Medical image-oriented 3D modeling data processing method and system

Abstract

The invention relates to the technical field of image data processing, in particular to a medical image-oriented 3D modeling data processing method and system, and the method comprises the steps: obtaining boundary contour information in continuous image frames in medical image data, recognizing a morphological change region between adjacent frames, calculating a morphological change rate, and judging the boundary smoothness of a medical image. And obtaining data space change characteristics. According to the invention, by analyzing the boundary contour information of the continuous image frames and combining the pixel texture and contrast characteristics, the morphological change area is identified, and the analysis capability of the image data space structure is improved. The key areas are screened according to the morphological change rate and the fluctuation intensity, dynamic optimal distribution of computing resources is achieved, and the processing precision of the complex morphological areas is improved. The task scheduling strategy adjusts the calculation batch and load according to the data characteristics, optimizes the utilization of calculation resources, and enhances the processing efficiency. The hierarchical storage and dynamic adjustment of the cache data improve the data access rate and accelerate the data loading and transmission.

Description

Technical Field

[0001] The present invention relates to the technical field of image data processing, and in particular to a 3D modeling data processing method and system for medical images. Background Art

[0002] The field of image data processing encompasses the research and application of a range of technologies for image acquisition, analysis, processing, storage, and transmission. These technologies primarily involve computer vision, image recognition, 3D modeling, and image enhancement. The core goal is to improve image quality and extract valuable information through image processing for practical applications. Core image data processing technologies include image segmentation, feature extraction, 3D reconstruction, image registration, and image compression, and are widely used in medical imaging, autonomous driving, security monitoring, virtual reality, and other fields.

[0003] 3D modeling data processing methods for medical imaging refer to technologies for three-dimensional reconstruction and processing of medical imaging data. These methods primarily focus on efficiently processing medical imaging data, specifically by enabling doctors to make accurate diagnoses through 3D modeling of medical imaging data such as CT and MRI. These methods address the challenge of extracting 3D information from medical images, involving techniques such as image data preprocessing, 3D reconstruction algorithms, and data optimization. Through specific image data processing methods, 2D image data is converted into visual 3D models to aid medical analysis.

[0004] Three-dimensional modeling of medical images mainly relies on traditional image reconstruction algorithms, which ignore the change information between consecutive image frames, resulting in limited ability to capture three-dimensional structural details. Morphological feature extraction methods are mostly based on single-frame data and lack dynamic analysis of morphological changes, which affects the accuracy of complex structure characterization. The computing resource allocation method does not set priorities for different morphological regions, resulting in an uneven distribution of computing load, and limited processing accuracy in high-complexity areas, affecting the overall modeling effect. The execution order of computing tasks is fixed and difficult to flexibly adjust according to data characteristics. The computing resource utilization rate is low, affecting processing efficiency. The data cache management method relies on static storage strategies and fails to dynamically adjust the storage hierarchy based on data access frequency, resulting in a decrease in data access rate and increased loading delay. During the three-dimensional mesh construction process, the node distribution and connection order are not optimized enough, resulting in uneven local mesh density, reducing the stability of the model and affecting the accuracy of three-dimensional visualization. Summary of the Invention

[0005] In order to solve the technical problems existing in the prior art, the embodiments of the present invention provide a 3D modeling data processing method and system for medical images. The technical solution is as follows:

[0006] A 3D modeling data processing method for medical images includes the following steps:

[0007] S1: Obtain boundary contour information in continuous image frames in medical imaging data, identify morphological change areas between adjacent frames, calculate the morphological change rate, determine the boundary smoothness of medical images, and obtain data spatial change characteristics;

[0008] S2: extracting the contour continuity and boundary sharpness of the morphological change area in the data spatial change feature, calculating the morphological fluctuation value within the unit volume, dividing the calculation priority according to the fluctuation value, and obtaining the data allocation result;

[0009] S3: According to the data allocation result, set the parallel execution order and execution batches, monitor the task execution status and adjust the computing load, optimize the task allocation execution time, readjust the execution batches, and obtain the data processing scheduling result;

[0010] S4: calling the cache read mode of the data processing scheduling result, adjusting the storage level of the cache queue according to the access frequency of the data block, calculating the throughput of the storage queue after the cache adjustment, and obtaining the data cache optimization result;

[0011] S5: Combining the data cache optimization result and the data processing scheduling result, refining the boundary grid of the image data, calculating the distribution density of the boundary grid nodes, optimizing the data output method, and obtaining the medical image three-dimensional data processing result.

[0012] As a further solution of the present invention, the data space change characteristics include boundary contour morphological parameters, morphological change rate index, boundary smoothness coefficient, and morphological change area distribution; the data allocation results include contour continuity parameters, boundary sharpness coefficient, unit volume morphological fluctuation value, and calculation priority level; the data processing scheduling results include parallel execution sequence list, execution batch division, calculation load distribution strategy, and task execution time optimization scheme; the data cache optimization results include cache read mode type, data block access frequency parameters, cache storage level adjustment scheme, and storage queue throughput measurement value; the medical image three-dimensional data processing results include boundary grid structure, grid node distribution density parameters, and data output format scheme.

[0013] As a further solution of the present invention, the steps for obtaining the data spatial variation characteristics are:

[0014] S101: Obtain boundary contour information of continuous image frames in medical imaging data, calculate pixel texture distribution and contrast within the boundary contour, identify gradient change areas of the boundary contour based on changes in pixel grayscale value distribution, screen out areas with obvious contour changes based on gradient change amplitude, and obtain boundary gradient change amount;

[0015] S102: Based on the boundary gradient change amount, analyze the morphological change area between adjacent image frames and calculate the morphological change rate using the formula:

[0016]

[0017] Among them, V c represents the rate of morphological change, G i represents the boundary gradient value of the i-th frame, n represents the total number of image frames, P i Represents the pixel position coordinates of the i-th frame, T i Represents the timestamp of the i-th frame;

[0018] Calculate the morphological change rate, filter the areas where the change rate exceeds the set threshold, and obtain the high change rate area;

[0019] S103: Call the high change rate area, calculate the change amplitude of its spatial structure morphology, calculate its smoothness index according to the boundary smoothness, and obtain the data space change characteristics in combination with the change characteristics of the spatial structure.

[0020] As a further solution of the present invention, the steps for obtaining the data allocation result are:

[0021] S201: Obtaining contour continuity data and boundary sharpness data of the morphological change area in the data space change feature, calculating the morphological fluctuation value within the unit volume, calling the morphological fluctuation values of all positions in the data space, and normalizing them to obtain a relatively balanced morphological fluctuation distribution, thereby generating a morphological fluctuation intensity value;

[0022] S202: Based on the morphological fluctuation intensity value, filter out regions exceeding a set fluctuation threshold, call the set morphological fluctuation threshold, calculate data points in the data space that meet the screening conditions, obtain all data regions exceeding the threshold, and generate high-fluctuation morphological regions;

[0023] S203: Based on the high-fluctuation morphological area, the calculation priority of different morphological change areas is divided according to the set calculation priority classification standard, and the data allocation amount of each priority area is calculated using the formula:

[0024]

[0025] Among them, R c Represents the data distribution result, W j represents the j-th level priority weight, S j Represents the number of computing units in the j-th level area, V j Represents the morphological fluctuation amplitude of the j-level area, L j represents the computational complexity factor of the j-th level region, T jrepresents the time span of the j-th level region, ∈ represents a minimum constant to prevent the denominator from being zero, and m represents the total number of priority levels;

[0026] The calculation obtains the computing resource requirements of each priority area, integrates the data, and generates data allocation results.

[0027] As a further solution of the present invention, the steps for obtaining the data processing scheduling result are:

[0028] S301: Based on the data allocation result, set the parallel execution order of computing tasks, obtain resource requirement data of all computing tasks to be executed, call the set computing priority information, arrange the task execution order, set the execution batch of each task, divide the computing tasks into different batches, and generate computing task execution batches;

[0029] S302: Based on the computing task execution batch, detect the task execution status of the computing unit, obtain the current task execution status data of each computing unit, calculate the current computing load, filter the computing units whose computing load exceeds the set threshold, and adjust the distribution of computing tasks, using the formula:

[0030]

[0031] Among them, L mod Represents the calculated load adjustment, T ex,k represents the current task execution time of the kth computing unit, R cap,k represents the remaining available computing resources of the kth computing unit, ∈ is a minimum value to prevent division by zero error, W ld,k represents the current load of the kth computing unit, W avg represents the average load of all computing units, and Q represents the total number of computing units;

[0032] Obtain the load adjustment amount of the computing unit by calculation, readjust the task allocation, and generate a computing load adjustment plan;

[0033] S303: Based on the computing load adjustment plan, the execution time of the task allocation plan is evaluated, the tasks whose execution time exceeds the set time threshold are screened, the task execution time data is called, the time excess ratio of the task execution is calculated, the execution batch is readjusted, the task scheduling plan is optimized, and the data processing scheduling result is generated.

[0034] As a further solution of the present invention, the steps for obtaining the data cache optimization result are:

[0035] S401: calling the data processing scheduling result, analyzing the cache reading pattern of the image data, obtaining the access frequency of the data block, performing statistics on the data access records, calculating the number of accesses to each data block within a set time window, screening the frequently accessed data blocks, adjusting the storage tier of the cache queue, and generating a cache queue storage tier allocation result;

[0036] S402: Based on the cache queue storage tier allocation result, filter the data blocks whose cache data access rate does not exceed the set access rate threshold, set the cache data access rate threshold to 200 MB / s, call the cache data access rate monitoring data, calculate the access rate of each data block, filter the data blocks whose access rate is lower than 200 MB / s, set the cache data update interval based on the screening result, adjust the data storage strategy, and optimize the cache update frequency using the formula:

[0037]

[0038] Among them, T cu Represents the update interval of cached data, D j Represents the storage size of the jth data block, V j represents the access rate of the jth data block, ∈ represents the minimum value to prevent division by zero error, U j represents the access weight of the jth data block, and M represents the total number of data blocks;

[0039] Calculate and obtain the update interval of cache data and generate cache update results;

[0040] S403: Based on the cache update result, calculate the throughput of the storage queue after the cache adjustment, analyze the change in data transmission rate, calculate the comparison value of the data transmission rate before and after the cache adjustment, optimize the data loading order, adjust the data loading order according to the adjusted cache storage status and data access characteristics, and finally generate a data cache optimization result.

[0041] As a further solution of the present invention, the steps for obtaining the medical image three-dimensional data processing results are:

[0042] S501: Combining the data cache optimization result and the data processing scheduling result, loading the image data of the computationally intensive area, loading the corresponding data according to the scheduling result of the computing task, analyzing the spatial distribution characteristics of the image data, extracting the coordinate information of the computationally intensive area, screening the image data blocks in the intensive area, and performing data format conversion, and finally generating an image data set of the computationally intensive area;

[0043] S502: Based on the image data set of the computationally intensive area, the boundary grid of the image data is refined, the topological relationship of the data structure is detected, the grid boundary information of the image data is extracted, and the distribution density between the grid nodes is calculated. Based on the spatial relationship of the nodes, the boundary grid is optimized and adjusted, and the node density difference of the three-dimensional contour grid is calculated using the formula:

[0044]

[0045] Among them, D μ represents the average deviation of the grid node density, σ q represents the node density of the qth grid cell, σ med represents the median node density of all grid cells, and X represents the total number of grid cells;

[0046] The average deviation of the grid node density is obtained by calculation to generate a three-dimensional contour grid structure;

[0047] S503: Based on the three-dimensional contour grid structure, adjust the grid distribution balance according to the connection order of the three-dimensional grid, analyze the topological structure of the grid, filter out areas where the connection relationship between grids is discontinuous, adjust the node alignment of adjacent grids, optimize the spatial consistency of the three-dimensional data, and optimize the data output method to generate medical image three-dimensional data processing results.

[0048] A 3D modeling data processing system for medical images, comprising:

[0049] The data space feature extraction module obtains the boundary contour information of continuous image frames of medical imaging data, analyzes the pixel texture distribution and contrast within the contour, identifies the morphological change areas of adjacent frames and calculates the change rate, filters out areas where the change rate exceeds a set threshold, determines the boundary smoothness and extracts the morphological change amplitude in the spatial structure to obtain the data space change characteristics;

[0050] The data allocation control module extracts the contour continuity and boundary sharpness of the morphological change area in the data spatial change characteristics, calculates the morphological fluctuation value within the unit volume, filters out the areas where the morphological fluctuation exceeds the set threshold, divides the calculation priority according to the screening results, calculates the computing resource allocation amount for each priority, and obtains the data allocation result;

[0051] The computing task scheduling module sets the parallel execution order of computing tasks according to the data allocation results, divides parallel task batches and detects the execution status of computing units, adjusts the computing load according to the detection results, evaluates the task allocation execution time and filters out tasks that exceed the set threshold, readjusts the execution batches, and obtains the data processing scheduling results;

[0052] The data cache optimization module calls the cache reading mode of the image data in the data processing scheduling result, obtains the data block access frequency and adjusts the cache queue level, filters the data blocks whose access rate does not exceed the set threshold, sets the cache data update interval and adjusts the data storage strategy based on the screening result, calculates the adjusted storage queue throughput, and obtains the data cache optimization result;

[0053] The three-dimensional data construction module combines the data cache optimization results and the data processing scheduling results, loads the image data of the computationally intensive area and refines the boundary grid, detects the topological relationship of the data structure and calculates the density of the three-dimensional contour grid nodes, adjusts the distribution balance according to the grid connection order and optimizes the data output method to obtain the three-dimensional data processing results of the medical image.

[0054] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0055] In the present invention, by analyzing the boundary contour information of continuous image frames and combining pixel texture and contrast features, morphological change areas are identified, thereby improving the ability to analyze the spatial structure of image data. Key areas are screened based on the morphological change rate and fluctuation intensity to achieve dynamic optimization allocation of computing resources and improve the processing accuracy of complex morphological areas. The task scheduling strategy adjusts the calculation batch and load according to data characteristics, optimizes computing resource utilization, and enhances processing efficiency. Hierarchical storage and dynamic adjustment of cached data improve data access rate and speed up data loading and transmission. Boundary mesh refinement and topology optimization improve the balance and resolution of three-dimensional models, providing efficient support for accurate modeling of medical images. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 is a flow chart of the method of the present invention;

[0057] Figure 2 This is a flow chart for obtaining data space change characteristics of the present invention;

[0058] Figure 3 A flowchart for obtaining data allocation results of the present invention;

[0059] Figure 4 This is a flow chart for obtaining data processing scheduling results of the present invention;

[0060] Figure 5 A flowchart for obtaining data cache optimization results of the present invention;

[0061] Figure 6 This is a flow chart for obtaining the results of medical image three-dimensional data processing according to the present invention. DETAILED DESCRIPTION

[0062] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0063] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as an "exemplary" in the present invention should not be interpreted as being preferred or advantageous over other embodiments or designs. Rather, the use of the word "exemplary" is intended to present concepts in a concrete manner. Furthermore, in the embodiments of the present invention, "and / or" can mean both or either of the two.

[0064] In the embodiments of the present invention, the terms "image" and "picture" may be used interchangeably. It should be noted that, when the distinction between them is not emphasized, their intended meanings are the same. The terms "of," "corresponding," and "corresponding" may be used interchangeably. It should be noted that, when the distinction between them is not emphasized, their intended meanings are the same.

[0065] In the embodiments of the present invention, sometimes a subscript such as W1 may be written as a non-subscript such as W1. When the difference is not emphasized, the meanings to be expressed are the same.

[0066] In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0067] See also Figure 1 The present invention provides a technical solution: a 3D modeling data processing method for medical images, comprising the following steps:

[0068] S1: Obtain boundary contour information in continuous image frames in medical imaging data, analyze the pixel texture distribution and contrast within the boundary contour, identify morphological change areas between adjacent image frames, calculate the morphological change rate, filter areas where the change rate exceeds the set change rate threshold, determine the boundary smoothness of the medical image, and extract the morphological change amplitude in the medical image spatial structure to obtain the data spatial change characteristics;

[0069] S2: Extract the contour continuity and boundary sharpness of the morphological change area in the data spatial change characteristics, calculate the morphological fluctuation value within the unit volume, filter out the areas where the morphological fluctuation exceeds the set fluctuation threshold, divide the calculation priority of different morphological change areas based on the screening results, calculate the computing resource allocation amount for each priority, and obtain the data allocation result;

[0070] S3: Based on the data allocation results, the parallel execution order of computing tasks is set, the execution batches of parallel tasks are divided, the task execution status of the computing unit is detected and the computing load is adjusted, the execution time of the task allocation plan is evaluated, tasks whose execution time exceeds the set time threshold are filtered out, and the execution batches are readjusted to obtain the data processing scheduling results;

[0071] S4: Calling the cache reading mode of the image data in the data processing scheduling result, obtaining the access frequency of the data block and adjusting the storage level of the cache queue, screening the data blocks whose cache data access rate does not exceed the set access rate threshold, setting the cache data update interval based on the screening result, and adjusting the data storage strategy, calculating the throughput of the storage queue after the cache adjustment, analyzing the change of the data transmission rate based on the throughput, optimizing the data loading order, and obtaining the data cache optimization result;

[0072] S5: Combine the data cache optimization results and data processing scheduling results to load the image data of the computationally intensive area, refine the boundary grid of the image data, detect the topological relationship of the data structure, calculate the node distribution density of the three-dimensional contour grid, adjust the grid distribution balance according to the connection order of the three-dimensional grid, and optimize the data output method to obtain the three-dimensional data processing results of medical images.

[0073] The data spatial change characteristics include boundary contour morphological parameters, morphological change rate index, boundary smoothness coefficient, and morphological change area distribution; the data allocation results include contour continuity parameters, boundary sharpness coefficient, unit volume morphological fluctuation value, and calculation priority level; the data processing scheduling results include parallel execution sequence list, execution batch division, calculation load distribution strategy, and task execution time optimization plan; the data cache optimization results include cache read mode type, data block access frequency parameters, cache storage level adjustment plan, and storage queue throughput measurement value; the medical imaging three-dimensional data processing results include boundary grid structure, grid node distribution density parameters, and data output format plan.

[0074] See also Figure 2 , the steps to obtain the data space change characteristics are:

[0075] S101: Obtain boundary contour information of continuous image frames in medical imaging data, calculate pixel texture distribution and contrast within the boundary contour, identify gradient change areas of the boundary contour based on changes in pixel grayscale value distribution, screen out areas with obvious contour changes based on gradient change amplitude, and obtain boundary gradient change amount;

[0076] Acquire continuous medical image frames, such as a cardiac ultrasound image sequence. For each frame of the image, apply an edge detection algorithm to extract the boundary contour of the organ or tissue. Commonly used edge detection algorithms include Canny, Sobel, etc. Suppose that in a frame of 512×512 pixels, the boundary contour of the left ventricle of the heart is successfully extracted. Next, calculate the pixel texture distribution and contrast within the boundary contour. To this end, first determine the pixel area inside the boundary contour and calculate the grayscale value distribution of the pixels in the area. For example, calculate the mean and standard deviation of the grayscale value to evaluate the uniformity and contrast of the texture. Suppose that in the left ventricle area, the grayscale value mean is 150 and the standard deviation is 20, which represents the texture distribution and contrast of the area. Based on these calculations, identify the gradient change area of the boundary contour and calculate the boundary gradient change. Suppose that in a certain frame, the boundary gradient change is 30, which represents the degree of boundary change, and obtain the boundary gradient change.

[0077] S102: Based on the boundary gradient change, analyze the morphological change area between adjacent image frames and calculate the morphological change rate using the formula:

[0078]

[0079] Among them, V c represents the rate of morphological change, G i represents the boundary gradient value of the i-th frame, n represents the total number of image frames, P i Represents the pixel position coordinates of the i-th frame, T i Represents the timestamp of the i-th frame;

[0080] Calculate the morphological change rate, filter the areas where the change rate exceeds the set threshold, and obtain the high change rate area;

[0081] Assuming that in a medical image sequence containing 10 frames, the boundary gradient values of each frame are 20, 25, 23, 28, 30, 35, 33, 40, 42, and 45 respectively, the gradient change is calculated as follows:

[0082] |25-20|+|23-25|+|28-23|+|30-28|+|35-30|+|33-35|+|40-33|+|42-40|+|45-42|=33;

[0083] Take the number of image frames n = 9, and calculate the mean gradient change:

[0084] Assuming that the pixel position coordinates of the corresponding frames are 5, 10, 15, 22, 30, 38, 45, 55, 63, and 70, and the timestamps are 0.0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, and 0.9, respectively, calculate the displacement speed between each frame:

[0085]

[0086] Compute the sum of squares:

[0087] 50 2 +50 2 +70 2 +80 2 +80 2 +70 2 +100 2 +80 2 +70 2 =2500+2500+4900+6400+6400+4900+10000+6400+4900=47400;

[0088] Take the square root:

[0089] Then the rate of change of shape: V c =3.67×217.7=798.1 pixels / second;

[0090] Assuming that the set change rate threshold is 600 pixels / second, the calculated V c =798.1 is greater than 600, so this area is marked as a high change rate area.

[0091] The results show that a certain area of the medical image data has undergone significant morphological changes in the time dimension, and its change rate exceeds the set change rate threshold. Therefore, this area may represent pathological changes or organ motion characteristics with important diagnostic value. This result is directly related to the high change rate area obtained in the step. Through this numerical result, we can further screen out areas in the medical image sequence whose movement speed exceeds the normal physiological range, and these areas can be used for subsequent pathological analysis or treatment plan formulation.

[0092] S103: Call the high change rate area, calculate the change amplitude of its spatial structure morphology, calculate its smoothness index based on the boundary smoothness, and combine the change characteristics of the spatial structure to obtain the data space change characteristics;

[0093] For areas with high change rates, the amplitude of morphological changes in the entire image sequence is calculated. For example, the area change of the area between different frames is calculated. Assuming that the area of a high change rate area is 50 square pixels in the first frame and 70 square pixels in the tenth frame, the amplitude of morphological change is 20 square pixels. Next, the smoothness index is calculated based on the boundary smoothness. The smoothness index can be evaluated by calculating the curvature change of the boundary. Assuming that the calculated smoothness index is 0.8, combined with the change characteristics of the spatial structure, the data space change characteristics are obtained.

[0094] See also Figure 3 , the steps to obtain the data allocation results are:

[0095] S201: Obtaining contour continuity data and boundary sharpness data of the morphological change area in the data space change feature, calculating the morphological fluctuation value within the unit volume, calling the morphological fluctuation values of all positions in the data space, and normalizing them to obtain a relatively balanced morphological fluctuation distribution, thereby generating a morphological fluctuation intensity value;

[0096] Suppose that in a medical image analysis process, the contour continuity of a lesion area in a CT image is measured to be 0.78, and the boundary sharpness is measured to be 0.85. Next, the morphological fluctuation value within the unit volume is calculated. Assuming that the volume of the lesion area is 120 cubic millimeters, the method for calculating its morphological fluctuation value is the product of contour continuity and boundary sharpness divided by the unit volume, that is, 0.78×0.85 / 120=0.005525. Subsequently, the area whose morphological fluctuation exceeds the set fluctuation threshold is screened. Assuming that the set fluctuation threshold is 0.005, then 0.005525>0.005 meets the screening conditions, which means The morphological fluctuations of the lesion area are large, and further attention may be needed. Finally, the calculation priorities of different morphological change areas are divided according to the screening results. Assuming that the area with a morphological fluctuation value greater than 0.006 is classified as high priority, the area between 0.004-0.006 is medium priority, and the area less than 0.004 is low priority, then this area belongs to medium priority. The amount of computing resources allocated to each priority is calculated. Assuming that the high priority area accounts for 30%, the medium priority accounts for 50%, and the low priority accounts for 20%, then this area is allocated 50% of the computing resources for more detailed morphological analysis, and finally the morphological fluctuation intensity value is obtained.

[0097] S202: Based on the morphological fluctuation intensity value, filter out areas exceeding a set fluctuation threshold, call the set morphological fluctuation threshold, calculate the data points that meet the screening conditions in the data space, obtain all data areas exceeding the threshold, and generate high-fluctuation morphological areas;

[0098] Assuming the morphological fluctuation threshold is 0.005, in an industrial image detection system, the morphological fluctuation intensity value of a material surface is measured to be 0.007. First, determine whether 0.007 is greater than the threshold value 0.005. Since 0.007>0.005, this area belongs to the high fluctuation area, which means that there may be obvious defects or unevenness on the material surface. Then, calculate the number of data points that meet the screening conditions in the data space. Assuming that the entire material surface is divided into 1000 sampling points, of which 120 points have a morphological fluctuation intensity value greater than 0.005, then filter out the 120 data points in the high fluctuation area, obtain all data areas that exceed the threshold, and perform statistical analysis on their spatial distribution. Assuming that the 120 high fluctuation points are distributed in 30% of the entire material surface area, mark this area as the key analysis object, and finally generate a high fluctuation morphological area.

[0099] S203: Based on the high-fluctuation morphological area, according to the set calculation priority classification standard, the calculation priority of the different morphological change areas is divided, and the data allocation amount of each priority area is calculated using the formula:

[0100]

[0101] Among them, R c Represents the data distribution result, W j represents the j-th level priority weight, S j Represents the number of computing units in the j-th level area, V j Represents the morphological fluctuation amplitude of the j-level area, L j represents the computational complexity factor of the j-th level region, T j represents the time span of the j-th level region, ∈ represents a minimum constant to prevent the denominator from being zero, and m represents the total number of priority levels;

[0102] Calculate and obtain the computing resource requirements of each priority area, integrate the data, and generate data allocation results;

[0103] Assume that in a biomedical image analysis task, there are three priority categories, where the weight W1 of the high-priority area is set according to the ratio of the importance of the area and the computational complexity. Statistics on the resource requirements of multiple image analysis tasks show that the computation time of the high-priority area usually accounts for 60% of the total time, so the weight is 0.6, the number of computational units S1 is 300, and the morphological fluctuation amplitude V1 is 1.2. The weight W2 of the medium-priority area is 0.3, based on the medium morphological fluctuation area accounting for 30% of the computation time, the number of computational units S2 is 500, and the morphological fluctuation amplitude V2 is 0.8. The weight W3 of the low-priority area is 0.1. Statistics show that the low-priority area has the smallest computational time. Only about 10%, the number of computing units S3 is 200, and the morphological fluctuation amplitude V3 is 0.5. In addition, the design complexity factors L1 = 50, L2 = 30, and L3 = 10. The computational complexity factors are derived from the computational load of historical computing tasks. The settings are based on the processing time of different levels of regions in the computing tasks. It is found that the average processing time of the high-priority region is 50 unit time, so L1 is set to 50, the average processing time of the medium-priority region is 30 unit time, and L2 is set to 30. The low-priority region has the least computing tasks, and L3 is set to 10. The time span T1 = 5, T2 = 10, and T3 = 15, and the minimum constant ∈ = 0.001, and the parameters are substituted into the formula:

[0104]

[0105] Calculate the values:

[0106]

[0107] R c =115.07+117.39+17.89+9.999+2.999+0.666=264.063;

[0108] Finally, the data distribution result R is calculated c =264.063. This result shows that under the set computing priority and resource allocation strategy, the computing resource demand matches the actual computing load, ensuring that high-priority areas receive more sufficient resource support while maintaining overall computing efficiency.

[0109] See also Figure 4 , the steps to obtain the data processing scheduling results are:

[0110] S301: Based on the data allocation result, set the parallel execution order of computing tasks, obtain resource requirement data of all computing tasks to be executed, call the set computing priority information, arrange the task execution order, set the execution batch of each task, divide the computing tasks into different batches, and generate computing task execution batches;

[0111] Obtain resource requirement data for all computing tasks to be executed, including the computing time required for the task, the storage space occupied, and the computational complexity. Set the execution order of the computing tasks, call the set computing priority information, and sort the tasks according to the task priority and resource requirement. The tasks are preliminarily arranged from large to small according to the computing resource requirement. For example, if the computing resources required for task A are 120 units, task B is 90 units, and task C is 60 units, then task A will be executed first, followed by task B and task C in turn. Then, the tasks are further sorted according to their urgency. For example, if task A has the highest computing resource requirement, but task C has the earliest deadline, then task C will be prioritized to the front of the execution batch. Set the execution batch for each task. Assuming that the system supports a maximum of 10 tasks to be executed in parallel, the sorted tasks will be allocated in batches. For example, tasks A, B, C, and D are allocated to the first batch, and tasks E, F, and G are allocated to the second batch. Finally, the computing task execution batch is generated.

[0112] S302: Based on the computing task execution batch, detect the task execution status of the computing unit, obtain the current task execution status data of each computing unit, calculate the current computing load, filter the computing units whose computing load exceeds the set threshold, and adjust the distribution of computing tasks. The formula is:

[0113]

[0114] Among them, L mod Represents the calculated load adjustment, T ex,k represents the current task execution time of the kth computing unit, R cap,k represents the remaining available computing resources of the kth computing unit, ∈ is a minimum value to prevent division by zero error, W ld,k represents the current load of the kth computing unit, W avg represents the average load of all computing units, and Q represents the total number of computing units;

[0115] Obtain the load adjustment amount of the computing unit by calculation, readjust the task allocation, and generate a computing load adjustment plan;

[0116] Assume that there are 4 computing units in the system, and their task execution times are T ex,1 =30s,T ex,2 =50s,T ex,3 =40s,T ex,4 = 60s, the remaining available computing resources of each computing unit are R cap,1 =200,R cap,2 =150,R cap,3 =180,R cap,4 =140, set the minimum value∈=0.001, and the current load of the computing unit is Wld,1 =80,W ld,2 =110,W ld,3 =90,W ld,4 =130, the average load of the computing unit is:

[0117]

[0118] Enter the formula to calculate:

[0119]

[0120] The results show that the current computing load imbalance is high and the load adjustment amount L mod The value is 20.42, indicating that there is a significant imbalance in the task allocation of computing units. The distribution of computing tasks needs to be readjusted. After further sorting, task reallocation operations can be performed based on the load differences of each computing unit to optimize the utilization efficiency of computing resources and finally generate a computing load adjustment plan.

[0121] S303: Based on the computational load adjustment plan, the execution time of the task allocation plan is evaluated, tasks whose execution time exceeds the set time threshold are screened, task execution time data is retrieved, the proportion of task execution time exceeding the threshold is calculated, the execution batch is readjusted, the task scheduling plan is optimized, and a data processing scheduling result is generated;

[0122] For example, the task execution time threshold is set to 50s. If the execution time of a task exceeds this threshold, it will be split into multiple subtasks and rescheduled to different batches. Assuming that the execution time of task X is 75s, which exceeds the threshold of 50s, task X will be split into task X1 (40s) and task X2 (35s), where X1 remains in the current batch and X2 is moved to the next batch for execution. The task scheduling plan is optimized, the adjusted task sequence is finally confirmed, and the new task batch arrangement is determined to finally generate the data processing scheduling result.

[0123] See also Figure 5 , the steps to obtain data cache optimization results are:

[0124] S401: Calling the data processing scheduling results, analyzing the cache reading pattern of the image data, obtaining the access frequency of the data blocks, collecting statistics on the data access records, calculating the number of accesses to each data block within a set time window, filtering the frequently accessed data blocks, adjusting the storage tier of the cache queue, and generating the cache queue storage tier allocation result;

[0125] The storage method of image data is detected, the access frequency of data blocks is obtained, the cache log file is read, and the number of times each data block has been accessed in the past period is counted. Assuming that the statistical period is set to 60 minutes, the access records of each data block within this period are organized into an access frequency matrix. Data blocks with higher access frequencies are prioritized and placed in the cache queue to reduce data access latency. For example, if the access frequency of image data block A is 100 times per minute and the access frequency of data block B is 30 times per minute, data block A is preferentially stored in the high-priority cache (L1 cache) and data block B is stored in the medium-priority cache (L2 cache). According to the set cache tier standard, the data blocks are classified into the L1, L2, or L3 cache queues. Subsequently, the storage strategy of different cache tiers is adjusted. For the L1 cache, a fast update strategy is adopted to ensure that frequently accessed data always remains in the cache. For the L3 cache, a delayed update strategy is adopted to reduce unnecessary data movement and improve cache space utilization. Finally, the cache queue storage tier allocation result is generated.

[0126] S402: Based on the cache queue storage tier allocation result, filter the data blocks whose cache data access rate does not exceed the set access rate threshold, call the cache data access rate monitoring data, calculate the access rate of each data block, filter the data blocks whose access rate is lower than the access rate threshold, and set the cache data update interval based on the screening result, adjust the data storage strategy, and optimize the cache update frequency using the formula:

[0127]

[0128] Among them, T cu Represents the update interval of cached data, D j Represents the storage size of the jth data block, V j represents the access rate of the jth data block, ∈ represents the minimum value to prevent division by zero error, U j represents the access weight of the jth data block, and M represents the total number of data blocks;

[0129] Calculate and obtain the update interval of cache data and generate cache update results;

[0130] Assume that the cache system contains five data blocks, with storage sizes of D1 = 500MB, D2 = 800MB, D3 = 600MB, D4 = 400MB, and D5 = 700MB. Their access rates are V1 = 180MB / s, V2 = 220MB / s, V3 = 150MB / s, V4 = 170MB / s, and V5 = 190MB / s, respectively. Set the minimum value ∈ = 0.001 to prevent division by zero errors. The access weights are U1 = 0.9, U2 = 0.7, U3 = 0.8, U4 = 0.6, and U5 = 0.85. Substitute them into the formula to calculate:

[0131]

[0132] The results show that the update interval of cached data is 3.32 seconds, reflecting the update frequency requirements of data blocks with different access rates and access weights in the cache. This value can be used to optimize the cache update mechanism, ensuring that high-access-weight data can be updated faster, while reducing unnecessary update operations for low-speed data, and ultimately generating a cache update strategy.

[0133] S403: Based on the cache update result, the throughput of the storage queue after the cache adjustment is calculated, the change in the data transmission rate is analyzed, and a comparison value of the data transmission rate before and after the cache adjustment is calculated. The data loading sequence is optimized, and the data loading sequence is adjusted according to the adjusted cache storage state and data access characteristics, and finally a data cache optimization result is generated.

[0134] Call the current throughput data of the cache storage system and calculate the total throughput before cache adjustment. Set the throughput calculation time window to 60 seconds. Within this time window, record the system's data read and write volume per second and calculate the total data transfer volume. Assume that the total throughput before adjustment is measured to be 4000MB / s. Then, perform cache adjustment operations to prioritize high-frequency access data blocks in the L1 cache, reducing the cache resource occupation of low-frequency access data blocks. The adjusted cache structure optimizes the data storage mode and improves the parallelism and access rate of data reads. Measure the adjusted data throughput again. Assume that the total throughput of the adjusted system is increased to 4600MB / s. Calculate the throughput improvement rate:

[0135] Based on this, we analyze the changes in data transfer rate and use the data transfer rate logs before and after the cache adjustment to calculate the change in average data transfer rate. Assuming that the average data transfer rate before the adjustment was 180 MB / s and increased to 210 MB / s after the adjustment, the data transfer rate increase is: ΔV = 210 - 180 = 30 MB / s.

[0136] This value reflects the impact of cache optimization on data transmission efficiency. Based on the adjusted cache storage status and data access characteristics, the data loading order is optimized to prioritize high-access data into the cache and ensure the continuity of the data flow. A dynamic data loading strategy is used to compare the read latency of data blocks before and after cache adjustment. Assuming that the access latency of high-frequency data blocks before cache optimization is 5ms and is reduced to 3.5ms after optimization, the latency reduction ratio is calculated as follows: A comprehensive analysis is conducted on the data access latency, throughput, and transmission rate before and after cache optimization to determine the optimal data storage strategy, adjust the cache cleaning mechanism, ensure the effectiveness of data updates, and ultimately generate data cache optimization results.

[0137] The results show that the cache optimization solution improves data throughput, increases the system data transmission rate by 15%, and reduces data access latency by 30%. It also optimizes the data loading order, ensures that frequently accessed data can be stored in the cache first, and reduces the cache space occupied by low-priority data, making cache resource allocation more reasonable and improving overall data access efficiency.

[0138] See also Figure 6 , the steps for obtaining the medical image three-dimensional data processing results are:

[0139] S501: Combining the data cache optimization results and the data processing scheduling results, loading the image data of the computationally intensive area, loading the corresponding data according to the scheduling results of the computing task, analyzing the spatial distribution characteristics of the image data, extracting the coordinate information of the computationally intensive area, screening the image data blocks in the intensive area, and performing data format conversion to finally generate an image dataset of the computationally intensive area;

[0140] Extract the storage path of the image data, and determine the high-priority data blocks based on the cache optimization scheme. The image data is stored in blocks, and the spatial coordinate information corresponding to the computationally intensive area is extracted. Computing resources are allocated through the data scheduling system, and the data sets to be processed are screened based on the task priority. Assuming that the system adopts a hierarchical storage strategy, the L1 cache stores high-frequency access data, the L2 cache stores medium-frequency access data, and the L3 cache stores low-frequency access data, then when loading data, data is first extracted from the L1 cache. If the data is missing, it is extracted from the L2 or L3 cache in turn. In order to ensure that the image data meets the subsequent computing requirements, the data blocks are formatted and aligned according to the regional division of the computing tasks. The spatial resolution of the data blocks is set, the boundary matching error between adjacent data blocks is calculated, and the alignment of the data storage is adjusted so that the image data in the computing area can be loaded efficiently, and finally an image data set of the computationally intensive area is generated.

[0141] S502: Based on the image dataset of the computationally dense area, refine the boundary grid of the image data, detect the topological relationship of the data structure, extract the grid boundary information of the image data, and calculate the distribution density between the grid nodes. Based on the spatial relationship of the nodes, optimize and adjust the boundary grid, and calculate the node density difference of the three-dimensional contour grid using the formula:

[0142]

[0143] Among them, Dμ represents the average deviation of the grid node density, σ q represents the node density of the qth grid cell, σ med represents the median node density of all grid cells, and X represents the total number of grid cells;

[0144] Assume that the three-dimensional grid system contains 10 grid cells and the node density of each grid cell is q (Unit: nodes / cubic millimeter) are as follows:

[0145] σ={10,12,8,15,14,11,13,9,16,10};

[0146] Calculate the median node density σ med :

[0147] Sort the data:

[0148] {8,9,10,10,11,12,13,14,15,16};

[0149] Take the median σ med :

[0150]

[0151] Compute the density deviation for each grid cell:

[0152] |10-11.5|=1.5, |12-11.5|=0.5, |8-11.5|=3.5,

[0153] |15-11.5|=3.5, |14-11.5|=2.5, |11-11.5|=0.5,

[0154] |13-11.5|=1.5, |9-11.5|=2.5, |16-11.5|=4.5,

[0155] |10-11.5|=1.5;

[0156] Calculate the mean deviation:

[0157]

[0158] The results show that the node density of the current mesh deviates by an average of 2.25 from the median, indicating an imbalance in the mesh node distribution. This value can be used as a basis for determining mesh optimization. A large value indicates that the mesh is locally overly concentrated or overly sparse, potentially leading to spatial distortion during image reconstruction. Subsequent mesh optimization can reduce this node density deviation, improve mesh uniformity, and ensure the computational accuracy of the 3D image data. Ultimately, these calculation results will be used to adjust the 3D mesh structure to achieve a more balanced mesh distribution, thereby optimizing the accuracy and integrity of the 3D contour mesh structure and generating a 3D contour mesh structure.

[0159] The average deviation of the grid node density is obtained by calculation to generate a three-dimensional contour grid structure;

[0160] S503: Based on the 3D contour grid structure, the grid distribution balance is adjusted according to the connection order of the 3D grids, the topological structure of the grids is analyzed, areas with discontinuous connection relationships between grids are screened, the node alignment of adjacent grids is adjusted, the spatial consistency of the 3D data is optimized, and the data output method is optimized to generate the 3D data processing results of the medical image;

[0161] Obtain the spatial distribution information of all grid cells, calculate the connection order of adjacent grids, compare the number of nodes of each grid cell with that of adjacent cells, identify areas with large changes in grid density, set a benchmark value for grid balance adjustment, and assume that the benchmark value is the average number of nodes of all grid cells. When the number of grid nodes in a certain area exceeds the benchmark value by more than 20%, the area is marked as an over-dense area. If it is less than 20% of the benchmark value, it is marked as an over-sparse area. During the adjustment process, for over-dense areas, reduce the subdivision level of the grid nodes in the area, and adjust the topological relationship of the boundary nodes to make them evenly distributed; for over-sparse areas, increase the grid subdivision level of the area, insert new grid nodes to improve spatial resolution, further analyze the adjusted grid connection order, ensure the balanced distribution of the adjusted grid, optimize the data output method, and select the appropriate storage format according to the computing requirements to finally generate the three-dimensional data processing results of medical images.

[0162] A 3D modeling data processing system for medical images, the system comprising:

[0163] The data space feature extraction module obtains the boundary contour information of continuous image frames of medical imaging data, analyzes the pixel texture distribution and contrast within the contour, identifies the morphological change areas of adjacent frames and calculates the change rate, filters out areas where the change rate exceeds a set threshold, determines the boundary smoothness and extracts the morphological change amplitude in the spatial structure to obtain the data space change characteristics;

[0164] The data allocation control module extracts the contour continuity and boundary sharpness of the morphological change area in the data space change characteristics, calculates the morphological fluctuation value within the unit volume, filters out the areas where the morphological fluctuation exceeds the set threshold, divides the calculation priority according to the screening results, calculates the computing resource allocation amount for each priority, and obtains the data allocation result;

[0165] The computing task scheduling module sets the parallel execution order of computing tasks based on the data allocation results, divides parallel task batches and detects the execution status of computing units, adjusts the computing load based on the detection results, evaluates the task allocation execution time and filters out tasks that exceed the set threshold, readjusts the execution batches, and obtains the data processing scheduling results;

[0166] The data cache optimization module calls the cache reading mode of the image data in the data processing scheduling result, obtains the data block access frequency and adjusts the cache queue level, filters the data blocks whose access rate does not exceed the set threshold, sets the cache data update interval and adjusts the data storage strategy based on the screening result, calculates the adjusted storage queue throughput, and obtains the data cache optimization result;

[0167] The three-dimensional data construction module combines the data cache optimization results with the data processing scheduling results, loads the image data of the computationally intensive area and refines the boundary grid, detects the topological relationship of the data structure and calculates the density of the three-dimensional contour grid nodes, adjusts the distribution balance according to the grid connection order and optimizes the data output method to obtain the three-dimensional data processing results of medical images.

[0168] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A 3D modeling data processing method for medical images, characterized in that: The following steps are involved: S1: Obtain boundary contour information in continuous image frames in medical imaging data, identify morphological change areas between adjacent frames, calculate the morphological change rate, determine the boundary smoothness of medical images, and obtain data spatial change characteristics; S2: extracting the contour continuity and boundary sharpness of the morphological change area in the data spatial change feature, calculating the morphological fluctuation value within the unit volume, dividing the calculation priority according to the fluctuation value, and obtaining the data allocation result; S3: According to the data allocation result, set the parallel execution order and execution batches, monitor the task execution status and adjust the computing load, optimize the task allocation execution time, readjust the execution batches, and obtain the data processing scheduling result; S4: calling the cache read mode of the data processing scheduling result, adjusting the storage level of the cache queue according to the access frequency of the data block, calculating the throughput of the storage queue after the cache adjustment, and obtaining the data cache optimization result; S5: Combining the data cache optimization result and the data processing scheduling result, refining the boundary grid of the image data, calculating the distribution density of the boundary grid nodes, optimizing the data output method, and obtaining the medical image three-dimensional data processing result.

2. The method for processing 3D modeling data for medical images according to claim 1, wherein: The data space change characteristics include boundary contour morphological parameters, morphological change rate index, boundary smoothness coefficient, and morphological change area distribution; the data allocation results include contour continuity parameters, boundary sharpness coefficient, unit volume morphological fluctuation value, and calculation priority level; the data processing scheduling results include parallel execution sequence list, execution batch division, calculation load distribution strategy, and task execution time optimization plan; the data cache optimization results include cache read mode type, data block access frequency parameters, cache storage level adjustment plan, and storage queue throughput measurement value; the medical image three-dimensional data processing results include boundary grid structure, grid node distribution density parameters, and data output format plan.

3. The method for processing 3D modeling data for medical images according to claim 1, wherein: The steps to obtain the data spatial change characteristics are: S101: Obtain boundary contour information of continuous image frames in medical imaging data, calculate pixel texture distribution and contrast within the boundary contour, identify gradient change areas of the boundary contour based on changes in pixel grayscale value distribution, screen out areas with obvious contour changes based on gradient change amplitude, and obtain boundary gradient change amount; S102: Based on the boundary gradient change amount, analyze the morphological change area between adjacent image frames and calculate the morphological change rate using the formula: Among them, V c represents the rate of morphological change, G i represents the boundary gradient value of the i-th frame, n represents the total number of image frames, P i Represents the pixel position coordinates of the i-th frame, T i Represents the timestamp of the i-th frame; Calculate the morphological change rate, filter the areas where the change rate exceeds the set threshold, and obtain the high change rate area; S103: Call the high change rate area, calculate the change amplitude of its spatial structure morphology, calculate its smoothness index according to the boundary smoothness, and obtain the data space change characteristics in combination with the change characteristics of the spatial structure.

4. The method for processing 3D modeling data for medical images according to claim 1, wherein: The steps to obtain data allocation results are: S201: Obtaining contour continuity data and boundary sharpness data of the morphological change area in the data space change feature, calculating the morphological fluctuation value within the unit volume, calling the morphological fluctuation values of all positions in the data space, and normalizing them to obtain a relatively balanced morphological fluctuation distribution, thereby generating a morphological fluctuation intensity value; S202: Based on the morphological fluctuation intensity value, filter out regions exceeding a set fluctuation threshold, call the set morphological fluctuation threshold, calculate data points in the data space that meet the screening conditions, obtain all data regions exceeding the threshold, and generate high-fluctuation morphological regions; S203: Based on the high-fluctuation morphological area, the calculation priority of different morphological change areas is divided according to the set calculation priority classification standard, and the data allocation amount of each priority area is calculated using the formula: Among them, R c Represents the data distribution result, W j represents the j-th level priority weight, S j Represents the number of computing units in the j-th level area, V j Represents the morphological fluctuation amplitude of the j-level area, L j represents the computational complexity factor of the j-th level region, T j represents the time span of the j-th level region, ∈ represents a minimum constant to prevent the denominator from being zero, and m represents the total number of priority levels; The calculation obtains the computing resource requirements of each priority area, integrates the data, and generates data allocation results.

5. The method for processing 3D modeling data for medical images according to claim 1, wherein: The steps to obtain data processing scheduling results are: S301: Based on the data allocation result, set the parallel execution order of computing tasks, obtain resource requirement data of all computing tasks to be executed, call the set computing priority information, arrange the task execution order, set the execution batch of each task, divide the computing tasks into different batches, and generate computing task execution batches; S302: Based on the computing task execution batch, detect the task execution status of the computing unit, obtain the current task execution status data of each computing unit, calculate the current computing load, filter the computing units whose computing load exceeds the set threshold, and adjust the distribution of computing tasks, using the formula: Among them, L mod Represents the calculated load adjustment, T ex,k represents the current task execution time of the kth computing unit, R cap,k represents the remaining available computing resources of the kth computing unit, ∈ is a minimum value to prevent division by zero error, W ld,k represents the current load of the kth computing unit, W avg represents the average load of all computing units, and Q represents the total number of computing units; Obtain the load adjustment amount of the computing unit by calculation, readjust the task allocation, and generate a computing load adjustment plan; S303: Based on the computing load adjustment plan, the execution time of the task allocation plan is evaluated, the tasks whose execution time exceeds the set time threshold are screened, the task execution time data is called, the time excess ratio of the task execution is calculated, the execution batch is readjusted, the task scheduling plan is optimized, and the data processing scheduling result is generated.

6. The method for processing 3D modeling data for medical images according to claim 1, wherein: The steps to obtain data cache optimization results are: S401: calling the data processing scheduling result, analyzing the cache reading pattern of the image data, obtaining the access frequency of the data block, performing statistics on the data access records, calculating the number of accesses to each data block within a set time window, screening the frequently accessed data blocks, adjusting the storage tier of the cache queue, and generating a cache queue storage tier allocation result; S402: Based on the cache queue storage tier allocation result, filter the data blocks whose cache data access rate does not exceed the set access rate threshold, set the cache data access rate threshold to 200 MB / s, call the cache data access rate monitoring data, calculate the access rate of each data block, filter the data blocks whose access rate is lower than 200 MB / s, set the cache data update interval based on the screening result, adjust the data storage strategy, and optimize the cache update frequency using the formula: Among them, T cu Represents the update interval of cached data, D j Represents the storage size of the jth data block, V j represents the access rate of the jth data block, ∈ represents the minimum value to prevent division by zero error, U j represents the access weight of the jth data block, and M represents the total number of data blocks; Calculate and obtain the update interval of cache data and generate cache update results; S403: Based on the cache update result, calculate the throughput of the storage queue after the cache adjustment, analyze the change in data transmission rate, calculate the comparison value of the data transmission rate before and after the cache adjustment, optimize the data loading order, adjust the data loading order according to the adjusted cache storage status and data access characteristics, and finally generate a data cache optimization result.

7. The method for processing 3D modeling data for medical images according to claim 1, wherein: The steps for obtaining the results of medical image 3D data processing are as follows: S501: Combining the data cache optimization result and the data processing scheduling result, loading the image data of the computationally intensive area, loading the corresponding data according to the scheduling result of the computing task, analyzing the spatial distribution characteristics of the image data, extracting the coordinate information of the computationally intensive area, screening the image data blocks in the intensive area, and performing data format conversion, and finally generating an image data set of the computationally intensive area; S502: Based on the image data set of the computationally intensive area, the boundary grid of the image data is refined, the topological relationship of the data structure is detected, the grid boundary information of the image data is extracted, and the distribution density between the grid nodes is calculated. Based on the spatial relationship of the nodes, the boundary grid is optimized and adjusted, and the node density difference of the three-dimensional contour grid is calculated using the formula: Among them, D μ represents the average deviation of the grid node density, σ q represents the node density of the qth grid cell, σ med represents the median node density of all grid cells, and X represents the total number of grid cells; The average deviation of the grid node density is obtained by calculation to generate a three-dimensional contour grid structure; S503: Based on the three-dimensional contour grid structure, adjust the grid distribution balance according to the connection order of the three-dimensional grid, analyze the topological structure of the grid, filter out areas where the connection relationship between grids is discontinuous, adjust the node alignment of adjacent grids, optimize the spatial consistency of the three-dimensional data, and optimize the data output method to generate medical image three-dimensional data processing results.

8. A 3D modeling data processing system for medical images, characterized in that: The method for processing 3D modeling data for medical images according to any one of claims 1 to 7 is executed, wherein the system comprises: The data space feature extraction module obtains the boundary contour information of continuous image frames of medical imaging data, analyzes the pixel texture distribution and contrast within the contour, identifies the morphological change areas of adjacent frames and calculates the change rate, filters out areas where the change rate exceeds a set threshold, determines the boundary smoothness and extracts the morphological change amplitude in the spatial structure to obtain the data space change characteristics; The data allocation control module extracts the contour continuity and boundary sharpness of the morphological change area in the data spatial change characteristics, calculates the morphological fluctuation value within the unit volume, filters out the areas where the morphological fluctuation exceeds the set threshold, divides the calculation priority according to the screening results, calculates the computing resource allocation amount for each priority, and obtains the data allocation result; The computing task scheduling module sets the parallel execution order of computing tasks according to the data allocation results, divides parallel task batches and detects the execution status of computing units, adjusts the computing load according to the detection results, evaluates the task allocation execution time and filters out tasks that exceed the set threshold, readjusts the execution batches, and obtains the data processing scheduling results; The data cache optimization module calls the cache reading mode of the image data in the data processing scheduling result, obtains the data block access frequency and adjusts the cache queue level, filters the data blocks whose access rate does not exceed the set threshold, sets the cache data update interval and adjusts the data storage strategy based on the screening result, calculates the adjusted storage queue throughput, and obtains the data cache optimization result; The three-dimensional data construction module combines the data cache optimization results and the data processing scheduling results, loads the image data of the computationally intensive area and refines the boundary grid, detects the topological relationship of the data structure and calculates the density of the three-dimensional contour grid nodes, adjusts the distribution balance according to the grid connection order and optimizes the data output method to obtain the three-dimensional data processing results of the medical image.

Images