Neurosurgical brain-puncture surgery robot and electronic device

Abstract

The application provides a neurosurgical brain puncture operation robot and electronic equipment, in the neurosurgical brain puncture operation robot, a control host is used for matching a brain CT image to be planned with an intracranial hematoma puncture path database constructed to determine a hematoma puncture planning path, and the mechanical arm is controlled according to the hematoma puncture planning path to drive the surgical instrument to reach the hematoma position of the brain.In the embodiment of the application, only the CT image scanned in the diagnosis process is used as the brain CT image to be planned, and no other additional image is needed, and the intracranial hematoma puncture path database constructed is matched, so that the planning of the intracranial hematoma puncture path is completed only using the brain tissue in a short time, the planning efficiency is improved, and moreover, the planning process does not need additional manual operation, and the implementation difficulty is smaller for young doctors and primary doctors with less clinical experience, and on this basis, the treatment efficiency of the patient with cerebral hemorrhage is effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer, in particular to a neurosurgery brain puncture operation robot and electronic equipment. BACKGROUND

[0002] In the neurosurgery application scenario, such as in the treatment process of cerebral hemorrhage, it is crucial to plan a reasonable hematoma puncture path. The current puncture path planning methods mainly include an iterative learning based method and a multi-constraint optimization method.

[0003] For the iterative learning based method, an initial path set is generated according to the reachable region of the needle tip, then the optimal path is selected based on the genetic simulated annealing algorithm, and finally the path is corrected according to the needle-tissue interaction model and the iterative learning algorithm. The method is mainly applied to a flexible needle. The above iterative process is directly considered on the premise that the flexible needle will have a large deformation in the puncture process. However, the cerebral hemorrhage surgery needs to puncture to the hematoma position through a rigid needle and place a drainage tube for drainage. Therefore, the existing iterative learning based method is not applicable to the planning of intracranial hematoma puncture path.

[0004] For the multi-constraint optimization method, the surgical risk is quantified as multiple constraint conditions and is given different weights. The surgical risk of different puncture paths is quantitatively analyzed by calculation to determine the optimal puncture path. However, in order to realize the quantification of multiple constraint conditions, it is necessary to combine multi-modal images. At the same time, the method also reconstructs multiple organs. As a result, the calculation is slow, time-consuming, and the efficiency of path planning is low, which is not suitable for emergency surgery such as neurosurgery. In addition, in this planning scheme, the doctor needs to have rich clinical experience for manual intervention. It is difficult for young doctors and primary doctors with less clinical experience to implement.

[0005] Therefore, there is an urgent need to provide a puncture path planning scheme that can overcome the above problems. SUMMARY

[0006] In view of the above problems, the present application is proposed, which provides a neurosurgery brain puncture operation robot and electronic equipment to at least solve the above problems.

[0007] One or more embodiments of the present application provide a neurosurgery brain puncture operation robot, which comprises a control host and a mechanical arm. The control host is used to match a brain CT image to be planned with a constructed intracranial hematoma puncture path database to determine a hematoma puncture planning path, and control the mechanical arm to drive a surgical instrument to reach the hematoma position of the brain according to the hematoma puncture planning path.

[0008] According to another aspect of the present application, there is provided an electronic device, comprising: one or more processors; and a memory storing a program;

[0009] Among them, the program includes instructions, which, when executed by the processor, enable the processor to perform the following steps: matching the brain CT image to be planned with the constructed intracranial hematoma puncture path database to determine the hematoma puncture planning path, and controlling the robotic arm to drive the surgical instrument to the hematoma position in the brain according to the hematoma puncture planning path.

[0010] In the embodiment of the present application, only the CT image scanned during the diagnosis process is required as the brain CT image to be planned, and no other additional images (MR, vascular images, etc.) are required to match it with the constructed intracranial hematoma puncture path database, thereby achieving the planning of the intracranial hematoma puncture path in a short time using only brain tissue, thereby improving the planning efficiency. Moreover, the planning process does not require additional manual operation, and for young doctors and grassroots doctors with less clinical experience, the implementation difficulty is relatively small. On this basis, the treatment efficiency of patients with cerebral hemorrhage is effectively improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0012] Figure 1 This is a structural schematic diagram of a neurosurgery brain puncture surgery robot for this application.

[0013] Figure 2 The process of performing contour evolution on the initial brain tissue model until the extracted brain tissue is obtained.

[0014] Figure 3 An exemplary flowchart for implementing registration according to an embodiment of the present application.

[0015] Figure 4 An electronic device is provided in an embodiment of the present application.

[0016] Figure 5 This is a flow chart of a method for planning a hematoma puncture path according to an embodiment of the present application. DETAILED DESCRIPTION

[0017] In order to make the person skilled in the art better understand the scheme of the present application, the technical scheme in the embodiments of the present application will be clearly and completely described below in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by the person skilled in the art without creative labor fall within the scope of protection of the present application.

[0018] In order to facilitate understanding, before the specific embodiments of the present application are described in detail, the application scenarios of the neurosurgical brain puncture surgery robot and electronic equipment of the present application are exemplarily described.

[0019] Figure 1 A structural schematic diagram of a neurosurgical brain puncture surgery robot according to the present application is shown in FIG. 1. As shown in FIG. 1, the neurosurgical brain puncture surgery robot comprises a control host 101 and a mechanical arm 102. The control host 101 is configured to match a to-be-planned brain CT image with a constructed intracranial hematoma puncture path database, to determine a hematoma puncture planning path, and to control the mechanical arm 102 to drive a surgical instrument to reach a hematoma position of the brain according to the hematoma puncture planning path. Figure 1 In the embodiments of the present application, only the CT image scanned in the diagnosis process is used as the to-be-planned brain CT image, and no other additional images (MR, blood vessel images, etc.) are needed. The to-be-planned brain CT image is matched with the constructed intracranial hematoma puncture path database, so that the planning of the intracranial hematoma puncture path is completed in a short time only using the brain tissue, the planning efficiency is improved, and moreover, the planning process does not need additional manual operation. For young doctors and primary doctors with less clinical experience, the implementation difficulty is relatively small, and on this basis, the treatment efficiency of the cerebral hemorrhage patient is effectively improved.

[0020]

[0021] Optionally, the intracranial hematoma puncture path database comprises a clinical entry point and a clinical target point corresponding to a hematoma puncture clinical path; and the control host is configured to perform the following steps to match the to-be-planned brain CT image with the constructed intracranial hematoma puncture path database, to determine the hematoma puncture planning path, comprising:

[0022] extracting a hematoma region from the to-be-planned brain CT image, and calculating a center position of the hematoma region, so as to take the center position as a target point of path planning;

[0023] transforming the target point of path planning into a standard space of the constructed brain CT image, to match with the clinical target points of all hematoma puncture clinical paths in the intracranial hematoma puncture path database;

[0024] ​Determine the clinical entry point of the matched clinical target point in the hematoma puncture clinical path, and transform it to the space of the brain CT image to be planned to obtain the entry point of the path planning.

[0025] To this end, in combination with this technical solution, the neurosurgical brain puncture surgery robot and its electronic equipment of the application bring significant technical benefits, which are specifically described as follows:

[0026] (1) Efficient and accurate path planning: By directly extracting the hematoma area from the brain CT image to be planned and calculating the center position as the target point of path planning, the application simplifies the pre-processing steps of path planning, avoids complex multi-modal image fusion and multi-organ reconstruction, thereby significantly improving the efficiency of path planning. At the same time, the target point is transformed to the standard space and matched with the clinical target point in the database to ensure the accuracy and reliability of the path planning, and based on the verified clinical path, the success rate of puncture surgery is improved.

[0027] (2) Reduce the computational burden and speed up the decision-making process: Traditional multi-constraint optimization methods often take a long time due to the involvement of a large amount of data processing and complex calculations. The application realizes fast matching by constructing and utilizing an intracranial hematoma puncture path database, greatly reducing the need for real-time calculation, so that puncture plans can be quickly developed in emergency situations, which is particularly important for neurosurgery which requires quick response.

[0028] (3) Reduce the difficulty of operation and improve the popularity: The path planning process of the application is highly automated, reducing the high dependence on the clinical experience of doctors. Even young doctors or primary doctors with less clinical experience can effectively perform puncture path planning with the assistance of the system, which not only reduces the difficulty of operation, but also promotes the application and promotion of advanced medical technology in a wider range.

[0029] (4) Combination of individualization and standardization: By matching the specific hematoma position of the patient with the standardized puncture path database, the application realizes the rapid generation of individualized treatment plans while ensuring the scientificity and safety of the treatment strategy. This combination takes into account the individual differences of patients and relies on verified clinical experience to improve the overall effectiveness of treatment.

[0030] Optionally, the control host is configured to perform the following steps to extract the hematoma area from the brain CT image to be planned:

[0031] Pretreat the brain CT image sample to obtain a pretreated brain CT image sample;

[0032] Enhance the pretreated brain CT image sample to obtain a plurality of enhanced brain CT image samples to train the constructed neural network model to obtain a hematoma extraction model;

[0033] Based on the hematoma extraction model, the hematoma area of each CT image to be planned is predicted respectively to obtain a predicted hematoma area;

[0034] The multiple predicted hematoma extracted from the multiple enhanced brain CT images are fused to obtain the extracted hematoma area.

[0035] Therefore, in combination with the optional technical solution, the technical benefits of the neurosurgical brain puncture surgery robot and its electronic device in extracting the hematoma area are as follows:

[0036] (1) By pre-processing the brain CT image samples, such as denoising, contrast correction, etc., the image quality can be significantly improved, providing clearer and more accurate input data for subsequent hematoma extraction. This helps to reduce false positives or false negatives caused by poor image quality, thereby improving the accuracy of hematoma area extraction.

[0037] (2) The pre-processed brain CT image samples are enhanced, which can highlight the features of the hematoma area, making it easier for the neural network model to learn the unique properties of the hematoma during training. Through the training of a large number of enhanced samples, the hematoma extraction model can more accurately identify and extract the hematoma area, further improving the accuracy.

[0038] (3) Multiple enhanced brain CT image samples: using multiple enhanced brain CT image samples for model training can enable the hematoma extraction model to learn more diverse hematoma shapes and features. This helps the model to maintain high recognition ability when facing different patients and different scanning conditions of brain CT images, thereby enhancing the robustness of the model.

[0039] (3) The predicted hematoma extracted from the multiple enhanced brain CT images is fused, which can comprehensively utilize the information of multiple images and reduce false positives caused by noise or artifacts in a single image. This fusion strategy further improves the robustness of hematoma area extraction.

[0040] Optionally, the pre-processed brain CT image samples are enhanced to obtain multiple enhanced brain CT image samples for training the constructed neural network model to obtain a hematoma extraction model; including: flipping the pre-processed brain CT image samples along the X, Y, Z axes of the voxel space to obtain a total of 8 enhanced brain CT image samples.

[0041] Therefore, in combination with this technical solution, the method of flipping the pre-processed brain CT image samples along the X, Y, Z axes of the voxel space to obtain a total of 8 enhanced brain CT image samples has the following technical benefits:

[0042] (1) Through the flipping operation, the original brain CT image sample is expanded into 8 different variants, which essentially greatly enriches the training data set through data augmentation techniques without increasing new patient data. This diversity helps the neural network model learn the characteristics of the hematoma at different angles and directions, thereby improving the model's recognition ability and generalization performance for hematoma.

[0043] (2) Deep learning models, especially neural networks, are prone to overfitting during training due to insufficient data or uneven data distribution. Through data augmentation, we provide the model with more diverse training samples, which helps the model better generalize during training, reduces the risk of overfitting, and enables the model to maintain good performance on unseen data.

[0044] (3) The flipping operation simulates different scanning directions and angles that may be encountered in actual clinical practice. By training the model to recognize these flipped images, the model becomes more robust to factors such as scanning direction and patient position, allowing it to accurately identify hematoma regions even when faced with images under different scanning conditions.

[0045] (4) Data augmentation generates new training samples through simple transformations based on the original data, which does not require additional data acquisition and labeling costs and can be achieved through computation alone. Therefore, it is an efficient way to utilize computational resources to improve model performance.

[0046] (5) By flipping the brain CT image in three-dimensional space, the model needs to learn and understand the shape and position changes of the hematoma in three-dimensional space during training. This helps the model better capture the spatial features of the hematoma and improve its recognition ability in three-dimensional space.

[0047] Optionally, the center position (X0, Y0, Z0) of the hematoma region is calculated based on the following formula:

[0048]

[0049]

[0050] where i, j, k are the coordinates of the voxels within the hematoma region along the X, Y, Z axes, M i,j,k is the gray value of the voxel at coordinates (i, j, k).

[0051] To this end, the condition M i,j,k > 0 in the formula ensures that only voxels within the hematoma region are calculated, effectively excluding the interference of background and non-hematoma regions, and improving the accuracy of the analysis.

[0052] Optionally, the target point of the path planning is transformed to the physical space coordinate (X c , Y c , Z c ) of the hematoma region in the constructed brain CT image standard space based on the following formula:

[0053]

[0054] wherein, A is a direction matrix of the brain CT image to be planned, represents a space matrix of the brain CT image to be planned, represents a center coordinate vector of the hematoma region, represents an origin coordinate vector of the brain CT image to be planned.

[0055] To this end, the accurate correspondence between the target point and the physical space coordinate of the hematoma region is achieved by using the accurate calculation of the direction matrix and the space matrix, thereby improving the accuracy of the puncture path.

[0056] Optionally, the target point of the path planning is transformed to the constructed brain CT image standard space, and is matched with the clinical target point of all hematoma puncture clinical paths in the intracranial hematoma puncture path database based on the following formula:

[0057]

[0058] wherein, (X db , Y db , Z db ) represents the coordinate of the clinical target point in the brain CT image standard space, (X c , Y c , Z c ) represents the coordinate of the target point of the path planning in the brain CT image standard space, and D ec represents the distance between the target point of the path planning and the clinical target point in the brain CT image standard space.

[0059] To this end, the accurate correspondence between the puncture path and the clinical target point is ensured by converting the target point coordinate of the path planning to the standard space aligned with the brain CT image.

[0060] Optionally, the control host is further configured to perform the following steps to transform the target point of the path planning to the constructed brain CT image standard space:

[0061] The standard brain tissue image and the clinical brain tissue image to be registered are registered to transform the clinical entry point and the clinical target point of the hematoma puncture clinical path of the clinical brain tissue image to be registered to the space where the standard brain tissue image is located.

[0062] To this end, in combination with this technical solution, by performing a series of steps to generate standard brain tissue images and clinical brain tissue images to be registered, and then used for puncture path planning, the following technical benefits can be brought:

[0063] (1) Unified reference framework: by obtaining standard brain CT images and taking them as brain CT image standard space, a unified reference framework is provided for puncture path planning. This helps to eliminate individual differences between different patient images, making the path planning results more consistent and comparable.

[0064] (2) Accurate brain tissue extraction: brain tissue extraction is performed on standard brain CT images and clinical brain CT images to be registered, resulting in clear and accurate brain tissue images. This helps to reduce noise and interference in the images, improving the accuracy and reliability of path planning.

[0065] (3) Improve registration accuracy: using accurately extracted standard brain tissue and clinical brain tissue images for registration can improve the accuracy and accuracy of registration. This helps to ensure that the clinical entry point and target point can be accurately transformed to the standard space, providing accurate positioning information for puncture path planning.

[0066] Optionally, the control host is also used to perform the following steps to generate the standard brain tissue images and the clinical brain tissue images to be registered:

[0067] Obtain standard brain CT images and take the image space where they are located as the brain CT image standard space;

[0068] Perform brain tissue extraction on the standard brain CT images to obtain standard brain tissue images;

[0069] Obtain clinical brain CT images to be registered, and perform brain tissue extraction on the clinical brain CT images to be registered to obtain clinical brain tissue images to be registered.

[0070] To this end, in combination with this technical solution, by performing a series of steps to generate standard brain tissue images and clinical brain tissue images to be registered, and then used for puncture path planning, the following technical benefits can be brought:

[0071] (1) Unified reference framework: by obtaining standard brain CT images and taking them as brain CT image standard space, a unified reference framework is provided for puncture path planning. This helps to eliminate individual differences between different patient images, making the path planning results more consistent and comparable.

[0072] (2) Accurate brain tissue extraction: brain tissue extraction is performed on standard brain CT images and clinical brain CT images to be registered, resulting in clear and accurate brain tissue images. This helps to reduce noise and interference in the images, improving the accuracy and reliability of path planning.

[0073] (3) Improve registration accuracy: using accurately extracted standard brain tissue and clinical brain tissue images for registration can improve the accuracy and accuracy of registration. This helps to ensure that the clinical entry point and target point can be accurately transformed into the standard space, providing accurate positioning information for the puncture path planning.

[0074] Optionally, the control host is specifically configured to perform the following steps to generate the standard brain tissue image and the clinical brain tissue image to be registered:

[0075] Obtain a clinical brain CT image library, which includes a plurality of clinical brain CT images;

[0076] Obtain a clinical brain CT image selected from the clinical brain CT image library as a standard brain CT image, and the space where the standard brain CT image is located as the brain CT image standard space;

[0077] The remaining clinical brain CT images in the clinical brain CT image library are used as the clinical brain CT images to be registered;

[0078] Brain tissue extraction is performed on the standard brain CT image to obtain a standard brain tissue image;

[0079] Brain tissue extraction is performed on the clinical brain CT image to be registered to obtain a clinical brain tissue image to be registered.

[0080] In combination with this technical solution, by performing a series of specific steps to generate a standard brain tissue image and a clinical brain tissue image to be registered, and then used for puncture path planning, the following significant technical benefits can be brought:

[0081] (1) Establish a unified standard:

[0082] A standard brain CT image is selected from the clinical brain CT image library, and the space where it is located is used as the brain CT image standard space. This provides a unified and consistent reference framework for all subsequent puncture path planning, ensuring the comparability and consistency of the planning results.

[0083] (2) Make full use of image resources: by constructing a clinical brain CT image library and selecting a standard image, existing image resources can be fully utilized. This not only improves the utilization rate of resources, but also provides rich data support for puncture path planning.

[0084] (3) Improve brain tissue extraction accuracy: brain tissue extraction is performed on the standard brain CT image and the clinical brain CT image to be registered, which can obtain more accurate and clear brain tissue images. This helps to reduce noise and interference in the image, and improves the accuracy and reliability of the path planning.

[0085] (3) Optimizing the registration process: Using accurately extracted standard brain tissue and clinical brain tissue images for registration can improve the accuracy and efficiency of registration. This helps ensure that the clinical entry point and target can be accurately transformed into the standard space, providing precise positioning information for puncture path planning.

[0086] (4) Enhance surgical safety and effectiveness: Accurate puncture path planning can ensure that important brain tissue and vascular structures are avoided during surgery, reducing surgical risks. By combining standard brain tissue images with clinical brain tissue images for path planning, the safety and success rate of surgery can be improved, resulting in better treatment effects and prognosis for patients.

[0087] Optionally, when constructing the intracranial hematoma puncture path database, the control host constructs the intracranial hematoma puncture path database based on the clinical entry points and clinical target points of the hematoma puncture clinical path corresponding to the clinical brain tissue image to be registered and the standard brain CT image in the brain CT image standard space.

[0088] Optionally, the control host uses the standard brain CT image or the clinical brain CT image to be registered as an object to be processed, and performs the following steps to extract brain tissue:

[0089] Determine the grayscale extremes of brain tissue;

[0090] Based on the brain tissue grayscale extreme value, brain tissue is extracted from the object to be processed.

[0091] Optionally, the control host is further configured to perform the following steps to determine the extreme grayscale value of brain tissue:

[0092] Converting the object to be processed into a voxel grayscale histogram of a brain image;

[0093] Determining a first voxel grayscale extreme value and a second voxel grayscale extreme value based on the grayscale histogram, wherein a sum of the second sorting ratio and the first sorting ratio is equal to 1;

[0094] Based on the first voxel grayscale extreme value and the second voxel grayscale extreme value, a brain tissue grayscale extreme value is determined.

[0095] To this end, by converting the to-be-processed object (a standard brain CT image or a clinical brain CT image to be registered) into a voxel gray histogram of the brain image, the gray distribution of different tissues in the image can be intuitively displayed. Based on the gray histogram, by setting a first sorting proportion and a second sorting proportion (the sum of which is 1), a first voxel gray extreme value (such as a minimum gray value) and a second voxel gray extreme value (such as a maximum gray value) can be determined. These two extreme values help to define the gray range of brain tissue. With the determined first voxel gray extreme value and second voxel gray extreme value, the brain tissue gray extreme value can be further accurately calculated, thereby providing an accurate gray threshold for brain tissue extraction.

[0096] Optionally, when determining the first voxel gray extreme value and the second voxel gray extreme value based on the gray histogram, the control host comprises the following steps:

[0097] Based on the gray histogram, the voxel gray values of the brain image are sorted to obtain a voxel gray value sequence;

[0098] Based on the set first sorting proportion, the voxel gray value sequence is intercepted to obtain the first voxel gray extreme value;

[0099] Based on the set second sorting proportion, the voxel gray value sequence is intercepted to obtain the second voxel gray extreme value.

[0100] To this end, by sorting the voxel gray values of the brain image and intercepting the voxel gray value sequence based on the set first sorting proportion and second sorting proportion, the first voxel gray extreme value and the second voxel gray extreme value can be accurately determined. These two extreme values jointly define the gray range of brain tissue, which helps to accurately distinguish brain tissue from non-brain tissue in the subsequent brain tissue extraction process. Since the gray distribution of brain CT images of different patients may differ, by setting adjustable first sorting proportion and second sorting proportion, different image conditions can be flexibly adapted. This flexibility ensures that brain tissue can be accurately extracted in different situations, improving the robustness of the technology.

[0101] Optionally, when determining the brain tissue gray extreme value based on the first voxel gray extreme value and the second voxel gray extreme value, the control host fuses the first voxel gray extreme value and the second voxel gray extreme value to obtain the brain tissue gray extreme value.

[0102] Optionally, the first voxel gray extreme value is smaller than the second voxel gray extreme value, and the specific size thereof can be flexibly determined according to the application scenario, which is not uniquely limited by the embodiments of the present application.

[0103] The first voxel gray scale extreme value and the second voxel gray scale extreme value are fused to obtain a brain tissue gray scale extreme value. For example, the fusion can be performed by linear interpolation. Based on the first voxel gray scale extreme value and the second voxel gray scale extreme value, the robust gray scale minimum value and the maximum value are obtained from the gray scale histogram, the interference of the skin, the bone and other tissues is excluded, and only the brain tissue is calculated, so that the calculation amount of the registration is greatly reduced, and the efficiency of the path planning is effectively improved.

[0104] Optionally, when the control host performs brain tissue extraction on the object to be processed based on the brain tissue gray scale extreme value, the following steps are included:

[0105] Based on the brain tissue gray scale extreme value, the basic parameters of the brain tissue are determined.

[0106] Based on the basic parameters of the brain tissue, an initial model of the brain tissue is constructed.

[0107] The initial model of the brain tissue is subjected to contour evolution to obtain the extracted brain tissue.

[0108] Therefore, by using the previously determined brain tissue gray scale extreme value, the control host can more accurately define the gray scale range of the brain tissue, thereby extracting the basic parameters of the brain tissue, such as size, shape, position, etc. These basic parameters provide accurate data support for subsequent construction of the initial model of the brain tissue, which helps to reduce errors in the extraction process. Based on the basic parameters of the brain tissue, the control host can construct an initial model of the brain tissue. This model is a preliminary description of the morphology and structure of the brain tissue, and provides a reliable starting point for subsequent contour evolution. The accuracy of the initial model is crucial for the final brain tissue extraction result, which can ensure that the evolution process is in the correct direction. By performing contour evolution on the initial model of the brain tissue, the control host can gradually approach the real brain tissue boundary, thereby extracting the accurate brain tissue image. The contour evolution process is a dynamic adjustment and optimization process, which can modify the initial model according to the actual situation of the image data, further improving the accuracy of the extraction.

[0109] Optionally, when the control host determines the basic parameters of the brain tissue based on the brain tissue gray scale extreme value, the following steps are included:

[0110] Each voxel on the object to be processed is traversed, and the gray scale value distribution intensity of the voxel in the X-axis, Y-axis and Z-axis directions is respectively counted based on the brain tissue gray scale extreme value, so as to determine the center coordinates of the brain tissue.

[0111] According to the total number of voxels on the object to be processed whose gray scale value is greater than the brain tissue gray scale extreme value and the physical size of a single voxel, the radius of a sphere when the center coordinates are the center of the sphere is calculated as the radius of the brain tissue.

[0112] To this end, by traversing each voxel on the object to be processed and based on the brain tissue gray value extremum statistics, the gray value distribution intensity of the voxel in the X-axis, Y-axis and Z-axis directions can be determined accurately. This method takes into account the gray value distribution characteristics of the brain tissue in three-dimensional space, avoids the deviation in a single direction, and thus improves the accuracy of brain tissue positioning. According to the total number of voxels on the object to be processed with a gray value greater than the brain tissue gray value extremum and the physical size of a single voxel, the radius of a sphere with the center coordinate as the center can be calculated as the radius of the brain tissue. This calculation method fully considers the actual gray value distribution and volume characteristics of the brain tissue, making the calculated radius more accurate and reliable. By determining the center coordinate and radius of the brain tissue, accurate initial parameters can be provided for the subsequent brain tissue extraction process, which helps to reduce the number of iterations and computational complexity in the extraction process and improve the extraction efficiency. Accurate center coordinate and radius parameters can guide the construction of the initial model of the brain tissue, making the model more consistent with the morphology and size of the actual brain tissue, which helps to converge to the true brain tissue boundary faster in the subsequent contour evolution process and improves the extraction accuracy.

[0113] Optionally, when the control host traverses each voxel on the object to be processed, respectively calculates the gray value distribution intensity of the voxel in the X-axis, Y-axis and Z-axis directions based on the brain tissue gray value extremum, and determines the center coordinate of the brain tissue, the control host comprises the following steps:

[0114] For each voxel on the object to be processed, the difference between the gray value of the voxel and the brain tissue gray value extremum is calculated as a first difference value, and the first difference value is compared with a second difference value, and the minimum value of the two is taken as the gray intensity value of the voxel, wherein the second difference value is the difference between the second voxel gray value extremum and the brain tissue gray value extremum;

[0115] The voxel gray intensity values are summed to obtain the total gray intensity distribution value of the voxels on the object to be processed;

[0116] For each voxel on the object to be processed, the gray value distribution intensity of the voxel in the X-axis direction is calculated according to the coordinate of each voxel in the X-axis direction, the physical size information of the voxel in the X-axis direction and the gray intensity value;

[0117] The gray value distribution intensities of all voxels in the X-axis direction are added to obtain the gray value distribution intensity of the voxels in the X-axis direction, and the ratio of the gray value distribution intensity to the total gray intensity distribution value is calculated as the center coordinate of the brain tissue in the X-axis direction;

[0118] According to the coordinates of each voxel in the Y-axis direction, the physical size information of the voxel in the Y-axis direction, and the gray intensity value, the gray value distribution intensity of each voxel on the object to be processed in the Y-axis direction is calculated.

[0119] The gray value distribution intensities of all voxels in the Y-axis direction are added to obtain the gray value distribution intensity of the voxel in the Y-axis direction, and the ratio of the gray value distribution intensity to the total gray intensity distribution value is calculated to obtain the center coordinate of the brain tissue in the Y-axis direction.

[0120] According to the coordinates of each voxel in the Z-axis direction, the physical size information of the voxel in the Z-axis direction, and the gray intensity value, the gray value distribution intensity of each voxel on the object to be processed in the Z-axis direction is calculated.

[0121] The gray value distribution intensities of all voxels in the Z-axis direction are added to obtain the gray value distribution intensity of the voxel in the Z-axis direction, and the ratio of the gray value distribution intensity to the total gray intensity distribution value is calculated to obtain the center coordinate of the brain tissue in the Z-axis direction.

[0122] To this end, by calculating the difference between the gray value of each voxel and the brain tissue gray extreme value, and comparing it with the second difference (the difference between the second voxel gray extreme value and the brain tissue gray extreme value), the minimum value is taken for non-negative truncation processing, the gray intensity value of the voxel can be obtained. This method considers the distribution range of the brain tissue gray value, making the calculation of the gray intensity more accurate. Not only the total gray intensity distribution value is calculated, but also the gray value distribution intensity of the voxel in the X-axis, Y-axis and Z-axis directions is calculated respectively. Through this comprehensive analysis, the center coordinates of the brain tissue in the three-dimensional space can be more accurately determined. By calculating the ratio of the gray value distribution intensity in each axis direction to the total gray intensity distribution value, the center coordinates of the brain tissue in each axis direction can be determined. This method avoids the deviation in a single direction and improves the accuracy of the center coordinate determination.

[0123] Optionally, when the control host calculates the sphere radius when the center coordinate is the center of the sphere according to the total number of voxels with a gray value greater than the brain tissue gray extreme value on the object to be processed and the physical size of a single voxel, the control host comprises the following steps:

[0124] The product of the total number of voxels with a gray value greater than the brain tissue gray extreme value on the object to be processed and the physical size of a single voxel is calculated to obtain the total volume of the voxels;

[0125] The total volume of the voxels is adjusted based on a set sphere shape factor, and the cube root obtained by the cube calculation is taken as the sphere radius.

[0126] To this end, the total volume of the voxels can be obtained by calculating the product of the total number of voxels with a gray value greater than the gray value extreme of the brain tissue on the object to be processed and the physical size of a single voxel. This method takes into account the volume characteristics of the brain tissue in the actual image data, making the calculation of the total volume of the voxels more accurate. A set of spherical shape factors is introduced to adjust the total volume of the voxels, because the actual shape of the brain tissue may not be a perfect sphere. By adjusting the shape factor, the shape of the brain tissue can be closer to the actual shape, thereby improving the accuracy of the calculation of the spherical radius. By performing an open cube calculation on the adjusted total volume of the voxels, the spherical radius can be directly obtained. This method is simple and clear, and the calculation result is accurate and reliable.

[0127] Optionally, the basic parameters of the brain tissue include the center coordinates of the brain tissue and the radius of the brain tissue, and the control host includes the following steps when constructing the initial model of the brain tissue based on the basic parameters of the brain tissue:

[0128] Obtaining the constructed plurality of polyhedrons, and dividing each polyhedron into a plurality of triangles;

[0129] For each triangle, adjusting the distance of each vertex thereof to the center coordinates to be close to half of the radius of the brain tissue, so as to take the region composed of the adjusted plurality of polyhedrons as the initial model of the brain tissue.

[0130] To this end, by adjusting the distance of each vertex of each triangle to the center coordinates to be close to half of the radius of the brain tissue, it can be ensured that the shape and size of the initial model are closer to the actual brain tissue. This method takes into account the spatial position and volume characteristics of the brain tissue, improving the accuracy of the initial model. Using a plurality of polyhedrons and dividing them into a plurality of triangles can more accurately represent the complex shape of the brain tissue. By combining these adjusted polyhedrons, a more accurate and realistic initial model of the brain tissue can be constructed.

[0131] Optionally, the control host includes the following steps when performing contour evolution on the initial model of the brain tissue to obtain the extracted brain tissue:

[0132] Based on each contour point on the initial model of the brain tissue, constructing contour evolution pulling forces along the X-axis, Y-axis and Z-axis directions;

[0133] Moving each contour point towards the edge of the sphere defined by the radius of the brain tissue, so that the contour evolution pulling forces along the X-axis, Y-axis and Z-axis directions constructed for each contour point reach equilibrium until a spherical region covering the sphere defined by the radius of the brain tissue is obtained.

[0134] To this end, by constructing contour evolution pulling forces along the X-axis, Y-axis and Z-axis directions based on each contour point on the initial model of the brain tissue,

[0135] The contour evolution pulling force in the Z-axis direction can realize accurate control of the contour evolution process, and this control method ensures that the contour evolution can proceed in the expected direction and speed, thereby improving the accuracy of brain tissue extraction. Moving each contour point towards the edge of the radius-defined sphere of the brain tissue until the contour evolution pulling force in the X-axis, Y-axis and Z-axis directions constructed for the contour point reaches equilibrium can ensure the stability and consistency of the contour evolution process, and this equilibrium state helps to avoid deviation and distortion in the contour evolution process, thereby obtaining more accurate brain tissue extraction results.

[0136] As shown in Figure 2 , the process of contour evolution for the brain tissue initial model until the extracted brain tissue is obtained, initially, the brain tissue initial model is a spherical region with the center coordinate as the center of the sphere and the radius being half of the radius of the brain tissue, and the region obtained after completing the contour evolution is not a complete spherical region, which is taken as the extracted brain tissue.

[0137] Optionally, the control host is configured to perform the following steps to register the standard brain tissue image and the clinical brain tissue image to be registered, so as to transform the clinical entry point and the clinical target point corresponding to the hematoma puncture clinical path of the clinical brain tissue image to be registered to the space where the standard brain tissue image is located:

[0138] Based on the feature points of the brain tissue anatomical structure, a plurality of registration control points and corresponding spline basis functions are constructed;

[0139] Based on all the registration control points and corresponding spline basis functions, an image registration model is constructed;

[0140] Based on the image registration model, the standard brain tissue image and the clinical brain tissue image to be registered are registered, so as to transform the clinical entry point and the clinical target point corresponding to the hematoma puncture clinical path of the clinical brain tissue image to be registered to the space where the standard brain tissue image is located.

[0141] To this end, by selecting the feature points of the brain tissue anatomical structure as the registration control points, it can be ensured that the registration process fully considers the morphological and structural features of the brain tissue, and this feature point-based registration method can more accurately reflect the corresponding relationship between the brain tissues, thereby improving the registration accuracy. Using spline basis functions to model the registration control points can more flexibly describe the deformation and displacement between the brain tissues, and the spline basis functions have good locality and smoothness, which can effectively handle the complex deformation in the brain tissue images, further improving the registration accuracy.

[0142] Figure 3 An exemplary flowchart for implementing registration by the embodiments of the present application is shown in Figure 3As shown in the figure, the standard brain tissue image is used as the fixed image, and the clinical brain tissue image to be registered is used as the moving image. The registration process is to deform the moving image to adapt to the fixed image.

[0143] See also Figure 3 , the overall process of registration is as follows:

[0144] (1) Pyramid is a pyramid module to achieve registration of different resolutions: first, coarse registration is performed on the low-resolution moving image, and then fine registration is performed on the full-resolution moving image to effectively improve the efficiency of registration.

[0145] (2) Sampler is a sampler. Under normal circumstances, it is unnecessary to loop through all voxels in the fixed image. Therefore, the fixed image is sampled by the sampler to obtain a subset, and the subset is aligned with the moving image to greatly improve the efficiency of the alignment.

[0146] (3) Interpolator is an interpolator. During the optimization process, numerical calculations are required at non-voxel positions of the moving image. Therefore, an interpolator is needed to perform grayscale interpolation to obtain the grayscale values ​​of non-voxel positions on the moving image.

[0147] (4) Metric is a similarity measure used to evaluate the similarity between the moving image and the fixed image. It serves as the cost function of the optimization problem in the registration process. The registration process is the process of minimizing the cost function. The transformation parameters are continuously optimized to minimize the metric between the moving image and the fixed image after transformation.

[0148] (5) Optimizer is an optimizer. In order to solve the optimization problem and obtain the optimal transformation parameters, an iterative optimization strategy is usually adopted. According to the optimizer calculation, it is continuously iterated until the metric reaches the minimum value or other stopping conditions are met.

[0149] (6) transform refers to the registration transformation type. The transformation type adopted by the present invention is the above-mentioned image registration model. To this end, in the above-mentioned optimizer, the control points are continuously optimized.

[0150] The detailed working process of the above-mentioned pyramid, sampler, interpolator, metric, and optimiser is prior art and will not be described in detail in the embodiments of this application. Figure 4An electronic device is provided for an embodiment of the present application. As shown in Figure 4 The electronic device includes one or more processors 401 and a memory 402 storing a program, wherein the program includes instructions that, when executed by the processor, cause the processor to perform the following steps: matching a brain CT image to be planned with a constructed intracranial hematoma puncture path database to determine a hematoma puncture planning path, and controlling the mechanical arm to carry a surgical instrument to the hematoma location of the brain according to the hematoma puncture planning path.

[0151] Figure 5 A hematoma puncture planning path planning method flowchart is provided for an embodiment of the present application. As shown in Figure 5 The method includes the following steps:

[0152] S501, obtaining a brain CT image to be planned;

[0153] S502, matching the brain CT image to be planned with a constructed intracranial hematoma puncture path database to determine a hematoma puncture planning path.

[0154] Optionally, in the hematoma puncture planning path planning method, the intracranial hematoma puncture path database includes a clinical entry point and a clinical target point corresponding to a hematoma puncture clinical path; and the matching of the brain CT image to be planned with the constructed intracranial hematoma puncture path database to determine the hematoma puncture planning path includes:

[0155] extracting a hematoma region from the brain CT image to be planned, and calculating a center position of the hematoma region to take the center position as a target point of path planning;

[0156] transforming the target point of path planning into a standard space of the constructed brain CT image to match with clinical target points of all hematoma puncture clinical paths in the intracranial hematoma puncture path database;

[0157] determining a clinical entry point of the hematoma puncture clinical path where the matched clinical target point is located, and transforming the clinical entry point into a space where the brain CT image to be planned is located to obtain an entry point of path planning.

[0158] Optionally, in the hematoma puncture planning path planning method, the extraction of the hematoma region from the brain CT image to be planned includes:

[0159] preprocessing a brain CT image sample to obtain a preprocessed brain CT image sample;

[0160] performing enhancement processing on the preprocessed brain CT image sample to obtain multiple enhanced brain CT image samples to train a constructed neural network model to obtain a hematoma extraction model;

[0161] Based on the hematoma extraction model, the hematoma region of each CT image to be planned is predicted respectively to obtain a predicted hematoma region;

[0162] The multiple predicted hematomas extracted from the multiple enhanced brain CT images are fused to obtain the extracted hematoma region.

[0163] Optionally, in the method for planning a hematoma puncture planning path, the preprocessed brain CT image sample is enhanced to obtain multiple enhanced brain CT image samples to train the constructed neural network model to obtain a hematoma extraction model, comprising: flipping the preprocessed brain CT image sample along the X, Y and Z axes of the space where the voxel is located to obtain a total of 8 enhanced brain CT image samples.

[0164] Optionally, in the method for planning a hematoma puncture planning path, the center position (Z0, Y0, Z0) of the hematoma region is calculated based on the following formula:

[0165]

[0166] Wherein, i, j, k are the coordinates of the voxel in the hematoma region along the X, Y and Z axes, M i,j,k is the gray value of the voxel at the coordinates (i, j, k).

[0167] Optionally, in the method for planning a hematoma puncture planning path, the target point of the path planning is transformed to the physical space coordinates (Z c , Y c , Z c ) of the hematoma region in the standard space of the constructed brain CT image based on the following formula:

[0168]

[0169] Wherein, is the direction matrix of the brain CT image to be planned, is the space matrix of the brain CT image to be planned, is the center coordinate vector of the hematoma region, is the origin coordinate vector of the brain CT image to be planned.

[0170] Optionally, in the method for planning a hematoma puncture planning path, the target point of the path planning is transformed to the standard space of the constructed brain CT image, and based on the following formula, the clinical target point of all hematoma puncture clinical paths in the intracranial hematoma puncture path database is matched:

[0171]

[0172] Wherein, (Zdb Y db Z db ) represents the coordinates of the clinical target point in the brain CT image standard space, (X c Y c Z b ) represents the coordinates of the path planning target point in the brain CT image standard space, D ec represents the distance between the path planning target point and the clinical target point in the brain CT image standard space.

[0173] Optionally, in the method for planning a hematoma puncture planning path, the transformation of the path planning target point into the constructed brain CT image standard space comprises:

[0174] registering the standard brain tissue image and the clinical brain tissue image to be registered to transform the clinical entry point and the clinical target point corresponding to the hematoma puncture clinical path of the clinical brain tissue image to be registered into the space in which the standard brain tissue image is located.

[0175] Optionally, in the method for planning a hematoma puncture planning path, the generation of the standard brain tissue image and the clinical brain tissue image to be registered comprises:

[0176] acquiring a standard brain CT image and taking the image space in which the standard brain CT image is located as the brain CT image standard space;

[0177] extracting brain tissue from the standard brain CT image to obtain a standard brain tissue image;

[0178] acquiring a clinical brain CT image to be registered and extracting brain tissue from the clinical brain CT image to be registered to obtain a clinical brain tissue image to be registered.

[0179] Optionally, in the method for planning a hematoma puncture planning path, the generation of the standard brain tissue image and the clinical brain tissue image to be registered comprises:

[0180] acquiring a clinical brain CT image library, the clinical brain CT image library comprising a plurality of clinical brain CT images;

[0181] acquiring a clinical brain CT image from the clinical brain CT image library as a standard brain CT image, and taking the space in which the standard brain CT image is located as the brain CT image standard space;

[0182] taking the remaining clinical brain CT images in the clinical brain CT image library as clinical brain CT images to be registered;

[0183] extracting brain tissue from the standard brain CT image to obtain a standard brain tissue image;

[0184] Brain tissue extraction is performed on the clinical brain CT image to be registered to obtain a clinical brain tissue image to be registered.

[0185] Optionally, in the planning method of the hematoma puncture planning path, the construction of the intracranial hematoma puncture path database includes: constructing the intracranial hematoma puncture path database based on the clinical entry points and clinical targets of the hematoma puncture clinical path corresponding to the clinical brain tissue image to be registered and the standard brain CT image in the standard brain CT image space.

[0186] Optionally, in the hematoma puncture planning path planning method, the standard brain CT image or the clinical brain CT image to be registered is used as the object to be processed, and the following steps are performed to extract brain tissue:

[0187] Determine the grayscale extremes of brain tissue;

[0188] Based on the brain tissue grayscale extreme value, brain tissue is extracted from the object to be processed.

[0189] Optionally, the control host is further configured to perform the following steps to determine the extreme grayscale value of brain tissue:

[0190] Converting the object to be processed into a voxel grayscale histogram of a brain image;

[0191] Determining a first voxel grayscale extreme value and a second voxel grayscale extreme value based on the grayscale histogram, wherein a sum of the second sorting ratio and the first sorting ratio is equal to 1;

[0192] Based on the first voxel grayscale extreme value and the second voxel grayscale extreme value, a brain tissue grayscale extreme value is determined.

[0193] Optionally, determining the first voxel grayscale extreme value and the second voxel grayscale extreme value based on the grayscale histogram includes:

[0194] Based on the grayscale histogram, sorting the voxel grayscale values ​​of the brain image to obtain a voxel grayscale value sequence;

[0195] intercepting the voxel grayscale value sequence based on a set first sorting ratio to obtain a first voxel grayscale extreme value;

[0196] The voxel grayscale value sequence is intercepted based on a set second sorting ratio to obtain a second voxel grayscale extreme value.

[0197] Optionally, in the planning method of the hematoma puncture planning path, determining the brain tissue grayscale extreme value based on the first voxel grayscale extreme value and the second voxel grayscale extreme value includes: fusing the first voxel grayscale extreme value and the second voxel grayscale extreme value to obtain the brain tissue grayscale extreme value.

[0198] Optionally, in the hematoma puncture planning path planning method, the brain tissue extraction of the object to be processed based on the brain tissue grayscale extreme value includes:

[0199] determining basic parameters of the brain tissue based on the brain tissue grayscale extreme value;

[0200] constructing an initial model of the brain tissue based on the basic parameters of the brain tissue;

[0201] The initial brain tissue model is subjected to contour evolution to obtain extracted brain tissue.

[0202] Optionally, in the hematoma puncture path planning method, determining basic parameters of the brain tissue based on the brain tissue grayscale extreme value includes:

[0203] Traversing each voxel on the object to be processed, and based on the grayscale extreme value of the brain tissue, respectively counting the grayscale value distribution intensity of the voxel in the X-axis, Y-axis and Z-axis directions to determine the center coordinates of the brain tissue;

[0204] According to the total number of voxels on the object to be processed whose grayscale values ​​are greater than the grayscale extreme value of the brain tissue and the physical size of a single voxel, the radius of the sphere when the central coordinate is the sphere center is calculated as the radius of the brain tissue.

[0205] Optionally, in the hematoma puncture planning path planning method, traversing each voxel on the object to be treated, and based on the grayscale extreme value of the brain tissue, respectively counting the grayscale value distribution intensity of the voxel in the X-axis, Y-axis, and Z-axis directions to determine the center coordinates of the brain tissue, includes:

[0206] For each voxel on the object to be processed, calculating the difference between its grayscale value and the grayscale extreme value of the brain tissue and taking it as a first difference, comparing it with the second difference, taking the minimum value therebetween, and performing non-negative truncation processing on the minimum value to serve as the grayscale intensity value of each voxel, wherein the second difference is the difference between the grayscale extreme value of the second voxel and the grayscale extreme value of the brain tissue;

[0207] Summing the voxel grayscale intensity values ​​to obtain an overall grayscale intensity distribution value of the voxels on the object to be processed;

[0208] For each voxel on the object to be processed, calculate the grayscale value distribution intensity of each voxel in the X-axis direction according to the coordinates of each voxel in the X-axis direction, the physical size information of the voxel in the X-axis direction, and the grayscale intensity value;

[0209] Summing up the intensity of the gray value distribution of all voxels in the X-axis direction, the intensity of the gray value distribution of the voxels in the X-axis direction is obtained, so as to calculate the ratio of the intensity of the gray value distribution to the total intensity of the gray value distribution, and the ratio is taken as the center coordinate of the brain tissue in the X-axis direction;

[0210] For each voxel on the object to be processed, the intensity of the gray value distribution of the voxel in the Y-axis direction is calculated according to the coordinates of each voxel in the Y-axis direction, the physical size information of the voxel in the Y-axis direction and the gray intensity value;

[0211] Summing up the intensity of the gray value distribution of all voxels in the Y-axis direction, the intensity of the gray value distribution of the voxels in the Y-axis direction is obtained, so as to calculate the ratio of the intensity of the gray value distribution to the total intensity of the gray value distribution, and the ratio is taken as the center coordinate of the brain tissue in the Y-axis direction;

[0212] For each voxel on the object to be processed, the intensity of the gray value distribution of the voxel in the Z-axis direction is calculated according to the coordinates of each voxel in the Z-axis direction, the physical size information of the voxel in the Z-axis direction and the gray intensity value;

[0213] Summing up the intensity of the gray value distribution of all voxels in the Z-axis direction, the intensity of the gray value distribution of the voxels in the Z-axis direction is obtained, so as to calculate the ratio of the intensity of the gray value distribution to the total intensity of the gray value distribution, and the ratio is taken as the center coordinate of the brain tissue in the Z-axis direction.

[0214] Optionally, in the method for planning a hematoma puncture planning path, when the center coordinate is the center of a sphere, the sphere radius is calculated according to the total number of voxels with a gray value greater than the gray value extreme of the brain tissue on the object to be processed and the physical size of a single voxel, which comprises:

[0215] The total volume of the voxels is obtained by calculating the product of the total number of voxels with a gray value greater than the gray value extreme of the brain tissue on the object to be processed and the physical size of a single voxel;

[0216] The total volume of the voxels is adjusted based on a set sphere shape factor, and the cube root obtained by the cube calculation is taken as the sphere radius.

[0217] Optionally, the basic parameters of the brain tissue include the center coordinate of the brain tissue and the radius of the brain tissue, and the brain tissue initial model is constructed based on the basic parameters of the brain tissue, which comprises:

[0218] A plurality of polyhedrons are obtained, and each polyhedron is divided into a plurality of triangles;

[0219] For each triangle, the distance of each vertex to the center coordinate is adjusted to be close to half of the radius of the brain tissue, and the region composed of the adjusted plurality of polyhedrons is taken as the brain tissue initial model.

[0220] Optionally, the profile evolution on the brain tissue initial model to obtain the extracted brain tissue comprises:

[0221] Based on each profile point on the brain tissue initial model, profile evolution pulling force along the X-axis, Y-axis and Z-axis directions is constructed;

[0222] Each profile point is moved towards the edge of the sphere defined by the radius of the brain tissue, so that the profile evolution pulling force along the X-axis, Y-axis and Z-axis directions constructed for each profile point reaches equilibrium until a spherical region covering the sphere defined by the radius of the brain tissue is obtained.

[0223] Optionally, the registration of the standard brain tissue image and the clinical brain tissue image to be registered to transform the clinical entry point and the clinical target point corresponding to the hematoma puncture clinical path of the clinical brain tissue image to be registered to the space under the standard brain tissue image comprises:

[0224] Based on the feature points of the brain tissue anatomy structure, a plurality of registration control points and corresponding spline basis functions are constructed;

[0225] Based on all the registration control points and corresponding spline basis functions, an image registration model is constructed;

[0226] Based on the image registration model, the standard brain tissue image and the clinical brain tissue image to be registered are registered to transform the clinical entry point and the clinical target point corresponding to the hematoma puncture clinical path of the clinical brain tissue image to be registered to the space under the standard brain tissue image.

[0227] Finally, it should be noted that: the above examples are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing examples, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A neurosurgery brain puncture robot, characterized in that: include: A control host and a robotic arm, wherein the control host is used to match the brain CT image to be planned with a constructed intracranial hematoma puncture path database to determine the hematoma puncture planning path, and control the robotic arm to drive the surgical instrument to the hematoma location in the brain according to the hematoma puncture planning path; The intracranial hematoma puncture pathway database includes clinical entry points and clinical targets corresponding to the hematoma puncture clinical pathway; The control host is used to perform the following steps to match the brain CT image to be planned with the constructed intracranial hematoma puncture path database to determine the hematoma puncture planning path: Extracting a hematoma area from the brain CT image to be planned, and calculating the center position of the hematoma area, so as to use the center position as a target point for path planning; transforming the target point of the path planning into the constructed brain CT image standard space to match the clinical target points of all hematoma puncture clinical pathways in the intracranial hematoma puncture pathway database; Determining a clinical entry point of the hematoma puncture clinical pathway where the matched clinical target is located, and transforming it to the space where the brain CT image to be planned is located, so as to obtain an entry point for pathway planning; The control host is further configured to execute the following steps to transform the target point of the path planning into the constructed brain CT image standard space: The standard brain tissue image and the clinical brain tissue image to be registered are registered to transform the clinical entry point and clinical target point of the hematoma puncture clinical pathway corresponding to the clinical brain tissue image to be registered into the space where the standard brain tissue image is located.

2. The neurosurgery brain puncture surgery robot according to claim 1, characterized in that: The control host is configured to perform the following steps to extract the hematoma area from the brain CT image to be planned: Preprocessing the brain CT image samples to obtain preprocessed brain CT image samples; performing enhancement processing on the preprocessed brain CT image samples to obtain a plurality of enhanced brain CT image samples for training the constructed neural network model to obtain a hematoma extraction model; Based on the hematoma extraction model, the hematoma area is predicted for each CT image to be planned, and the predicted hematoma area is obtained; A plurality of predicted hematomas extracted from a plurality of enhanced brain CT images are fused to obtain the extracted hematoma region.

3. The neurosurgery brain puncture surgery robot according to claim 2, characterized in that: The preprocessed brain CT image samples are enhanced to obtain multiple enhanced brain CT image samples for training the constructed neural network model to obtain a hematoma extraction model, including: flipping the preprocessed brain CT image samples along the X, Y, and Z axes of the voxel space to obtain a total of 8 enhanced brain CT image samples.

4. The neurosurgery brain puncture surgery robot according to claim 1, characterized in that: The center position of the hematoma area is calculated based on the following formula ( , , ): , , , Wherein, i, j, k are the coordinates of the voxels in the hematoma area along the X, Y, and Z axes, is the grayscale value of the voxel at coordinate (i, j, k).

5. The neurosurgery brain puncture surgery robot according to claim 1, characterized in that: Based on the following formula, the target point of the path planning is transformed into the physical space coordinates of the hematoma area in the constructed brain CT image standard space ( , , ); in, is the direction matrix of the brain CT image to be planned, represents the spatial matrix of the brain CT image to be planned, represents the center coordinate vector of the hematoma area, represents the origin coordinate vector of the brain CT image to be planned.

6. The neurosurgery brain puncture surgery robot according to claim 1, characterized in that: The control host is further configured to execute the following steps to generate the standard brain tissue image and the clinical brain tissue image to be registered: Acquire a standard brain CT image and use the image space where the image is located as the standard brain CT image space; performing brain tissue extraction on the standard brain CT image to obtain a standard brain tissue image; A clinical brain CT image to be registered is acquired, and brain tissue is extracted from the clinical brain CT image to be registered to obtain a clinical brain tissue image to be registered.

7. The neurosurgery brain puncture operation robot according to claim 6, characterized in that: The control host is configured to perform the following steps to register the standard brain tissue image and the clinical brain tissue image to be registered, so as to transform the clinical entry point and clinical target of the hematoma puncture clinical pathway corresponding to the clinical brain tissue image to be registered into the space where the standard brain tissue image is located: Based on the characteristic points of brain tissue anatomical structure, multiple registration control points and corresponding spline basis functions are constructed; Construct an image registration model based on all registration control points and corresponding spline basis functions; Based on the image registration model, the standard brain tissue image and the clinical brain tissue image to be registered are registered to transform the clinical entry point and clinical target of the hematoma puncture clinical pathway corresponding to the clinical brain tissue image to be registered into the space where the standard brain tissue image is located.

8. An electronic device, characterized in that: include: one or more processors; as well as Memory for storing programs; The program includes instructions that, when executed by the processor, perform the following steps: matching the brain CT image to be planned with a constructed intracranial hematoma puncture path database to determine a hematoma puncture planning path, and controlling the robotic arm to drive the surgical instrument to the hematoma location in the brain according to the hematoma puncture planning path; The intracranial hematoma puncture pathway database includes clinical entry points and clinical targets corresponding to the hematoma puncture clinical pathway; The processor performs the following steps to match the brain CT image to be planned with the constructed intracranial hematoma puncture path database to determine the hematoma puncture planning path: Extracting a hematoma area from the brain CT image to be planned, and calculating the center position of the hematoma area, so as to use the center position as a target point for path planning; transforming the target point of the path planning into the constructed brain CT image standard space to match the clinical target points of all hematoma puncture clinical pathways in the intracranial hematoma puncture pathway database; Determining a clinical entry point of the hematoma puncture clinical pathway where the matched clinical target is located, and transforming it to the space where the brain CT image to be planned is located, so as to obtain an entry point for pathway planning; The processor is further configured to execute the following steps to transform the target point of the path planning into the constructed brain CT image standard space: The standard brain tissue image and the clinical brain tissue image to be registered are registered to transform the clinical entry point and clinical target point of the hematoma puncture clinical pathway corresponding to the clinical brain tissue image to be registered into the space where the standard brain tissue image is located.

Images