An image data processing method and an intelligent laser device

Abstract

The present invention relates to the field of laser technology, and in particular to an image processing method for image data and an intelligent laser device. The method includes: acquiring multi-modal image data of a first position of a user; reconstructing the multi-modal image data of the first position into a 3D image, and automatically identifying and acquiring basic information of a specified area according to the 3D image; inputting the basic information of the specified area into a pre-constructed scoring model to obtain a status score of the specified area of the user, and acquiring relevant parameters according to the status score of the specified area of the user and a pre-set processing parameter setting rule; calculating and generating an accurate path of a robotic arm according to the 3D image and the position of the specified area, the robotic arm reaching the specified area along the accurate path, and operating a laser processing head connected to the robotic arm according to the relevant parameters. The present invention uses the image processing method for image data and the intelligent laser device to solve the problems of insufficient accuracy and consistency in the operation of the laser device, and realizes accuracy, consistency, and minimally invasive nature.

Description

Technical Field

[0001] The present invention relates to the field of laser technology, and in particular to an image processing method for image data and an intelligent laser device. Background Art

[0002] Nowadays, laser technology has been widely used in the medical field. Soon after the first ruby ​​laser came out, it was quickly used in ophthalmology to treat fundus diseases. Soon after, the birth of carbon dioxide laser was also quickly introduced into surgical operations for precise surgical cutting. With the continuous advancement of optical transmission technology and the advent of ND:YAG laser, it was quickly used in the endoscopic treatment of gastrointestinal diseases. With the continuous improvement of solid laser technology based on the yttrium aluminum garnet series, such as thulium Tm, holmium Ho, erbium Er, neodymium Nd, etc., and the emergence of high-power semiconductor laser technology, laser medical equipment has gradually achieved miniaturization and intelligence. These advances have brought a new climax to the clinical application of lasers, and laser devices have been widely used in clinical practice.

[0003] However, existing oral and skin laser devices rely entirely on manual operation, which is affected by the operating habits of different users and cannot achieve accuracy and consistency in operation. They lack effectiveness and safety. Existing oral and skin laser devices lack digital standard operating specifications, and there are large differences between the plans of different operators, making it difficult to ensure accuracy and consistency. At the same time, existing laser devices cannot achieve digital path planning, and manual operation is difficult to achieve accuracy.

[0004] Therefore, there is an urgent need for an image processing method for image data and an intelligent laser device. Summary of the invention

[0005] 1. Technical issues to be resolved

[0006] In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides an image processing method and an intelligent laser device for image data, which solve the technical problems in the prior art that the accuracy and consistency of laser operation are low and it is difficult to achieve precise operation.

[0007] (II) Technical solution

[0008] In order to achieve the above object, the main technical solutions adopted by the present invention include:

[0009] In a first aspect, the present invention provides an image processing method for image data, comprising:

[0010] S100, obtaining multimodal image data of a user at a first position; the first position is inside the oral cavity or the skin to be measured;

[0011] S200. Reconstruct the multimodal image data at the user's first position into a 3D image, and automatically identify and obtain the basic information of the specified area based on the 3D image;

[0012] The basic information of the specified area includes: the position of the specified area, the size of the specified area, the depth of the specified area, and the irregularity of the specified area;

[0013] S300. Input the size of the specified area, the depth of the specified area, and the irregularity of the specified area in the basic information of the specified area into the pre-constructed scoring model to obtain the specified area status score of the user, and obtain the relevant parameters according to the specified area status score of the user and the pre-set processing parameter setting rules;

[0014] S400. Calculate and generate the precise path of the robotic arm based on the 3D image and the position of the specified area. The robotic arm reaches the specified area along the precise path, and operates the laser processing head connected to the robotic arm according to the relevant parameters.

[0015] Optionally, in S200, automatically identifying and obtaining the basic information of the specified area based on the 3D image includes:

[0016] S210. Train the specified area basic information acquisition model using the training dataset to obtain the trained specified area basic information acquisition model;

[0017] S220. Input the 3D image into the trained specified area basic information acquisition model to obtain the basic information of the specified area;

[0018] The specified area basic information acquisition model is a convolutional neural network model;

[0019] The training dataset is the past 3D image data and the basic information of the corresponding specified area.

[0020] Optionally, S210 specifically includes:

[0021] Input the past 3D image data and the basic information of the corresponding specified area into the specified area basic information acquisition model, and use the cross-entropy function and the Adam optimizer with a learning rate of 0.001 until the cross-entropy regression loss function converges to obtain the trained specified area basic information acquisition model.

[0022] Optionally, S300 includes:

[0023] S310. Input the size of the specified area, the depth of the specified area, and the irregularity of the specified area in the basic information of the specified area into the pre-constructed scoring model to obtain the first specified area status score of the user's specified area;

[0024] S320. Normalize the first score of the specified area status to obtain the score of the specified area status;

[0025] S330. Obtain relevant parameters according to the score of the specified area status and the pre-set processing parameter setting rules.

[0026] Optionally, S310 includes:

[0027] Input the specified area size, specified area depth, and specified area irregularity into the following formula to obtain the first score of the user-specified area status:

[0028] ;

[0029] where ω s is the influence weight of the specified area size, ω D is the influence weight of the specified area depth, ω F represents the influence weight of the specified area shape, S is the specified area size, and S max is the maximum reference value set for the specified area size, k s is the non-linear strength coefficient, D is the specified area depth, and D max is the set maximum reference depth, β D is the steepness of the saturation curve, F is the specified area irregularity, and Score is the first score of the specified area status.

[0030] Optionally, S320 includes:

[0031] Input the first score of the specified area status into the following formula for normalization to obtain the score of the specified area status:

[0032] ;

[0033] where Final Score is the score of the specified area status, ω s is the influence weight of the specified area size, ω D is the influence weight of the specified area depth, ω F represents the influence weight of the specified area shape, k s is the non-linear strength coefficient, D max is the set maximum reference depth, F max is the set maximum irregularity.

[0034] Optionally, S330 includes:

[0035] Input the score of the specified area status into the following formula to obtain relevant parameters:

[0036] ;

[0037] Among them, T is a relevant parameter, and T min is the minimum value of the relevant parameter, and T max is the maximum value of the relevant parameter, and FinalScore is the status score of the specified area;

[0038] The relevant parameters include laser power, laser operation duration, and laser irradiation area.

[0039] Optionally, in the S330,

[0040] The minimum value of the laser power is 0.5W, and the maximum value is 36W;

[0041] The minimum value of the laser operation duration is 1s, and the maximum value is 30 min;

[0042] The minimum value of the laser irradiation area is 0.5 mm 2 , and the maximum value is 49 cm 2 .

[0043] Optionally, the S400 specifically includes:

[0044] S410. Based on the specified area position and 3D image, use the cubic spline interpolation algorithm to generate the continuous path of the robotic arm in three-dimensional space;

[0045] S420. According to the shortest path algorithm and the continuous path of the robotic arm in three-dimensional space, obtain the precise path of the robotic arm operation. The robotic arm moves to the specified area along the precise path, and operates the laser processing head connected to the robotic arm according to the relevant parameters.

[0046] In a second aspect, an embodiment of the present invention provides an intelligent laser device, and the intelligent laser device is used to execute an image processing method for image data as described in any one of the first aspects. The intelligent laser device includes:

[0047] A robotic arm, a laser processing head, a sensor assembly, a laser generator, a central processing assembly, and a display screen;

[0048] The robotic arm, the laser processing head, the sensor assembly, the laser generator, the central processing assembly, and the display screen are respectively electrically connected to the central processing assembly;

[0049] The robotic arm and the laser processing head are connected by a rotating joint with a micro motor.

[0050] (3) Beneficial effects

[0051] The beneficial effects of the present invention are as follows: For an image data processing method and an intelligent laser device of the present invention, by using 3D images to obtain basic information of a specified area and scoring the user's situation through a scoring model, and obtaining relevant parameters according to the score. Compared with the prior art, it can achieve the accuracy and consistency of laser operations, realize the personalization of operation plans, increase the suitability of operation plans, and improve the safety and effectiveness of laser operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 It is a schematic flowchart of an image data processing method according to Embodiment 1 of the present invention;

[0053] Figure 2 It is a schematic flowchart of an image data processing method according to Embodiment 2 of the present invention;

[0054] Figure 3 It is a schematic diagram of a simple structure of an intelligent laser device according to Embodiment 5 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0055] In order to better explain the present invention for easy understanding, the present invention will be described in detail below with reference to the drawings and through specific embodiments.

[0056] An image data processing method and an intelligent laser device proposed in an embodiment of the present invention achieve the accuracy and consistency of the operation of the laser device, and solve the problems of insufficient accuracy, precision, and consistency of existing devices. In this application, the user's situation is scored through a scoring model, and relevant parameters are obtained according to the score, realizing the accuracy and consistency of laser operation, the personalization of operation plans, increasing the suitability of operation plans, and improving the effectiveness and safety of laser operation.

[0057] In order to better understand the above technical solution, the exemplary embodiments of the present invention will be described in more detail below with reference to the drawings. Although the exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided so that the present invention can be understood more clearly and thoroughly, and the scope of the present invention can be fully conveyed to those skilled in the art.

[0058] Embodiment 1

[0059] See Figure 1 , an image data processing method of this embodiment includes:

[0060] Step S100, obtaining multi-modal image data of the first position of the user; the first position is inside the oral cavity or the skin to be measured;

[0061] Step S200: Reconstruct the multi-modal image data at the first user position into a 3D image, and automatically identify and obtain the basic information of the specified area based on the 3D image;

[0062] The basic information of the specified area includes: the position of the specified area, the size of the specified area, the depth of the specified area, and the irregularity of the specified area;

[0063] Step S300: Input the size of the specified area, the depth of the specified area, and the irregularity of the specified area in the basic information of the specified area into a pre-constructed scoring model to obtain the specified area status score of the user, and obtain relevant parameters according to the specified area status score of the user and the pre-set processing parameter setting rules;

[0064] Step S400: Calculate and generate the precise path of the robotic arm according to the 3D image and the position of the specified area. The robotic arm reaches the specified area along the precise path, and operates the laser processing head connected to the robotic arm according to the relevant parameters.

[0065] An image data processing method in this embodiment realizes the accuracy and consistency of the processing of multi-modal image data of the oral cavity and skin, realizes the personalization of the operation plan, increases the suitability of the operation plan, and improves the effectiveness and safety of laser operation.

[0066] Embodiment 2

[0067] See Figure 2 , an image data processing method in this embodiment includes:

[0068] Step A100: Obtain the multi-modal image data at the first user position; the first position is inside the oral cavity or the skin to be tested;

[0069] Step A200: Reconstruct the multi-modal image data at the first user position into a 3D image, and automatically identify and obtain the basic information of the specified area based on the 3D image;

[0070] The basic information of the specified area includes: the position of the specified area, the size of the specified area, the depth of the specified area, and the irregularity of the specified area;

[0071] Step A300: Input the size of the specified area, the depth of the specified area, and the irregularity of the specified area in the basic information of the specified area into a pre-constructed scoring model to obtain the specified area status score of the user, and obtain relevant parameters according to the specified area status score of the user and the pre-set processing parameter setting rules;

[0072] Step A400: Calculate and generate the precise path of the robotic arm based on the 3D image and the position of the specified area. The robotic arm reaches the specified area along the precise path and operates the laser processing head connected to the robotic arm according to relevant parameters.

[0073] Specifically, in A200, automatically identifying and obtaining the basic information of the specified area based on the 3D image includes:

[0074] A210: Train the basic information acquisition model of the specified area using the training data set to obtain the trained basic information acquisition model of the specified area;

[0075] A220: Input the 3D image data into the trained basic information acquisition model of the specified area to obtain the basic information of the specified area;

[0076] The basic information acquisition model of the specified area is a convolutional neural network model;

[0077] The training data set is the past 3D image data and the basic information of the corresponding specified area.

[0078] A210 specifically includes:

[0079] Input the past 3D image data and the basic information of the corresponding specified area into the basic information acquisition model of the specified area, and use the cross-entropy function and the Adam optimizer with a learning rate of 0.001 until the cross-entropy regression loss function converges to obtain the trained basic information acquisition model of the specified area.

[0080] Meanwhile, introduce GS preprocessing during the training process to enhance the model's ability to recognize the spatial details of irregular boundaries;

[0081] Furthermore, the steps to construct and train the basic information acquisition model of the specified area are as follows:

[0082] Data preprocessing and augmentation: First, collect past 3D images of the oral cavity and skin. These data come from multiple sources to ensure diversity. Subsequently, preprocess the data, including denoising, normalization, data augmentation (such as rotation, scaling, flipping, etc.) and GS preprocessing, to enhance the generalization ability of the model;

[0083] Among them, GS preprocessing is to convert the 3D image into a sparse Gaussian point cloud representation to reduce noise interference and enhance edge details;

[0084] The basic information acquisition model of the specified area in this embodiment combines GS preprocessing on the basis of the convolutional neural network, which can make up for the shortcoming of the convolutional neural network in grasping the global shape, effectively enhance the recognition of the specified area, improve boundary characterization and efficient data utilization.

[0085] Label Refinement and Quality Control: Each 3D image is equipped with detailed basic information labels for the specified regions. These labels are manually marked by experienced doctors and undergo multiple levels of review to ensure their accuracy;

[0086] Model Architecture: A 3D Convolutional Neural Network (3D CNN) is used as the model for obtaining basic information about the specified regions. Additionally, an attention mechanism is further introduced to enable the model to focus on the key regions in the image, namely the specified regions, thereby improving the detection accuracy.

[0087] Training Optimization: The cross-entropy function and the Adam optimizer are used, with a learning rate of 0.001 until the cross-entropy regression loss function converges, to obtain a trained model for obtaining basic information about the specified regions. At the same time, an early stopping mechanism is introduced to avoid overfitting, and data augmentation techniques are used to enhance the robustness of the model.

[0088] The learning rate is also dynamically adjusted during training. The learning rate is:

[0089] ;

[0090] where η t represents the learning rate at the t-th iteration, η0 is the initial learning rate, set to 0.001, is the term that decays with the iteration t. Here, α is the decay coefficient, and as the training progresses, this term gradually decreases, thus causing the learning rate to gradually decrease; is the term that dynamically adjusts the learning rate according to the change rate of the loss function and the gradient norm, represents the change in the loss function at the t-th iteration, is the L2 norm of the gradient at the t-th iteration, and β is the adjustment coefficient.

[0091] Furthermore, to enhance the interpretability of the model, the model for obtaining basic information about the specified regions in this embodiment also includes a set of visualization tools that can overlay the basic information of the specified regions on the original 3D image, thereby visually observing the position and characteristics of the specified regions.

[0092] In this embodiment, A300 specifically includes:

[0093] A310. Input the size, depth, and irregularity of the specified region in the basic information of the specified region into a pre-constructed scoring model to obtain the first score of the specified region status of the user-specified region;

[0094] Specifically:

[0095] Input the size, depth, and irregularity of the specified region into the following formula to obtain the first score of the specified region status of the user-specified region:

[0096] ;

[0097] Among them, ω s is the influence weight of the specified area size, ω D is the influence weight of the specified area depth, ω F represents the influence weight of the specified area shape, S is the specified area size, S max is the maximum reference value set for the specified area size, k s is the non-linear strength coefficient, D is the specified area depth, D max is the maximum reference depth set, β D is the steepness of the saturation curve, F is the specified area irregularity, and Score is the first score of the specified area state.

[0098] Among them, the range of the specified area irregularity is 0 - 1, where 0 represents a completely regular shape and 1 represents an extremely irregular shape.

[0099] A320. Normalize the first score of the specified area state to obtain the specified area state score;

[0100] Input the first score of the specified area state into the following formula for normalization to obtain the specified area state score:

[0101] ;

[0102] Among them, Final Score is the specified area state score, ω s is the influence weight of the specified area size, ω D is the influence weight of the specified area depth, ω F represents the influence weight of the specified area shape, k s is the non-linear strength coefficient, D max is the maximum reference depth set, F max is the maximum irregularity set.

[0103] A330. Obtain relevant parameters according to the specified area state score and the pre-set processing parameter setting rules.

[0104] Input the specified area state score into the following formula to obtain relevant parameters:

[0105] ;

[0106] Among them, T is the relevant parameter, T min is the minimum value of the relevant parameter, T maxis the maximum value of relevant parameters, and FinalScore is the status score of the specified area. The relevant parameters include laser power, laser operation duration, and laser irradiation area;

[0107] The minimum value of the laser power is 0.5W, and the maximum value is 36W;

[0108] The minimum value of the laser operation duration is 1s, and the maximum value is 30min;

[0109] The minimum value of the laser irradiation area is 0.5mm 2 , and the maximum value is 49cm 2 .

[0110] In this embodiment, step S400 specifically includes:

[0111] A410. Based on the specified area position and 3D image, use the cubic spline interpolation algorithm to generate a continuous path of the robotic arm in three-dimensional space;

[0112] A420. According to the shortest path algorithm and the continuous path of the robotic arm in three-dimensional space, obtain the accurate path of the robotic arm operation. The robotic arm moves to the specified area along the accurate path, and operates the laser processing head connected to the robotic arm according to the relevant parameters.

[0113] In the specific implementation process, in order to ensure that the robotic arm can safely and efficiently reach the specified area position and operate the laser processing head on the robotic arm, it is first necessary to plan the operation path of the robotic arm based on the specific position of the specified area and the 3D image. In this process, the cubic spline interpolation algorithm can be used to create a smooth and continuous path. The specific steps are as follows:

[0114] Convert the specified area position into the reference coordinate system of the robotic arm operation;

[0115] Preprocess the original 3D image, including denoising, segmentation, feature extraction, etc., to clearly identify the specified area and the important structures around it;

[0116] In the 3D image, select a series of key points according to clinical requirements. These key points include the starting point, the ending point, and the intermediate control points set to avoid important organs or structures. These points are used as the nodes of the cubic spline interpolation; at the same time, ensure that all selected key points are in the same coordinate system, usually the base coordinate system of the robotic arm. If necessary, apply the necessary coordinate transformation matrix to adjust the position of the points;

[0117] For each pair of adjacent key points, define a cubic polynomial function. The form of the function is:

[0118] ;

[0119] Among them, t is a parameterized time variable, and i represents the i-th curve segment;

[0120] Set appropriate boundary conditions, such as the velocity or acceleration at the starting point and the ending point being zero (natural spline), or specifying the first derivative or second derivative at certain points (clamping conditions). This helps to ensure the smoothness and coherence of the path;

[0121] Using the known positions of key points and boundary conditions, establish a system of linear equations and solve for the coefficients a i 、b i 、c i 、d i of each polynomial, that is, obtain the continuous path of the robotic arm in three-dimensional space.

[0122] Furthermore, after obtaining the continuous path of the robotic arm in three-dimensional space, the path needs to be further optimized, that is, according to the shortest path algorithm and the continuous path of the robotic arm in three-dimensional space, obtain the precise path of the robotic arm's operation. Specifically, use the Dijkstra algorithm or other shortest path algorithms suitable for three-dimensional environments to find the optimal path between the starting point and the target point.

[0123] In addition, in order to make the robotic arm operate smoothly, when obtaining the precise path of the robotic arm's operation, the Bezier curve is also used to smooth the path, that is, connect the discrete path points into a continuous trajectory.

[0124] Among them, the Bezier curve is a parametric curve that defines a shape through control points. Before using the Bezier curve to smooth the path, it is necessary to first obtain the path nodes of the robotic arm's operation. This path is usually composed of a series of discrete nodes, namely:

[0125] ;

[0126] In this embodiment, a cubic Bezier curve is used to connect adjacent path nodes. Assuming that the connecting nodes are P i and P i+1 , it is determined according to the following steps:

[0127] Calculate the tangent direction at each node, and the tangent direction can be approximated by the vector between adjacent nodes. For example, for the node P i , its tangent direction vector can be approximated as: , when i = 0, ;

[0128] For the cubic Bezier curve connecting the nodes P i and P i+1 , if the four control points are P i 、C i，1, C i，2 , P i+1 , where C i，1 , C i，2 Determined according to the above tangent direction and proportional coefficient, assuming the proportional coefficient is k (usually between 0.2-0.5), then:

[0129] ;

[0130] ;

[0131] For each segment of P i , C i，1 , C i，2 , P i+1 A cubic Bezier curve can be generated by uniformly sampling the parameter t∈[0,1]. For example, divide t from 0 to 1 into N equally spaced points t j =j / N, j=0,1,2,…,N, then the points on the curve can be calculated by the cubic Bezier curve formula:

[0132] ;

[0133] Q j is a point on the curve, t j To divide t from 0 to 1 into N equally spaced points, P i , C i，1 , C i，2 , P i+1 There are four control points.

[0134] By connecting the points generated by the Bezier curves of all segments, we get a smooth path.

[0135] At this stage, a PID controller (proportional-integral-differential controller) is also used to accurately monitor and dynamically adjust the operating state of the robot arm. The PID controller continuously receives data from sensors, such as the position, speed, and acceleration of the robot arm, and adjusts the output signal accordingly to correct the behavior that deviates from the set path. At the same time, the PID controller can also be used to adjust the relevant parameters of the laser processing head during processing to adapt to the individual differences of different users and real-time changes.

[0136] In summary, steps S410 and S420 can ensure that the robotic arm reaches the designated area position accurately according to the predetermined plan, while ensuring the safety and reliability of the operation.

[0137] Through the above method, a smooth, safe and efficient path from the starting position to the target designated area position can be obtained for the robotic arm to perform the laser operation task.

[0138] A method for image processing of image data in this embodiment ensures the accuracy and consistency of image processing of multi-modal image data of the oral cavity and skin, realizes the personalization and suitability of the operation plan, and improves the effectiveness and safety of laser operation.

[0139] Embodiment 3

[0140] A method for image processing of image data in this embodiment includes:

[0141] Step C100: Obtain multi-modal image data of the first position of the user; the first position is the inside of the oral cavity or the skin to be measured.

[0142] Step C200: Reconstruct the multi-modal image data of the first position of the user into a 3D image, and automatically identify and obtain basic information of the specified area according to the 3D image.

[0143] The basic information of the specified area includes: the position of the specified area, the size of the specified area, the depth of the specified area, and the irregularity of the specified area.

[0144] Step C300: Input the size of the specified area, the depth of the specified area, and the irregularity of the specified area in the basic information of the specified area into a pre-constructed scoring model to obtain the specified area status score of the user, and obtain relevant parameters according to the specified area status score of the user and the pre-set processing parameter setting rules.

[0145] Step C400: Calculate and generate the exact path of the robotic arm according to the 3D image and the position of the specified area. The robotic arm reaches the specified area along the exact path, and operates the laser processing head connected to the robotic arm according to the relevant parameters.

[0146] Specifically, in step C200, automatically identifying and obtaining the basic information of the specified area according to the 3D image includes:

[0147] Step C210: Train a specified area basic information acquisition model with a training data set to obtain a trained specified area basic information acquisition model.

[0148] Step C220: Input the 3D image into the trained specified area basic information acquisition model to obtain the basic information of the specified area.

[0149] The specified area basic information acquisition model is a convolutional neural network model.

[0150] The training data set is past 3D image data and the basic information of the corresponding specified area.

[0151] In the specific implementation process, training the specified area basic information acquisition model specifically includes the following steps:

[0152] Data collection: To ensure that the model has good generalization ability, 3D images of past users and relevant information about the corresponding specified regions are collected. These images come from a wide range of sources, covering users of different ages, genders, and ethnicities, and also include images under various skin or oral conditions, such as skin ulcers, dental caries, periodontitis, malocclusion, etc. The data is sourced from hospitals and clinics.

[0153] Data denoising processing:

[0154] The Non-Local Means Filtering method is adopted. This method averages pixel values by finding similar local regions in the image, thereby effectively removing noise.

[0155] Normalization processing: The denoised images are normalized so that the image data has zero mean and unit variance.

[0156] Data augmentation: To increase the diversity of training data and improve the generalization ability of the model, this embodiment adopts various data augmentation techniques, including rotation, scaling, flipping, etc. At the same time, an elastic deformation augmentation method is introduced. Elastic deformation is achieved by applying a random elastic force field to the image, causing local deformation of the image. Specifically, a Gaussian kernel is used to generate a random displacement field, and then the image is interpolated according to the displacement field to obtain the augmented image data.

[0157] GS preprocessing: Convert the 3D image into a sparse Gaussian point cloud representation. First, the image is voxelized and divided into small voxels. Then, for each voxel, the coordinates of its center point and the corresponding intensity value are calculated. Next, a Gaussian function is used to fit these voxel center points to obtain a sparse Gaussian point cloud. In this way, noise interference can be reduced and edge details can be enhanced.

[0158] Label annotation: Each 3D image is equipped with detailed basic information labels for the specified region, which are manually annotated by experienced doctors. The annotated content includes features such as the location, size, and irregularity of the specified region. For example, for the specified region of a tooth, the doctor will mark the boundaries of the tooth crown and root, as well as the pulp cavity inside the tooth. To ensure the accuracy of the labels, the annotated labels will also undergo multi-level reviews to ensure their accuracy and consistency.

[0159] Model architecture: A 3D convolutional neural network is adopted as the basic architecture of the specified region basic information acquisition model. 3DCNN can directly process 3D image data and automatically extract features in the image through components such as convolutional layers, pooling layers, and fully connected layers. In the convolutional layer, 3D convolutional kernels are used for convolutional operations, and its calculation formula is:

[0160] ;

[0161] Among them, x is the input 3D image, ω is the 3D convolution kernel, b is the bias term, y is the output after convolution, i, j, k are the coordinates of the 3D image data, m, n, l are the coordinates of the sliding window of the convolution kernel in three dimensions, and M, N, L are the sizes of the convolution kernel in three dimensions (i.e., the size of the convolution kernel).

[0162] Attention mechanism: To enable the model to focus on the key regions in the image, that is, the specified regions, an attention mechanism is introduced. Specifically, during the feature extraction process of 3D CNN, an attention module is added. This module generates an attention weight map based on the input feature map, and then multiplies the attention weight map with the feature map to enhance the feature representation of the key regions. The generation of the attention weight map can be calculated by the following formula:

[0163] ;

[0164] Among them, F is the input feature map, W1 and W2 are learnable weight matrices, σ is the sigmoid function, A is the attention weight map, and tanh is the hyperbolic tangent activation function.

[0165] Training optimization:

[0166] The cross-entropy function is used as the loss function, and the Adam optimizer is used to update the model parameters. The calculation formula of the cross-entropy loss function is:

[0167] ;

[0168] Among them, y i is the true label, is the predicted output of the model, that is, the basic information of the specified region, and N is the number of samples. The learning rate of the Adam optimizer is set to 0.001. By continuously iterating and updating the model parameters, the cross-entropy loss function converges. γ i is the sample weight coefficient, which is used to assign weights to each sample according to the rarity or importance of the sample. For key region samples with a small number (such as some rare feature regions), larger weights can be assigned to improve the learning effect of the model on these samples.

[0169] At the same time, in the above formula, τ is the balance parameter, is the KL divergence term, and τ is used to balance the importance of the cross-entropy term and the KL divergence term. According to the specific task and data characteristics, corresponding adjustments can be made. And the introduction of the KL divergence term is used to measure the prediction distribution Q i and the prior only distribution P iDifferences, for example, in the oral cavity field, there is prior anatomical knowledge about structures such as teeth and gums. For example, the shape, position distribution, etc. of normal teeth have certain regularities. P i Constructed based on these prior knowledge, Q i is the predicted distribution of the above-specified region basic information acquisition model for this sample. By minimizing the KL divergence between these two, it is possible to guide the prediction result of the model to be more in line with prior knowledge and enhance the rationality and accuracy of the model prediction.

[0170] In the specific implementation process, the learning rate of the Adam optimizer can also be dynamically adjusted during the training process. The specific acquisition method is as follows:

[0171] ;

[0172] where η t represents the learning rate at the t-th iteration, η0 is the initial learning rate, set to 0.001, is the term that decays with the iteration t. Among them, α is the decay coefficient. As the training progresses, this term will gradually decrease, so that the learning rate also gradually decreases. This can ensure that the learning rate is larger in the initial stage of training and can converge quickly; while the learning rate is smaller in the later stage of training to avoid missing the optimal solution.

[0173] is the term for dynamically adjusting the learning rate according to the change rate of the loss function and the gradient norm, represents the change amount of the loss function at the t-th iteration, is the L2 norm of the gradient at the t-th iteration, and β is the adjustment coefficient used to control the influence degree of this dynamic adjustment term.

[0174] Furthermore, to avoid the situation of overfitting of the model, an early stopping mechanism is introduced. During the training process, the data set is divided into a training set and a validation set. After each training cycle ends, the loss value of the model on the validation set is calculated. If the loss value on the validation set no longer decreases for multiple consecutive cycles, the training is stopped, and the model at this time is selected as the finally trained model.

[0175] Input the pre - processed 3D image into the trained model for obtaining basic information of the specified area. The model will automatically identify and output the basic information of the specified area according to the input image, including the position, size, depth, and irregularity of the specified area. To enhance the interpretability of the model, the model for obtaining basic information of the specified area in this embodiment also includes a set of visualization tools. This tool can overlay the basic information of the specified area on the original 3D image and visually display the position and its features of the specified area in different ways such as colors and transparencies. For example, for the specified area of teeth, the carious area can be represented in red and the healthy area in green. Doctors can visually observe the pathological conditions of teeth through the visualization tool, providing strong support for treatment.

[0176] In the image - processing method of image data in this embodiment, by training the model for obtaining basic information of the specified area, the model for obtaining basic information of the specified area becomes more accurate and perfect, so as to obtain more accurate basic information of the specified area, thereby improving the accuracy and effectiveness of subsequent laser operations.

[0177] Embodiment 4

[0178] The image - processing method of image data in this embodiment is any one of the image - processing methods in Embodiment 1, Embodiment 2, and Embodiment 3, which will not be elaborated here.

[0179] The user in this embodiment uses the image - processing method of image data in this embodiment for processing. The specific steps are as follows:

[0180] Use imaging devices such as dermoscopes, ultrasounds, MRI (magnetic resonance imaging), OCT (optical coherence tomography) to obtain multi - modal image data of the skin and related tissues to be measured of the user, and use imaging devices such as ultrasounds, dental X - ray machines, panoramic X - ray machines, computed tomography, OCT (optical coherence tomography), oral scanners, and oral endoscopes to obtain multi - modal image data of the interior of the user's oral cavity.

[0181] Taking the image - processing method of the interior oral cavity image data as an example in this embodiment, after obtaining the multi - modal image data of the interior of the user's oral cavity, align the above - mentioned multi - modal image data of the interior of the user's oral cavity or skin, and use 3D Slicer (medical image layer - by - layer fusion) for 3D reconstruction to obtain the 3D image data of the user;

[0182] Input the 3D image data of the user into the trained model for obtaining basic information of the specified area to obtain the following basic information of the specified area:

[0183] Position of the specified area: Gingiva of the first molar in the left mandible;

[0184] Specified area size: volume is approximately 4 mm 3 ;

[0185] Specified area depth: the distance from the gum surface to the deepest part is 2.5 mm;

[0186] Specified area shape: the irregularity score is 0.7;

[0187] Input the specified area size, specified area depth, and specified area irregularity into the following formula to obtain the first score of the user-specified area status:

[0188] ;

[0189] where ω s is the influence weight of the specified area size, set to 0.5, ω D is the influence weight of the specified area depth, set to 0.3, ω F represents the influence weight of the specified area shape, set to 0.2, S is the specified area size, S max is the maximum reference value set for the specified area size, which is 10 mm 3 , k s is the non-linear strength coefficient, set to 2, D is the specified area depth, D max is the set maximum reference depth, which is 5 mm, β D is the steepness of the saturation curve, F is the specified area irregularity, and Score is the first score of the specified area status.

[0190] According to the above formula, the first score of the user's specified area status can be obtained as 1.267;

[0191] Input the first score of the specified area status into the following formula for normalization to obtain the specified area status score:

[0192] ;

[0193] where Final Score is the specified area status score, ω s is the influence weight of the specified area size, ω D is the influence weight of the specified area depth, ω F represents the influence weight of the specified area shape, k s is the non-linear strength coefficient, D max is the set maximum reference depth, F max is the set maximum irregularity.

[0194] From this, the specified area status score can be obtained as 26.22 points.

[0195] According to the specified area status score, substitute it into the following formula:

[0196] ;

[0197] Among them, T is a relevant parameter, T min is the minimum value of the relevant parameter, T max is the maximum value of the relevant parameter, and FinalScore is the status score of the specified area.

[0198] From this, the relevant parameters can be obtained as follows: laser power: 1.42 W; laser operation duration: 11.56 min; laser irradiation area: 3 mm 2 .

[0199] The above relevant parameters are used to guide the subsequent image processing process to ensure that the processing is both safe and effective.

[0200] Example 5

[0201] Refer to Figure 3 , an intelligent laser device in this embodiment executes an image data processing method according to any one of Embodiment 1, Embodiment 2, Embodiment 3, and Embodiment 4. An intelligent laser device includes:

[0202] A robotic arm, a laser processing head, a sensor assembly, a laser generator, a central processing assembly, and a display screen;

[0203] The robotic arm, the laser processing head, the sensor assembly, the laser generator, the central processing assembly, and the display screen are respectively electrically connected to the central processing assembly;

[0204] The robotic arm and the laser processing head are connected through a rotating joint with a micro motor.

[0205] In this embodiment, the robotic arm includes a control component and a plurality of electric gears. A power source is provided on the electric gears, and the laser processing head is respectively connected to an electric gear; the control component is arranged on the electric gear, and while controlling the rotation of the corresponding electric gear, the control component can precisely control the running direction and displacement of the laser processing head.

[0206] Furthermore, the robotic arm in this embodiment adopts a multi-degree-of-freedom design and, combined with a precise sensor assembly, can achieve micron-level control of the laser processing head.

[0207] At the same time, a replaceable optical fiber is integrated at the end of the laser processing head. The optical fiber has a micro-magnet attached near the tip. The magnetic field is driven by an external coil to achieve high-speed micro-swing of the optical fiber, thereby completing the two-dimensional scanning control of the laser beam. Based on this design architecture, the laser processing head only needs to be positioned to the working position through macroscopic movement, while the two-dimensional scanning and fine trajectory description are achieved by the magnetically controlled optical fiber mechanism. This electromechanical separation drive method not only ensures the system's large-range motion capability, but also ensures the dynamic response performance during precision scanning.

[0208] In the specific implementation process, the intelligent laser device also includes an alarm device. When the laser processing head exceeds the safe distance from the top of the designated area, the alarm device sounds an alarm. The safe distance is 5mm. An optical fiber transmission system is integrated between the laser processing head and the laser generator of this embodiment. The laser generator sends the laser to the laser processing head through the optical fiber transmission system to ensure efficient transmission and precise focusing of the laser energy. The laser processing head can achieve 360-degree rotation without dead angles through a micro motor connected by a rotating joint to meet the laser processing needs of different positions.

[0209] In addition, the laser processing head can perform reciprocating motion or circular motion around a designated area according to a set path.

[0210] The sensor components of this embodiment include: temperature sensors, optical sensors and other types of sensors; among them, the temperature sensor is installed on the laser processing head, and can monitor the temperature changes during the processing process in real time. Once the temperature exceeds the safe range, it will immediately feed back the signal to the central processing component, and the central processing component will adjust the power of the laser transmitter or suspend processing accordingly; the optical sensor is responsible for monitoring the propagation and reflection of the laser, and by analyzing the intensity, wavelength and other parameters of the reflected light, it determines whether the laser acts accurately on the target processing area and evaluates the processing effect.

[0211] The sensor component is connected to the central processing component through a special signal acquisition circuit, which can amplify, filter and process the weak signals collected by the sensor, and then transmit the processed signals accurately to the central processing component.

[0212] During the specific implementation process, the display screen uses an LCD screen and supports touch operation. Doctors can manually optimize and adjust relevant parameters through the touch screen.

[0213] In this embodiment, the intelligent laser device further includes:

[0214] The storage device is used to store past user 3D images, basic information of the designated area, status scores of the designated area and related parameters, and the storage device is stored in an encrypted storage manner.

[0215] The storage device of this embodiment adopts advanced encryption technology to ensure the secure storage of users' 3D images, basic information of specified areas, status scores of specified areas, and related parameters. At the same time, it supports cloud backup function to prevent data loss and facilitate remote consultation and data sharing.

[0216] An intelligent laser device in this embodiment reduces the labor intensity of doctors, increases the accuracy of operations, ensures the security of users' data, and improves the comfort of users.

[0217] In the description of the present invention, it should be understood that the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the present invention, "a plurality" means two or more, unless otherwise specifically defined.

[0218] In the present invention, unless otherwise clearly defined and limited, the terms "installed", "connected", "connected", "fixed", etc. should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or integrated; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium; it can be the connection inside two components or the interaction relationship between two components. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0219] In the present invention, unless otherwise clearly defined and limited, the first feature being "on" or "under" the second feature can be that the first and second features are in direct contact, or the first and second features are indirectly in contact through an intermediate medium. Moreover, the first feature being "above", "over" and "on" the second feature can be that the first feature is directly above or obliquely above the second feature, or simply means that the first feature has a higher horizontal height than the second feature. The first feature being "under", "below" and "beneath" the second feature can be that the first feature is directly below or obliquely below the second feature, or simply means that the first feature has a lower horizontal height than the second feature.

[0220] In the description of this specification, the descriptions of terms such as "one embodiment", "some embodiments", "embodiment", "example", "specific example", or "some examples" mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0221] Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention.

Claims

1. An image processing method for image data, characterized in that, Including: S100. Obtain multi-modal image data of the user's first position; the first position is inside the oral cavity or the skin to be measured. S200. Reconstruct the multi-modal image data of the user's first position into a 3D image, and automatically identify and obtain the basic information of the specified area according to the 3D image. The basic information of the specified area includes: the position of the specified area, the size of the specified area, the depth of the specified area, and the irregularity of the specified area. The basic information of the specified area is obtained by training and recognizing the 3D image based on the basic information acquisition model of the specified area. During the training and recognition process of the 3D image by the basic information acquisition model of the specified area, GS preprocessing is performed on the 3D image to convert the 3D image into a sparse Gaussian point cloud representation. The basic information acquisition model of the specified area is a convolutional neural network model. S300. Input the size of the specified area, the depth of the specified area, and the irregularity of the specified area in the basic information of the specified area into a pre-constructed scoring model to obtain the specified area status score of the user. According to the specified area status score of the user and the pre-set processing parameter setting rules, obtain relevant parameters. S400. Calculate and generate the precise path of the robotic arm according to the 3D image and the position of the specified area. The robotic arm reaches the specified area along the precise path, and operates the laser processing head connected to the robotic arm according to the relevant parameters. The S300 includes: S310. Input the size of the specified area, the depth of the specified area, and the irregularity of the specified area in the basic information of the specified area into a pre-constructed scoring model to obtain the first specified area status score of the user's specified area. S320. Normalize the first specified area status score to obtain the specified area status score. S330. Obtain relevant parameters according to the specified area status score and the parameter setting rules. The S310 includes: Input the size of the specified area, the depth of the specified area, and the irregularity of the specified area into the following formula to obtain the first specified area status score of the user's specified area: ； Among them, ω s is the influence weight of the specified area size, ω D is the influence weight of the specified area depth, ω F represents the influence weight of the specified area form, S is the specified area size, S max is the maximum reference value set for the specified area size, k s is the non-linear strength coefficient, D is the specified area depth, D max is the maximum reference depth set, β D is the steepness of the saturation curve, F is the specified area irregularity, Score is the first score of the specified area state; The S320 includes: Input the first specified area status score into the following formula for normalization processing to obtain the specified area status score: ； Among them, Final Score is the specified area status score, ω s is the influence weight of the specified area size, ω D is the influence weight of the specified area depth, ω F represents the influence weight of the specified area shape, k s is the non-linear strength coefficient, D max is the set maximum reference depth, F max is the set maximum irregularity.

2. The image data processing method according to claim 1, characterized in that, In the S200, automatically identifying and obtaining the basic information of the specified area according to the 3D image includes: S210. Train the basic information acquisition model of the specified area using the training data set to obtain the trained basic information acquisition model of the specified area. S220. Input the 3D image into the trained basic information acquisition model of the specified area to obtain the basic information of the specified area. The training data set is the past 3D image data and the basic information of the corresponding specified area.

3. The image data processing method according to claim 2, wherein The S210 specifically includes: Input the past 3D image data and the basic information of the corresponding specified area into the basic information acquisition model of the specified area, and use the cross-entropy function and the Adam optimizer with a learning rate of 0.001 until the cross-entropy regression loss function converges to obtain the trained basic information acquisition model of the specified area.

4. The image data processing method according to claim 1, characterized in that The S330 includes: Input the specified area status score into the following formula to obtain relevant parameters: ； Among them, T is a relevant parameter, and T min is the minimum value of the relevant parameter, and T max is the maximum value of the relevant parameter, and Final Score is the status score of the specified area; The relevant parameters include laser power, laser operation duration, and laser irradiation area.

5. The image data processing method according to claim 4, characterized in that In S330, The minimum value of the laser power is 0.5W, and the maximum value is 36W; The minimum value of the laser operation duration is 1s, and the maximum value is 30min; The minimum value of the laser irradiation area is 0.5 mm 2 , and the maximum value is 49 cm 2 .

6. The image data processing method according to claim 1, characterized in that, S400 specifically includes: S410. Based on the specified area position and 3D image, use the cubic spline interpolation algorithm to generate the continuous path of the robotic arm in three-dimensional space; S420. According to the shortest path algorithm and the continuous path of the robotic arm in three-dimensional space, obtain the precise path of the robotic arm operation. The robotic arm moves to the specified area along the precise path, and operates the laser processing head connected to the robotic arm according to the relevant parameters.

7. An intelligent laser device, characterized in that, The intelligent laser device is used to execute the image processing method of the image data according to any one of claims 1 to 6. The intelligent laser device includes: A robotic arm, a laser processing head, a sensor assembly, a laser generator, a central processing assembly, and a display screen; The robotic arm, the laser processing head, the sensor assembly, the laser generator, the central processing assembly, and the display screen are respectively electrically connected to the central processing assembly; The robotic arm and the laser processing head are connected by a rotating joint with a micro motor.

Images