal of clinical design: to develop an implant restoration plan that meets both aesthetic requirements and promotes long-term biological stability. The implant placement parameters include implant depth and angle, while the restoration morphology parameters include the restoration margin contour and crown contact point location. Furthermore, the planning of implant placement parameters and the design of restoration morphology parameters are constrained by the requirement to preserve the biological width of the oral soft tissue.

[0067] In a specific clinical design implementation, the physician operates on implantation design software loaded with this high-fidelity 3D model.

[0068] When planning implant placement parameters, the model accurately reflects the height of the gingival papilla and the soft tissue thickness of the alveolar ridge, allowing for more precise vertical positioning by the dentist. Based on the model data, the software automatically marks a virtual "biological width" area approximately 2 mm wide around the implant. When the dentist drags the virtual implant in the 3D view, if the implant platform is positioned too close to the alveolar ridge crest, encroaching on this "biological width" area, the software will immediately issue an alert, such as highlighting the area in red and displaying a warning: "Warning: Encroachment on biological width may lead to bone resorption and soft tissue recession." This forces the dentist to place the implant in a deeper, more securely osseointegrated location, such as placing the implant platform 3 mm below the future gingival margin, leaving sufficient space for soft tissue attachment.

[0069] This high-precision soft tissue model also plays a crucial role in designing the morphological parameters of restorations. When designing the emergence profile of the crown, dentists no longer rely on guesswork based on an overly smoothed, blurred gingival cuff, but can clearly see the actual, corrected gingival morphology. They can design a smooth, concave emergence profile that gently guides and supports the gingival tissue, especially the gingival papilla, creating a full, natural, and easy-to-clean shape. For example, when designing the contact point of the crown, it can be set at an appropriate height to ensure sufficient support for the underlying gingival papilla, preventing its recession and achieving optimal aesthetic results.

[0070] By using the core biological principle of "preserving biological breadth" as a hard constraint on software design, and combining it with the accurate data provided by high-fidelity models, this application directly translates the technological precision advantage into safety and predictability in clinical treatment, truly achieving the ultimate goal of technology serving medicine.

[0071] In some embodiments, this application also provides a 3D modeling and design system 100 for implants based on oral scan data, capable of performing the above-described process. For example... Figure 2 As shown in the illustration, this application also provides an exemplary structural diagram of a 3D implant modeling and design system based on oral cavity scan data. The system includes: The data acquisition module 10 is used to receive the raw point cloud data of the oral soft tissue region collected by the intraoral scanner, and to acquire device measurement deviation information containing geometric measurement offset that describes the intraoral scanner under specific lighting conditions, as well as bio-optical response information reflecting the optical response characteristics of the oral soft tissue. Feature analysis module 20 is used to perform geometric morphological feature analysis and optical reflection feature analysis on each local region in the original point cloud data to obtain feature analysis results; The geometric deformation recognition module 30 is used to compare the feature analysis results with the equipment measurement deviation information and the bio-optical response information, and to identify the systematic geometric deformation in the original point cloud data caused by the combined effect of scanner deviation and biological characteristics through preset judgment rules. The geometric correction module 40 is used to perform geometric correction according to the type of geometric deformation, using a corresponding differentiated correction strategy to obtain corrected point cloud data. The 3D model building module 50 is used to build a 3D oral cavity model by adjusting the surface reconstruction algorithm parameters based on the corrected point cloud data to preserve the fine anatomical structure of the oral soft tissue; and to plan the implantation parameters and design the prosthesis morphology parameters based on the 3D oral cavity model.

[0072] This system solidifies the aforementioned methodology into specific hardware or software functional modules, providing physical support for the technology's implementation. In a typical implementation, the system could be a set of dedicated software installed on a high-performance computer workstation, which connects to the intraoral scanner via a data cable.

[0073] At the software level, the data acquisition module can be an interface program responsible for communicating with the scanner driver. It not only receives the massive stream of raw point cloud data transmitted from the scanner, but also loads, at system startup, a device measurement deviation information file matching the serial number of the currently connected scanner from a specified path on the hard drive, as well as a general bio-optical response information database file.

[0074] The feature analysis module can be a multi-threaded computing engine. It receives point cloud data from the data acquisition module and performs a series of calculations, such as registration and time series analysis, according to the aforementioned multi-frame analysis method, to extract stable and reliable geometric and optical features.

[0075] The geometric deformation recognition module is the core decision-making unit of the system. It can be designed as a rule-based expert system or a trained machine learning model. It receives the output of the feature analysis module and performs complex comparisons and logical judgments with the two major information databases loaded by the data acquisition module, ultimately outputting a determination of whether systematic geometric deformation exists in each local area and the type of deformation.

[0076] The geometric correction module is an algorithm library. It contains various pre-built correction algorithms, such as non-uniform scaling algorithms and non-rigid deformation algorithms based on radial basis functions. This module calls the corresponding algorithms to accurately correct the specified point cloud region according to the instructions from the geometric deformation recognition module.

[0077] The 3D model building module is the generator of the final result. It receives the corrected point cloud data output from the geometry correction module and calls the Poisson surface reconstruction algorithm, which implements dynamic parameter adjustment, to generate the final high-fidelity 3D model. Simultaneously, this module also integrates implant design functionality, allowing physicians to interactively plan implants and prostheses on the generated model.

[0078] Through the collaborative work of its various modules, the system can automatically complete the entire process from data collection, analysis, identification, correction to modeling and design, transforming complex theoretical algorithms into efficient clinical tools that doctors can use.

[0079] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for three-dimensional modeling and design of implants based on oral cavity scan data, characterized in that, include: The system receives raw point cloud data of the oral soft tissue region acquired by an intraoral scanner, and obtains device measurement deviation information containing geometric measurement offset, which describes the intraoral scanner under specific lighting conditions, as well as bio-optical response information reflecting the optical response characteristics of the oral soft tissue. Geometric morphological feature analysis and optical reflection feature analysis are performed on each local region in the original point cloud data to obtain feature analysis results; The feature analysis results are compared with the device measurement deviation information and the bio-optical response information. By using preset judgment rules, the systematic geometric deformation in the original point cloud data caused by the combined effect of scanner deviation and biological characteristics is identified. Based on the type of geometric deformation, a corresponding differential correction strategy is adopted to perform geometric correction and obtain corrected point cloud data; Based on the corrected point cloud data, the surface reconstruction algorithm parameters are adjusted to preserve the fine anatomical structure of the oral soft tissue, and a three-dimensional oral model is constructed; and based on the three-dimensional oral model, implant placement parameters are planned and prosthesis morphology parameters are designed.

2. The implant three-dimensional modeling and design method according to claim 1, characterized in that, The steps of performing geometric morphological feature analysis and optical reflection feature analysis on each local region in the original point cloud data to obtain the feature analysis results include: During the oral cavity scan, the original point cloud data is acquired in multiple frames at continuous time points to obtain the geometric morphological features and optical reflection features of the local area at different time points. Identify the geometric features of stable reference regions in the original point cloud data; The geometric features and optical reflection features of the local area collected at each time point are registered with the geometric features of the stable reference area to eliminate the influence of overall displacement and obtain the registered local area geometric features and optical reflection features. Time series analysis is performed on the registered local region's geometric morphology and optical reflection features to identify their change trajectories at continuous time points; and based on the amplitude and frequency of the change trajectory, the feature differences caused by minute irregular movements are distinguished from actual physiological structural changes or scanner errors, thereby obtaining feature analysis results.

3. The implant three-dimensional modeling and design method according to claim 2, characterized in that, The step of comparing the feature analysis results with the device measurement deviation information and the bio-optical response information, and identifying the systematic geometric deformation in the original point cloud data caused by the combined effect of scanner deviation and biological characteristics through preset judgment rules, includes: The local geometric features and optical reflection features of the original point cloud data at multiple time points are acquired, and the geometric features of stable regions in the original point cloud data are identified as reference benchmarks. The local geometric features include surface curvature variation features and surface normal direction consistency features; the optical reflection features include light intensity variation features and reflectivity variation features. By comparing the local geometric features and optical reflection features at each time point with the device measurement deviation information, the bio-optical response information, and the reference benchmark, the type of geometric deformation that may exist at that time point is identified. The type of geometric deformation is a regular geometric distortion type produced by the combined effect of scanner deviation and biological characteristics, including at least one of stretching deformation and torsional deformation. By tracking the changing trends of geometric deformation patterns identified in the same local area at different time points and recognizing the regular evolution of the geometric deformation patterns at continuous time points, and by using preset judgment rules, systematic geometric deformations can be identified.

4. The implant three-dimensional modeling and design method according to claim 1, characterized in that, The step of performing geometric correction using a corresponding differential correction strategy based on the type of geometric deformation includes: A mapping relationship library between geometric deformation types and correction strategies is established. For tensile deformation, a non-uniform scaling correction algorithm is configured to restore the original geometric proportions. For torsional deformation, a local non-rigid deformation correction algorithm based on radial basis functions is configured to restore the natural anatomical shape. If the geometric deformation is determined to be a stretching deformation, the main stretching direction of the local area where the stretching deformation is located is identified, and a reverse non-uniform scaling correction algorithm is applied along this direction to correct the deformation, so that the coordinates of each point in the local area where the stretching deformation is located are adjusted according to the preset shrinkage ratio to restore the original geometric proportions. If the geometric deformation is determined to be a tortuous deformation, multiple control points in the local area where the tortuous deformation is located are selected, the target position of each control point is determined according to the geometric trend of the surrounding healthy tissue, and then the local non-rigid deformation correction algorithm based on radial basis function is used for correction.

5. The implant three-dimensional modeling and design method according to claim 1, characterized in that, The step of obtaining device measurement deviation information, which describes the geometric measurement offset generated by the intraoral scanner under specific lighting conditions, includes: When the reference light source illuminates multiple calibration points with known optical reflection characteristics set on the standard model, the reflection spectrum data of each calibration point is collected, and the reflection spectrum data is compared with the preset pure spectrum reference of the calibration point to identify the spectral drift of the reference light source. Based on the spectral drift, the driving parameters of the reference light source are adjusted to compensate for the spectral drift of the reference light source; When the calibration point is illuminated by the scanner's own light source, reflectance spectral data is collected. Differential analysis is performed on the spectral data collected under the illumination of the scanner's own light source and the spectral data collected under the illumination of the compensated reference light source to obtain the differential analysis results. Based on the differential analysis results, optical noise introduced by the decrease in the spectral purity of the scanner's own light source and the photosensitive reaction characteristics of the coating on the standard model body is removed; The optical signal after removing optical noise is compared with a preset digital reference model. Based on the comparison results, the pure geometric deviation data points are quantified and recorded as the measurement deviation information of the device.

6. The implant three-dimensional modeling and design method according to claim 1, characterized in that, The steps for obtaining bio-optical response information reflecting the optical response characteristics of oral soft tissue include: The original point cloud data is divided into regions to obtain multiple sub-regions; Perform local pre-scanning on each sub-region and collect reflectance spectral data of each sub-region under multiple specific wavelength illuminations; Based on the reflectance spectral data of each sub-region, the tissue type and physiological state of each sub-region are classified to obtain the classification results; Based on the classification results of each sub-region, a set of optical response parameters matching the sub-region is extracted from the bio-optical property database as the bio-optical response information.

7. The implant three-dimensional modeling and design method according to claim 1, characterized in that, The step of adjusting the surface reconstruction algorithm parameters based on the corrected point cloud data to preserve the fine anatomical structure of the oral soft tissue and constructing a three-dimensional model of the oral cavity includes: A 3D oral cavity model was reconstructed using the Poisson surface reconstruction algorithm based on the corrected point cloud data. During the reconstruction process, the reconstruction parameters were dynamically adjusted according to the anatomical importance of different regions in the corrected point cloud data. The anatomical importance was determined based on whether the region contained the fine anatomical structures required for implant design. Fine anatomical structures included gingival papillae and mucosal folds, while non-fine anatomical structures were flat soft tissue areas. For areas identified as fine anatomical structures such as gingival papillae or mucosal folds, a first grid sampling density and a first smoothness coefficient are set; For flat soft tissue regions, a second grid sampling density and a second smoothing intensity coefficient are set; The first grid sampling density is higher than the second grid sampling density; the first smoothing intensity coefficient is lower than the second smoothing intensity coefficient.

8. The implant three-dimensional modeling and design method according to claim 3, characterized in that, The preset judgment rule is as follows: when the point cloud data of a local area simultaneously meets the following conditions, the point distribution is sparse, the curvature changes are abnormal, the normal direction is inconsistent, the light intensity change pattern matches the abnormal scattering pattern of the optical deviation data points, and the regional optical characteristics are consistent with the bio-optical response characteristics of the patient's atrophied gingiva, the area is judged to be a systematic geometric deformation area.

9. The implant three-dimensional modeling and design method according to claim 1, characterized in that, The implantation parameters include implantation depth and implantation angle, and the prosthesis morphology parameters include prosthesis margin contour and crown contact point position. In the planning of implantation parameters and the design of prosthesis morphology parameters, the preservation of the biological width of oral soft tissue is taken as a constraint.

10. A three-dimensional modeling and design system for implants based on oral cavity scan data, characterized in that, The system includes: The data acquisition module is used to receive raw point cloud data of the oral soft tissue region collected by the intraoral scanner, and to acquire device measurement deviation information containing geometric measurement offset that describes the intraoral scanner under specific lighting conditions, as well as bio-optical response information reflecting the optical response characteristics of the oral soft tissue. The feature analysis module is used to perform geometric morphological feature analysis and optical reflection feature analysis on each local region in the original point cloud data to obtain feature analysis results; The geometric deformation recognition module is used to compare the feature analysis results with the device measurement deviation information and the bio-optical response information, and to identify the systematic geometric deformation in the original point cloud data caused by the combined effect of scanner deviation and biological characteristics through preset judgment rules. The geometric correction module is used to perform geometric correction based on the type of geometric deformation using corresponding differentiated correction strategies, and obtain corrected point cloud data. The 3D model construction module is used to construct a 3D oral cavity model by adjusting the surface reconstruction algorithm parameters based on the corrected point cloud data to preserve the fine anatomical structure of the oral soft tissue; and to plan implantation parameters and design prosthesis morphology parameters based on the 3D oral cavity model.

Images