Rapid CT positioning navigation method and system

Abstract

The invention relates to a rapid CT positioning navigation method and system. The method comprises the following steps: firstly, acquiring CT image data of a target patient, and establishing a three-dimensional positioning model based on the image data; thirdly, segmenting and extracting the three-dimensional positioning model to obtain non-standard plane features; and if the geometric parameter value of the plane feature exceeds a preset range, generating a dynamic adjustment interval of the scanning angle. Then, establishing a space mapping relation between a scanning angle and an anatomical section by utilizing an algorithm, and generating an initial scanning path; and then, inputting the initial scanning path into the trained convolutional neural network model to obtain an optimized scanning path parameter. And performing tomography according to the scanning path parameter and the dynamic adjustment interval to obtain an image sequence. And finally, inputting the image sequence into a deep learning classification model to obtain a morphological feature parameter set. Dynamic tracking and accurate diagnosis of complex illness conditions are achieved, and the accuracy, efficiency and diagnosis reliability of CT scanning are improved.

Description

Technical Field

[0001] The present invention belongs to the field of medical image processing, and particularly relates to a fast CT positioning and navigation method and system. Background Art

[0002] With the development of medical image processing technology, fast CT positioning and navigation technology has emerged. With the progress of medical technology, the clinical requirements for the accuracy and efficiency of CT scans continue to rise. However, in the face of non-standard plane imaging of complex anatomical structures, traditional fixed-angle scanning methods have frequent problems. For example, when imaging special parts such as the cranial base slope and intervertebral disc, the deviation between the conventional scanning plane and the optimal display angle is large. Relying solely on the operator's experience to manually adjust, it is very easy to cause image distortion and omission of structural details, and it is difficult to completely capture the lesion area. Summary of the Invention

[0003] Based on this, in view of the above technical problems, it is necessary to provide a fast CT positioning and navigation method and system that improve the accuracy, efficiency, and diagnostic reliability of CT scans.

[0004] In a first aspect, the present application provides a fast CT positioning and navigation method, including:

[0005] Obtaining CT image data of a target patient; establishing a three-dimensional positioning model of the corresponding anatomical section of the target tissue and organ based on the image data.

[0006] Segmenting and extracting the three-dimensional positioning model according to a preset gray threshold to obtain non-standard plane features of the anatomical section; if the geometric parameter value of the non-standard plane feature exceeds the preset range, generating a dynamic adjustment interval for the scanning angle.

[0007] Using a geometric optimization path generation algorithm to establish a spatial mapping relationship between the scanning angle and the anatomical section, and generating an initial scanning path.

[0008] Inputting the initial scanning path into a trained convolutional neural network model to obtain optimized scanning path parameters.

[0009] Performing tomographic scanning according to the scanning path parameters and the dynamic adjustment interval to obtain an image sequence of the non-standard plane.

[0010] Performing noise reduction and enhancement processing on the image sequence and inputting it into a deep learning classification model to obtain a set of morphological feature parameters of the lesion area; if the set of morphological feature parameters meets the preset conditions, reconstructing the three-dimensional positioning model.

[0011] In one embodiment, segmenting and extracting the three-dimensional positioning model according to a preset gray threshold to obtain non-standard plane features of the anatomical structure, including:

[0012] Threshold segmentation of the three-dimensional positioning model is performed according to a preset grayscale threshold to obtain a grayscale segmentation result corresponding to the anatomical structure features.

[0013] Extract non-standard plane feature points according to the grayscale segmentation result to obtain the spatial distribution of the feature points.

[0014] Judge planarity based on the spatial distribution and perform feature extraction to obtain a feature data set of the non-standard plane.

[0015] Input the feature data set into the trained geometric parameter calculation model to obtain geometric parameter values; if the geometric parameter values exceed the preset range, generate a dynamic adjustment interval for the scanning angle and update the preset grayscale threshold according to the dynamic adjustment interval.

[0016] Use a clustering algorithm to classify the feature data set to obtain the main feature regions of the anatomical structure.

[0017] Analyze according to the spatial positions of the main feature regions to obtain the feature boundary information of the non-standard plane.

[0018] Input the feature boundary information into the non-standard plane feature model to obtain the non-standard plane features of the anatomical structure.

[0019] In one embodiment, a spatial mapping relationship between the scanning angle and the anatomical section is established using a geometric optimization path generation algorithm to generate an initial scanning path, including:

[0020] Calculate according to the three-dimensional positioning model and the scanning angle to obtain a spatial mapping matrix.

[0021] Use a geometric optimization algorithm to perform iterative calculations on the spatial mapping matrix to obtain an optimized spatial mapping relationship.

[0022] According to the optimized spatial mapping relationship, generate the initial parameters of the scanning path.

[0023] Generate an initial scanning path based on the initial parameters using a path planning algorithm.

[0024] Judge the accuracy of the initial scanning path based on a preset threshold. If the preset threshold is not reached, adjust the scanning angle and recalculate the spatial mapping relationship to obtain an updated initial scanning path.

[0025] In one embodiment, input the initial scanning path into the trained convolutional neural network model to obtain optimized scanning path parameters, including:

[0026] Obtain the coordinate sequence data of the initial scanning path; the coordinate sequence data includes the spatial distribution characteristics of the scanning trajectory.

[0027] Input the standardized coordinate sequence data into the trained convolutional neural network model for feature extraction to obtain path feature vectors.

[0028] Calculate path optimization parameters based on the path feature vectors to obtain dynamically adjusted parameters.

[0029] Generate optimized scan path parameters based on the preset path constraint conditions and the dynamically adjusted parameters.

[0030] In one embodiment, the path optimization parameters are calculated by the following formula:

[0031]

[0032] where P represents the path optimization parameter, α represents the global adjustment coefficient, β i represents the feature weight coefficient, v i represents the path feature vector component, γ represents the dynamic adjustment coefficient, ω j represents the local weight coefficient, d j represents the path dynamic parameter component.

[0033] In one embodiment, perform noise reduction and enhancement processing on the image sequence and input it into the deep learning classification model to obtain a set of morphological feature parameters of the lesion area, including:

[0034] Use a noise reduction algorithm to process the image sequence for noise reduction and perform enhancement processing on the denoised image to obtain a noise reduction and enhanced image sequence.

[0035] Input the noise reduction and enhanced image sequence into the trained deep learning classification model for analysis to obtain a lesion judgment result; the lesion judgment result is used to determine whether there is a lesion area in the image; if there is a lesion area, extract the morphological features of the lesion area; the morphological features include area, shape, and edge.

[0036] Construct according to the extracted morphological features to obtain a set of morphological feature parameters of the lesion area; the set of morphological feature parameters includes the area value of the lesion area, the shape type, the smoothness of the edge, the roughness quantization value, the perimeter value, the major and minor axis dimensions, and the aspect ratio parameter.

[0037] In a second aspect, the present application also provides a fast CT positioning and navigation system, which includes:

[0038] A model construction module, configured to obtain the CT image data of the target patient; establish a three-dimensional positioning model of the corresponding anatomical section of the target tissue and organ based on the image data.

[0039] A feature extraction module, configured to segment and extract a three-dimensional positioning model according to a preset gray threshold to obtain non-standard plane features of an anatomical section; if the geometric parameter values of the non-standard plane features exceed a preset range, a dynamic adjustment interval of the scanning angle is generated.

[0040] A positioning optimization module, configured to establish a spatial mapping relationship between the scanning angle and the anatomical section by using a geometric optimization path generation algorithm to generate an initial scanning path; and further configured to input the initial scanning path into a trained convolutional neural network model to obtain optimized scanning path parameters.

[0041] A lesion analysis module, configured to perform tomographic scanning according to the scanning path parameters and the dynamic adjustment interval to obtain an image sequence of a non-standard plane; perform noise reduction and enhancement processing on the image sequence and input it into a deep learning classification model to obtain a set of morphological feature parameters of the lesion area; if the set of morphological feature parameters meets a preset condition, reconstruct the three-dimensional positioning model.

[0042] In a third aspect, the present application further provides a computer device, including a memory and a processor, where the memory stores a computer program, and when the processor executes the computer program, the foregoing method is implemented.

[0043] In a fourth aspect, the present application further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the foregoing method is implemented.

[0044] For the foregoing rapid CT positioning and navigation method and system, first, CT image data of a target patient is obtained, and a three-dimensional positioning model of an anatomical section corresponding to a target tissue and organ is established based on the image data. Next, the three-dimensional positioning model is segmented and extracted according to a preset gray threshold to obtain non-standard plane features of the anatomical section; if the geometric parameter values of the non-standard plane features exceed a preset range, a dynamic adjustment interval of the scanning angle is generated. Subsequently, a spatial mapping relationship between the scanning angle and the anatomical section is established by using a geometric optimization path generation algorithm to generate an initial scanning path. Then, the initial scanning path is input into a trained convolutional neural network model to obtain optimized scanning path parameters. Then, tomographic scanning is performed according to the scanning path parameters and the dynamic adjustment interval to obtain an image sequence of a non-standard plane. Finally, noise reduction and enhancement processing are performed on the image sequence and input into a deep learning classification model to obtain a set of morphological feature parameters of the lesion area; if the set of morphological feature parameters meets a preset condition, the three-dimensional positioning model is reconstructed. It realizes dynamic tracking and accurate diagnosis of complex conditions, improves the accuracy, efficiency, and diagnostic reliability of CT scanning, and strongly promotes the progress of medical imaging diagnosis technology. Description of the Drawings

[0045] To more clearly illustrate the technical solutions in the embodiments of the present application or the related art, the following will briefly introduce the drawings required for the description of the embodiments or the related art. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0046] Figure 1 It is a flowchart of a fast CT positioning and navigation method provided by an embodiment of the present invention.

[0047] Figure 2 It is a flowchart of segmenting and extracting a three-dimensional positioning model according to a preset gray threshold to obtain non-standard plane features of anatomical structures provided by an embodiment of the present invention.

[0048] Figure 3 It is a flowchart of establishing a spatial mapping relationship between scanning angles and anatomical sections using a geometric optimization path generation algorithm to generate an initial scanning path provided by an embodiment of the present invention.

[0049] Figure 4 It is a flowchart of inputting the initial scanning path into a trained convolutional neural network model to obtain optimized scanning path parameters provided by an embodiment of the present invention.

[0050] Figure 5 It is a structural block diagram of a fast CT positioning and navigation system provided by an embodiment of the present invention. Detailed implementation manners

[0051] In order to make the purpose, technical solutions and advantages of the present application more clear, the following further details the present application in conjunction with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0052] First, the implementation environment of the embodiments of the present application is described. Exemplarily, the implementation environment includes a CT scanning device, a computer processing device, a positioning and navigation device, and a data transmission device.

[0053] In a fast CT positioning and navigation method and system, the CT scanning device is responsible for collecting image data, and transmits the large amount of image data obtained through a high-speed data transmission cable or a stable network connection to the computer processing device. The computer processing device analyzes and processes the data, and the processed data result is then transmitted to the positioning and navigation device in a wired or wireless manner through the data transmission device. During the operation of the positioning and navigation device, the position information tracked by it is also fed back to the computer processing device through the data transmission device to achieve accurate positioning and navigation. Each device works in cooperation with such a connection relationship to ensure the efficient operation of the system.

[0054] During the scanning process of the CT scanning device, the X-ray source rotates around the patient's body, emitting X-ray beams from multiple different angles to penetrate the human body. The detectors closely surround the patient, precisely capturing the attenuated ray signals after passing through the human body and converting the signals into digital signals recognizable by the computer. This device has excellent high-resolution imaging performance, capable of clearly showing the fine structures of internal organs, bones, and the specific morphology of diseased tissues in the human body. At the same time, its scanning speed is extremely fast, and a large number of tomographic image data can be obtained in a very short time.

[0055] The computer processing device is equipped with a powerful central processing unit (CPU) and a graphics processing unit (GPU), which enables it to efficiently and parallelly process the massive image data transmitted from the CT scanning device. With the help of specially developed image processing algorithms and positioning and navigation software, the computer processing device sequentially performs noise reduction processing on the image data to remove interference information, conducts segmentation operations to separate different tissues and structures, and implements registration processes to ensure the spatial consistency between images. Key anatomical structures and target position information are accurately extracted therefrom. In addition, the computer processing device has a large-capacity memory, which can ensure the rapid reading and writing of data, and is paired with high-speed storage devices for long-term proper storage of image data and processing results, providing a solid computing foundation and data support for accurate calculations and real-time feedback of positioning and navigation.

[0056] When using optical positioning technology, the positioning and navigation device captures the position information of the reflective markers pasted on the surgical instruments or the patient's body through cameras arranged at specific positions, and calculates the spatial position of the target based on the principle of triangulation. When using electromagnetic positioning technology, it utilizes the emission and reception of electromagnetic fields, and determines its position and direction based on the induction signals of the positioning sensors in the electromagnetic field. The positioning and navigation device continuously tracks the spatial position and attitude changes of the surgical instruments and the targets in the patient's body, and transmits the acquired real-time information to the computer processing device. After processing, it is presented to the doctor in an intuitive and visual form, helping the doctor achieve precise positioning during the operation, and significantly improving the accuracy and safety of the surgical operation.

[0057] The data transmission device is responsible for connecting the CT scanning device, the computer processing device, and the positioning and navigation device. It can use a high-speed wired network to ensure the rapid and accurate transmission of a large amount of image data from the CT scanning device to the computer processing device. Or it can choose a stable wireless network, which can also ensure the smooth interaction of data such as positioning information and processing results between devices in a complex medical environment. With reliable transmission performance, the data transmission device effectively avoids data delay and loss, maintains the efficient and stable operation of the entire system, and ensures the seamless connection of collaborative work between devices.

[0058] Combined with the above implementation environment, the application scenarios of the embodiments of this application are described.

[0059] A fast CT positioning and navigation method and system provided by an embodiment of the present application. Exemplarily, the fast CT positioning and navigation method and system provided by an embodiment of the present application can be applied to at least one of the following scenarios including but not limited to the following scenarios.

[0060] First, the fast CT positioning and navigation method and system are applied to the surgical navigation scenario. In the surgical navigation application scenario, the CT scanning device first scans the patient to quickly obtain high-resolution body tomographic image data, clearly showing the anatomical structure of the surgical area, including bones, blood vessels, organs, etc. The data is transmitted to the computer processing device in a high-speed and stable manner through the data transmission device. With its powerful computing power, the computer processing device runs specific algorithms to process the image data, such as three-dimensional reconstruction, to construct a three-dimensional model of the surgical area and accurately mark the key anatomical structures and target positions. During the operation, the positioning and navigation device uses optical or electromagnetic technology to track the position and posture of the surgical instrument in real time and feeds the information back to the computer processing device in real time. The computer processing device compares the instrument position with the pre-constructed model and transmits the latest navigation information to the surgical operation terminal through the data transmission device, presenting it to the surgeon in a visual form to assist the doctor in precise operation, avoiding important tissues and blood vessels, and ensuring the safe and efficient progress of the operation.

[0061] Second, the fast CT positioning and navigation method and system are applied to the disease diagnosis scenario. For disease diagnosis, the CT scanning device quickly acquires tomographic image data of a specific part of the patient's body. The data transmission device quickly transmits the data to the computer processing device, and the computer processing device uses image processing algorithms and diagnostic software to perform multi-dimensional analysis on the images, such as enhancing the contrast to highlight the lesion area and performing feature extraction to assist in identifying abnormal tissues. In this scenario, the positioning and navigation device can assist the doctor in accurately positioning the suspected lesion area, and through marking and guiding, help the doctor quickly lock the target area in the image. The computer processing device integrates the analysis results and positioning information to generate a detailed diagnostic report, providing strong evidence for the doctor to judge the disease type, development degree, etc., and improving the accuracy and efficiency of disease diagnosis.

[0062] Third, the fast CT positioning and navigation method and system are applied to the interventional therapy scenario. In the application of interventional therapy, the CT scanning device first scans the patient to provide clear anatomical basic image data for interventional therapy. The data transmission device timely transmits the data to the computer processing device, which processes the image according to the treatment requirements, plans the interventional path, simulates the treatment process, and determines key parameters such as the optimal puncture point, needle insertion angle, and depth. During the treatment process, the positioning and navigation device real-time tracks the position of the interventional instrument to ensure that it accurately reaches the target treatment area along the planned path. The computer processing device continuously compares and adjusts the instrument position information fed back by the positioning and navigation device with the pre-planned path, and conveys the adjustment information to the operating doctor through the data transmission device.

[0063] In one embodiment, as Figure 1 shown, the present application provides a fast CT positioning and navigation method, which may include the following steps:

[0064] Step S101, obtain the CT image data of the target patient; establish a three-dimensional positioning model of the corresponding anatomical section of the target tissue and organ based on the image data.

[0065] With the aid of an advanced CT scanning device, comprehensively scan the target patient to accurately obtain its CT image data. The data covers rich and detailed information inside the patient's body. Immediately afterwards, using professional medical image processing software and algorithms, based on the obtained image data, through complex mathematical operations and modeling techniques, construct a three-dimensional positioning model of the corresponding anatomical section of the target tissue and organ. This model can intuitively and stereoscopically present the morphology, position of the target part, and its spatial relationship with the surrounding tissues.

[0066] Step S102, segment and extract the three-dimensional positioning model according to a preset gray threshold to obtain the non-standard plane features of the anatomical section; if the geometric parameter values of the non-standard plane features exceed the preset range, generate a dynamic adjustment interval for the scanning angle.

[0067] Perform segmentation and extraction operations on the constructed three-dimensional positioning model according to the set gray threshold. Precisely extract the non-standard plane features of the anatomical section from the complex model data. Subsequently, for the extracted features, carefully measure their geometric parameter values and compare them with the pre-set reasonable range. Once it is found that the geometric parameter values of the non-standard plane features exceed the preset range, the system will comprehensively consider various factors and generate a dynamic adjustment interval for the scanning angle that adapts to it.

[0068] Step S103, use the geometric optimization path generation algorithm to establish the spatial mapping relationship between the scanning angle and the anatomical section, and generate the initial scanning path.

[0069] Use the geometric optimization path generation algorithm to deeply analyze the complex spatial structure characteristics of the anatomical section. Through the precise calculation of the scanning angle and various key parameters of the anatomical section, a rigorous spatial mapping relationship between the two is established. Based on the constructed spatial mapping relationship, through a series of logical operations and rule constraints, an initial scanning path that meets the requirements of medical imaging is generated.

[0070] Step S104: Input the initial scanning path into the trained convolutional neural network model to obtain the optimized scanning path parameters.

[0071] First, input the initially generated initial scanning path completely into the pre-trained convolutional neural network model. This model has undergone repeated training with a large amount of medical image data and the corresponding optimal scanning path parameters, and has powerful feature recognition and data analysis capabilities. After receiving the initial scanning path, the model uses its internal complex convolutional, pooling and other operation units to deeply mine and analyze various features in the path. Through multiple rounds of calculations and optimizations, the optimized scanning path parameters are finally output, which can significantly improve the accuracy and effectiveness of CT scanning when dealing with complex anatomical structures.

[0072] Step S105: Perform tomographic scanning according to the scanning path parameters and the dynamic adjustment interval to obtain an image sequence of the non-standard plane.

[0073] When the optimized scanning path parameters and the previously generated dynamic adjustment interval are obtained, the key information is input into the control system of the CT scanning device. The scanning device accurately adjusts its mechanical structure and scanning angle according to the received scanning path parameters, and strictly performs the scanning action according to the established path. During the scanning process, the device will flexibly correct the scanning angle in real time according to the dynamic adjustment interval to ensure that the non-standard plane area of the anatomical structure can be covered in all directions. Finally, through continuous tomographic scanning operations, an image sequence reflecting the detailed information of this non-standard plane is obtained.

[0074] Step S106: Perform noise reduction and enhancement processing on the image sequence and input it into the deep learning classification model to obtain a set of morphological feature parameters of the lesion area; if the set of morphological feature parameters meets the preset conditions, reconstruct the three-dimensional positioning model.

[0075] First, use an image algorithm to perform noise reduction and enhancement processing on the acquired image sequence, remove the noise interference generated during the scanning process, and at the same time strengthen the key details in the image to improve the image quality. Subsequently, input the processed image sequence into a pre-trained deep learning classification model, analyze the image frame by frame, accurately identify and extract the lesion area. Through further calculation, a set of morphological feature parameters covering the area, shape, edge, etc. of the lesion area is obtained. Immediately afterwards, compare this set of parameters with the preset conditions. If the preset conditions are met, it indicates that there may be deviations in the current three-dimensional positioning model or further optimization is required. At this time, reconstruct the three-dimensional positioning model to ensure the accuracy of the diagnosis.

[0076] The above-mentioned fast CT positioning and navigation method first obtains the CT image data of the target patient and establishes a three-dimensional positioning model of the corresponding anatomical section of the target tissue and organ based on this image data. Then, perform segmentation and extraction on the three-dimensional positioning model according to the preset gray threshold to obtain the non-standard plane features of the anatomical section; if the geometric parameter values of the non-standard plane features exceed the preset range, generate a dynamic adjustment interval for the scanning angle. Subsequently, use the geometric optimization path generation algorithm to establish the spatial mapping relationship between the scanning angle and the anatomical section, and generate the initial scanning path. After that, input the initial scanning path into the trained convolutional neural network model to obtain the optimized scanning path parameters. Then, perform tomographic scanning according to the scanning path parameters and the dynamic adjustment interval to obtain the image sequence of the non-standard plane. Finally, perform noise reduction and enhancement processing on the image sequence and input it into the deep learning classification model to obtain a set of morphological feature parameters of the lesion area; if the set of morphological feature parameters meets the preset conditions, reconstruct the three-dimensional positioning model. Realize the dynamic tracking and accurate diagnosis of complex conditions, improve the accuracy, efficiency and diagnostic reliability of CT scanning, and strongly promote the progress of medical imaging diagnosis technology.

[0077] In one embodiment, as Figure 2 shown, performing segmentation and extraction on the three-dimensional positioning model according to the preset gray threshold to obtain the non-standard plane features of the anatomical structure may include the following steps:

[0078] Step S201, perform threshold segmentation on the three-dimensional positioning model according to the preset gray threshold to obtain the gray segmentation result corresponding to the anatomical structure features.

[0079] Step S202, extract non-standard plane feature points according to the gray segmentation result to obtain the spatial distribution of the feature points.

[0080] Step S203, judge the planarity based on the spatial distribution and perform feature extraction to obtain the feature data set of the non-standard plane.

[0081] Step S204: Input the feature data set into the trained geometric parameter calculation model to obtain geometric parameter values; if the geometric parameter values exceed the preset range, generate a dynamic adjustment interval for the scanning angle and update the preset gray threshold according to the dynamic adjustment interval.

[0082] Step S205: Use a clustering algorithm to classify the feature data set to obtain the main feature regions of the anatomical structure.

[0083] Step S206: Analyze according to the spatial positions of the main feature regions to obtain the feature boundary information of the non-standard plane.

[0084] Step S207: Input the feature boundary information into the non-standard plane feature model to obtain the non-standard plane features of the anatomical structure.

[0085] First, perform threshold segmentation on the constructed three-dimensional positioning model according to the preset gray threshold to accurately obtain the gray segmentation result corresponding to the anatomical structure features. Based on this result, further extract the non-standard plane feature points and clarify the spatial distribution of the feature points. Subsequently, use the spatial distribution features to judge the planarity and perform feature extraction to construct the feature data set of the non-standard plane. Then, input the feature data set into the trained geometric parameter calculation model to obtain the corresponding geometric parameter values; once the geometric parameter values exceed the preset range, the system automatically generates a dynamic adjustment interval for the scanning angle and updates the preset gray threshold according to this interval. Then use the clustering algorithm to classify the feature data set to determine the main feature regions of the anatomical structure. Through in-depth analysis of the spatial positions of the main feature regions, obtain the feature boundary information of the non-standard plane. Finally, input the feature boundary information into the non-standard plane feature model to obtain the non-standard plane features of the anatomical structure.

[0086] By deeply mining the non-standard plane features, the adaptability and accuracy of CT scans in the face of complex anatomical structures are greatly improved. At the same time, the classification of the feature data set and the extraction of the boundary information help to more clearly define the lesion area, provide comprehensive and accurate references for subsequent medical diagnosis and treatment plan formulation, and significantly improve the reliability and effectiveness of medical image diagnosis.

[0087] In one embodiment, as Figure 3 shown, use the geometric optimization path generation algorithm to establish the spatial mapping relationship between the scanning angle and the anatomical section, and generate the initial scanning path, which may include the following steps:

[0088] Step S301: Calculate according to the three-dimensional positioning model and the scanning angle to obtain the spatial mapping matrix.

[0089] Step S302: Use a geometric optimization algorithm to iteratively calculate the spatial mapping matrix to obtain an optimized spatial mapping relationship.

[0090] Step S303: Generate initial parameters of the scanning path according to the optimized spatial mapping relationship.

[0091] Step S304: Generate an initial scanning path based on the initial parameters using a path planning algorithm.

[0092] Step S305: Judge the accuracy of the initial scanning path based on a preset threshold. If the preset threshold is not reached, adjust the scanning angle and recalculate the spatial mapping relationship to obtain an updated initial scanning path.

[0093] First, calculations are carried out based on the three-dimensional positioning model and the scanning angle to obtain the spatial mapping matrix. Immediately afterwards, a geometric optimization algorithm is used to perform multiple rounds of iterative calculations on this spatial mapping matrix, gradually optimizing and finally obtaining a more accurate spatial mapping relationship. Based on this optimized spatial mapping relationship, the initial parameters of the scanning path are generated. Subsequently, using a path planning algorithm, an initial scanning path is successfully generated based on the initial parameters. Finally, the accuracy of the initial scanning path is judged according to the preset threshold. Once the accuracy of the initial scanning path does not reach the preset threshold, the scanning angle will be adjusted in a timely manner, and the spatial mapping relationship will be recalculated, thereby obtaining an updated initial scanning path.

[0094] Through a complex calculation and optimization process, the degree of fit between the scanning path and the actual requirements of the anatomical structure can be effectively improved. The accurate spatial mapping relationship and the repeatedly verified scanning path greatly improve the accuracy and efficiency of CT scanning, ensuring that the scanning process can cover the target anatomical area more comprehensively and meticulously, and obtaining high-quality image data.

[0095] In one embodiment, as Figure 4 shown, inputting the initial scanning path into a trained convolutional neural network model to obtain optimized scanning path parameters may include the following steps:

[0096] Step S401: Obtain coordinate sequence data of the initial scanning path; the coordinate sequence data includes the spatial distribution characteristics of the scanning trajectory.

[0097] Step S402: Input the standardized coordinate sequence data into a trained convolutional neural network model for feature extraction to obtain a path feature vector.

[0098] Step S403: Calculate path optimization parameters according to the path feature vector to obtain dynamic adjustment parameters.

[0099] Step S404: Generate optimized scanning path parameters based on preset path constraint conditions and dynamic adjustment parameters.

[0100] First, comprehensively collect the coordinate sequence data of the initial scan path. Immediately afterwards, input the standardized coordinate sequence data into a pre-trained convolutional neural network model to obtain a path feature vector that can accurately reflect the characteristics of the scan path. Subsequently, based on the obtained path feature vector, use a professional algorithm to calculate and obtain path optimization parameters for adjusting the scan path, namely dynamic adjustment parameters. Finally, combine the preset path constraint conditions and the obtained dynamic adjustment parameters, and generate optimized scan path parameters through complex operations.

[0101] On the one hand, through in-depth mining and analysis of the coordinate sequence data of the initial scan path, the characteristics of the scan path can be accurately grasped, providing a strong basis for subsequent optimization. On the other hand, using a convolutional neural network model for feature extraction and parameter calculation significantly improves the scientificity and accuracy of scan path optimization. The optimized scan path parameters can enable the CT scanning device to better meet the actual needs of complex anatomical structures when performing scanning tasks, effectively improving the accuracy and efficiency of scanning.

[0102] In one embodiment, the path optimization parameters can be calculated through the following formula:

[0103]

[0104] Where P represents the path optimization parameter, α represents the global adjustment coefficient, β i represents the feature weight coefficient, v i represents the path feature vector component, γ represents the dynamic adjustment coefficient, ω j represents the local weight coefficient, d j represents the path dynamic parameter component.

[0105] By combining and calculating the path feature vector component and the path dynamic parameter component according to their respective corresponding coefficients, comprehensive and accurate path optimization parameters are finally obtained. This formula comprehensively considers various factors such as global and local, static features and dynamic changes, making the calculation results more reasonable and comprehensive. In practical applications, the path optimization parameters obtained by means of this formula can make the planning of the CT scan path more scientific and accurate, better adapting to complex and variable anatomical structures. This helps to improve the overall quality of CT scans, obtain clearer and more accurate medical images, provide stronger data support for doctors to diagnose diseases, and play a key role in improving the level of medical image diagnosis.

[0106] In one embodiment, the steps for performing noise reduction and enhancement processing on an image sequence and inputting it into a deep learning classification model to obtain a set of morphological feature parameters of the lesion area may include the following:

[0107] Step S501, process the image sequence using a noise reduction algorithm for denoising and enhance the denoised image to obtain a denoised and enhanced image sequence.

[0108] Step S502, input the denoised and enhanced image sequence into a trained deep learning classification model for analysis to obtain a lesion judgment result; the lesion judgment result is used to determine whether there is a lesion area in the image; if there is a lesion area, extract the morphological features of the lesion area; the morphological features include area, shape, and edge.

[0109] Step S503, construct based on the extracted morphological features to obtain a set of morphological feature parameters for the lesion area; the set of morphological feature parameters includes the area value of the lesion area, shape type, smoothness of the edge, roughness quantization value, perimeter value, major axis and minor axis dimensions, and aspect ratio parameter.

[0110] Use a professional noise reduction algorithm to perform denoising operations on the collected image sequence, remove various noise interferences in the image, and then further perform enhancement processing on the denoised image to finally obtain a high-quality denoised and enhanced image sequence. Immediately afterwards, input this denoised and enhanced image sequence into a pre-trained deep learning classification model for in-depth analysis. After the model runs, it outputs a lesion judgment result, which is used to clarify whether there is a lesion area in the image. Once it is determined that there is a lesion area, immediately extract the morphological features of the lesion area, and the features cover key aspects such as area, shape, and edge. Finally, based on the extracted morphological features, generate a set of morphological feature parameters for the lesion area. This set contains rich information such as the area value of the lesion area, specific shape type (such as circular, elliptical, irregular, etc.), smoothness or roughness quantization value of the edge, perimeter value, major axis and minor axis dimensions, and aspect ratio parameter.

[0111] Through the denoising and enhancement processing of the image sequence, the image quality is significantly improved, providing a clear and accurate data basis for subsequent lesion analysis, and helping the deep learning classification model to more accurately identify the lesion area. The detailed extraction of the morphological features of the lesion area and the construction of the parameter set can provide comprehensive and quantitative lesion information for doctors, assisting doctors to more accurately judge the development degree and characteristics of the condition, so as to formulate more scientific and effective treatment plans.

[0112] In one embodiment, as Figure 5 shown, the present application also provides a fast CT positioning and navigation system, and the system includes:

[0113] A model construction module 601, configured to obtain CT image data of a target patient; establish a three-dimensional positioning model of the corresponding anatomical section of the target tissue and organ based on the image data.

[0114] A feature extraction module 602 is configured to segment and extract a three-dimensional positioning model according to a preset gray threshold to obtain non-standard plane features of an anatomical section; if the geometric parameter values of the non-standard plane features exceed a preset range, a dynamic adjustment interval of the scanning angle is generated.

[0115] A positioning optimization module 603 is configured to establish a spatial mapping relationship between the scanning angle and the anatomical section by using a geometric optimization path generation algorithm to generate an initial scanning path; it is also configured to input the initial scanning path into a trained convolutional neural network model to obtain optimized scanning path parameters.

[0116] A lesion analysis module 604 performs tomographic scanning according to the scanning path parameters and the dynamic adjustment interval to obtain an image sequence of a non-standard plane; performs noise reduction and enhancement processing on the image sequence and inputs it into a deep learning classification model to obtain a set of morphological feature parameters of the lesion area; if the set of morphological feature parameters meets a preset condition, the three-dimensional positioning model is reconstructed.

[0117] In the above-mentioned fast CT positioning and navigation method and system, a model construction module obtains CT image data of a target patient and constructs a three-dimensional positioning model of the corresponding anatomical sections of the target tissue and organ based on the data; a feature extraction module segments and extracts the three-dimensional positioning model according to a preset gray threshold to obtain non-standard plane features of the anatomical section. If the geometric parameter values of the non-standard plane features exceed a preset range, a dynamic adjustment interval of the scanning angle is generated; a positioning optimization module uses a geometric optimization path generation algorithm to establish a spatial mapping relationship between the scanning angle and the anatomical section, thereby generating an initial scanning path. At the same time, the initial scanning path is input into a trained convolutional neural network model to obtain optimized scanning path parameters; a lesion analysis module performs tomographic scanning according to the scanning path parameters and the dynamic adjustment interval to obtain an image sequence of a non-standard plane, performs noise reduction and enhancement processing on the image sequence and inputs it into a deep learning classification model to obtain a set of morphological feature parameters of the lesion area. If the set of morphological feature parameters meets a preset condition, the three-dimensional positioning model is reconstructed. It realizes the dynamic tracking and accurate diagnosis of complex conditions, improves the accuracy, efficiency and diagnostic reliability of CT scanning, and strongly promotes the progress of medical imaging diagnosis technology.

[0118] It should be understood that although the steps in the flowcharts involved in the above-described embodiments are shown in sequence according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear indication in this article, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-described embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be executed alternately or in turn with at least a part of the steps or stages in other steps or other steps.

[0119] In one embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the steps of a fast CT positioning and navigation method and system as described above.

[0120] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps in the above-described method embodiments.

[0121] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to the partial descriptions of the method embodiments. The device embodiments described above are only illustrative. The components described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the present disclosure solution. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0122] The above-described embodiments only represent several implementation manners of the embodiments of the present application. The description is relatively specific and detailed, but it should not be construed as a limitation on the patent scope of the embodiments of the application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the embodiments of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the embodiments of the present application.

Claims

1. A fast CT positioning and navigation method, characterized in that: The method comprises: Acquire CT image data of a target patient; and establish a three-dimensional positioning model of the target tissue and organ corresponding to the anatomical section based on the image data; Segmenting and extracting the three-dimensional positioning model according to a preset grayscale threshold value to obtain non-standard plane features of the anatomical section; if the geometric parameter value of the non-standard plane feature exceeds a preset range, generating a dynamic adjustment interval of the scanning angle; Using a geometric optimization path generation algorithm to establish a spatial mapping relationship between the scanning angle and the anatomical section to generate an initial scanning path; Inputting the initial scanning path into the trained convolutional neural network model to obtain optimized scanning path parameters; Perform tomography according to the scanning path parameters and the dynamic adjustment interval to obtain an image sequence of a non-standard plane; The image sequence is subjected to noise reduction and enhancement processing and input into a deep learning classification model to obtain a set of morphological feature parameters of the lesion area; if the set of morphological feature parameters meets preset conditions, the three-dimensional positioning model is reconstructed.

2. The method according to claim 1, characterized in that: The segmentation and extraction of the three-dimensional positioning model according to a preset grayscale threshold to obtain non-standard plane features of the anatomical structure includes: Threshold segmentation of the three-dimensional positioning model according to a preset grayscale threshold value to obtain a grayscale segmentation result corresponding to the anatomical structure feature; Extracting non-standard plane feature points according to the grayscale segmentation result to obtain the spatial distribution of the feature points; Based on the spatial distribution, planarity is determined and features are extracted to obtain a feature data set of a non-standard plane; Inputting the feature data set into a trained geometric parameter calculation model to obtain a geometric parameter value; if the geometric parameter value exceeds a preset range, generating a dynamic adjustment interval of the scanning angle and updating the preset grayscale threshold according to the dynamic adjustment interval; Using a clustering algorithm to classify the feature data set to obtain the main feature areas of the anatomical structure; Analyze the spatial position of the main feature area to obtain feature boundary information of the non-standard plane; The feature boundary information is input into a non-standard plane feature model to obtain non-standard plane features of the anatomical structure.

3. The method according to claim 1, characterized in that The method of using a geometric optimization path generation algorithm to establish a spatial mapping relationship between the scanning angle and the anatomical section to generate an initial scanning path includes: Calculating according to the three-dimensional positioning model and the scanning angle to obtain a spatial mapping matrix; Iteratively calculating the spatial mapping matrix using a geometric optimization algorithm to obtain an optimized spatial mapping relationship; Generating initial parameters of a scanning path according to the optimized spatial mapping relationship; Generate an initial scanning path using a path planning algorithm based on the initial parameters; The accuracy of the initial scanning path is judged based on a preset threshold, and if the preset threshold is not reached, the scanning angle is adjusted and the spatial mapping relationship is recalculated to obtain an updated initial scanning path.

4. The method according to claim 1, characterized in that The step of inputting the initial scanning path into a trained convolutional neural network model to obtain optimized scanning path parameters includes: Acquire coordinate sequence data of the initial scanning path; the coordinate sequence data includes spatial distribution characteristics of the scanning trajectory; Inputting the standardized coordinate sequence data into a trained convolutional neural network model for feature extraction to obtain a path feature vector; Calculate path optimization parameters according to the path feature vector to obtain dynamic adjustment parameters; Based on the preset path constraint conditions and the dynamic adjustment parameters, optimized scanning path parameters are generated.

5. The method according to claim 1, characterized in that The path optimization parameter is calculated by the following formula: Among them, P represents the path optimization parameter, α represents the global adjustment coefficient, β i represents the feature weight coefficient, v i represents the path characteristic vector component, γ represents the dynamic adjustment coefficient, ω j represents the local weight coefficient, d j Represents the path dynamic parameter component.

6. The method according to claim 1, characterized in that The image sequence is subjected to noise reduction and enhancement processing and input into a deep learning classification model to obtain a set of morphological feature parameters of the lesion area, including: Processing the image sequence with a denoising algorithm to perform denoising and enhancing the denoised image to obtain a denoised and enhanced image sequence; Input the denoised and enhanced image sequence into a trained deep learning classification model for analysis to obtain a lesion judgment result; the lesion judgment result is used to judge whether there is a lesion area in the image; if the lesion area exists, extract the morphological features of the lesion area; the morphological features include area, shape, and edge; The morphological feature set of the lesion area is constructed based on the extracted morphological features, and the morphological feature parameter set includes the area value, shape type, edge smoothness, roughness quantification value, perimeter value, major and minor diameter dimensions and aspect ratio parameter of the lesion area.

7. A fast CT positioning navigation system, characterized in that: The system comprises: A model building module is used to obtain CT image data of a target patient; and to establish a three-dimensional positioning model of the target tissue and organ corresponding to the anatomical section based on the image data; A feature extraction module is used to segment and extract the three-dimensional positioning model according to a preset grayscale threshold to obtain non-standard plane features of the anatomical section; if the geometric parameter value of the non-standard plane feature exceeds a preset range, a dynamic adjustment interval of the scanning angle is generated; A positioning optimization module, used to establish a spatial mapping relationship between the scanning angle and the anatomical section using a geometric optimization path generation algorithm to generate an initial scanning path; and also used to input the initial scanning path into a trained convolutional neural network model to obtain optimized scanning path parameters; The lesion analysis module performs tomography according to the scanning path parameters and the dynamic adjustment interval to obtain an image sequence of a non-standard plane; performs noise reduction and enhancement processing on the image sequence and inputs the image sequence into a deep learning classification model to obtain a set of morphological feature parameters of the lesion area; if the set of morphological feature parameters meets preset conditions, the three-dimensional positioning model is reconstructed.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images