Transcranial ultrasound treatment fixation method and helmet based on individual skull image 3D printing

Abstract

The present application belongs to the technical field of ultrasonic physiotherapy, and discloses a transcranial ultrasound treatment fixing method and a helmet based on individual skull image 3D printing. The main process is as follows: MR and CT imaging technology is used to scan the head of a patient to collect data, the system marks the brain lesion of the patient according to the collection result, and then obtains a tomographic optimized data set for training an artificial intelligence model, and finally realizes automatic generation of a solid helmet model with the help of artificial intelligence. The helmet is made by a 3D printer according to the generated model, and at the same time, the installation hole of the ultrasound treatment head is generated in the process of making the helmet, which is used to install the ultrasonic transducer, so as to ensure that the power size, frequency amplitude, incident angle of the ultrasonic wave and the distance between the ultrasonic transducer and the affected area reach the best state, forming customized treatment for different patients. In addition, the helmet model is also provided with a nose frame, which is placed on the patient's nose bridge when worn, which can prevent the helmet from displacement and inclination, thereby ensuring the accuracy and reliability of the helmet installation of the ultrasonic head.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ultrasonic physiotherapy, in particular to a transcranial ultrasound treatment fixing method and a helmet based on individual skull image 3D printing. BACKGROUND

[0002] The core advantage of ultrasonic technology is non-invasive, precise and reversible, which shows certain practical value in the treatment and rehabilitation of brain diseases, and can realize non-invasive penetration of the skull and precise targeting of the thalamic nucleus or lesion area, thereby reducing damage to the surrounding tissue.

[0003] However, in actual application, effective fixing means are often lacking to implement ultrasonic physiotherapy, especially for children patients, as their minds are not mature, and their movement often causes the position of the ultrasonic transducer emitting ultrasonic waves to deviate during a long physiotherapy process.

[0004] Therefore, the present application provides a new fixing means, which can significantly improve the effect of ultrasonic physiotherapy by combining the use of 3D printing technology. SUMMARY

[0005] Therefore, the present application provides a transcranial ultrasound treatment fixing method and a helmet based on individual skull image 3D printing to solve the problem of reduced ultrasonic physiotherapy effect caused by the lack of guidance for young children in the prior art.

[0006] In order to achieve the above-mentioned purpose, the present application provides the following technical solutions:

[0007] According to the first aspect of the present application;

[0008] The transcranial ultrasound treatment fixing method based on individual skull image 3D printing disclosed by the present application comprises:

[0009] The patient's head is scanned by CT technology or MR imaging technology to collect data, and soft tissue contrast images are obtained by different sequences to form continuous two-dimensional slice images, and head tomographic data is obtained;

[0010] The head tomographic data is stored in DICOM format, and the DICOM sequence is imported into professional medical image software, and the data is labeled manually to determine the lesion area on the two-dimensional slice image, while ensuring that the area boundary is closed, and then the tomographic optimization data set is obtained;

[0011] The outer contour line of the lesion area and the head outer contour line in the tomographic optimization data set are extracted, and are stacked and interpolated according to the layer thickness, so as to obtain the surface mesh model data of the skull and the lesion, and are saved in STL format;

[0012] The surface grid model data is corrected using Blender software, and the processing process includes smoothing the grid, filling the holes, reducing the noise points, and generating a wearing model based on the surface grid model. According to the lesion position, a plurality of installation holes are opened on the outside of the wearing model to face the lesion position, thereby forming the 3D printing model data;

[0013] The 3D printing model data is put into a printer to manufacture an entity helmet, and then an ultrasonic transceiver is installed on the installation hole of the entity helmet. The entity helmet is worn on the patient's head, and the ultrasonic transceiver is signal-connected with a data collector worn by the patient;

[0014] The ultrasonic transceiver emits a treatment wave to act on the patient's head, and at the same time of releasing the treatment wave, a detection wave is intermittently released to collect data;

[0015] After the detection wave is radiated to the lesion site and reflected to the ultrasonic transceiver, a detection signal is formed, which is introduced into the data collector to form detection data and saved.

[0016] Further, the soft tissue contrast image is obtained by different sequences, and a single sequence takes 2-10 minutes.

[0017] Further, the layer thickness of the magnetic resonance imaging scanning the patient's head is 1mm-2mm, and the resolution is 0.5mm-1mm.

[0018] Further, the 3D printing model data is put into a printer to manufacture an entity helmet, and after the patient tries it on for the first time, the position of the installation hole of the entity helmet is thinned.

[0019] Further, the number of ultrasonic transceivers is several, and the data collector transmits the detection signal to the tissue imager at regular intervals, and generates a human tissue structure image of the lesion area by analyzing the detection signal.

[0020] Further, the process of opening a plurality of installation holes on the outside of the wearing model to face the lesion position includes:

[0021] From the perspective of the wearing model, a projection line of the outer contour line of the lesion area is generated;

[0022] An equidistant line is generated based on the projection line;

[0023] The equidistant line is projected onto the outer surface of the wearing model to form an original line, and then a plurality of equidistant sections are set based on the axis of the wearing model. The equidistant sections intersect with the original line to form source points;

[0024] The source points and the center points of the lesion surface grid model are connected, and the installation hole can be set along the connection direction.

[0025] Further, the step of automatically generating fault optimization data set by using big data is further included, and specifically includes:

[0026] The patient head fault data is prepared, denoised and normalized to form picture training data, and the position height of the fault is extracted to form a lesion judgment data set and a position label.

[0027] The intelligent agent is trained, the patient of the same age group is selected, the lesion judgment data set and the continuous position label are input into the initial model, so as to learn and generate a decision-making body, the decision-making body can identify the region where the lesion is located according to the color feature and position information of the lesion judgment data set, and output a decision-making result.

[0028] The model is optimized, the judgment result is screened by manual operation, and the correct and incorrect results of the decision-making body are distinguished, the results are put into the decision-making body for iteration, and the decision-making strategy is optimized by a reward function.

[0029] Offline verification, test the model with historical data, compare the manual operation results, and if the error is less than one ten-thousandth, gradually migrate to the real system, otherwise, set the artificial takeover threshold and prepare for artificial disposal problems.

[0030] Further, the printer for manufacturing the entity helmet is a light-curing 3D printer.

[0031] Further, after the ultrasonic transceiver is installed on the mounting hole of the entity helmet, the ultrasonic transceiver is connected with the coupling agent container through the peristaltic pump and transports the coupling agent along the ultrasonic transceiver to the surface of the patient's skin.

[0032] According to the second aspect of the application,

[0033] The application discloses a helmet manufactured by a transcranial ultrasound treatment fixing method based on individual skull image 3D printing, including an entity helmet manufactured by inputting 3D printing model data into a printer, a nose frame fixedly arranged at the outer edge of the entity helmet and clamped and fixed on the patient's nasal bone.

[0034] Further, the entity helmet is provided with a strap at the bottom.

[0035] The application has the following advantages:

[0036] The application discloses a transcranial ultrasound treatment fixing method based on individual skull image 3D printing and a helmet, and the core method is that the position of a patient's lesion and the shape of the head are determined through nuclear magnetic resonance scanning, and the detection result is converted into a solid model, and through 3D printing technology, one-to-one customized design is realized for individual patients, so that the comfort of the patient wearing can be effectively improved, and the ultrasound transducer is installed on the basis to emit ultrasonic waves, and the lesion part is treated, and in the treatment process, the diagnosis and treatment of the lesion rehabilitation can also be realized synchronously, so that the method can be used as a new fixing means, and the ultrasound treatment effect is obviously improved. BRIEF DESCRIPTION OF DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced below. Obviously, the drawings in the following description are only exemplary, and for those skilled in the art, other implementation drawings can also be obtained according to the provided drawings without creative labor.

[0038] The structures, proportions, sizes and the like shown in the specification are only used to cooperate with the content disclosed in the specification, to be understood and read by those skilled in the art, and do not define the limiting conditions for the implementation of the present application, so they do not have technical significance. Any modification of structure, change of proportion relationship or adjustment of size, which does not affect the effect and purpose that can be achieved by the present application, should still fall within the scope of the technical content disclosed by the present application.

[0039] Figure 1 A flowchart of the transcranial ultrasound treatment fixing method based on individual skull image 3D printing provided by the present application is provided.

[0040] Figure 2 An automatic opening and installation hole flowchart provided by the present application is provided.

[0041] Figure 3 An automatically generated tomographic optimization data set flowchart provided by the present application is provided.

[0042] Figure 4 A solid helmet wearing side view provided by the present application is provided.

[0043] Figure 5 A solid helmet wearing perspective view provided by the present application is provided. DETAILED DESCRIPTION

[0044] The following specific embodiments illustrate the embodiments of the present application, and those skilled in the art can easily understand other advantages and effects of the present application from the disclosure. Obviously, the described embodiments are part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0045] Please refer to Figures 1-3 The disclosed individual skull image 3D printing-based transcranial ultrasound treatment fixing method adopts 3D printing technology combined with artificial intelligence technology, and can provide customized physiotherapy services for patients one by one. The technical solutions of the present application will be described in conjunction with collective embodiments, and the specific solutions include:

[0046] In a specific embodiment of the present disclosure, specifically, in the S1 step, the patient's head is scanned for data acquisition by magnetic resonance imaging (MR) or computed tomography (CT) imaging technology, and soft tissue contrast images are obtained by different sequences to generate continuous two-dimensional slice images, and head tomographic data is obtained, which is a slice image of a certain position of the patient's brain. The head tomographic data is stored in DICOM format, and the DICOM sequence is imported into professional medical image software, so that it can be operated on an electronic display. In the S2 step, the data is manually labeled, and the lesion area is obtained on the two-dimensional slice image, and the region boundary is ensured to be closed to obtain a tomographic optimized data set. Specifically, the doctor manually outlines the lesion area on the touch screen and ensures that the outlined boundary is closed, thereby obtaining the tomographic optimized data set. Continue to perform the S3 step, extract the outer contour line of the lesion area in the tomographic optimized data set and the head outer contour line, stack and interpolate according to the layer thickness, thereby obtaining the surface mesh model data of the skull and the lesion, and saving in STL format, and the STL format file is three-dimensional model data. In the S4 and S5 steps, the surface mesh model data is processed by using the Blender software, and the processing process includes smoothing the mesh, filling the holes, reducing the noise points, etc., and the wearing model is generated based on the surface mesh model. At the same time, according to the lesion position, a plurality of installation holes are opened outside the wearing model to face the lesion position, thereby forming the 3D printing model data. Finally, in the S6 and S7 steps, the 3D printing model data is put into the printer to manufacture the entity helmet, and then the ultrasonic transceiver is installed on the installation hole of the entity helmet. The entity helmet is worn on the patient's head, the ultrasonic transceiver is signal connected with the data collector worn by the patient, the image color (gray scale / texture) and position (height label) are combined for multi-modal input, at the same time, the ultrasonic transceiver emits treatment waves to act on the patient's head, and at the same time of releasing the treatment waves, the detection waves are intermittently released for data collection. On this basis, after the detection waves are radiated to the lesion site and reflected to the ultrasonic transceiver to form a detection signal, the detection signal is imported into the data collector to form detection data and saved.

[0047] In this embodiment, in the step S1, the soft tissue contrast images are obtained by different sequences, and the single sequence takes 2-10 minutes. And the layer thickness of the magnetic resonance imaging scanning the patient's head is 1mm~2mm, and the resolution is 0.5mm~1mm.

[0048] In this embodiment, when performing the S5 step, the 3D printing model data needs to be put into the printer to manufacture the entity helmet, and after the patient tries it on for the first time, the position of the installation hole of the entity helmet is thinned. In this embodiment, since the entity helmet is a customized product and cannot be reused, the printer for manufacturing the entity helmet is preferably a light-cured 3D printer to reduce the manufacturing cost.

[0049] In this embodiment, when implementing the S6 step, after the ultrasonic transceiver is mounted on the mounting hole of the entity helmet, the ultrasonic transceiver is connected with the coupling agent container through the peristaltic pump and delivers the coupling agent along the ultrasonic transceiver to the skin surface of the patient. In the S7 step, the number of ultrasonic transceivers is installed, and the data collector transmits the detection signal to the computer for analysis at regular time intervals, and the detection signal of different strengths generated due to the difference in acoustic impedance of different tissues is analyzed, so that the lesion part can be ultrasonically scanned from multiple directions to form a human tissue structure image of the lesion.

[0050] In one specific implementation of the present application, as Figure 2 and Figure 4 , according to the position of the lesion, a plurality of mounting holes are automatically opened outside the wearing model and directly opposite the position of the lesion, and the process is realized by artificial intelligence, specifically including, in the S51 step, a projection line of the outer contour line of the lesion area is generated from the overhead direction of the wearing model. In the S52 and S53 steps, equidistant lines are generated based on the projection line. The equidistant line spacing can be adjusted according to the diameter of the ultrasonic transceiver and the treatment range (such as 5mm spacing), and then the equidistant line is projected onto the outer surface of the wearing model to form the original line. Then, a plurality of equal division sections are set based on the axis of the wearing model, the equal division sections intersect with the original line to form source points. In the S54 step and the S55 step, the equidistant line is projected onto the outer surface of the wearing model to form the original line, and then a plurality of equal division sections are set based on the axis of the wearing model, and the number of equal division sections depends on the size of the lesion (usually 6-12 sections to ensure uniform coverage). The equal division sections intersect with the original line to form source points, and finally, the source points and the center points of the surface grid model of the lesion are connected, and the mounting holes can be set along the connecting direction, and the connecting direction needs to be perpendicular to the surface of the skull, so as to avoid ultrasonic refraction loss.

[0051] As Figure 3In a specific embodiment of the present disclosure, the step of automatically generating fault optimization data set by artificial intelligence using big data is further included. Specifically, in the S81 step, first, data is prepared, non-local mean (NLM) or wavelet transform is used to remove image noise, and pixel value is standardized to the range of [0, 1], so as to realize normalization. It should be noted that the patient's head DICOM fault sequence needs to contain the lesion mark, therefore, in the specific operation, the patient's head fault data needs to form picture training data, and the height coordinates (Z-axis position) of each fault layer are recorded to form the lesion judgment data set and the labeled data set (image + lesion position mask). Then in the S82 step and the S83 step, the intelligent agent is trained, the patients of the same age group are selected, the lesion judgment data set and the continuous position label are input to the initial model, the purpose is to form the spatial attention mechanism and enhance the position sensitivity, so as to learn and generate the decision-making subject, the decision-making subject is a convolutional neural network (CNN), which is mainly used for image feature extraction. In the S84 step, the decision-making subject is suitable for judging the lesion area according to the color and position of the lesion judgment data set, and outputting the judgment result obtained by decision-making. The key points of the training strategy are to select the lesion data set of the patients of the same age group to avoid the adverse effects caused by age deviation, and then output the lesion area probability graph. After obtaining the preliminary model, the model needs to be optimized, and the optimization strategy is to use the loss function: Dice Loss + CELoss to optimize the small lesion detection. Specifically, in the S85 step, first, the judgment result is screened by artificial selection, and the results of the decision-making subject are distinguished between correct and incorrect, the results are put into the decision-making subject for iteration, the decision-making strategy is optimized by the reward function, and on this basis, in the S86 step, offline verification is performed again, the model is tested by historical data, the S87 step is executed as a judgment condition, and the artificial operation result is compared, if the error is less than one ten-thousandth, the S88 step is executed to gradually migrate to the real system, otherwise, the artificial takeover threshold is set, and artificial disposal problems are prepared.

[0052] In this embodiment, the Dice coefficient measures the degree of overlap between the predicted segmentation (Pred) and the true label (GT), and is more sensitive to small lesions. The loss function form is:

[0053] ;

[0054] : the probability of the predicted pixel i in the lesion (Softmax output, range [0, 1]);

[0055] : true label (1=lesion, 0=background);

[0056] : smoothing term.

[0057] CE Loss provides pixel-level classification gradient, enhances model convergence stability:

[0058] ;

[0059] Combining the advantages of both, weighted reconciliation, get joint loss function:

[0060] ;

[0061] Wherein, the weight Usually set to 0.5 to balance Dice and CE, or adjust according to specific tasks, for example, small lesion detection can increase the weight It should be noted that the installation hole is set on the basis of automatically generating a tomographic optimization data set and generating an entity helmet model by artificial intelligence, compared with the prior art, the CT technology and MR technology can be used to intelligently generate an entity helmet model adapted to the patient, thereby effectively improving the physiotherapy efficiency and reducing the pain of the patient.

[0062] Based on the same inventive concept, the application also discloses a helmet, which is manufactured by the transcranial ultrasound treatment fixing method based on individual skull image 3D printing as above, comprising an entity helmet manufactured by using 3D printing model data into a printer, wherein, as Figure 4 And Figure 5 The number of ultrasonic transceivers installed on the entity helmet is not less than three, and a strap is further arranged at the bottom of the entity helmet to prevent the helmet from falling off. The outer edge of the entity helmet is fixedly provided with a nose frame, and the nose frame is clamped and fixed on the patient's nasal bridge bone, so that the helmet will not be displaced or inclined during wearing, and the nose frame is used as a positioning element to ensure the fit and accuracy after wearing the helmet, so that the radiation effect of the ultrasonic transceiver is accurate and reliable.

[0063] Although the application has been described in detail in the foregoing with general description and specific embodiments, some modifications or improvements can be made on the basis of the application, which is obvious to those skilled in the art. Therefore, these modifications or improvements made on the basis of not deviating from the spirit of the application, all belong to the scope of protection claimed by the application.

Claims

1. A transcranial ultrasound treatment fixation method based on 3D printing of individual skull images, characterized in that, include: Data is collected by scanning the patient's head with CT or MR imaging technology, and soft tissue contrast images are obtained through different sequences. Continuous two-dimensional slice images are formed according to the slice thickness to obtain head tomographic data. Head tomographic data is stored in DICOM format, and the DICOM sequence is imported into professional medical imaging software. The data is manually annotated to obtain the lesion area on the two-dimensional slice image and ensure that the area boundary is closed, thereby obtaining the tomographic optimized dataset. Extract the outer contour lines of the lesion region and the outer contour lines of the patient's head from the tomographic optimization dataset, perform stacking interpolation according to the layer thickness, thereby obtaining the surface mesh model data of the skull and lesion, and save it in STL format; The surface mesh model data is modified using Blender software. The process includes smoothing the mesh, filling holes, and reducing noise. A wearable model is then generated using the surface mesh model as a base. Several mounting holes are made on the outside of the wearable model, which are aligned with the lesion location, thus forming the 3D printing model data. The 3D printed model data is fed into a printer to manufacture a physical helmet. Then, an ultrasonic transceiver is installed on the mounting holes of the physical helmet. The physical helmet is worn on the patient's head, and the ultrasonic transceiver is connected to a data acquisition device worn by the patient. The process of creating several mounting holes on the outside of the wearable model, directly opposite the location of the lesion, according to the location of the lesion includes: From the top view of the wearable model, generate the projection lines of the outer contour of the lesion area; Generate equidistant lines based on the projection lines; The equidistant lines are projected onto the outer surface of the wearable model to form the original line. Then, multiple equally divided sections are set with the axis of the wearable model as a reference. The equally divided sections intersect with the original line to form the source point. Connect the source point to the center point of the mesh model on the lesion surface, and then set the mounting holes along the direction of the connecting line.

2. The transcranial ultrasound treatment and fixation method based on individual skull image 3D printing according to claim 1, characterized in that, Soft tissue contrast images were acquired using different sequences, with each sequence taking 2–10 minutes.

3. The transcranial ultrasound treatment and fixation method based on individual skull image 3D printing according to claim 1, characterized in that, The MR imaging scan of the patient's head has a slice thickness of 1mm to 2mm and a resolution of 0.5mm to 1mm.

4. The transcranial ultrasound treatment fixation method based on individual skull image 3D printing according to claim 1, characterized in that, The 3D printing model data is fed into a printer to manufacture a physical helmet, and after the patient's first fitting, the helmet is thinned at the location of the mounting holes.

5. The transcranial ultrasound treatment fixation method based on individual skull image 3D printing according to claim 1, characterized in that, The ultrasonic transceiver is of several types, and the data acquisition unit periodically transmits the detection signal to the tissue imager and analyzes the detection signal to form an image of the lesion human tissue structure.

6. The transcranial ultrasound treatment and fixation method based on individual skull image 3D printing according to claim 1, characterized in that, It also includes the process of automatically generating tomographic optimization datasets using big data, specifically including: Prepare data, reduce noise, and normalize patient head tomographic data to form image training data, and extract the position and height of the tomography to form a lesion judgment dataset and location labels; The intelligent agent is trained by selecting patients of the same age group and inputting a lesion judgment dataset and continuous location labels into an initial model to learn and generate a decision-making agent. The decision-making agent is adapted to determine the region of the lesion based on the color and location of the lesion judgment dataset and output the judgment result obtained from the decision. The model is optimized by manually filtering the judgment results and distinguishing between correct and incorrect judgments by the decision-making body. The results are then fed into the decision-making body for iteration, and the decision-making strategy is optimized through a reward function. Offline verification involves testing the model with historical data and comparing the results with those obtained through manual operation. If the error is less than one ten-thousandth, the model is gradually migrated to the real system. Otherwise, a threshold for manual intervention is set, and preparations are made for manual handling of the problem.

7. The transcranial ultrasound treatment fixation method based on individual skull image 3D printing according to claim 4, characterized in that, The printer used to manufacture the physical helmet is a photopolymer 3D printer.

8. The transcranial ultrasound treatment fixation method based on individual skull image 3D printing according to claim 4, characterized in that, After the ultrasonic transceiver is installed on the mounting hole of the physical helmet, the ultrasonic transceiver is connected to the coupling agent container through a peristaltic pump and delivers the coupling agent to the patient's skin surface along the ultrasonic transceiver.

Images