Precise acupuncture positioning method under ultrasonic guidance

Abstract

The invention discloses an accurate acupuncture positioning method under ultrasonic guidance, which relates to the technical field of medical health and comprises the following steps: acquiring a two-dimensional or three-dimensional ultrasonic image of a target area by utilizing ultrasonic image acquisition equipment; the Young modulus of the target area is calculated by measuring the shear wave propagation speed of the tissue of the target area; a sensor module is adopted to monitor the interaction force between the puncture needle and the tissue in real time, and mechanical data are obtained; according to Young modulus and modulus variance obtained through calculation of the shear wave elastography image and mechanical data fed back by a sensor, a time sequence feature vector is constructed; performing dynamic modeling and prediction on the tissue state in the puncture process by using a long-short-term memory neural network; the path offset of the needle head is judged, a correction signal is generated through a PID controller, and the posture or direction of the needle head is adjusted in real time; the problem of artifacts in a traditional ultrasonic image can be solved, more accurate acupuncture path planning is provided, and the method adapts to the change of tissue elasticity in real time.

Description

Technical Field

[0001] The present invention relates to the technical field related to medical health, and specifically to a method for accurate needle puncture positioning under ultrasound guidance. Background Art

[0002] Biopsy of small deep liver tumors under ultrasound guidance is a method commonly used for diagnosing liver diseases (such as liver cancer, metastatic tumors, etc.). Through ultrasound guidance, doctors can accurately obtain tumor tissue samples in the deep and inaccessible parts of the liver.

[0003] In the clinical operation of biopsy of small deep liver tumors under ultrasound guidance, due to image artifacts such as enhanced echo and acoustic shadow at the tissue interface, the imaging of the puncture needle in the key area is not clear, resulting in a significant increase in intraoperative positioning errors; in addition, the differences in elasticity and hardness of different tissues will affect the depth and path of needle puncture. Ultrasound images are mainly used for positioning, but they cannot effectively evaluate and adapt to tissue elasticity changes in real time. Especially at the junction of soft tissue and hard tissue, it is difficult to control the needle puncture depth, which is prone to errors. Summary of the Invention

[0004] To solve the defects existing in the prior art, the present invention provides a method for accurate needle puncture positioning under ultrasound guidance.

[0005] To solve the above technical problems, the present invention provides the following technical solutions:

[0006] The present invention provides a method for accurate needle puncture positioning under ultrasound guidance, including:

[0007] S1: Ultrasound image acquisition and processing: Use an ultrasound image acquisition device to obtain two-dimensional or three-dimensional ultrasound images of the target area. The ultrasound images are used to preliminarily plan the puncture path and provide structural visual information of the target area, including the positioning of organ boundaries, tumors, and blood vessels;

[0008] S2: Adopt shear wave elastography technology. By measuring the shear wave propagation speed of the tissue in the target area, calculate the Young's modulus of the target area, so as to obtain the elastic distribution information of this area. This information is used to evaluate the hardness differences of different tissues;

[0009] S3: Use a sensor module to real-time monitor the interaction force between the puncture needle and the tissue, and obtain mechanical data for real-time feedback on the movement state of the needle tip;

[0010] S4: Construction of time series feature vectors and dynamic modeling. According to the Young's modulus and modulus variance calculated from the shear wave elastography images, as well as the mechanical data feedback by the sensor, construct time series feature vectors to describe the tissue elasticity changes and the movement state of the needle tip during the puncture process;

[0011] S5: Dynamically model and predict the tissue state during the puncture process using a long short-term memory neural network;

[0012] S6: Judge the path offset of the needle based on the prediction structure in step S5, and generate a correction signal through a PID controller to adjust the attitude or direction of the needle in real time.

[0013] As a preferred technical solution of the present invention, the ultrasonic image acquisition device is a two-dimensional, three-dimensional or four-dimensional ultrasonic device, which is used to provide image data of the target area at different levels.

[0014] As a preferred technical solution of the present invention, the Young's modulus E is calculated by the following formula:

[0015] E = ρ·V s 2 ;

[0016] where ρ is the density of the tissue and V s is the shear wave propagation speed.

[0017] As a preferred technical solution of the present invention, the mechanical data includes the contact force and the bending moment;

[0018] The mechanical data of the contact force Fz is described by the following relationship:

[0019] F z = k·x;

[0020] where F z is the contact force, k is the sensitivity coefficient of the sensor, and x is the sensor displacement;

[0021] The mechanical data of the bending moment Mx is described by the following relationship:

[0022] M x = τ·x;

[0023] where M x is the bending moment and τ is the sensitivity coefficient of the sensor.

[0024] As a preferred technical solution of the present invention, the time series feature vector includes the average elastic modulus E i of the current region, the modulus variance σ E , the contact force Fz, the bending moment Mx, the spatial gradient of the elastic modulus and the time increment Δt, which are used to describe the state of tissue elasticity and needle movement;

[0025] Therefore, the feature vector x t at each time point t is expressed as:

[0026]

[0027] As a preferred technical solution of the present invention, the long short-term memory neural network obtains the time series feature vector x of tissue elasticity and needle movement t to predict the change of the puncture path.

[0028] As a preferred technical solution of the present invention, the control signal of the PID controller can be expressed as:

[0029]

[0030] where e(t) is the path error, K p 、K i and K d are the proportional gain, integral gain and derivative gain respectively.

[0031] As a preferred technical solution of the present invention, the needle puncture precise positioning method further includes the steps of:

[0032] S7: Using the puncture safety window judgment model, when the predicted path offset exceeds the safety range, a path correction signal is issued.

[0033] As a preferred technical solution of the present invention, the puncture safety window judgment model determines whether the predicted path offset exceeds the safety range by defining the maximum offset threshold δ max :

[0034] |δ|≤δ max .

[0035] As a preferred technical solution of the present invention, the safety window judgment model also includes evaluating the relative relationship between the posture of the puncture needle and the boundary of the target area. When the puncture path approaches the boundary of the target area, a correction signal is automatically generated to avoid damaging the target tissue.

[0036] The beneficial effects of the present invention are:

[0037] The needle puncture precise positioning method under ultrasonic guidance of the present invention can overcome the artifact problem in traditional ultrasonic images, provide more accurate needle puncture path planning, adapt to the change of tissue elasticity in real time, and reduce the positioning error and puncture depth control problem that may occur during the operation. By dynamically modeling and real-time path correction, the accuracy and safety of liver deep small tumor tissue biopsy can be effectively improved. Description of the Drawings

[0038] The drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention, and do not constitute a limitation to the present invention.

[0039] In the accompanying drawings:

[0040] Figure 1 This is a schematic flow chart of the needle puncture precise positioning method under ultrasonic guidance of the present invention. Detailed implementation manners

[0041] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only for the purpose of illustrating and explaining the present invention, and are not used to limit the present invention.

[0042] Embodiment 1

[0043] As Figure 1 shown, the needle puncture precise positioning method under ultrasonic guidance includes:

[0044] S1: Ultrasonic image acquisition and processing: Use an ultrasonic image acquisition device to obtain two-dimensional or three-dimensional ultrasonic images of the target area. The ultrasonic images are used to preliminarily plan the puncture path and provide structural visual information of the target area, including the positioning of organ boundaries, tumors, and blood vessels; through image processing methods, accurately extract the anatomical structure information of the target area to provide a basis for optimizing the puncture path;

[0045] These image data are not only used to preliminarily plan the puncture path, but also can accurately extract the anatomical structure information of the target area, including the positioning of the liver boundary, tumor, and blood vessels, providing an important basis for optimizing the subsequent needle puncture path.

[0046] Generally speaking, through image processing techniques, such as edge detection, image smoothing, and denoising, improve the clarity of ultrasonic images to help improve the visibility of tumors. Specifically, use image segmentation algorithms, such as thresholding, region growing, or deep learning networks, to automatically segment the target areas (such as tumors, blood vessels, liver tissues) in the image, which can accurately locate the boundaries and positions of tumors and provide data support for puncture path planning. This is prior art and will not be elaborated here.

[0047] S2: Adopt shear wave elastography technology. By measuring the shear wave propagation speed of the tissue in the target area, calculate the Young's modulus of the target area, so as to obtain the elastic distribution information of this area. This information is used to evaluate the hardness differences of different tissues and support the planning of the puncture path;

[0048] The hardness differences of different tissues can be accurately evaluated with the support of this technology, providing a reference for puncture path planning. Especially in the area where soft and hard tissues meet, it can effectively reduce errors.

[0049] Shear waves are emitted by an ultrasound device, and their propagation speed is measured. The faster the propagation speed, the harder the tissue. Generally, tumor tissues are harder than the surrounding normal liver tissues. Young's modulus can reflect the hardness difference between the tumor and the surrounding normal tissues, helping doctors better understand the tissue structure and ensuring that the puncture path avoids areas of overly hard or overly soft tissues. By analyzing the tissue elasticity distribution, the scope and internal structure of the tumor can be judged, providing a basis for optimizing the puncture path.

[0050] S3: The sensor module is used to monitor the interaction force between the puncture needle and the tissue in real time, obtain mechanical data, and be used for real-time feedback of the movement state of the needle tip; this step ensures that the movement of the puncture needle can be precisely controlled, avoiding incorrect paths or over-penetration;

[0051] This step ensures that the movement of the puncture needle can be precisely controlled, avoiding the formation of incorrect paths or over-penetration. Especially in complex tissue structures, it can greatly reduce the puncture error.

[0052] It should be noted that the sensor module usually consists of multiple sensor units, and these units can measure different types of forces and strains. Specifically, these sensor arrays can include the following types of sensors: force sensors (used to measure the vertical or horizontal forces when the needle contacts the tissue, such as the contact force F z ), strain sensors (used to detect the bending or deformation of the needle, so as to calculate the bending moment M x ) between the needle and the tissue, piezoelectric sensors (generating voltage signals when subjected to mechanical forces, thereby sensing the contact force and bending strain between the needle and the tissue), multi-point sensor arrays (installing multiple sensor units on the surface or around the puncture needle to form an array for obtaining force or strain data at different positions), etc.

[0053] S4: Construction of time-series feature vectors and dynamic modeling. According to the Young's modulus and modulus variance calculated from the shear wave elastography images, as well as the mechanical data fed back by the sensors, time-series feature vectors are constructed to describe the tissue elasticity changes and the movement state of the needle tip during the puncture process;

[0054] These information are input into a deep learning model, which can adjust the puncture path in real time and perform dynamic modeling and prediction, enhancing the real-time adaptability to the acupuncture process.

[0055] S5: Using a long short-term memory neural network to perform dynamic modeling and prediction on the tissue state during the puncture process, which can accurately predict the changes in the puncture path, thereby further optimizing the puncture process and reducing errors.

[0056] The LSTM neural network can capture the interaction pattern between the needle and the tissue during the puncture process by learning historical data. By training the model, the LSTM can predict the deviation of the puncture path, further adjust the operation strategy, and can also accurately predict the movement trajectory of the needle, the change of tissue hardness, etc. through learning the time series data, so as to ensure that the needle moves precisely along the predetermined path.

[0057] S6: Judge the path deviation amount of the needle based on the prediction structure in step S5, and generate a correction signal through the PID controller to adjust the attitude or direction of the needle in real time. This function plays a crucial role in avoiding the deviation of the puncture path and further improves the accuracy of the operation.

[0058] According to the prediction results of the LSTM network on the tissue state and the needle path during the puncture process, calculate the path deviation amount of the puncture needle. If there is a deviation, the system will issue a warning and automatically make adjustments. The PID controller generates a correction signal according to the deviation amount, and by adjusting the attitude or direction of the puncture needle, ensures that the needle accurately reaches the target position, avoiding mis-puncture or over-penetration.

[0059] In summary, the present invention solves the positioning error problem of traditional ultrasound-guided needle puncture operations through means such as precise image processing, elastography, real-time mechanical data feedback, and intelligent prediction, and shows significant advantages especially in the biopsy of small liver tumors deep in the body. They effectively overcome the puncture difficulties at the junction of hard and soft tissues, optimize the puncture path control, thereby improving the accuracy and safety of the puncture.

[0060] Furthermore, the ultrasound image acquisition device is a two-dimensional, three-dimensional or four-dimensional ultrasound device, which is used to provide image data of the target area at different levels.

[0061] Two-dimensional ultrasound device: It can display the longitudinal or transverse sectional images of the target area in real time, and is suitable for conventional clinical puncture operations. The relative relationship between the puncture needle and the target is judged through the experience of the operator.

[0062] Three-dimensional ultrasound device: It can generate a three-dimensional structure image of the target area, enabling the operator to obtain more complete spatial anatomical information, which is helpful for formulating a more reasonable needle insertion path, especially suitable for areas with complex structures or dense surrounding tissues and organs.

[0063] Four-dimensional ultrasound device: On the basis of three-dimensional imaging, it adds a real-time dynamic dimension, can continuously capture and display the dynamic changes of tissues during the puncture process, realizes dynamic tracking and precise navigation, and is of great significance for real-time monitoring of the tip position.

[0064] By using the above-mentioned ultrasonic image acquisition devices of different types in combination, not only the accuracy of target area recognition is improved, but also the safety and effectiveness of the acupuncture process are significantly enhanced. Especially during the puncture operation, the doctor can perform preliminary positioning based on the two-dimensional image, observe the spatial position of the target organ or lesion through the three-dimensional image, and make dynamic adjustments with the help of the four-dimensional image, ultimately realizing the "visualized, controllable, and refined" accurate puncture path planning and implementation.

[0065] Furthermore, the Young's modulus E is calculated by the following formula:

[0066] E = ρ·V s 2 ;

[0067] where ρ is the density of the tissue and V s is the shear wave propagation velocity.

[0068] Under ultrasonic guidance, ultrasonic waves can provide real-time elastic information of tissues, and the Young's modulus value is deduced through the shear wave propagation velocity. This is particularly important for acupuncture positioning. For example, in tumor treatment, different tumor tissues have different stiffness characteristics. Using this method can help doctors accurately locate the target lesion area, thereby improving the accuracy of acupuncture treatment, avoiding accidental injury to surrounding normal tissues. Combining ultrasonic guidance and Young's modulus calculation can provide a more accurate and personalized treatment plan during the acupuncture process, reducing treatment errors and improving the treatment effect of patients.

[0069] The acquisition methods of the parameter ρ include:

[0070] Empirical value: According to different types of tissues (such as muscle, fat, bone, etc.), use the standard density values given in existing literature;

[0071] References: Cite the density values of specific tissue types given in some medical literature. For example, the bone density can refer to the average value obtained from CT scans or DEXA scans;

[0072] Volume estimation: Combine the tissue volume calculated by medical imaging techniques (such as MRI or CT) and calculate it with the known reference density value.

[0073] Furthermore, the mechanical data includes contact force and bending moment;

[0074] The mechanical data of the contact force F z is described by the following relationship:

[0075] F z = k·x;

[0076] where F zFz is the contact force, representing the force applied when the needle or probe contacts the tissue. k is the sensitivity coefficient of the sensor, representing the response degree of the sensor to displacement. x is the sensor displacement, representing the displacement change that occurs during the contact process of the sensor.

[0077] The mechanical data of the bending moment Mx is described by the following relationship:

[0078] M x = τ·x;

[0079] Where, M x is the bending moment, representing the bending effect generated on the acupuncture device due to external forces during acupuncture. The bending moment is usually related to the acupuncture angle, the stiffness of the material, and the magnitude of the external force. By measuring the bending moment, it is possible to determine whether the needle has deviated from the predetermined path and help the doctor accurately adjust the acupuncture direction. τ is the sensitivity coefficient of the sensor, representing the response ability of the sensor to the bending moment. The x in this formula also represents the displacement of the sensor. In this case, the displacement is usually caused by bending deformation.

[0080] Furthermore, the time series feature vector includes the average elastic modulus E i of the current region, the modulus variance σ E , the contact force Fz, the bending moment Mx, the spatial gradient of the elastic modulus and the time increment Δt, which are used to describe the state of tissue elasticity and needle movement;

[0081] Therefore, the feature vector x t at each time point t is expressed as:

[0082]

[0083] The puncture process is a continuous, dynamic, and interactive process. The elasticity of the tissue and the state of the puncture needle are constantly changing. In order to achieve high-precision and real-time path prediction and adjustment, all key variables during the puncture process need to be abstracted into a sequence of feature vectors for input into a deep learning model for modeling and prediction.

[0084] These variables comprehensively reflect the physical properties of the target tissue and the dynamic interaction state of the puncture needle, and have time continuity, making them suitable for input into a time series neural network model (such as LSTM) for modeling.

[0085] Furthermore, the long short-term memory neural network (LSTM) obtains the time series feature vector x t of tissue elasticity and needle movement to predict changes in the puncture path.

[0086] First, a sequence of time series feature vectors needs to be given:

[0087] {x1, x2,......, x T};

[0088] where x t is the input feature vector at the t-th time step, and T is the total length of the sequence.

[0089] The goal of the long short-term memory neural network model is to calculate and update the hidden state h t and the memory cell C t at each moment t. The LSTM is particularly suitable for tasks that require remembering and updating long-term information. For example, in the method of precise needle insertion under ultrasound guidance, as time goes by, the tissue elasticity and the state of the needle movement will change. The LSTM can predict the change of the puncture path according to these temporal features.

[0090] Next, we will explain each key part of the LSTM in more detail:

[0091] Forget gate

[0092] Determines how much of the old memory (C t-1 from the previous moment) to discard, that is, how much the memory state of the previous moment affects the current moment. In the LSTM, the memory at each moment is updated based on the memory state of the previous moment. In precise needle insertion, the tissue elasticity and the position of the needle will change continuously over time. Therefore, through the forget gate, the LSTM can selectively forget the historical information that is irrelevant to the current state and focus on the new and more important input information. The calculation formula is:

[0093] f t = σ(Wf · [h t-1 , x t + bf f );

[0094] where f t is the output of the forget gate, with a value range of (0, 1). The closer it is to 0, the more old memories are discarded. σ is the Sigmoid activation function, W f is the weight matrix of the forget gate, bf is the bias term of the forget gate, and [h t-1 , x t is the concatenation of the hidden state of the previous moment and the current input;

[0095] For example, in practical applications, if the tissue state of the previous moment has no influence on the prediction of the current moment (such as obvious changes in the tissue), the forget gate will "forget" the memory of the previous moment to avoid negative impacts on the current moment.

[0096] Input gate

[0097] Determine how much of the current input needs to be stored in the memory unit. It controls the "importance" of the current input information, thereby updating the state of the memory unit. For the accurate needle positioning under ultrasound guidance, the input gate determines the "importance" of information such as tissue elasticity and needle position at the current moment, and then determines their contribution to the memory unit. The calculation formula:

[0098] i t =σ(W i ·[h t-1 ,x t +b i );

[0099] Among them, i t is the output of the input gate, representing the proportion of the current input updated to the memory unit. W i is the weight matrix of the input gate, and b i is the bias term of the input gate;

[0100] For example, the needle position at the current moment may be more important than the previous tissue elasticity. At this time, the input gate will determine that more input information is stored in the memory unit.

[0101] Candidate memory content

[0102] Generate potential new memory content according to the current input to help decide whether to store the current input information in the memory unit. This part generates new candidate memory content through non-linear transformation. The calculation formula:

[0103] C t ′=tanh(WC·[h t-1 ,x t +b C );

[0104] Among them, C t ′ is the candidate memory content, whose range is between [-1, 1]. W C is the weight matrix of the candidate memory, and b C is the bias term of the candidate memory;

[0105] In needle positioning, the candidate memory can take into account the complexity of the input information, such as the relative position between the tissue elasticity and the needle at the current moment, and generate new memory content through a non-linear activation function (such as tanh).

[0106] Memory unit update

[0107] Fuse the old memory C t-1 and the new candidate memory C t′, The forget gate determines how much of the old memory to discard, while the input gate determines how much new information to store in the memory. This is the core part of the LSTM, which allows the model to dynamically update the memory and capture long-term dependencies when processing sequential data at each moment. The calculation formula is:

[0108] C t = f t · C t-1 + i t · C t ′;

[0109] Among them, C t is the memory cell state at the current moment, C t-1 is the memory cell state at the previous moment, f t is the output of the forget gate, which determines the proportion of the old memory to be retained;

[0110] In the acupuncture precise positioning task, the memory cell will comprehensively consider the needle position and tissue elasticity information at the current moment to update the state of the memory cell. For example, the tissue elasticity at a certain moment may cause a change in the needle trajectory, and the memory cell needs to be updated in a timely manner to reflect this change.

[0111] Output gate

[0112] Determines the amount of information finally output to the hidden state, that is, the final hidden state. It controls which information will be passed to the next moment or used as the final output at the current moment. The calculation formula is:

[0113] o t = σ(Wo · [h t-1 , x t + b o );

[0114] Among them, o t is the output of the output gate, which controls the output proportion of the final hidden state. W o is the weight matrix of the output gate, b o is the bias term of the output gate;

[0115] In the acupuncture precise positioning task, the output gate will determine whether the needle movement trajectory and tissue elasticity information at the current moment are passed to the next moment, affecting future predictions.

[0116] Hidden state update

[0117] Calculate the final output hidden state according to the current memory state. It is the core information finally passed by the network to the next moment. Through the update of the memory cell state by a non-linear function (such as tanh), the LSTM can convert long-term memory into the output at the current moment, further helping to predict the movement path of the needle. The calculation formula is:

[0118] h t = o t ·tanh(C t );

[0119] where h t is the hidden state at the current moment, and tanh(C t ) non-linearly scales the memory cell state. In acupuncture needle positioning, the hidden state represents the expected trajectory or position of the current needle, which affects the decision-making at the next moment.

[0120] Through the above six steps, the LSTM network can update the memory cell according to the current input and the state at the previous moment at each moment and output the hidden state. In this way, the LSTM can capture long-term dependencies and predict the change of the puncture path according to the tissue elasticity and the temporal characteristics of the needle movement. The functions and formulas of each gate complement each other to ensure that the LSTM can effectively process complex temporal data.

[0121] Furthermore, the control signal of the PID controller can be expressed as:

[0122]

[0123] where u(t) is the control signal, representing the adjustment signal output by the controller. In this application, the control signal u(t) is the control signal output by the PID controller according to the path error to control the movement direction and speed of the needle. e(t) is the path error, representing the deviation between the current system state and the target state. Specifically, in ultrasound-guided acupuncture needle positioning, e(t) may represent the difference between the current position of the needle and the desired position, which may include lateral error and depth error;

[0124] K p , K i and K d are the proportional gain, integral gain, and derivative gain respectively. K p controls the intensity of the proportional term. The role of the proportional term is to adjust the control signal according to the current error. The larger the error, the stronger the adjustment signal output by the controller. K i controls the intensity of the integral term. The integral term reduces the small errors accumulated in the long term by accumulating the errors. This term helps to eliminate the long-term steady-state error of the system. K d controls the intensity of the derivative term. The derivative term focuses on the rate of change of the error and is used to predict the future trend of the error, so as to adjust the control signal in advance to avoid the system reacting too slowly or over-adjusting.

[0125] In the acupuncture precise positioning method under ultrasonic guidance, the PID controller plays a crucial role. It ensures that the needle can accurately reach the target position through real-time feedback of path errors and the combination of three control methods. The reasonable adjustment of the three gains of proportional, integral, and differential is the key to achieving precise control.

[0126] Furthermore, the acupuncture precise positioning method further includes the steps:

[0127] S7: Using the puncture safety window judgment model, when the predicted path offset exceeds the safe range, the system can automatically send a correction signal to avoid damage to key areas such as the liver and tumors caused by puncture errors. This safety window judgment model ensures the safety of the puncture path and minimizes the intraoperative risk. Therefore, the puncture safety window judgment model needs to continuously evaluate the offset of the current acupuncture path and determine whether the offset exceeds the safe range.

[0128] Furthermore, the puncture safety window judgment model determines whether the predicted path offset exceeds the safe range by defining a maximum offset threshold δ max :

[0129] |δ| ≤ δ max .

[0130] In the puncture safety window judgment model, by defining a maximum offset threshold δmax, the allowable error range of the acupuncture path is determined. If the predicted path offset |δ| exceeds this threshold, then it is considered that the current path has exceeded the safe range. At this time, the system will send a path correction signal to prompt the doctor to adjust the acupuncture angle or depth to ensure that the operation does not deviate from the safe path.

[0131] Definition of the maximum offset threshold δmax: It represents the maximum allowable offset of the acupuncture path within the safe range. This threshold can be adjusted according to specific clinical needs and different operation sites. For example, near some important organs, the maximum offset threshold may be set smaller to ensure the safety of the operation.

[0132] Path offset δ: The path offset refers to the difference between the predicted acupuncture path and the expected path. This difference may be caused by various factors, such as ultrasonic image errors, patient body position changes, and minor changes during the operation.

[0133] When the system detects that the path offset exceeds the maximum offset threshold δmax, it immediately sends a path correction signal. These signals can be notified to the doctor in various ways, such as image prompts, sound alarms, vibration feedback, etc. The purpose is to remind the doctor to immediately correct the acupuncture path to avoid further mispuncture or injury.

[0134] Way to correct the path: Based on the correction signal, the doctor can adjust the acupuncture angle, depth, or direction to ensure that the acupuncture path can return to the expected safe path. For example, the real-time feedback of ultrasound images can be used to assist in adjusting the position of the needle tip.

[0135] Furthermore, the safety window judgment model also includes an evaluation of the relative relationship between the posture of the puncture needle and the boundary of the target area. When the puncture path approaches the boundary of the target area, a correction signal is automatically generated to avoid damaging the target tissue.

[0136] In ultrasound-guided acupuncture operations, the "safety window" usually refers to a virtual area that is technically considered relatively safe. The puncture needle can enter the target area without damaging important tissues. The safety window judgment model analyzes ultrasound images and their relationship with the posture of the puncture needle to real-time judge whether there are potential risks and issue warning signals.

[0137] Evaluation of the needle tip posture: The posture of the puncture needle (such as angle, direction, offset, etc.) directly affects its path when entering the tissue. If the posture of the puncture needle is incorrect, it may cause accidental damage to the boundary of the target area or excessive puncture depth, thus causing unnecessary injuries.

[0138] Recognition and dynamic evaluation of the target area boundary: The boundary of the target area (such as a tumor, lesion, or specific treatment area) is a key factor in the puncture operation. The safety window judgment model can real-time identify the boundary of the target area through ultrasound image analysis and dynamically evaluate whether the puncture needle is approaching these boundaries. If the needle tip is about to enter the risk area, the system will generate a correction signal to prompt the operator to adjust the posture in time.

[0139] Significance of the correction signal: The generation of the correction signal is a key technology of this method. This signal is usually achieved by combining computer vision and machine learning algorithms with ultrasound image data. When the puncture path approaches the boundary of the target area, the system will real-time calculate the relative position between the puncture needle and the boundary of the target area and generate a correction signal according to the preset safety threshold. This signal can be an audible alarm, vibration feedback, or a prompt mark displayed on the image.

[0140] Specific measures to avoid damage: The role of the correction signal is not only to give an early warning but also to guide the operator to take measures, such as adjusting the angle of the needle or changing the depth of the puncture path. This automated feedback mechanism greatly improves the safety and accuracy of the puncture operation and reduces the potential damage to the target area and surrounding tissues.

[0141] For example, during tumor biopsy, the puncture needle needs to precisely enter the tumor without damaging the normal tissues or vital organs surrounding the tumor. The safety window judgment model can dynamically evaluate the relationship between the needle and the tumor boundary through ultrasound images and send correction signals in a timely manner to ensure the accuracy and safety of the puncture operation.

[0142] Finally, it should be noted that the above are only preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. An ultrasound-guided needle puncture precise positioning method, characterized in that, Including: S1: Ultrasonic image acquisition and processing: Use an ultrasonic image acquisition device to obtain two-dimensional or three-dimensional ultrasonic images of the target area. The ultrasonic images are used for preliminary planning of the puncture path and provide structural visual information of the target area, including the positioning of organ boundaries, tumors, and blood vessels; S2: Adopt shear wave elastography technology. By measuring the shear wave propagation speed of the tissue in the target area, calculate the Young's modulus of the target area, so as to obtain the elastic distribution information of this area. This information is used to evaluate the hardness differences of different tissues; S3: Use a sensor module to monitor the interaction force between the puncture needle and the tissue in real time, and obtain mechanical data for real-time feedback on the movement state of the needle tip; S4: Construction of time series feature vectors and dynamic modeling. According to the Young's modulus and modulus variance calculated from the shear wave elastography images, as well as the mechanical data feedback from the sensor, construct time series feature vectors to describe the tissue elasticity changes and the movement state of the needle tip during the puncture process; S5: Use a long short-term memory neural network to dynamically model and predict the tissue state during the puncture process; S6: Based on the prediction result in step S5, judge the path offset of the needle tip, and generate a correction signal through a PID controller to adjust the attitude or direction of the needle tip in real time.

2. The ultrasonic-guided acupuncture precise positioning method according to claim 1, wherein The ultrasonic image acquisition device is a two-dimensional, three-dimensional or four-dimensional ultrasonic device, which is used to provide image data of the target area at different levels.

3. The ultrasonic-guided acupuncture precise positioning method according to claim 1, wherein The Young's modulus E is calculated by the following formula: E = ρ·V s 2 ; where ρ is the density of the tissue and V s is the shear wave propagation velocity.

4. The ultrasound-guided acupuncture precise positioning method according to claim 1, wherein The mechanical data includes contact force and bending moment; The mechanical data of the contact force Fz is described by the following relationship: F z = k·x; Among them, F z is the contact force, k is the sensitivity coefficient of the sensor, and x is the displacement of the sensor; The mechanical data of the bending moment Mx is described by the following relationship: M x = τ·x; Among them, M x is the bending moment, and τ is the sensitivity coefficient of the sensor.

5. The ultrasonic-guided acupuncture precise positioning method according to claim 1, wherein The timing feature vector includes the average elastic modulus E of the current region i , the modulus variance σ E , the contact force Fz, the bending moment Mx, the spatial gradient of the elastic modulus and the time increment Δt, which are used to describe the state of tissue elasticity and needle movement; Therefore, the feature vector x at each time point t t is represented as:

6. The ultrasonic-guided acupuncture precise positioning method according to claim 5, characterized in that The long short-term memory neural network obtains the time series feature vector x of tissue elasticity and needle movement t to predict the change of the puncture path.

7. The ultrasound-guided acupuncture precise positioning method according to claim 1, wherein The control signal of the PID controller can be expressed as: where e(t) is the path error, and K p , K i and K d are the proportional gain, integral gain, and derivative gain, respectively.

8. The ultrasonic-guided acupuncture precise positioning method according to claim 1, characterized in that, The method for accurate needle puncture positioning further includes the step: S7: Use a puncture safety window judgment model to issue a path correction signal when the predicted path offset exceeds the safety range.

9. The ultrasonic-guided acupuncture precise positioning method according to claim 8, wherein The puncture safety window judgment model determines that the predicted path offset exceeds the safety range by defining a maximum offset threshold δ max as follows: |δ| ≤ δ max .

10. The ultrasonic-guided acupuncture precise positioning method according to claim 8, characterized in that The safety window judgment model also includes evaluating the relative relationship between the attitude of the puncture needle and the boundary of the target area. When the puncture path approaches the boundary of the target area, a correction signal is automatically generated to avoid damaging the target tissue.

Images