A three-dimensional cross-comparison analysis method for the precision of digital surgical guide-guided dental implantation

Abstract

This invention provides a three-dimensional cross-comparison analysis method for the accuracy of digital surgical guide-guided dental implant surgery, relating to the field of oral medicine. It solves the problems of unintuitive accuracy evaluation, complex operation, and large subjective errors in digital surgical guide-guided dental implant surgery. The method includes: acquiring preoperative design position images and postoperative actual position images of the implant before and after the patient undergoes dental implant surgery; performing three-dimensional reconstruction on the preoperative design position images and postoperative actual position images to determine the three-dimensional positional relationship between the designed implant and the actual implant, and calculating the three-dimensional cross-comparison ratio (CCR) value; and analyzing and evaluating the accuracy of the patient's dental implant based on the calculated CCR value. The evaluation index of this invention is the three-dimensional cross-comparison ratio between the preoperative design position and the postoperative actual position of the implant, which allows for precise quantification of implant accuracy and facilitates objective quality control of postoperative results.

Description

Technical Field

[0001] This invention relates to the field of oral medicine and is applied to the process of evaluating the precision of dental implants, specifically to a three-dimensional cross-comparison analysis method for the precision of dental implants guided by digital surgical guides. Background Technology

[0002] In the field of modern dentistry, digitally guided dental implant surgery has become a widely used technique. Compared to traditional free-hand methods, this technique significantly improves the precision and predictability of the surgery. Traditional dental implant surgery relies primarily on the surgeon's experience and skill level to determine the implant's position, angle, and depth. Because of variations in each surgeon's proficiency, this often leads to a degree of uncertainty in the surgical outcome, affecting the final treatment result.

[0003] With the introduction of digital surgical guides, the ideal placement of implants can be precisely planned using detailed preoperative 3D imaging data (such as CT or CBCT scans). Surgical guides created based on this data provide clear guidance for drilling and implantation steps during the actual surgery, significantly reducing the impact of human factors on the outcome. This method not only improves the success rate of individual implants but also enhances the relative positional relationships between multiple implants, which is particularly beneficial for complex cases.

[0004] However, despite the significant improvement in the quality of dental implant surgery achieved by digital surgical guides, accurately assessing postoperative precision remains a challenge in clinical practice. Current methods for evaluating the effectiveness of digitally guided dental implant surgery primarily involve comparing key indicators such as shoulder deviation, root deviation, depth deviation, and angle deviation between the preoperative design and the actual postoperative results. However, while some existing software offers built-in measurement tools, their operation is relatively complex and not intuitive, making it difficult for non-professionals to quickly master the techniques. Furthermore, selecting appropriate cross-sections for analysis often relies on personal experience, a process inherently subjective and prone to deviations in results due to differences in operator skill.

[0005] Therefore, it is particularly necessary to develop a simpler, easier-to-use, and more objective method for evaluating postoperative accuracy. Such a new method needs to be digital and automated to reduce errors caused by human intervention and provide clear and concise evaluation results to help doctors and patients better understand the surgical outcome. Summary of the Invention

[0006] Based on the current state of the art, the purpose of this invention is to address the problems of unintuitive accuracy evaluation, complex operation, and large subjective errors in digital surgical guide-guided dental implant surgery. Therefore, a three-dimensional cross-comparison analysis method for the accuracy of digital surgical guide-guided dental implant surgery is proposed. The evaluation index of this invention is the three-dimensional cross-comparison ratio between the preoperative designed position and the actual postoperative position of the implant. This method allows for precise quantification of implant accuracy, facilitating objective quality control of postoperative results and promoting the progress and development of digital dentistry.

[0007] The present invention employs the following technical solutions to achieve its objective:

[0008] A three-dimensional crossover ratio analysis method for the accuracy of digital surgical guide-guided dental implantation includes: acquiring preoperative design position images and postoperative actual position images of the implants before and after the patient undergoes dental implant surgery; performing three-dimensional reconstruction on the preoperative design position images and postoperative actual position images to determine the three-dimensional positional relationship between the designed implants and the actual implants, and calculating the three-dimensional crossover ratio value between the designed implants and the actual implants; and analyzing and evaluating the accuracy of the patient's dental implantation based on the calculated three-dimensional crossover ratio value.

[0009] Specifically, the methods for obtaining preoperative design images and postoperative actual images are as follows: After the patient undergoes dental implant surgery, the postoperative CBCT image data of the patient is obtained. After processing the postoperative CBCT image data through dental digital design software, the postoperative implant and jawbone data of the patient are obtained. The corresponding implant design data of the patient is obtained. The implant design data, postoperative implant and jawbone data are imported into dental digital software for fitting processing to obtain the preoperative design image and postoperative actual image of the patient's implant.

[0010] Specifically, the fitted images are displayed as axial section images using dental digital software. Corresponding to the preoperative design position image and the postoperative actual position image, multiple images of the implant are extracted and saved as the basis for the three-dimensional reconstruction of the designed implant and the actual implant.

[0011] Furthermore, after obtaining the preoperative design location image and the postoperative actual location image of the patient's implant, the same image preprocessing process is performed, including: mask extraction, color curve extraction, concave surface processing, and filling.

[0012] Specifically, for all image files in the preoperative design location image and postoperative actual location image that have been acquired, each image file is cropped and scaled, and then the mask of the preset area is extracted by the file index. The mask is applied to the cropped and scaled image and saved to complete the mask extraction.

[0013] When extracting color curves, the color space of all image files is converted from RGB to HSV. HSV thresholds for multiple colors are set, and color masks are generated. Index values ​​are selected by file indexing, and convolution kernels of different sizes are selected based on the index values. Dilation and erosion operations are performed on the preset color masks to obtain binary images. After thinning the binary images to extract the skeleton, connected regions are marked for the remaining color regions, and the bounding box attributes of the corresponding regions of the implants are extracted. Finally, the multiple color masks are merged, and the regions corresponding to each color are retained, thus completing the extraction of color curves from the image.

[0014] For the binary mask corresponding to the binary image, the concave regions in the binary mask are processed by dilating the binary mask and detecting its contour to fill the external concave regions.

[0015] Furthermore, the same three-dimensional reconstruction process is performed on the preoperative planned position image and the postoperative actual position image, including: cyclic processing of the image dataset, image reading and matching, geometric transformation operation, and saving of reconstruction results; wherein, the geometric transformation operation is used to form a rotated image structure for these two types of images.

[0016] Specifically, the preoperative planned location images and postoperative actual location images are preprocessed to form two corresponding datasets. Each dataset includes multiple images that have been processed in a loop. A target array is set to store 3D data. Each slice in the 3D data corresponds to each processed image in the dataset. Images are loaded in a preset order, converted to grayscale mode, and then cropped to match the width of the target array.

[0017] During 3D reconstruction, the rotation method of the 3D data stored in the target array is determined by defining the idx value. When the index value is less than the preset value, the coordinates of each row of pixels in the corresponding image are transformed by sine and cosine rotation according to a specific angle to simulate the rotation in the third quadrant. After determining the rotation center and the half-width of the image, the rotation angle is calculated, and the x and y coordinates of the data are updated. When the index value is greater than or equal to the preset value, the rotation in the corresponding direction is performed directly according to the angle offset. The updated data is assigned to the target array according to the new coordinates to form the rotated image structure. The two reconstruction results, corresponding to the preoperative designed position image and the postoperative actual position image, are saved separately.

[0018] Furthermore, after the 3D reconstruction is completed, the 3D intersection ratio between the designed implant and the actual implant is calculated based on the 3D reconstruction results. At the same time, the 3D positional relationship between the designed implant and the actual implant is presented through rendering. For the calculated 3D intersection ratio, the larger the value, the higher the precision of the dental implant in the patient's mouth.

[0019] Specifically, when calculating the 3D intersection-union ratio, the 3D data corresponding to the designed implant and the actual implant in the 3D reconstruction results are first converted into binary masks. The intersection region and the union region of these two binary masks are calculated in turn to obtain the intersection prime number and the union prime number. The 3D intersection-union ratio is the result of dividing the intersection prime number by the union prime number.

[0020] For the 3D data of the designed implant and the actual implant, obtain the corresponding volume data files of the 3D reconstruction results; stitch the two volume data files along the width axis to form the stitched volume data, and add the two volume data files element by element to form the superimposed volume data; the stitched volume data and the superimposed volume data are reserved for use during rendering;

[0021] Finally, while presenting the 3D intersection-union ratio values ​​to the outside world, two rendering scenes are created to display the spliced ​​volume data and the overlay volume data respectively; the rendering effect of the rendering scenes is controlled by the transfer function, and the rendering scenes are presented to the outside world synchronously.

[0022] The present invention also provides a computer system, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the aforementioned three-dimensional cross-comparison analysis method for the accuracy of digital guide-guided dental implantation.

[0023] In summary, due to the adoption of this technical solution, the beneficial effects of this invention are as follows:

[0024] This invention can precisely quantify the difference between the preoperative designed location and the actual postoperative location of implants, providing clinicians with a more intuitive and easily understood evaluation standard. Specifically, this invention uses a three-dimensional intersection-union ratio (IUU) as the evaluation index. This not only simplifies the traditional process of manually selecting specific cross-sections for comparison but also avoids result bias caused by subjective judgment differences among different operators. This method makes the entire evaluation process more automated and standardized, improving work efficiency while ensuring the consistency and reliability of results.

[0025] The evaluation system resulting from the application of the method in this invention has broad application prospects. It can not only serve as an important tool for internal quality control in hospitals, helping medical institutions better monitor and manage surgical outcomes and promote the overall improvement of medical service levels, but also contribute to driving the entire industry towards a higher level of digital transformation. With the adoption of this evaluation standard, greater breakthroughs are expected in improving the success rate of dental implant surgery in the future. It is evident that this invention, through three-dimensional cross-analysis of the precision of dental implantation under digital surgical guide guidance, achieves a more accurate and objective quantitative evaluation of implant surgery results, laying the foundation for further exploration and development of more efficient and precise oral medical technologies. Attached Figure Description

[0026] Figure 1 This is a schematic diagram of the overall steps of the method of the present invention;

[0027] Figure 2 A schematic diagram of the patient's implant;

[0028] Figure 3 A schematic diagram of the patient's jawbone;

[0029] Figure 4 A schematic diagram showing the comparison between preoperative planned position images and postoperative actual position images;

[0030] Figure 5 To design a three-dimensional positional relationship diagram between the implant and the actual implant;

[0031] Figure 6 This is a schematic diagram of the numerical calculation results of the three-dimensional intersection-union ratio in the method of the present invention;

[0032] Figure 7 This is a schematic diagram of the image preprocessing process in the method of the present invention;

[0033] Figure 8 This is a schematic diagram of the image three-dimensional reconstruction process in the method of the present invention;

[0034] Figure 9 This is a schematic diagram illustrating the process of external rendering and 3D intersection-union ratio calculation in the method of this invention. Detailed Implementation

[0035] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The parts of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0036] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0037] Example 1

[0038] A three-dimensional cross-comparison analysis method for improving the accuracy of digital surgical guide-guided dental implantation. Figure 1 The overall flow of this method is shown and can be viewed simultaneously. The specific steps of this method are described below:

[0039] S1. After the patient undergoes dental implant surgery, postoperative image data is obtained, such as postoperative CBCT images. These images are then input into dental digital design software such as EXOCAD, the image threshold range is adjusted, and additional hard tissue is removed to obtain and save the patient's postoperative implant and jawbone data. (Implant details can be found in the documentation.) Figure 2 The diagram shows the jawbone. Figure 3 The illustration.

[0040] S2. Import the implant and jawbone data extracted from EXOCAD software, along with the corresponding patient's implant design data (in .3oxz files for 3Shape software), into digital dental software like 3Shape for fitting, to obtain preoperative design images and postoperative actual implant position images, such as... Figure 4 As shown.

[0041] S3. In the 3Shape software, the fitted image type is displayed as a mid-axis section image type; corresponding to the preoperative design position image and the postoperative actual position image, after extracting all 180 section images of the implant, save them to the specified path.

[0042] S4. Perform 3D reconstruction on the images saved in the specified path to determine the 3D positional relationship between the designed implant and the actual implant. (See attached image.) Figure 5 The three-dimensional intersection-to-exchange ratio of the designed implant and the actual implant was calculated and can be referred to [reference needed]. Figure 6 Based on the calculated three-dimensional crossover ratio, the precision of dental implantation in patients can be objectively and quantitatively analyzed and evaluated.

[0043] Example 2

[0044] Based on Example 1, this example provides a detailed description of the preprocessing workflow for the image stored in the specified path in step S4 of the method. (See also...) Figure 7 As illustrated, this embodiment uses a Python programming environment and related code to implement the relevant functions. The sub-steps are as follows:

[0045] S4-1-1, Image Preprocessing:

[0046] This function processes all image files in the specified image directory, reading each image sequentially and then cropping and scaling it. It selects a tag value using the file index and passes it to the `extract_colored_curves` function to extract a mask for a specific region. The mask is then applied to the cropped and scaled image and saved.

[0047] S4-1-2. Extract the color curve:

[0048] This is the core of the extract_colored_curves function code. In this example, it mainly extracts green, brown, and dark blue curves from the image to achieve the corresponding functional effects.

[0049] First, the image is converted from RGB to HSV color space to more easily define the color range. Then, HSV thresholds are set for green, sepia, and dark blue, generating color masks. Different sized convolutional kernels are selected based on tag values ​​to perform dilation and erosion operations on the sepia mask, removing small noise areas and filling in small holes. Next, skeletonization is used to thin the binary image to extract the skeleton. The mask_green part targets the green regions, performing connected region labeling and extracting the bounding box attributes of specific regions. Finally, the masks are merged to preserve each color region.

[0050] S4-1-3, Processing concave surfaces:

[0051] To process the recessed areas in the binary mask, the outer recesses are filled by expanding the mask and detecting the contour.

[0052] S4-1-4, Filling:

[0053] This step is used in practice to fill the blank areas between labels in a given list.

[0054] Example 3

[0055] Based on any of the above embodiments, this embodiment provides a detailed description of the 3D reconstruction process of the image stored in the specified path in step S4 of the method. (See also...) Figure 8 As illustrated, this embodiment uses a Python programming environment and related code to implement the relevant functions. The sub-steps are as follows:

[0056] S4-2-1, Processing the dataset in a loop:

[0057] After preprocessing the preoperative planned location images and the postoperative actual location images, two corresponding datasets were formed, denoted as base_path_1 and base_path_2. Each dataset includes multiple images that have undergone iterative processing. A target array data_total was set to store the 3D data, with a shape of (667, 681, 681), and each slice corresponds to each processed image in the dataset.

[0058] S4-2-2, Image Reading and Processing:

[0059] Images are loaded in filename order and converted to grayscale mode (cv2.imread(...,0)). The images are then cropped to a width of 681 (data[:,0:681]), ensuring that the width matches the target array data_total.

[0060] S4-2-3, Geometric Transformation Processing:

[0061] The processing logic is divided into two parts, with the rotation method determined by defining the idx value. The idx value is used to track the index of the current image file, i.e., the index value when traversing each image file. In this embodiment, the idx value is used to control the rotation angle of the image. For the 180 sliced ​​images captured in step S3, the first 90 images and the last 90 images are rotated differently. Furthermore, after idx is converted into an angle (using the math.radians(idx) function or the math.radians(idx-90) function), it is used to calculate the subsequent sine and cosine values ​​to achieve the image rotation, as follows:

[0062] When the idx value is less than 90, the coordinates of each row of pixels in the corresponding image are transformed by sine and cosine rotation according to the angle to simulate rotation in the third quadrant. After determining the rotation center e and the image half-width q, the rotation angle is calculated using sin_1_degree and cos_1_degree, and the x and y coordinates of the data are updated.

[0063] When the idx value is greater than or equal to 90, the rotation in the corresponding direction is performed directly based on the angle offset, starting from idx-90. The updated data is then assigned to the target array according to the new coordinates, forming the rotated image structure.

[0064] S4-2-4. Save the results:

[0065] When idx == 0, the target array data_total is saved to the save_path_1 dataset; otherwise, it is saved to the save_path_2 dataset, generating two .npy files.

[0066] In this embodiment, the two reconstruction results corresponding to the preoperative planned position image and the postoperative actual position image are saved separately through the above steps.

[0067] Example 3

[0068] Based on any of the above embodiments, this embodiment provides a detailed description of the three-dimensional intersection-union ratio numerical calculation and the three-dimensional positional relationship diagram in step S4 of the method. (See also...) Figure 9 As illustrated, this embodiment uses a Python programming environment and related code to implement the relevant functions. The sub-steps are as follows:

[0069] S4-3-1, Calculation of IoU coefficient:

[0070] In the Python programming environment, the `iou_coefficient` function is used to calculate the 3D intersection-over-union ratio (IoU coefficient) between two volumes of data. First, the 3D data corresponding to the designed implant and the actual implant in the 3D reconstruction results are converted into binary masks, which are boolean arrays of type `uint8`.

[0071] The np.logical_and function is used to calculate the intersection region of the two masks; the np.logical_or function is used to calculate the union region of the two masks; after obtaining the intersection and union prime numbers, the IoU coefficient is the result of dividing the intersection prime number by the union prime number.

[0072] S4-3-2, Loading and Processing 3D Data:

[0073] Load two volume data files; this example uses total1_new.npy and total2_new.npy.

[0074] Two volume data files are concatenated along the width axis (horizontally) to form a concatenated volume data `data_cat`; the two volume data files are added element-wise to form an overlay volume data `data_total_3`, used to display their overlapping area. The concatenated and overlay volume data are reserved for use during rendering.

[0075] S4-3-3, Display the three-dimensional intersection-over-union (IoU) results:

[0076] After calling the `iou_coefficient` function to calculate the IoU coefficients between the two masks, the results are printed out. See [link / reference]. Figure 6 .

[0077] S4-3-4, Mayavi volume rendering:

[0078] Use the Mayavi library to create two rendering scenes: fig1 to display the stitched volume data data_cat, and fig2 to display the overlapping volume data data_total_3; use vol1 and vol2 to represent the volume data of these two scenes, respectively.

[0079] The opacity transfer function (_otf) and color transfer function (_ctf) are used to control the rendering effect, grading the volume data from black (transparent) to white (fully visible). Finally, the `mlab.show()` function is used to display the Mayavi scene window, synchronously presenting the rendered scene to the outside world. (See also...) Figure 5 .

Claims

1. A three-dimensional cross-reference analysis method for improving the accuracy of digital surgical guide-guided dental implantation, characterized in that: Before and after the patient undergoes dental implant surgery, images of the preoperative planned position and the actual postoperative position of the implant are acquired. After three-dimensional reconstruction of the preoperative planned position and the actual postoperative position images, the three-dimensional positional relationship between the designed implant and the actual implant is determined, and the three-dimensional intersection-union ratio of the designed implant and the actual implant is calculated. Based on the calculated three-dimensional intersection-union ratio, the accuracy of the patient's dental implant is analyzed and evaluated. The same 3D reconstruction process is performed on both the preoperative planned position image and the postoperative actual position image, including: cyclic processing of the image dataset, image reading and matching, geometric transformation operation, and saving of reconstruction results; among which, the geometric transformation operation is used to form a rotated image structure for these two types of images. After preprocessing the preoperative planned location images and the postoperative actual location images, two corresponding datasets are formed. Each dataset includes multiple images that have been processed in a loop. A target array is set to store 3D data. Each slice in the 3D data corresponds to each processed image in the dataset. Images are loaded in a preset order, converted to grayscale mode, and then cropped to match the width of the target array. During 3D reconstruction, the rotation method of the 3D data stored in the target array is determined by defining the idx value. When the index value is less than the preset value, the coordinates of each row of pixels in the corresponding image are rotated by sine and cosine according to a specific angle to simulate the rotation in the third quadrant. After determining the rotation center and the half-width of the image, the rotation angle is calculated, and the x and y coordinates of the data are updated. When the index value is greater than or equal to the preset value, the rotation in the corresponding direction is performed directly according to the angle offset. The updated data is assigned to the target array according to the new coordinates to form the rotated image structure. The two reconstruction results, corresponding to the preoperative designed position image and the postoperative actual position image, are saved separately. After the 3D reconstruction is completed, the 3D intersection-union ratio (IUU) of the designed implant and the actual implant is calculated based on the 3D reconstruction results. When calculating the 3D IUU, the 3D data corresponding to the designed implant and the actual implant in the 3D reconstruction results are first converted into binary masks. The intersection region and the union region of these two binary masks are calculated in turn to obtain the intersection prime number and the union prime number. The 3D IUU is the result of dividing the intersection prime number by the union prime number. For the 3D data of the designed implant and the actual implant, obtain the corresponding volume data files of the 3D reconstruction results; stitch the two volume data files along the width axis to form the stitched volume data, and add the two volume data files element by element to form the superimposed volume data; the stitched volume data and the superimposed volume data are reserved for use during rendering; Finally, while presenting the 3D intersection-union ratio values ​​to the outside world, two rendering scenes are created to display the spliced ​​volume data and the overlay volume data respectively; the rendering effect of the rendering scenes is controlled by the transfer function, and the rendering scenes are presented to the outside world synchronously.

2. The three-dimensional intersection-union analysis method according to claim 1, characterized in that, The methods for obtaining preoperative design images and postoperative actual images are as follows: After the patient undergoes dental implant surgery, the postoperative CBCT image data of the patient is obtained. After processing the postoperative CBCT image data through dental digital design software, the postoperative implant and jawbone data of the patient are obtained. The corresponding implant design data of the patient is obtained. The implant design data, postoperative implant and jawbone data are imported into dental digital software for fitting processing to obtain the preoperative design image and postoperative actual image of the patient's implant.

3. The three-dimensional intersection-union analysis method according to claim 2, characterized in that: The fitted images are displayed as axial section images using dental digital software. Multiple images of the implant are extracted from the preoperative design image and the postoperative actual position image and saved as the basis for the three-dimensional reconstruction of the designed implant and the actual implant.

4. The three-dimensional intersection-union analysis method according to claim 1, characterized in that: After obtaining the preoperative design location image and the postoperative actual location image of the patient's implant, the same image preprocessing process is performed on each, including: mask extraction, color curve extraction, concave surface processing, and filling.

5. The three-dimensional intersection-union analysis method according to claim 4, characterized in that: For all image files in the preoperative design location image and postoperative actual location image that have been acquired, each image file is cropped and scaled, and then the mask of the preset region is extracted by the file index. The mask is applied to the cropped and scaled image and saved to complete the mask extraction. When extracting color curves, the color space of all image files is converted from RGB to HSV. HSV thresholds for multiple colors are set, and color masks are generated. Index values ​​are selected by file indexing, and convolution kernels of different sizes are selected based on the index values. Dilation and erosion operations are performed on the preset color masks to obtain binary images. After thinning the binary images to extract the skeleton, connected regions are marked for the remaining color regions, and the bounding box attributes of the corresponding regions of the implants are extracted. Finally, the multiple color masks are merged, and the regions corresponding to each color are retained, thus completing the extraction of color curves from the image. For the binary mask corresponding to the binary image, the concave regions in the binary mask are processed by dilating the binary mask and detecting its contour to fill the external concave regions.

6. The three-dimensional intersection-union analysis method according to claim 1, characterized in that: The three-dimensional positional relationship between the designed implant and the actual implant is presented through rendering. For the calculated three-dimensional intersection-to-union ratio, the larger the value, the higher the precision of the dental implant in the patient's mouth.

7. A computer system comprising a memory, a processor, and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the three-dimensional cross-comparison analysis method for the accuracy of digital guide-guided dental implantation as described in any one of claims 1 to 6.

Images