Low-intensity focused ultrasound neurocraniocerebral treatment system based on data twinning

Abstract

The invention belongs to the technical field of therapeutic apparatuses, and particularly relates to a low-intensity focused ultrasound neurocraniocerebral treatment system based on data twinning, which comprises the following steps: S1, binding a head band with a treatment head to the brain of a person to be treated, and twinning port data of the therapeutic apparatus to a dummy in real time to realize twinning real-time simulation and reproduction actions, s2, acquiring multi-modal data in a treatment process; and S3, constructing an interactive digital twinborn model synchronous with physical treatment. The low-intensity focused ultrasound neurocraniocerebral treatment system based on data twinning has the effect of complete mapping from a physical principle to a clinical endpoint through a digital twinning model. The Weservel equation of the sound propagation model can accurately predict the distribution of a nonlinear sound field in a complex craniocerebral environment, and the effect of indicating which sound is hit is achieved; the Pennes equation of the bio-thermal equation can dynamically predict spatio-temporal evolution of the three-dimensional temperature field, and the effect of accurately controlling temperature and time is achieved.

Description

Technical Field

[0001] This invention relates to the field of therapeutic device technology, and in particular to a low-intensity focused ultrasound neuro-cranial therapy system based on data twins. Background Technology

[0002] A cranial therapy device is a type of medical device that stimulates the brain through physical means (such as ultrasound or ultrasonic waves) to improve neurological function or treat brain diseases. It is primarily used in neuromodulation and is suitable for the rehabilitation treatment of neurological disorders such as stroke, Parkinson's disease, depression, and aphasia.

[0003] In existing cranial therapy devices, most treatments involve a one-to-one, experience-based comparison of the device's results with standard patterns on a control board. While this allows for data assessment, the data displayed by each individual's brain when stimulated by microcurrents varies, especially among different patients with different conditions. Therefore, controlling the energy emitted by the device becomes crucial. The destination and control of the emitted energy, as well as feedback on the final treatment effect, can only be known through experience or patient feedback. Consequently, the data from these treatments is often not useful for subsequent patients, leading to waste and hindering the control of treatment time and effectiveness. Summary of the Invention

[0004] Based on existing technical problems, this invention proposes a low-intensity focused ultrasound neurocranial therapy system based on data twins.

[0005] The present invention proposes a low-intensity focused ultrasound neuro-cranial treatment system based on data twins, comprising S1, binding a head-mounted device with a treatment head to the brain of the patient, and simultaneously twinning the port data of the treatment device onto the mannequin in real time to achieve real-time simulation and reproduction of the actions.

[0006] S2. Acquire multimodal data during the treatment process.

[0007] S3. Construct an interactive digital twin model that is synchronized with physical therapy.

[0008] S4. Input the real-time monitoring data into the neural network model, and use the neural network to learn the physical model and find the optimal treatment coordinates.

[0009] S5, Real-time Simulation and Visualization Layer; Enables low-latency simulation updates and 3D / 4D visualization.

[0010] Preferably, the multimodal data in S2 includes:

[0011] Medical images are used to generate anatomical geometry models specific to the patient's head.

[0012] Ultrasonic parameters include frequency, intensity, focal length, and emission timing.

[0013] When a transcranial ultrasound transducer is worn on the head of the patient, it generates treatment data: transducer focal coordinates. Coordinates of the target treatment point Low-intensity focused ultrasound power ,frequency Launch duration Quickly predict the temperature distribution across the entire computational domain. and / or damage index .

[0014] Preferably, the sound intensity range of the therapeutic device in S1 is 0.3~3w / cm².

[0015] Preferably, the S3 digital twin model includes:

[0016] The geometric model is based on MRI / CT reconstruction of the skull, brain tissue, blood vessels, and target area.

[0017] The geometric model is used to digitize the head information of the patient to simulate information that closely resembles real brain information.

[0018] Preferably, the digital twin model further includes a physical model.

[0019] The physical models include an acoustic propagation model based on the Westervelt equations, a biothermal equation based on the Pennes equations, and a thermal damage model of the simulated temperature field based on the Arrhenius integral.

[0020] The Westervelt equation determines the energy flow direction emitted by the transcranial ultrasound transducer.

[0021] The Pennes equation controls the temperature distribution over time and space of ultrasound waves after they are absorbed and converted into heat in the brain tissue.

[0022] The Arrhenius score provides feedback on the actual treatment effect on the brain.

[0023] Preferably, the Westervelt equation is used to determine the energy flow direction emitted by the transcranial ultrasound transducer:

[0024] ;

[0025] in, Sound pressure (Pa) is the position. and time The function;

[0026] The speed of sound for small signals (m / s) is the speed of sound in the linear case.

[0027] The static density of the medium (kg / m³);

[0028] This is the sound diffusion coefficient (m² / s), which is directly related to sound absorption. For many biological tissues, ;in, It is the amplitude absorption coefficient (Np / m). It is angular frequency;

[0029] This is the Laplace operator term, describing the propagation of sound waves in space.

[0030] This is the inertial / linear wave term, describing the inertial effect of sound pressure changing with time. Together with the first term, it forms the linear lossless wave equation. .

[0031] For absorption / dissipation, this is a third-order time derivative term that approximates the acoustic energy absorption caused by thermoviscous dissipation and molecular relaxation.

[0032] This is a nonlinear term describing the finite amplitude effect: due to the sound pressure itself Appearing in the square term In this case, the equation becomes nonlinear.

[0033] Preferably, the Pennes equation;

[0034]

[0035] in, Position in the organization In time Temperature (°C);

[0036] Tissue density;

[0037] For the specific heat capacity of the tissue;

[0038] For the organization's thermal conductivity;

[0039] It serves as a metabolic heat source; This refers to heat dissipation caused by blood perfusion. It serves as an external heat source.

[0040] And setting up a metabolic heat source The reason is that the capillary bed and the brain tissue reach thermal equilibrium. At this time, blood perfusion acts like a distributed radiator, so metabolic heat sources need to be considered. .

[0041] in, Blood perfusion rate is the most critical, patient-specific, and variable parameter; These are the density and specific heat capacity of blood, respectively. Arterial temperature, if tissue temperature Higher than arterial blood temperature ,but A negative value indicates that the blood is carrying away heat, which is the most important self-regulating heat dissipation mechanism of biological tissues.

[0042] Preferably, the Arrhenius integral;

[0043]

[0044] in, For position Place, when the time comes The cumulative damage integral up to date;

[0045] when =1 is usually defined as the critical threshold for irreversible coagulative necrosis;

[0046] when At that time, it is generally considered that there is no significant damage;

[0047] when At that time, it is usually considered to be complete necrosis;

[0048] Frequency factor (unit: ), representing the frequency of molecular collisions. The values ​​are extremely large, typically in the range of... Magnitude;

[0049] Activation energy (unit: J / mol) represents the energy barrier that a molecule needs to overcome to undergo denaturation. Typical values ​​are in the range of... Between J / mol;

[0050] This is the universal gas constant;

[0051] Absolute temperature, location In time Temperature; It is the integral variable.

[0052] Preferably, since the Arrhenius integral is a summation of exponential functions, the calculation needs to be optimized:

[0053] Assuming the time step is ,but:

[0054] .

[0055] Preferably, S4 includes;

[0056] S41. Using the hemispherical shape of the skull as the transducer focal coordinates The movable points are marked with latitude and longitude on the skull surface. The latitude and longitude coordinates are used as the sound wave emission points of the transcranial ultrasound transducer, and the distance between adjacent latitude and longitude points is the simulated coordinate distance for each transcranial ultrasound transducer treatment. Then, using the coordinates of the target treatment point that needs to be treated. Enter it.

[0057] S42, Transducer focal coordinates for each step This serves as the simulated coordinate for the next transcranial ultrasound transducer 11 treatment. The sound waves emitted from the simulated coordinates of each transcranial ultrasound transducer 11 are then mapped to the target treatment point coordinates. The input is used as the endpoint to simulate treatment in S3, obtaining the transducer focus coordinates. Coordinates of the target treatment point Mapping relationship between → .

[0058] S43. Map each relationship → The generated data constitutes a simulated dataset. in, The data is input into a neural network model for learning, and then the appropriate treatment coordinates are selected based on the actual set conditions.

[0059] The beneficial effects of this invention are as follows:

[0060] 1. Through digital twin models, a complete mapping from physical principles to clinical endpoints is achieved. The Westervelt equation of the sound propagation model can accurately predict the distribution of nonlinear sound fields in complex cranial environments, achieving precise targeting; the Pennes equation of the biothermal equation can dynamically predict the spatiotemporal evolution of three-dimensional temperature fields, achieving precise temperature and timing control; and the Arrhenius integral of the simulated temperature field thermal damage model can quantify complex temperature-time history into intuitive damage probabilities, achieving effect visualization.

[0061] 2. By inputting the data into the neural network model, the optimal treatment coordinates are solved by back-optimization using a physics-neural network hybrid model. Attached Figure Description

[0062] Figure 1 This is a three-dimensional view of the main unit of a low-intensity focused ultrasound neuro-cranial treatment system based on data twin proposed in this invention.

[0063] Figure 2 This is a schematic diagram of brain activity in a low-intensity focused ultrasound neuro-cranial therapy system based on data twins proposed in this invention.

[0064] Figure 3 This is a schematic diagram of a transcranial ultrasound transducer for a low-intensity focused ultrasound neuro-cranial therapy system based on data twins proposed in this invention.

[0065] Figure 4 This is a schematic diagram illustrating the reverse solution of the optimal treatment coordinates of the transcranial ultrasound transducer in a low-intensity focused ultrasound neurocranial therapy system based on data twins proposed in this invention.

[0066] In the diagram: 1. Output signal monitoring port A; 2. Output signal monitoring port B; 3. Channel A signal output port; 4. Channel B signal output port; 5. Switch and power adjustment knob; 6. Touch screen; 7. Treatment head signal input port; 8. Shielded flexible cable; 9. Transducer fixing cap; 10. Adaptive telescopic column; 11. Transcranial ultrasound transducer; 12. Skull; 13. Brain tissue; 14. Blood vessels; 15. Target area. Detailed Implementation

[0067] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0068] Reference Figures 1-4 A low-intensity focused ultrasound neuro-cranial therapy system based on data twins includes S1, which involves attaching a head-mounted device with a treatment head to the brain of the patient, and simultaneously twinning the data from the treatment device's ports onto a mannequin in real time to achieve real-time simulated and reproduced movements; the treatment device is a head-mounted device, such as... Figure 1 As shown, the main unit of the therapeutic instrument consists of an output signal monitoring port A1, an output signal monitoring port B2, an A-channel signal output port 3, a B-channel signal output port 4, a switch and power adjustment knob 5, and a touch screen display 6.

[0069] The treatment head consists of a treatment head signal input port 7, a shielded flexible cable 8, a transducer fixing cap 9, an adaptive telescopic column 10, and a transcranial ultrasound transducer 11. The transcranial ultrasound transducer 11 is worn on the head of the patient to achieve the effect of transcranial ultrasound therapy.

[0070] like Figures 1-2 As shown, the treatment device here is a transcranial low-intensity focused ultrasound (TCUS) device. It uses a piezoelectric ceramic transducer as the ultrasound source, focusing low-intensity (0.3~3 W / cm²) ultrasound waves onto the cranial region (the focal point area is less than 20 mm × 10 mm). This causes mechanical stretching of the neuronal cell membrane, altering cell membrane tension, thereby promoting the opening of mechanosensitive ion channels, leading to transmembrane ion flow, and ultimately generating action potentials, thus regulating neural activity. The electrical safety of this device is tested according to the methods specified in GB 9706.1-2020. After connecting all parts of the device, the device is fixed to the test area of ​​the simulated human body. A body temperature sensor is placed at the contact point with the simulated human body. Using the maximum ultrasound stimulation intensity mode, the simulated human body is continuously stimulated for 24 hours, and the device status and the temperature indicators of the simulated human body are continuously observed. At this time, the TUS test results should meet the electrical safety requirements of GB 9706.1, and it should have leakage current protection and overload protection functions. After 24 hours of continuous use, it should have no adverse effects on the human body, meet the safety requirements of YY 0607-2015, and meet the evaluation indicators for human safety after testing in GB / T16886.1-2022, GB / T 16886.5-2017, and GB / T 16886.10-2024.

[0071] S2. Acquire multimodal data during the treatment process. Multimodal data includes:

[0072] Medical imaging (MRI / CT) is used to generate a patient's head-specific anatomical geometry model. Ultrasound parameters, such as frequency, intensity, focal length, and emission timing, are also used.

[0073] When the transcranial ultrasound transducer 11 is worn on the head of the patient, it generates treatment data, such as the transducer focal coordinates. Coordinates of the target treatment point Low-intensity focused ultrasound power ,frequency Launch duration Quickly predict the temperature distribution across the entire computational domain. and / or damage index .

[0074] Among them, low-intensity focused ultrasound (HIFU): using acoustic and electronic methods, a low-energy ultrasound beam is focused into a narrow area in a medium, forming an ultrasound wave with an intensity capable of influencing certain physical, chemical, and biological effects. The acoustic intensity range of the therapeutic device in this application is 0.3~3 W / cm².

[0075] Sound intensity: refers to the intensity of a sound field at a given moment. The sound energy passing through a unit area perpendicular to the velocity direction of a particle at a certain point in a sound field per unit time is called the instantaneous sound intensity. In a steady-state sound field, the sound intensity is the average value of the instantaneous sound intensity over a certain time interval. Specifically, the instantaneous sound intensity is:

[0076] ;

[0077] Hourly heat source items: ;

[0078] in, The absorption coefficient represents the harmonics generated by nonlinear effects over time. Absorption becomes stronger (with increasing frequency), therefore The calculation must be based on the full sound pressure level including harmonics. ;

[0079] For time; For location; Sound pressure (Pa); The speed of sound for small signals (m / s) is the speed of sound in the linear case. The static density of the medium is (kg / m³).

[0080] Multimodal data also includes real-time monitoring data, such as temperature, sound field, and EEG / fMRI feedback, as well as patient physiological response data, such as heart rate and body movement.

[0081] S3. Construct an interactive digital twin model that is synchronized with physical therapy.

[0082] Digital twin models include:

[0083] Geometric model: Skull 12, brain tissue 13, blood vessels 14, and target area 15 reconstructed based on MRI / CT.

[0084] The geometric model is used to digitize the head information of the patient to simulate information that closely resembles the real brain.

[0085] After the transcranial ultrasound transducer 11 emits sound waves, some of them will be absorbed or blocked by the skull 12, brain tissue 13, and blood vessels 14. At this point, a physical model is needed to determine the energy flow direction, work done, and effect that finally reaches the target area 15. The physical model includes sound propagation models (such as the Westervelt equation or k-Wave simulation), biological heat equations (Pennes equation), and thermal damage models simulating temperature fields (such as Arrhenius integral).

[0086] S31. In this embodiment, due to the strong distortion effect of the skull on ultrasound waves, without an acoustic model, it is impossible to know whether the energy emitted by the transcranial ultrasound transducer 11 is truly focused on the target point (such as the ventral intermediate nucleus of the thalamus) within the skull. It may deviate, deform, or even generate multiple hot spots, leading to off-target damage or poor therapeutic effect. Therefore, to avoid the safety risks and unreliable therapeutic effects caused by relying on experience or blind testing, the acoustic propagation model uses the Westervelt equation to determine the energy flow direction emitted by the transcranial ultrasound transducer 11: ;

[0087] in, Sound pressure (Pa) is the position. and time The function;

[0088] The speed of sound for small signals (m / s) is the speed of sound in the linear case.

[0089] The static density of the medium (kg / m³);

[0090] This is the sound diffusion coefficient (m² / s), which is directly related to sound absorption. For many biological tissues, ;in, It is the amplitude absorption coefficient (Np / m). It is angular frequency;

[0091] This is the Laplace operator term. It describes the propagation (diffusion) of sound waves in space. This is the core of the classical wave equation;

[0092] This is the inertial / linear wave term. It describes the inertial effect of sound pressure changing with time. Together with the first term, it forms the linear lossless wave equation. ;

[0093] This is the absorption / dissipation term. It's a third-order time derivative term that approximates the sound energy absorption caused by thermoviscous dissipation and molecular relaxation. This is one of the main physical mechanisms by which sound energy is converted into heat. This term causes the sound wave amplitude to attenuate with propagation distance.

[0094] This is a nonlinear term, the core of Westervelt's equations. It describes the finite amplitude effect: due to the sound pressure itself... Appearing in the square term In this case, the equation becomes nonlinear.

[0095] Physically, when sound waves propagate, the compressed phase in the high-pressure section propagates faster than the rarefied phase in the low-pressure section. This is because the local sound velocity... ;in, It is the particle velocity; this causes the waveform to gradually distort as it propagates, with the sine wave "becoming steeper at the leading edge and gentler at the trailing edge," which may eventually form a shock wave. Nonlinear effects generate harmonics (such as second and third harmonics), and these high-frequency components are more easily absorbed by tissues, thereby significantly enhancing local heating efficiency and making the char area sharper.

[0096] S32. In order to understand the temperature distribution over time and space after ultrasound waves are absorbed and converted into heat in the brain tissue, and to avoid the problem of insufficient or excessive target temperature in target area 15, the Pennes equation is adopted in this embodiment for biothermal equation:

[0097]

[0098] in, Position in the organization In time Temperature (°C);

[0099] Tissue density;

[0100] For the specific heat capacity of the tissue;

[0101] For the organization's thermal conductivity;

[0102] It serves as a metabolic heat source; This refers to heat dissipation caused by blood perfusion. It serves as an external heat source.

[0103] And setting up a metabolic heat source The reason is that the capillary bed and the brain tissue reach thermal equilibrium. At this time, blood perfusion acts like a distributed radiator, so metabolic heat sources need to be considered. ;

[0104] in, Blood perfusion rate is the most critical, patient-specific, and variable parameter; These are the density and specific heat capacity of blood, respectively. Arterial temperature, if tissue temperature Higher than arterial blood temperature ,but A negative value indicates that the blood is carrying away heat, which is the most important self-regulating heat dissipation mechanism of biological tissues.

[0105] In addition, the energy deposition factor of FUS also needs to be considered: ;

[0106] in, The ultrasonic absorption coefficient of the tissue is frequency-dependent. ); The spatial distribution of ultrasonic intensity is derived from independent sound field simulation.

[0107] Finally, the Penns equation and the sound field equation are coupled unidirectionally: usually, the sound field is calculated first. This heat source is then input into the Pennes equation. This decoupling is reasonable when the temperature change is small.

[0108] S33. To prevent the sound waves emitted by the transcranial ultrasound transducer 11 from causing damage to the brain tissue, and also to better monitor or provide feedback on the actual therapeutic effect of the emitted energy on the brain, the thermal damage model in this embodiment uses the Arrhenius integral:

[0109]

[0110] in, For position Place, when the time comes The cumulative damage integral up to date;

[0111] when =1 is usually defined as the critical threshold for irreversible coagulative necrosis;

[0112] when At that time, it is generally considered that there is no significant damage;

[0113] when At that time, it is usually considered to be complete necrosis;

[0114] Frequency factor (unit: ), representing the frequency of molecular collisions. The values ​​are extremely large, typically in the range of... Magnitude;

[0115] Activation energy (unit: J / mol) represents the energy barrier that a molecule needs to overcome to undergo denaturation. Typical values ​​are in the range of... Between J / mol;

[0116] This is the universal gas constant;

[0117] Absolute temperature (unit: K). Location In time Temperature;

[0118] Note: The Kelvin temperature scale must be used because the exponent is extremely sensitive to temperature units. Conversion: ;

[0119] The variable is the integral variable, and the unit is time. .

[0120] Observe the temperature index item in, denominator It is the average thermal energy of the molecules; The ratio determines the ease or difficulty of the reaction; the larger the ratio, the slower the reaction rate. In order to prevent temperature from causing damage to the brain, the sound intensity range of the therapeutic device in this application is 0.3~3w / cm².

[0121] By setting up an Arrhenius integral model, compared to simplified models that use threshold concepts such as "equivalent minutes above 43°C" or "instantaneous damage at 54°C," the Arrhenius integral model is applicable to any temperature-time history and is a model for individualized, precise dose control. In real-time digital twin systems, the Arrhenius integral serves as an endpoint predictor and safety controller. This plays a crucial role in digital twins for craniocerebral treatment.

[0122] 1. Real-time calculation and visualization;

[0123] The system obtains the temperature field by solving the Penns equation. After that, real-time parallel computing is required. To achieve real-time computing and visualization, visualization is typically performed by... The isosurface is rendered as a semi-transparent red area, overlaid on the patient's MRI anatomical image. This visually tells the doctor "how the predicted necrotic core is growing," with the target center... Values, and adjacent key functional areas (such as the internal capsule). value.

[0124] 2. As a basis for decision-making regarding treatment termination;

[0125] Treatment goal: To target the target area (such as the Vim nucleus). Achieve a target ablation level of 1.0–4.0 (ensuring effective ablation), while also ensuring ablation of all non-target areas (especially white matter tracts and ventricular walls). Much less than 1.0 (usually <0.1).

[0126] Real-time alerts: When sensitive areas When the predicted value exceeds the safety threshold (e.g., 0.5), the system will issue an alarm, and medical staff can immediately stop treatment or adjust the parameters.

[0127] 3. Synergy with MRI temperature measurement data;

[0128] MRI thermometry provides sparse, delayed temperature data. .

[0129] Data assimilation: These data were used to correct the temperature field predicted by the Penns equation. .

[0130] Damage recalculation: based on more accurate correction Recalculate This can significantly improve the reliability of damage extent prediction.

[0131] Retrospective analysis: After treatment, a "final damage map" was calculated using complete, assimilated historical temperature data. This map serves as the gold standard for postoperative assessment and is compared with subsequent follow-up MRI (which shows the actual necrotic area) to validate and optimize model parameters.

[0132] S34. Since the Arrhenius integral is a summation of exponential functions, the calculation needs to be optimized:

[0133] Assuming the time step is ,but

[0134]

[0135] Specific computational optimizations include:

[0136] 1. Table lookup method: Index calculation is expensive. It can be calculated in advance.

[0137] For lookup tables with temperature ranges (e.g., 37°C-100°C), interpolation lookup is performed directly during real-time calculations.

[0138] 2. Parallel computing: for each spatial grid point The computation is independent, making it ideal for parallel processing on GPUs.

[0139] 3. Selective calculation: High-precision calculation is performed only in the region of interest (the target point and a few centimeters around it), while a coarse grid is used in other areas or calculation is only started when the temperature exceeds a threshold (such as 39°C).

[0140] The Arrhenius integral is located at the very end of the prediction chain in digital twins, and is also the link closest to clinical decision-making. Its specific process is as follows:

[0141] Therapeutic device parameters → Acoustic ROM → Sound field →Heat source →Thermal ROM→Temperature Field → Arrhenius integral → Damage field It can provide objective and quantitative data support for the most critical decision of "when to stop treatment," thereby achieving a controllable risk outcome.

[0142] Digital twin models offer a complete mapping from physical principles to clinical endpoints. The Westervelt equation of the sound propagation model can accurately predict the distribution of nonlinear sound fields in complex brain environments, achieving precise targeting. The Pennes equation of the biothermal equation can dynamically predict the spatiotemporal evolution of three-dimensional temperature fields, achieving precise temperature and timing control. The Arrhenius integral of the simulated temperature field thermal damage model can quantify complex temperature-time histories into intuitive damage probabilities, enabling visualization of the effects.

[0143] S4. Input the real-time monitoring data into the neural network model, use the neural network to learn the physical model and find the optimal treatment coordinates, so as to correct or predict the transducer focus coordinates. and the coordinates of the target treatment point The effect is achieved in the following way:

[0144] like Figure 4 As shown, when the sound waves emitted by the transcranial ultrasound transducer 11 pass through the skull 12, brain tissue 13, and blood vessels 14 and reach the target point in the target area 15, after processing by the Westervelt equation of the sound propagation model, the Pennes equation of the biothermal equation, and the Arrhenius integral of the simulated temperature field thermal damage model, multiple lines from the transducer focal coordinates are generated based on the cranial CT image. Reaching the target treatment point coordinates The treatment pathway can be set internally within the system:

[0145] S41, using the 12 hemispheres of the skull as the transducer focal coordinates The movable points are marked with latitude and longitude on the surface of the skull 12. The latitude and longitude coordinates are used as the sound wave emission points of the transcranial ultrasound transducer 11, and the distance between adjacent latitude and longitude points is the simulated coordinate distance for each transcranial ultrasound transducer 11 treatment. Then, using the coordinates of the target treatment point that needs to be treated. Enter it;

[0146] S42, Transducer focal coordinates for each step This serves as the simulated coordinate for the next transcranial ultrasound transducer 11 treatment. The sound waves emitted from the simulated coordinates of each transcranial ultrasound transducer 11 are then mapped to the target treatment point coordinates. The input is used as the endpoint to simulate treatment in S3, obtaining the transducer focus coordinates. Coordinates of the target treatment point Mapping relationship between → ;

[0147] S43. Map each relationship → The generated data constitutes a simulated dataset. in, The data is input into a neural network model for learning, and then the appropriate treatment coordinates are selected based on the actual set conditions.

[0148] The actual settings can be prioritized, such as by treatment effect or treatment time.

[0149] By inputting data into a neural network model and using a hybrid physics-neural network model for inverse optimization, the optimal treatment coordinates can be solved.

[0150] S5, Real-time Simulation and Visualization Layer; Enables low-latency simulation updates and 3D / 4D visualization:

[0151] Dynamic contour maps / isosurfaces of temperature and sound pressure fields. Damage area prediction overlaid on anatomical structures. Display of treatment equipment status and target area alignment.

[0152] Interaction and control interface; allows medical staff to adjust parameters and see prediction results, and even achieve closed-loop optimization control.

[0153] It enables human-computer interaction and allows for diverse treatments through the setting of specific conditions without damaging the human brain.

[0154] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A low-intensity focused ultrasound neuro-cranial therapy system based on data twins, characterized in that: Including S1, attaching a headband with a treatment head to the brain of the patient, and simultaneously twinning the port data of the treatment device to the mannequin in real time to achieve real-time simulation and reproduction of the actions. S2. Acquire multimodal data during the treatment process; S3. Construct an interactive digital twin model that is synchronized with physical therapy; S4. Input the real-time monitoring data into the neural network model, and use the neural network to learn the physical model and find the optimal treatment coordinates; S5, Real-time Simulation and Visualization Layer; Enables low-latency simulation updates and 3D / 4D visualization.

2. The low-intensity focused ultrasound neuro-cranial therapy system based on data twins according to claim 1, characterized in that: The multimodal data in S2 includes: Medical images are used to generate anatomical geometry models specific to the patient's head; Ultrasonic parameters include frequency, intensity, focal length, and emission timing; When the transcranial ultrasound transducer (11) is worn on the head of the patient, it generates treatment data: transducer focal coordinates. Coordinates of the target treatment point Low-intensity focused ultrasound power ,frequency Launch duration Quickly predict the temperature distribution across the entire computational domain. and / or damage index .

3. The low-intensity focused ultrasound neuro-cranial therapy system based on data twins according to claim 1, characterized in that: The sound intensity range of the therapeutic device in S1 is 0.3~3w / cm².

4. The low-intensity focused ultrasound neuro-cranial therapy system based on data twins according to claim 2, characterized in that: The S3 digital twin model includes: The geometric model is based on MRI / CT reconstruction of the skull (12), brain tissue (13), blood vessels (14), and target area (15). The geometric model is used to digitize the head information of the patient to simulate information that closely resembles real brain information.

5. The low-intensity focused ultrasound neuro-cranial therapy system based on data twins according to claim 4, characterized in that: The digital twin model also includes a physical model; The physical models include a sound propagation model based on the Westervelt equations, a biothermal equation based on the Pennes equations, and a simulated temperature field thermal damage model based on the Arrhenius integral. The Westervelt equation determines the energy flow direction emitted by the transcranial ultrasound transducer 11; The Pennes equation controls the temperature distribution over time and space of ultrasound waves after they are absorbed and converted into heat in the brain tissue. The Arrhenius score provides feedback on the actual treatment effect on the brain.

6. A low-intensity focused ultrasound based on data twins according to claim 5 The neurocranial therapy system is characterized by: The Westervelt equation is used to determine the energy flow direction emitted by the transcranial ultrasound transducer (11): ； in, Sound pressure (Pa) is the position. and time The function; The speed of sound for small signals (m / s) is the speed of sound in the linear case. The static density of the medium (kg / m³); This is the sound diffusion coefficient (m² / s), which is directly related to sound absorption. For many biological tissues, ;in, It is the amplitude absorption coefficient (Np / m). It is angular frequency; For the Laplace operator, it describes the propagation of sound waves in space; This is the inertial / linear wave term, describing the inertial effect of sound pressure changing with time. Together with the first term, it forms the linear lossless wave equation. ; For absorption / dissipation, this is a third-order time derivative term that approximates the sound energy absorption caused by thermoviscous dissipation and molecular relaxation. This is a nonlinear term describing the finite amplitude effect: due to the sound pressure itself Appearing in the square term In this case, the equation becomes nonlinear.

7. A low-intensity focused ultrasound neuro-cranial therapy system based on data twins according to claim 6, characterized in that: The Pennes equation; 8. Among them, Position in the organization In time Temperature (°C); Tissue density; For the specific heat capacity of the tissue; For the organization's thermal conductivity; It serves as a metabolic heat source; This refers to heat dissipation caused by blood perfusion. External heat source; And setting up a metabolic heat source The reason is that the capillary bed and the brain tissue reach thermal equilibrium. At this time, blood perfusion acts like a distributed radiator, so metabolic heat sources need to be considered. ; in, Blood perfusion rate is the most critical, patient-specific, and variable parameter; These are the density and specific heat capacity of blood, respectively. Arterial temperature, if tissue temperature Higher than arterial blood temperature ,but A negative value indicates that the blood is carrying away heat, which is the most important self-regulating heat dissipation mechanism of biological tissues.

9. A low-intensity focused ultrasound neuro-cranial therapy system based on data twins according to claim 6, characterized in that: The Arrhenius integral; 10. Among them, For position Place, when the time comes The cumulative damage integral up to date; when =1 is usually defined as the critical threshold for irreversible coagulative necrosis; when At that time, it is generally considered that there is no significant damage; when At that time, it is usually considered to be complete necrosis; Frequency factor (unit: ), representing the frequency of molecular collisions. The values ​​are extremely large, typically in the range of... Magnitude; Activation energy (unit: J / mol) represents the energy barrier that a molecule needs to overcome to undergo denaturation. Typical values ​​are in the range of... Between J / mol; This is the universal gas constant; Absolute temperature, location In time Temperature; It is the integral variable.

11. A low-intensity focused ultrasound neuro-cranial therapy system based on data twins according to claim 8, characterized in that: Since the Arrhenius integral is a summation of exponential functions, the calculation needs to be optimized: Assuming the time step is ,but: 。 12. A low-intensity focused ultrasound neuro-cranial therapy system based on data twins according to claim 6, characterized in that: S4 includes; S41, using the hemispherical shape of the skull (12) as the coordinates of the transducer focal point. The movable points are marked with latitude and longitude on the surface of the skull (12). The latitude and longitude coordinates are used as the sound wave emission points of the transcranial ultrasound transducer (11). The distance between adjacent latitude and longitude points is the simulated coordinate distance for each transcranial ultrasound transducer (11) treatment. Then, using the coordinates of the target treatment point that needs to be treated. Enter it; S42, Transducer focal coordinates for each step This serves as the simulated coordinate for the next transcranial ultrasound transducer 11 treatment. The sound waves emitted from the simulated coordinates of each transcranial ultrasound transducer 11 are then mapped to the target treatment point coordinates. The input is used as the endpoint to simulate treatment in S3, obtaining the transducer focus coordinates. Coordinates of the target treatment point Mapping relationship between → ; S43. Map each relationship → The generated data constitutes a simulated dataset. in, The data is input into a neural network model for learning, and then the appropriate treatment coordinates are selected based on the actual set conditions.

Images