Medical data multi-dimensional perception and integrated processing system
By employing data sensing, adaptive filtering, and format standardization, the system addresses the issues of flexibility and standardization in medical data processing systems. It enables efficient acquisition, noise reduction, and unified representation of multidimensional data, thereby improving the accuracy and compatibility of medical analysis and clinical applications.
Patent Information
- Application Number
- CN202511857807.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-10
- Publication Date
- 2026-03-03
AI Technical Summary
Existing medical data processing systems are not flexible enough when processing different types of medical data, have difficulty adapting to multi-dimensional data fusion, have limited noise reduction effects, and their data formatting processing cannot meet industry standards, making it difficult to directly interface with other medical systems.
The system employs a data sensing unit to collect multidimensional medical data and synchronize timestamps, an adaptive filtering algorithm to eliminate noise, a data integration unit to extract features and perform cross-dimensional fusion, and a data output unit to perform format standardization processing to generate a unified data representation that conforms to the DICOM standard.
It enables efficient acquisition, noise reduction, and format standardization of multidimensional medical data, ensuring data quality and consistency, supporting cross-system integration, and improving the accuracy and efficiency of medical analysis and clinical applications.
Smart Images
Figure CN121601128A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of medical data processing, and more specifically, to a multi-dimensional perception and integrated processing system for medical data. Background Technology
[0002] With the continuous development of medical technology, the types of data generated by modern medical equipment are becoming increasingly diverse, mainly including physiological signals, imaging data, and laboratory indicators. These data sources are characterized by high dimensionality, heterogeneity, and noise interference, posing numerous challenges to data processing, fusion, and analysis. Especially in medical data analysis, the reliability of data quality directly affects the accuracy of diagnostic results. Therefore, efficient noise reduction, feature extraction, and cross-dimensional data fusion technologies have become crucial.
[0003] Currently, although some systems can process and analyze medical data, they generally suffer from the following problems: First, the processing methods for different types of medical data are not flexible enough to meet the needs of multi-dimensional data fusion; second, the noise reduction effect is limited, making it difficult to effectively remove noise interference from complex data; and third, the data formatting process usually fails to meet industry standards, making it difficult to directly interface with other medical systems. Therefore, there is an urgent need for a system that can efficiently collect, process, fuse, and standardize the output of multi-dimensional medical data to achieve accurate medical analysis and clinical applications. Summary of the Invention
[0004] The purpose of this invention is to provide a multi-dimensional perception and integrated processing system for medical data, which solves the problems of existing medical data processing and analysis systems. These systems are not flexible enough in processing different types of medical data, making it difficult to adapt to the needs of multi-dimensional data fusion. They also have limited noise reduction effects, making it difficult to effectively remove noise interference from complex data. Furthermore, the formatting of data usually fails to meet industry standards, and it is difficult to directly interface with other medical systems.
[0005] This invention achieves the above objectives through the following technical solution: a multi-dimensional sensing and integrated processing system for medical data, the system comprising: Data sensing unit, noise cancellation unit, data integration unit, and data output unit; The data sensing unit is used to collect multidimensional raw medical data to be processed; The noise cancellation unit is used to perform noise cancellation processing on the multidimensional medical raw data; The data integration unit is used to extract features and fuse cross-dimensional data into the noise-reduced multidimensional medical data to generate a unified data representation. The data output unit is used to standardize the format of the unified data representation and finally output the processed data that can be directly used for medical analysis or clinical applications.
[0006] Furthermore, the multidimensional raw medical data collected by the data sensing unit covers at least one dimension corresponding to a medical data type; The medical data types include any one or more combinations of physiological signals, imaging data, and laboratory indicators; During the data collection process, the timestamps corresponding to the data collection of each dimension are recorded synchronously to achieve time alignment of the data of each dimension.
[0007] Furthermore, the noise cancellation unit employs an adaptive filtering algorithm to perform noise cancellation processing; The adaptive filtering algorithm removes noise from multidimensional medical raw data through a process of initializing filtering parameters, constructing filtering input vectors, calculating filtering output values, solving error signals, dynamically adjusting filtering weights, and iterative judgment.
[0008] Furthermore, when initializing the filtering parameters, the adaptive filtering algorithm sets an initial value for the filtering weight vector, and the dimension of the filtering weight vector is adapted to the filter order. When constructing the filter input vector, the sampled values of the medical data to be processed at the current moment and the sampled values of the medical data at historical neighboring moments are selected based on a preset sliding time window to form the filter input vector.
[0009] Furthermore, when the adaptive filtering algorithm calculates the filtered output value, it obtains the filtered output at the current moment by performing a weighted operation between the filtered weight vector and the filtered input vector. When solving for the error signal, based on the preset medical data pure signal reference model, the difference between the filtered output and the signal value of the pure signal reference model at the corresponding time is calculated as the error signal.
[0010] Furthermore, when the adaptive filtering algorithm dynamically adjusts the filter weights, it uses a minimum mean square class adaptive algorithm to update the filter weight vector, and introduces a step size factor in the update process. The step size factor can be dynamically adjusted based on the mean square value of the error signal, or a fixed value can be used, and the value of the step size factor satisfies the convergence condition of the filtering algorithm.
[0011] Furthermore, the noise cancellation unit is configured with appropriate filtering parameters for different types of medical data noise; The filtering parameters include the filter order and a preset error threshold for stopping the iteration; When a sudden change in the type of noise is detected in medical data, the initial values of the filter weight vector are recalibrated to adapt to the noise processing requirements after the change.
[0012] Furthermore, when the data integration unit extracts features from the denoised multidimensional medical data: Extract the temporal features of the data from each dimension, wherein the temporal features include any one or more combinations of mean, variance, and kurtosis. At the same time, based on the type of medical data and application requirements, other non-temporal features that can characterize data features can be extracted.
[0013] Furthermore, the data integration unit calculates the weight coefficients of the extracted features in each dimension through an attention mechanism. The calculation of the weight coefficients is based on the dispersion or importance evaluation index of the data features in each dimension. Based on the calculated weight coefficients, the features of each dimension are weighted and fused to generate a unified data representation with a unified data structure.
[0014] Furthermore, when the data output unit performs format standardization processing on the unified data representation: Map unified data representations to a pre-defined range of standard medical data values; The mapped representation data is formatted and encapsulated according to medical data exchange standards. The final packaged data file contains data identification information, data processing timestamps, and standardized unified data representation, and the data exchange standard includes the DICOM standard or other industry-standard medical data exchange standards.
[0015] The beneficial effects of this invention are as follows: 1. Through the data sensing unit, the system can efficiently collect multi-dimensional medical data, including physiological signals, imaging data, laboratory indicators and other data types. It also achieves time alignment of data in each dimension through synchronized timestamps, ensuring the timeliness and consistency of the data.
[0016] 2. An adaptive filtering algorithm is used to eliminate noise in multidimensional medical data. Precise filtering is applied to different types of noise to ensure the purity of the data, thus providing a reliable foundation for subsequent analysis.
[0017] 3. Through the data integration unit, the system can extract time-domain features and other non-time-domain features from the denoised multidimensional medical data, and perform weighted fusion by combining the attention mechanism to generate a unified data representation, thereby improving the effectiveness of the data and the accuracy of cross-dimensional data fusion.
[0018] 4. The data output unit performs format standardization processing on the generated unified data representation to ensure that the data can conform to the DICOM standard or other industry-standard medical data exchange standards, which facilitates seamless integration with other medical systems and supports a wider range of application scenarios.
[0019] 5. The system can automatically adjust noise cancellation parameters and feature extraction methods according to the different characteristics of medical data, and has strong adaptability, which can meet the processing needs of different types of medical data. Attached Figure Description
[0020] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a flowchart of the overall processing of the system of the present invention; Figure 2 This is a flowchart of the adaptive filtering process of the noise cancellation unit of the present invention; Figure 3 This is a flowchart of the feature fusion process of the data integration unit of the present invention. Detailed Implementation
[0021] The present application will now be described in further detail with reference to the accompanying drawings. It should be noted that the following specific embodiments are only used to further illustrate the present application and should not be construed as limiting the scope of protection of the present application. Those skilled in the art can make some non-essential improvements and adjustments to the present application based on the above application content.
[0022] Example 1: Please see Figure 1-3 This invention provides a technical solution: a multi-dimensional sensing and integrated processing system for medical data, the system comprising: Data sensing unit, noise cancellation unit, data integration unit and data output unit; The data sensing unit is used to collect multidimensional raw medical data to be processed. Among them, acquisition is the process of obtaining multidimensional medical raw data to be processed through specific equipment or technical means; in the multidimensional medical raw data to be processed, multidimensional means that the data has multiple different dimensions or attributes; medical raw data is data obtained directly from the medical environment without any processing or manipulation, and may contain noise, errors or incomplete information. The noise cancellation unit uses an adaptive filtering algorithm to process noise in the multidimensional medical raw data; In adaptive filtering algorithms, the "adaptive" aspect refers to the algorithm's ability to automatically adjust its parameters based on the characteristics of the input signal to adapt to different noise environments and data features. Compared to filtering algorithms with fixed parameters, adaptive filtering algorithms offer greater adaptability and flexibility. Filtering algorithms are mathematical methods used to remove noise or interference from signals. In medical data processing, filtering algorithms can remove random noise, electromagnetic interference, and other contaminants from multidimensional medical raw data, improving data quality and accuracy. Noise cancellation processing utilizes adaptive filtering algorithms to process multidimensional medical raw data, removing noise components and making the data purer and more reliable. The data integration unit performs feature extraction and cross-dimensional data fusion on the noise-reduced multidimensional medical data to generate a unified data representation. Among these, features are representative and distinguishable information in the data; feature extraction is the identification and extraction of meaningful features from multidimensional medical data after noise removal, for subsequent data analysis and processing. Through feature extraction, the dimensionality of the data can be reduced, improving data processing efficiency; cross-dimensionality involves data from multiple different dimensions; data fusion is the integration and merging of data from different dimensions to form a unified data representation. In medical data processing, cross-dimensional data fusion can combine data from multiple dimensions such as basic patient information, physiological indicators, and diagnostic information, providing more comprehensive information for medical analysis and clinical applications; unified data representation is the data representation form that comprehensively reflects the characteristics of multidimensional medical data after feature extraction and cross-dimensional data fusion. This unified data representation facilitates subsequent data analysis and processing. The data output unit performs format standardization processing on the unified data representation and outputs processed data that can be directly used for medical analysis or clinical applications. Here, format refers to the organization and representation of data; different medical systems and applications may use different data formats. Standardization processing transforms a unified data representation into a universal, standard data format to ensure compatibility and interoperability between different systems and applications. Processed data that can be directly used for medical analysis or clinical applications is data that has undergone format standardization processing, has a unified format and good quality, and can be directly used by medical analysis software or clinical application systems to support medical decision-making and clinical treatment.
[0023] It should be noted that during use, the data perception unit collects multidimensional raw medical data, comprehensively acquiring various aspects of patient information and providing rich material for subsequent processing. The noise elimination unit uses an adaptive filtering algorithm, which can automatically adjust parameters according to data characteristics, accurately remove noise, improve data quality, and avoid noise interference affecting analysis results. The data integration unit performs feature extraction and cross-dimensional fusion, which can uncover the potential value of the data, generate a unified representation, and make the data more logical and correlated, facilitating comprehensive analysis. Finally, the data output unit performs format standardization processing, making the data adaptable to different medical analysis and clinical application systems, enhancing compatibility and universality. The entire system, from data acquisition to output, is optimized layer by layer, efficiently processing medical data and providing accurate, comprehensive, and directly usable data support for medical decision-making, thereby improving the quality and efficiency of medical services.
[0024] In one embodiment, the data sensing unit collects multidimensional raw medical data as follows:
[0025] in This indicates the number of dimensions in medical data, encompassing different data types such as physiological signals, image pixels, and laboratory indicators. Indicates the sampling length of single-dimensional data; Data collection timestamps are recorded synchronously during the collection process. This ensures that the data across all dimensions is aligned in time.
[0026] This design represents the collected multidimensional medical raw data in a specific format and records timestamps synchronously. Presenting the data in a clear mathematical form clearly defines the number of data dimensions and the sampling length of each dimension, covering multiple types of data dimensions and ensuring data comprehensiveness. Synchronously recording timestamps ensures that the data of each dimension is aligned in time, enabling accurate analysis of the relationship between data changes over time in subsequent processing. This avoids data analysis errors caused by time asynchrony, providing a reliable foundation for precision medicine analysis and enabling doctors to make more scientific diagnostic and treatment decisions based on accurate time series data.
[0027] In one embodiment, the noise cancellation unit employs an adaptive filtering algorithm for noise cancellation processing, including: Initialize the adaptive filter parameters, and set the filter weight vector as follows:
[0028] in Indicates the number of iterations. Let be the filter order and initial value. ; Construct the filter input vector by selecting the current data to be processed and historical neighborhood data based on a sliding time window to form the input vector:
[0029] in For the first Sampled values of a certain dimension of multidimensional medical raw data at any given time; Calculate the filtered output value and obtain the current filtered output value by weighted summation:
[0030] Calculate the error signal based on a clean signal reference model of medical data. The error is obtained as follows:
[0031] in Generated by interpolation fitting from a standard medical data sample library; The filter weights are dynamically adjusted using a least mean square adaptive algorithm to update the weight vector. The update formula is as follows:
[0032] in Let be the step size factor, satisfying , The largest eigenvalue of the autocorrelation matrix of the input vector; Iteratively execute the above steps until the error signal is received. mean square value Less than the preset threshold , Stop weight updates and output noise-reduced medical data. .
[0033] This design employs an adaptive filtering algorithm in the noise cancellation unit, which includes steps such as initializing parameters, constructing the input vector, calculating the output value and error signal, and dynamically adjusting weights. The initialization parameters provide the starting point for the algorithm. The input vector is constructed by using a sliding time window to cover historical data, enhancing its adaptability to noise. The output value is calculated by weighted summation, and the error is calculated by combining it with a clean signal reference model, which can accurately measure the filtering effect. The dynamic adjustment of weights is based on the least mean square adaptive algorithm, which continuously optimizes the filter, effectively eliminates noise, improves data quality, and provides cleaner data for subsequent medical analysis.
[0034] In one embodiment, step size factor The dynamic adjustment formula is:
[0035] in The initial step size has a range of values. , This represents the mean square value of the error signal at the initial iteration time; When a sudden change in noise type is detected, the initial weight vector is recalibrated using the following formula.
[0036] in This is the estimated mean square value of the noise signal after the abrupt change.
[0037] This design employs a dynamic adjustment formula for the step size factor. The step size factor affects the weight update speed and algorithm convergence. The initial step size setting range ensures the initial stability of the algorithm. The dynamic adjustment formula adjusts the step size according to the change in the mean square value of the error signal. When the error is large, the step size is increased to speed up convergence, and when the error is small, the step size is decreased to improve accuracy. When a sudden change in noise type is detected, the initial value of the weight vector is recalibrated. This design can quickly adapt to noise changes, enabling the filter to maintain good performance in complex noise environments, effectively eliminating various types of noise, and improving the accuracy and reliability of medical data.
[0038] In one embodiment, the parameter optimization strategy of the noise cancellation unit for different types of medical data noise is as follows: To address power frequency interference noise in physiological electrical signals, the filter order is set. Preset threshold ; To address Gaussian noise in medical imaging data, the filter order is set. Preset threshold ; To address random fluctuations in test data, adjust the filter order. Preset threshold .
[0039] This design sets different parameters for different types of medical data noise, as different types of medical data have different noise characteristics, such as power frequency interference in physiological electrical signals, Gaussian noise in medical images, and random fluctuation noise in laboratory data. By setting the filter order and preset threshold according to the noise characteristics, the adaptive filtering algorithm can eliminate the corresponding noise more accurately. An appropriate filter order can effectively capture noise characteristics, and the preset threshold can accurately determine the convergence of the filter, improve the noise elimination effect, avoid over-filtering or under-filtering, and ensure that different types of medical data retain their original characteristics after noise removal, which is beneficial for subsequent medical analysis.
[0040] In one embodiment, the feature extraction process of the data integration unit includes: Data after noise removal in each dimension Calculate time-domain features, including the mean:
[0041] variance:
[0042] Peak factor:
[0043] The weight coefficients of each feature dimension are calculated using an attention mechanism:
[0044] Feature fusion based on weighting coefficients generates a unified data representation:
[0045] This design allows the data integration unit's feature extraction process to include calculating temporal features, calculating weight coefficients through an attention mechanism, and performing feature fusion. Calculating temporal features such as mean, variance, and peak factor comprehensively reflects the basic characteristics of data across all dimensions. The attention mechanism calculates weight coefficients, allocating weights according to the importance of features across dimensions to highlight key features. Based on these weight coefficients, feature fusion generates a unified data representation, integrating multidimensional data information, uncovering potential data correlations, and providing more representative and comprehensive data for medical analysis. This helps doctors gain a more comprehensive and accurate understanding of patient conditions and develop personalized treatment plans.
[0046] In one embodiment, the format standardization process of the data output unit includes: Unified data representation Mapped to a preset medical data standard range The mapping formula is:
[0047] in, and These are the maximum and minimum values corresponding to the representation in the medical data standard library, respectively; According to the DICOM medical data exchange standard, The format is encapsulated, and the final data file containing data identifiers, processing timestamps, and standardized representations is output.
[0048] This design maps the unified data representation to a standard range and encapsulates it according to the DICOM standard. Mapping to a standard range unifies the data scope, eliminates the influence of different data units, and facilitates data comparison and analysis. Encapsulation according to the DICOM medical data exchange standard ensures the compatibility and interoperability of data between different medical systems and devices, facilitates data sharing and transmission, and outputs a final data file containing data identifiers, processing timestamps, and standardized representations. This provides complete and accurate information for medical analysis, improves medical work efficiency and data utilization value, and promotes the informatization and standardization of the medical industry.
[0049] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0050] The above embodiments provide a detailed description of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.
Claims
1. A multi-dimensional sensing and integrated processing system for medical data, characterized in that, The system includes: Data sensing unit, noise cancellation unit, data integration unit, and data output unit; The data sensing unit is used to collect multidimensional raw medical data to be processed; The noise cancellation unit is used to perform noise cancellation processing on the multidimensional medical raw data; The data integration unit is used to extract features and fuse cross-dimensional data into the noise-reduced multidimensional medical data to generate a unified data representation. The data output unit is used to standardize the format of the unified data representation and finally output the processed data that can be directly used for medical analysis or clinical applications.
2. The medical data multidimensional sensing and integrated processing system according to claim 1, characterized in that: The multidimensional raw medical data collected by the data sensing unit covers at least one dimension corresponding to a medical data type. The medical data types include any one or more combinations of physiological signals, imaging data, and laboratory indicators; During the data collection process, the timestamps corresponding to the data collection of each dimension are recorded synchronously to achieve time alignment of the data of each dimension.
3. The medical data multidimensional sensing and integrated processing system according to claim 1, characterized in that: The noise cancellation unit uses an adaptive filtering algorithm to perform noise cancellation processing. The adaptive filtering algorithm removes noise from multidimensional medical raw data through a process of initializing filtering parameters, constructing filtering input vectors, calculating filtering output values, solving error signals, dynamically adjusting filtering weights, and iterative judgment.
4. The medical data multidimensional sensing and integrated processing system according to claim 3, characterized in that: When initializing the filtering parameters, the adaptive filtering algorithm sets the initial value of the filtering weight vector, and the dimension of the filtering weight vector is adapted to the filter order. When constructing the filter input vector, the sampled values of the medical data to be processed at the current moment and the sampled values of the medical data at historical neighboring moments are selected based on a preset sliding time window to form the filter input vector.
5. The medical data multidimensional sensing and integrated processing system according to claim 3, characterized in that: When the adaptive filtering algorithm calculates the filtered output value, it obtains the filtered output at the current time by performing a weighted operation between the filtered weight vector and the filtered input vector. When solving for the error signal, the difference between the filtered output and the signal value at the corresponding time of the pure signal reference model is calculated based on the preset medical data pure signal reference model as the error signal.
6. The medical data multidimensional sensing and integrated processing system according to claim 3, characterized in that: When the adaptive filtering algorithm dynamically adjusts the filter weights, it uses the least mean square class adaptive algorithm to update the filter weight vector, and introduces a step size factor in the update process. The step size factor can be dynamically adjusted based on the mean square value of the error signal, or a fixed value can be used, and the value of the step size factor satisfies the convergence condition of the filtering algorithm.
7. The medical data multidimensional sensing and integrated processing system according to claim 3, characterized in that: The noise cancellation unit is configured with appropriate filtering parameters for different types of medical data noise. The filtering parameters include the filter order and a preset error threshold for stopping the iteration; When a sudden change in the type of noise is detected in medical data, the initial values of the filter weight vector are recalibrated to adapt to the noise processing requirements after the change.
8. The medical data multi-dimensional sensing and integrated processing system according to claim 1, characterized in that, When the data integration unit extracts features from the denoised multidimensional medical data: Extract the temporal features of the data from each dimension, wherein the temporal features include any one or more combinations of mean, variance, and kurtosis. At the same time, based on the type of medical data and application requirements, other non-temporal features that can characterize data features can be extracted.
9. The medical data multidimensional sensing and integrated processing system according to claim 8, characterized in that: The data integration unit calculates the weight coefficients of the extracted features in each dimension through an attention mechanism. The calculation of the weight coefficients is based on the dispersion or importance evaluation index of the data features in each dimension. Based on the calculated weight coefficients, the features of each dimension are weighted and fused to generate a unified data representation with a unified data structure.
10. The medical data multidimensional sensing and integrated processing system according to claim 1, characterized in that, When the data output unit performs format standardization processing on the unified data representation: Map unified data representations to a pre-defined range of standard medical data values; The mapped representation data is formatted and encapsulated according to medical data exchange standards. The final packaged data file contains data identification information, data processing timestamps, and standardized unified data representation, and the data exchange standard includes the DICOM standard or other industry-standard medical data exchange standards.
Citation Information
Patent Citations
Dysphagia detection and automatic scoring system based on high-resolution acoustic sensor
CN120048494A