CN115937656A - Medical image-oriented data processing method and system and electronic equipment - Google Patents

Abstract

The invention provides a data processing method, a system and electronic equipment for medical images, which relate to the technical field of medical images, and the method comprises the following steps: training the data analysis decision model by using a training database; generating a target data analysis execution flow by utilizing the trained data analysis decision model according to the data analysis requirement; data analysis requirements include: a processing target for medical image data; the target data analysis execution stream comprises a plurality of medical image data analysis links, relevant parameters of all links and a corresponding execution sequence; the method has the advantages that the obtained actual medical image data are processed based on the target data analysis execution flow, the final processing result is generated, and automatic analysis and processing of the medical image data are achieved.

Description

Medical image-oriented data processing method and system and electronic equipment

Technical Field

The present invention relates to the field of medical image technologies, and in particular, to a data processing method and system for medical images, and an electronic device.

Background

With the rapid development of artificial intelligence, big data, neuroimaging technology, etc., people are increasingly researching the mechanisms of brain development, aging and nervous system diseases, and further, people need to understand the constituent elements and the relationship among cognition, emotion and diseases more systematically, wherein brain calculation including multi-modal neuroimaging data analysis is the central driving force.

In combination with different research targets, the data processing for neuroimaging includes many links with different targets, such as preprocessing, feature extraction, statistical analysis, modeling, and the like. In these data processing links, a large number of software packages are generally involved, and according to incomplete statistics, neuroimaging related software is at least hundreds. In contrast, data processing personnel need to deeply understand algorithms, parameters, result formats, meanings, design methods and the like contained in the software to correctly and effectively apply the software for image processing; the threshold is higher and the difficulty is higher for researchers in neuroscience; meanwhile, each neuroimaging data processing link often includes a plurality of sub-links, and different software needs to be combined to form a data processing stream, so that the matching and the like among the software need to be deeply known, which is also a difficult and time-consuming problem. Similarly, the above-described problem also exists in data processing of medical images other than neuroimages. That is to say, the existing data processing scheme for medical images has the problems of high processing difficulty and low processing efficiency.

Disclosure of Invention

The invention aims to provide a data processing method, a data processing system and electronic equipment for medical images, so as to solve the technical problems of high processing difficulty and low processing efficiency in the prior art.

In a first aspect, an embodiment of the present invention provides a data processing method for medical images, including: training the data analysis decision model by using a training database; the training database is generated based on a medical image data processing knowledge base; the medical image data processing knowledge base comprises: a data processing target and a data analysis link;

generating a target data analysis execution flow by using the trained data analysis decision model according to the data analysis requirement; the data analysis requirements include: a processing target for medical image data; the target data analysis execution stream comprises a plurality of medical image data analysis links, relevant parameters of all links and corresponding execution sequences;

and processing the acquired actual medical image data based on the target data analysis execution flow to generate a final processing result so as to realize automatic analysis and processing of the medical image data.

In some possible embodiments, the method further comprises: constructing a medical image data processing knowledge base; the framework of the medical image data processing knowledge base comprises a plurality of top-level categories and a plurality of sub-level categories; each current sublevel category is obtained by subdividing a previous sublevel category of the current sublevel category; the above top layer categories include: data processing target, data analysis link, object data, data storage, server, file and task.

In some possible embodiments, the method further comprises: constructing a training database based on the medical image data processing knowledge base; the training database includes: data analysis requirements and a data analysis execution flow; the data analysis requirement is a coded text sequence expression of a processing target of the medical image data; the data analysis execution flow includes: a text sequence expression of a component corresponding to at least one medical image data analysis link; each of the above components includes at least one module for representing a specific algorithm of a corresponding link.

In some possible embodiments, the method further comprises: constructing a computing facility decision model; the above-mentioned computational facility decision model comprises: a correspondence rule and a machine learning model; generating corresponding rules among user data processing requirements, data storage and a server according to the medical image data processing knowledge base; training and generating a machine learning model based on relevant data in the server operation process; the machine learning model includes: a data storage decision model and a server decision model; the input of the data storage decision model is the user data processing requirement, and the output is a data storage target; the input of the server decision model is the user data processing requirement, and the output is a server target; determining a target computing facility based on the computing facility decision model; the target computing facility is used for executing a data analysis link of the target data analysis execution flow.

In some possible embodiments, the step of determining the target computing facility based on the computing facility decision model comprises: determining a data storage target based on the data storage decision model; the data storage target includes: storage type, capacity, and target storage device; the target storage equipment is used for storing initial data to be processed and data generated by each data analysis link; determining a server target based on the server decision model; the above server object includes: a target server type and a target server; the target server is used for executing the calculation of each data analysis link.

In some possible embodiments, the method further comprises: generating a processing strategy related to the target data analysis execution flow by utilizing a trained data analysis decision model according to the data analysis requirement; the processing strategy comprises a data calculation strategy and a data analysis strategy; the data analysis policy is used for indicating a mode for executing each analysis link in the target data analysis execution flow; the data analysis strategy comprises the following steps: individual analysis, voting analysis, multidimensional analysis; the data calculation strategy is used for expressing a data calculation mode in the analysis link; the data calculation method comprises the following steps: CPU calculation, GPU calculation and mixed calculation.

In some possible embodiments, the method further comprises: and filtering the data calculation strategy and the data analysis strategy respectively based on predefined rules among the data storage target, the server target and the processing strategy to generate the filtered data calculation strategy and the filtered data analysis strategy.

In some possible embodiments, the medical image data analysis process includes: any of preprocessing, feature extraction, statistical analysis, machine learning modeling, deep learning modeling, visualization, result annotation and processing strategies; based on the target data analysis execution flow, the acquired actual medical image data is processed, and the method comprises the following steps: sequentially executing the medical image data analysis links on the acquired actual medical image data according to the corresponding sequence, and generating a processing result of each link; calculating a data quality control index according to a processing result of the data analysis link; the data control index is used for evaluating the availability of data; and generating result evaluation parameters according to the final processing result.

In a second aspect, an embodiment of the present invention provides a data processing system for medical images, where the system includes:

the model training module is used for training the data analysis decision model by utilizing a training database; the training database is generated based on a medical image data processing knowledge base; the medical image data processing knowledge base comprises: a data processing target and a data analysis link;

the execution flow generation module is used for generating a target data analysis execution flow by utilizing the trained data analysis decision model according to the data analysis requirement; the data analysis requirements include: a processing target for medical image data; the target data analysis execution stream comprises a plurality of medical image data analysis links, relevant parameters of all links and corresponding execution sequences;

and the processing result generation module is used for analyzing the execution flow based on the target data, processing the acquired actual medical image data and generating a final processing result so as to realize automatic analysis and processing of the medical image data.

In a third aspect, an embodiment of the present invention provides an electronic device, including a memory and a processor, where the memory stores a computer program operable on the processor, and the processor implements the steps of the method according to any one of the first aspect when executing the computer program.

The invention provides a data processing method, a system and electronic equipment for medical images, wherein the method comprises the following steps: training the data analysis decision model by using a training database; generating a target data analysis execution flow by utilizing the trained data analysis decision model according to the data analysis requirement; data analysis requirements include: a processing target for medical image data; the target data analysis execution stream comprises a plurality of medical image data analysis links, relevant parameters of all links and corresponding execution sequences; the method has the advantages that the obtained actual medical image data are processed based on the target data analysis execution flow, the final processing result is generated, and automatic analysis and processing of the medical image data are achieved.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the embodiments or the prior art descriptions will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.

Fig. 1 is a schematic flowchart of a medical image-oriented data processing method according to an embodiment of the present invention;

fig. 2 is a schematic flowchart of a medical image-oriented data processing method according to an embodiment of the present invention;

fig. 3 is a schematic structural diagram of a medical image-oriented data processing system according to an embodiment of the present invention;

fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. The components of embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations.

Thus, the following detailed description of the embodiments of the present invention, as presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be obtained by a person skilled in the art without inventive step based on the embodiments of the present invention, are within the scope of protection of the present invention.

It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined and explained in subsequent figures. Some embodiments of the invention are described in detail below with reference to the accompanying drawings. The embodiments described below and the features of the embodiments can be combined with each other without conflict.

With the rapid development of artificial intelligence, big data, neuroimaging technology, etc., people are increasingly researching the mechanisms of brain development, aging and nervous system diseases, and further, people need to understand the constituent elements and the relationship among cognition, emotion and diseases more systematically, wherein brain calculation including multi-modal neuroimaging data analysis is the central driving force. In combination with different research targets, the data processing aiming at the neural image comprises various links with different targets, such as preprocessing, feature extraction, statistical analysis, modeling and the like. In these data processing links, a number of software packages are typically involved, such as: freeseason, FSL (FMRIB software library), SPM (Statistical parametric mapping), AFNI (Analysis functional neuroImage), tensorflow, pyrrch, and the like. According to incomplete statistics, neuroimage related software is at least hundreds.

For this reason, data processing personnel need to deeply understand algorithms, parameters, result formats, meanings, design methods and the like contained in the software, so as to correctly and effectively apply the software to perform image processing; the threshold is higher and the difficulty is higher for researchers in neuroscience; meanwhile, each neuroimaging data processing link often includes a plurality of sub-links, and different software needs to be combined to form a data processing stream, so that the matching and the like among the software need to be deeply known, which is also a difficult and time-consuming problem. Similarly, the above-described problem also exists in data processing of medical images other than neuroimages. That is to say, the existing data processing scheme for medical images has the problems of high processing difficulty and low processing efficiency. Although some neuroimaging data processing flow software for specific targets exist at present, the problem still remains unsolved, which still needs the user to process the data after configuring parameters and selecting methods on a command line interface on the basis of logic in deep understanding software.

Based on this, embodiments of the present invention provide a data processing method, a system and an electronic device for medical images, so as to alleviate technical problems of high processing difficulty and low processing efficiency in the prior art.

To facilitate understanding of the present embodiment, first, an aspect disclosed in the present embodiment of the invention is provided

The data processing method of medical image is described in detail with reference to fig. 1, which is a flow chart of a data processing method for medical image 5, and the method can be executed by an electronic device and mainly includes

Next step S110 to step S130:

s110: training the data analysis decision model by using a training database; the training database is generated based on a medical image data processing knowledge base; the medical image data processing knowledge base comprises: object of data processing

A marking and data analysis link;

in the present embodiment, a medical image data processing knowledge base is first constructed. Medical image data processing

The framework of the knowledge base comprises a plurality of top-level categories and a plurality of sub-level categories; each current sublevel category is obtained by subdividing a previous sublevel category of the current sublevel category; the top level categories include: data processing target, data analysis link, object data, data storage, server, file and task.

Wherein, the framework of the knowledge base is a multi-level and multi-link category system, and as a specific example 5, the top-level category includes: processing targets, object data, data storage, servers, computational processes (i.e., data analysis links), files, tasks, and the like. According to a knowledge base framework, a evidence-based literature thought is adopted, and contents such as processing targets, object data, data storage, servers, calculation processes and the like supported by clear high-quality evidence are recorded, so that a knowledge base is established.

Then, in this embodiment, a training database is constructed based on the medical image data processing knowledge base; the 0 training database includes: data analysis requirements and a data analysis execution flow; the data analysis requirement is a coded text sequence expression of a processing target of the medical image data; the data analysis execution flow includes: a text sequence expression of a component corresponding to at least one medical image data analysis link; each component includes at least one module for representing a specific algorithm of a corresponding link.

As a specific example, the data analysis requirements mainly include disease, focus-5 after normalization constitutes a coded text sequence, and computational efficiency and performance requirements-will be valid according to a predetermined rule

The rate and performance requirements correspond to the very high, medium, etc. options-and constitute the encoded literal sequence. The data analysis execution flow consists of one or more components, wherein the components mainly refer to preprocessing, feature extraction, statistical analysis, machine learning modeling, deep learning modeling, visualization, result annotation, processing strategies and the like; the component is composed of one or more modules, which refer to a specific algorithm applied to a specific calculation, such as a head movement correction algorithm (there may be a plurality) in preprocessing; from this perspective, the data analysis execution flow is a directed graph, the nodes are modules, and the edges are sequential relationships. Training data as per 7:1.5: the proportion of 1.5 is divided into training, verifying and testing sets.

S120: generating a target data analysis execution flow by using the trained data analysis decision model according to the data analysis requirement; data analysis requirements include: a processing target for medical image data; the target data analysis execution stream comprises a plurality of medical image data analysis links, relevant parameters of all links and a corresponding execution sequence;

in this embodiment, the medical image data analysis process may include: any of preprocessing, feature extraction, statistical analysis, machine learning modeling, deep learning modeling, visualization, result annotation, and processing strategies. And the target data analysis execution flow may generally further include the corresponding: software tools, algorithms, parameter values, etc.

S130: and processing the acquired actual medical image data based on the target data analysis execution flow to generate a final processing result so as to realize automatic analysis and processing of the medical image data.

In one embodiment, the method further comprises:

s21: constructing a computing facility decision model; the computing facility decision model includes: a correspondence rule and a machine learning model;

s22: generating a corresponding rule between a user data processing requirement and data storage and a server according to a medical image data processing knowledge base;

s23: training and generating a machine learning model based on relevant data in the server operation process;

s24: determining a target computing facility based on the computing facility decision model; the target computing facility is used for executing a data analysis link of the target data analysis execution flow.

In this embodiment, the step of training and generating the machine learning model based on the relevant data in the server operation process in S23 specifically includes: and respectively constructing a data storage decision model and a server decision model by using a machine learning method based on related data in the server operation process. The input of the data storage decision model is the user data processing requirement (such as data processing purpose, data itself, data processing expectation and the like), and the output is the data storage target; the input of the server decision model is the user data processing requirement (such as data processing purpose, data itself, data processing expectation and the like), and the output is the server target.

In this embodiment, the step of determining the target computing facility based on the computing facility decision model in S24 specifically includes: (S231) determining a data storage target based on the data storage decision model; (S232) determining a server objective based on the server decision model.

Wherein the data storage target comprises: storage type, capacity, and target storage device; the target storage equipment is used for storing initial data to be processed and data generated in each data analysis link; the server object includes: a target server type and a target server; the target server is used for executing the calculation of each data analysis link.

In this embodiment, format check is performed on data stored in the target storage device, and format conversion is performed on data that does not conform to the agreed format.

That is, data specified in the user information configuration is transferred to storage according to the determined storage type, capacity, and specific storage selection. And checking the data object transmitted to the destination position according to a predefined data organization mode (such as bids), if the data does not conform to the agreed format, converting the format of the data object according to the agreed format, and storing the data.

In one embodiment, the method further comprises: (S233) generating a processing strategy related to the target data analysis execution flow by using the trained data analysis decision model according to the data analysis requirement; the processing strategy comprises a data calculation strategy and a data analysis strategy; the data analysis strategy is used for expressing the mode of executing each analysis link in the target data analysis execution flow; the data analysis strategy comprises the following steps: individual analysis, voting analysis, multidimensional analysis, and the like; the data calculation strategy comprises a data calculation mode in an analysis link; the data calculation mode comprises the following steps: CPU calculation, GPU calculation and mixed calculation.

As a specific example, the data analysis process includes different manners of performing analysis processes such as individual analysis, voting analysis, multidimensional analysis (weighting), and the like according to different analysis policies, and the data analysis process may be divided into different manners of calculating such as cpu calculation, gpu calculation, hybrid calculation, and the like according to different calculation policies. The above policies are automatically recommended during the generation of the data analysis execution stream (including reference to literature or by default settings such as individual analysis, cpu calculation, etc.) and filtered through rules between the storage server and the policies (determined in advance for determining analysis, calculation policies in a particular storage and server environment).

In one embodiment, the method further comprises: and respectively filtering the data calculation strategy and the data analysis strategy based on predefined rules among the data storage target, the server target and the processing strategy to generate the filtered data calculation strategy and the filtered data analysis strategy.

In the above S130, the processing the acquired actual medical image data based on the target data analysis execution flow includes: s31, sequentially executing the medical image data analysis links on the acquired actual medical image data according to the corresponding sequence, and generating a processing result of each link; s32, calculating a data quality control index according to a processing result of the data analysis link; the data control index is used for evaluating the availability of data; s33 generates result evaluation parameters for the final processing result. Evaluating the final processing result from multiple dimensions through different result evaluation parameters; and evaluating whether the data is available or not by calculating a data quality control index.

The result evaluation parameters are mainly used for objectively evaluating the data analysis execution result; and the data quality control is mainly evaluated from the whole process of data analysis (carried out along with the data analysis process), so that a process quality basis is provided for reasonably looking at results. As a specific example, the evaluation parameters may include: significance level in statistical analysis, correlation coefficient size, regression coefficient size, accuracy of machine learning model, recall, F1, ROC, AUC, etc.

The result of data quality control is a dimensionless number (for each data object, etc.), called the data quality control index (0-100), which is divided into several ranges according to the size of the index: less than 60 times the data is not available (data in this state will not participate in the data analysis process), 60-80 times the data is available, and more than 80 times the data is better.

In the process of calculating the data quality control index, for example, for the structural image, first, feature data of image data, including parameter differences (such as repetition time), signal-to-noise ratio, structural image and template differences, brain tissue symmetry, gray matter outside the template, differences between calculation results of different tools, and the like, are extracted, after the data are normalized and normalized (normalized and normalized according to a gold standard structural image determined in advance and values thereof), weighted calculation (weights are determined in advance by a delphire method) is performed on each data, and the data are normalized to be between 0 and 100.

The calculation result of the data analysis, the result evaluation, the data quality control index, and the like are stored in a position designated by the user with reference to a predetermined format (for example, bids) pattern.

The embodiment of the invention provides an intelligent data processing method for medical images, which combines the data characteristics of medical images (such as neuroimages), realizes the automatic analysis and processing process of data through the steps of knowledge base construction, information configuration, calculation facility determination, data acquisition, data analysis, result evaluation, result acquisition and the like, reduces the difficulty of data processing, improves the efficiency of data processing, ensures the quality of data processing, and reduces the workload of data processing. As a specific example, referring to fig. 2, the method may specifically include the following steps 1 to 7.

Step 1, knowledge base construction

In order to support the neural image-oriented intelligent data processing calculation, firstly, a neural image intelligent data processing knowledge base framework is constructed, and a multi-level multi-link method is adopted for construction. The taxonomy relationships of the knowledge base framework are as follows: the top level categories include: processing targets, object data, data storage, servers, computational processes (i.e., data analysis links), files, tasks, and the like. The secondary category is subdivided by the top category, such as: categories under the treatment target include: disease, focus, efficiency, performance, etc.; categories under object data include: original, transformed, intermediate, final, etc.; categories under data storage include: physical storage, logical storage, and the like; the categories under the server include: virtual machines, physical servers; categories under the calculation process include: preprocessing, feature extraction, statistical analysis, machine learning modeling (shallow learning), deep learning modeling, visualization, data quality control, result annotation, auxiliary tools, processing strategies and the like; categories under the file include: configuration files, data files, log files, etc.; categories under the task include: file transfer tasks, data processing tasks, and the like.

The third category is obtained by subdividing the second category, such as: categories under disease may include: dementia, autism, etc.; categories under points of interest may include: cognition, social cognition, mood, quality of life, somatic function, and the like; the categories under efficiency may include computation time, occupied space, etc.; categories under performance may include accuracy, recall, level of significance, etc.; categories under the raw data include: demographic characteristics, group, modality, field strength, sequence, additional characteristics, relevant data, and the like; categories under the conversion/intermediate results include: steps, inputs, algorithms, results, etc.; categories under the final result include: statistical analysis results (including tested/grouped, algorithm, results), model modeling results (including tested/grouped, algorithm, model), and the like; categories under physical storage include: file storage, small file storage, object storage and relation storage; categories under logical storage include: local, remote, primary, secondary, restricted, unrestricted, etc.; categories under the physical server include: a cpu server, a gpu server, a cluster server, etc.; the categories under the pretreatment comprise format conversion, time point removal, time layer correction, head movement correction, standardization, regression covariates, linear drift removal, smoothing, filtering, time point deletion, generation and the like; categories under feature extraction include: statistical indexes, graph theory indexes, structural networks, functional networks, feature selection, dimension reduction and the like; categories under statistical analysis may include: differences, correlations, regressions, etc.; categories under machine learning modeling may include: classification, clustering, integration, etc.; categories under deep learning may include: based on cnn model, based on rnn model, based on attention model, integrated model, etc.; the categories under the data quality control can comprise quality control indexes, quality control methods and the like; categories under result annotation may include annotation methods, annotation algorithms, and the like; the categories under the auxiliary tool can comprise format conversion, anonymization and the like, and the categories under the processing strategy can comprise an analysis strategy, a calculation strategy and the like; the categories under the data files can comprise original files, conversion files, intermediate process files, calculation result files and the like; categories under the data processing task may include: the data processing method comprises a data conversion task, a preprocessing task, a feature extraction task, a statistical analysis task, a machine learning task, a deep learning task and the like.

The category of the fourth level, the fifth level and the like can be obtained by sequentially decomposing downwards, for example, the cognitive category can be further subdivided into: computing, reasoning, problem solving, decision making, perception, memory, attention, visual space, execution, learning, language; the social cognitive categories can be further subdivided into: interpersonal relationships, social adaptation, etc.; the mood categories can be further subdivided into: anxiety, depression, hostility, dullness, confusion, and the like; the graph theory index category can be further subdivided into node degree, shortest path, clustering coefficient, global efficiency, rich hub and the like; the structural network categories may be further subdivided into: white matter fibers, structural covariances, structural causality, etc.; the functional network category can be further subdivided into dynamic state, resting state, undirected state, directed state and the like; the quality control index categories can be further subdivided into: signal-to-noise ratio, difference between the structure image and the template, brain tissue symmetry, gray matter outside the template, difference between tool results, etc.; the category of the quality control method can be further subdivided into a supervision method, unsupervised clustering and the like; the analysis strategy category can be further subdivided into independent analysis, voting analysis, multi-dimensional analysis and the like; the category of the calculation strategy can be further subdivided into cpu calculation, gpu calculation, mixed calculation and the like; and so on until inseparable.

In addition, each subdivision category under the category of the calculation process comprises detailed parameters such as input, output, parameters, algorithms, tools, steps, templates (if any) and the like; each subdivision category under the storage category comprises subdivision categories such as types, capacities, average access rates and the like; each subdivision category under the server category comprises subdivision categories such as a cpu/gpu core number, a cpu/gpu frequency, an internal memory, a gpu display memory, a current load and the like; each subdivision category under the object data category comprises subdivision categories such as total data capacity, format, small file ratio and the like.

The relation category in the knowledge base framework mainly comprises the is-a, the attribute relation and the like. According to the knowledge base framework, a evidence-based literature thought is adopted, and a knowledge base is established for content records such as processing targets, object data, data storage, servers, calculation processes and the like supported by clear high-quality evidence (knowledge contents such as entities, relations and the like can be integrally obtained by adopting an information extraction method or a manual arrangement method, and the established knowledge base is essentially a knowledge map); in addition, the operation case related data in the operation process of the service system can be extracted from the log file and added into the knowledge base after manual examination and confirmation.

Step 2, information configuration

A graphical interface is provided for the user to express the relevant requirements of the neural image data processing in order to bring convenience to the user to carry out the neural image relevant data processing to the maximum extent. The content in the graphical interface for the user to express the data processing related requirements can be divided into three categories, which are respectively: data processing purposes, data itself, data processing expectations; referring to the knowledge base framework, the data processing object content required here mainly includes: for diseases (which may be empty), points of interest, etc., the data itself includes: data size, format, modality, current location (and directory hierarchy-such as subject, group, etc.), etc., and the data processing expects relevant content including: expected data processing time, intended footprint, expected performance metrics, etc. (training, validation, test data sets required in modeling are also specified herein).

In addition, the physical hardware information of the service system may be based on a graph in the embodiment

The interface is obtained by the highest administrator initial settings. The physical hardware information of the service system can be divided into: storing 5 information and executing server information; the storage information here includes the type, location, ip, capacity, access rate, access key, etc. of storage; the execution server comprises a type, a position, an ip, a cpu/gpu core number, a cpu/gpu frequency, an internal storage, a gpu display memory, a current load (which can be dynamically and automatically obtained in the process), an access key and the like.

Step 3, determining a computing facility

0 to achieve the data processing goal of the user, this step requires determining the type, capacity and facilities of the storage

Volume storage, server type and specific server (all computing facilities within the user authority), etc.

It is considered that the knowledge base content in the initial state may not be sufficient to build a good computational facility decision model (machine learning model). Thus, here the construction of the decision model comprises two parts: a rule part and a machine learning model part.

5 first, based on the related expert knowledge, establishing the related requirements and storage of user processing data, and establishing the server

The related decision is made according to the rules; then, based on the relevant data in the server operation process, a machine learning method is adopted to construct a relevant model of user data processing relevant requirements, storage and the server (the data for training is from the content in a knowledge base and is confirmed by expert labeling), and if the model is effective

If the relevant requirements are met, the rules are replaced; in the method, an algorithm 0 adopted in model construction comprises integration based on a decision tree and a neural network; input of storage models for correlation of user data processing

The method comprises the following steps: data processing purpose, data itself, data processing expectation and the like (coding related parameters), the output of the model is storage selection, and the loss function of the storage model simultaneously considers the targets of storage balance, minimum transmission time and the like; the input of the server model is the relevant requirements of user data processing: the data processing purpose, the data itself, the data processing expectation, etc. (encoding the relevant parameters), and the storage selection, etc., and the output is the server selection, and the loss function of the server model considers the targets of the server balance load, the minimum execution time, etc.

The determined calculation facility result is automatically displayed in a graphic mode, and a user can make a self-customization modification selection aiming at the information.

In addition, the present embodiment can also provide setting options such as addition, change, and the like of a new storage, a server, and the like.

Step 4, acquiring data

And transmitting the data specified in the user information configuration to the determined storage according to the storage type, the capacity and the specific storage selection determined in the steps. In the transmission process, the high-capacity neuroimaging data can be transmitted through an http protocol, or can be rapidly transmitted through protocols such as ftp or sftp. According to the specific storage determined by the above steps, if the number of physical storage locations exceeds 1, that is, when data is stored in a distributed manner, the data needs to be divided into different storage locations according to the determined capacity.

For the data object transmitted to the destination storage, checking according to a predefined data organization mode (such as bids), and if the data object conforms to an agreed format, finishing data acquisition through checking; and if the data does not conform to the agreed format, converting the format of the data object according to the agreed format, and finishing the data acquisition. The data format agreed here provides a data base for subsequent flexible selection, automatic execution of data processing streams, and a range of data sharing.

Step 5, analyzing data; this step includes two substeps: data analysis execution flow and parameter determination and data analysis execution process.

(1) Data analysis execution flow and parameter determination

According to the relevant requirements of the user on data processing in information configuration, including diseases, concern points, data conditions and the like, a specific execution process sequence of the data analysis flow is obtained based on the established data analysis decision model, and the specific execution process sequence comprises preprocessing, feature extraction, statistical analysis, machine learning modeling, deep learning modeling, visualization, result annotation, processing strategies and the like.

Firstly, constructing a training database to obtain a data analysis decision model, wherein training data come from contents in a knowledge base and are confirmed by expert labeling, and the specific data mainly comprise analysis requirements (mainly comprise diseases and concerns, a coded character sequence is formed after standardization, and the efficiency and performance requirements are calculated, the efficiency and performance requirements are corresponded to extremely high, medium and other options according to a predetermined rule, and the coded character sequence is formed) and a data analysis execution stream; the data analysis execution flow consists of one or more components, wherein the components mainly refer to preprocessing, feature extraction, statistical analysis, machine learning modeling, deep learning modeling, visualization, result annotation, processing strategies and the like; the component is composed of one or more modules, which refer to a specific algorithm applied to a specific calculation, such as a head movement correction algorithm (there may be a plurality) in preprocessing; from this point of view, the data analysis execution flow is a directed graph, the nodes are modules, and the edges are in a sequential relationship; the character sequence required for data analysis and the character sequence of the data analysis execution stream (both subjected to normalization processing) are expressed in the following manner; and the training data are compared with the following data according to the ratio of 7:1.5: the proportion of 1.5 is divided into training, verifying and testing sets.

The expression of the encoded literal sequence required for analysis is formed as follows: the method comprises the steps of standardizing diseases, attention points and the like to form a coded character sequence, corresponding efficiency and performance requirements to options such as extremely high, high and medium according to a predetermined rule by calculating efficiency, performance requirements and the like to form the coded character sequence, and then combining Wen Zixu columns together to form a unified character sequence (a connection symbol is comma), for example: (Alzheimer's disease, memory, high efficiency, high performance). The expression of the data analysis execution flow literal sequence is formed as follows: the main part is a component, the component includes modules, specific algorithms and related parameter setting recommendations (connected by commas), the components are represented by a hierarchical method, the components are represented by "- >" to represent a sequential relationship, the modules are represented by "= >" to represent a sequential relationship, and a parallel relationship is represented by commas ", for example: (

(pretreatment)

(

(removal time point, algorithm =, first =10, …) = > (…)

)

)

- > (feature extraction (…) = > …)

->(…)

)

Correspondingly, the input of the data analysis decision model is a literal sequence expression composed of relevant requirements of data analysis, namely, diseases, concerns, calculation efficiency, performance requirements and the like (after standardization). The output of the data analysis decision model is a literal sequence expression composed of components (i.e., preprocessing, feature extraction, statistical analysis, machine learning modeling, deep learning modeling, visualization, result annotation, processing strategies, and the like, as well as the specific modules that constitute the components, and the like). The model uses sequence-sequence generative models such as rnn, or T5, etc. in conjunction with attention.

And obtaining a data analysis execution flow by the data analysis decision model according to the data analysis requirement of the user. Presetting software related preset parameters in the data analysis execution flow according to evidence-based documents, and marking sources; the preset parameters related to the data object are detected by software in advance (including the user preset part).

And the data analysis execution flow result obtained by the data analysis decision model is expressed on a user interaction interface in a graphic mode, and user self-customization selection is provided.

In addition, the present embodiment can also provide setting options such as addition, change, and the like of new modules, components, execution flows, and the like.

(2) Data analysis execution process

After the user customizes and confirms the data analysis execution flow, the data analysis execution flow is started to be executed according to the user setting (based on the determined server, the task of the corresponding category is started), and the data analysis process (such as data preprocessing, feature extraction, statistical analysis, machine learning modeling and the like which are generally included) is automatically completed.

The intermediate result files and the result files in the data analysis process are stored according to a predetermined format (for example, refer to bids) (the results are labeled according to dates, software names, etc. for distinction).

It should be noted that the analysis policy and the computation policy are important factors in the data analysis execution flow execution process (the storage and server determined in the above steps are used as implicit parameters to influence the analysis and computation policy, for example, when a plurality of servers are distributed, weighted multidimensional analysis is recommended). According to different analysis strategies, the data analysis process comprises different execution analysis processes such as individual analysis, voting analysis, multidimensional analysis (weighting) and the like, and according to different calculation strategies, the data analysis process can be divided into different calculation modes such as cpu calculation, gpu calculation, mixed calculation and the like. The above policies are automatically recommended during the generation of the data analysis execution stream (including reference to literature or by default settings such as individual analysis, cpu calculation, etc.) and filtered through rules between the storage server and the policies (determined in advance for determining analysis, calculation policies in a particular storage and server environment).

Step 6, analysis and evaluation

The step mainly comprises the parts of data quality control, result evaluation and the like. The result evaluation mainly comprises the steps of objectively evaluating a data analysis execution result; and the data quality control is mainly evaluated from the whole process of data analysis (carried out along with the data analysis process), so that a process quality basis is provided for reasonably looking at results.

The evaluation of the results herein allows objective recognition of the results themselves, mainly by providing different evaluation parameters. The evaluation parameters comprise significance level, correlation coefficient size, regression coefficient size, accuracy rate, recall rate, F1, ROC, AUC and the like in statistical analysis.

The result of data quality control is a dimensionless number (for each data object, etc.), called the data quality control index (0-100), which is divided into several ranges according to the size of the index: less than 60 times the data is not available (data in this state will not participate in the data analysis process), 60-80 times the data is available, and more than 80 times the data is better.

In the process of calculating the data quality control index, for example, for the structural image, first, feature data of image data, including parameter differences (such as repetition time), signal-to-noise ratio, structural image and template differences, brain tissue symmetry, gray matter outside the template, differences between calculation results of different tools, and the like, are extracted, after the data are normalized and normalized (normalized and normalized according to a gold standard structural image determined in advance and values thereof), weighted calculation (weights are determined in advance by a delphire method) is performed on each data, and the data are normalized to be between 0 and 100.

The calculation result of the data analysis, the result evaluation, the data quality control index, and the like are stored in a position designated by the user with reference to a predetermined format (for example, bids) pattern.

Step 7, obtaining results

After the data calculation and analysis processes such as data analysis, analysis and evaluation, data quality control and the like are completed, the results can be checked and downloaded through the graphical interface provided by the method, and the result data is utilized for further analysis and use.

Compared with the prior art, the embodiment of the invention has the following characteristics: (1) The overall process of preprocessing, feature extraction, statistical analysis, machine learning modeling, deep learning modeling and the like can be selected and completed in a one-stop manner through configuring the target of related data processing, the data related condition and the like on an interface; (2) An inference method based on the combination of knowledge and learning and a standardized character sequence expression method are combined, so that automatic recommendation of a data analysis process can be supported; (3) Based on a specially designed evidence-based medical knowledge base, automatic recommendation of parameter setting in the analysis process can be supported; (4) On the basis of automatic recommendation, supporting the customization selection of data analysis execution flows, components and modules in a hot plug mode; (5) Through methods such as automatic selection recommendation of data processing hardware facilities, automatic recommendation of a data analysis process, automatic setting and recommendation of related parameters, labeling of related reference documents and the like, the difficulty of neural image data processing is reduced, and the working efficiency and the working quality are improved; (6) By setting a uniform data format (including a data directory and the like), on one hand, a foundation is laid for automatically executing data processing analysis and the like, and meanwhile, data and result sharing in a certain range is facilitated.

In addition, an embodiment of the present invention further provides a data processing system for medical images, and referring to fig. 3, the system includes:

a model training module 310, configured to train a data analysis decision model using a training database; the training database is generated based on a medical image data processing knowledge base; the medical image data processing knowledge base comprises: a data processing target and a data analysis link;

an execution flow generation module 320, configured to generate a target data analysis execution flow according to the data analysis requirement by using the trained data analysis decision model; data analysis requirements include: a processing target for medical image data; the target data analysis execution stream comprises a plurality of medical image data analysis links, relevant parameters of all links and a corresponding execution sequence;

the processing result generating module 330 is configured to analyze the execution flow based on the target data, process the acquired actual medical image data, and generate a final processing result, so as to implement automatic analysis and processing of the medical image data.

The medical image-oriented data processing system provided by the embodiment of the application can be specific hardware on the device or software or firmware installed on the device. The system provided by the embodiment of the present application has the same implementation principle and technical effect as the foregoing method embodiment, and for the sake of brief description, reference may be made to the corresponding contents in the foregoing method embodiment where no part of the embodiment of the apparatus is mentioned. It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the foregoing systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again. The medical image-oriented data processing system provided by the embodiment of the application has the same technical characteristics as the medical image-oriented data processing method provided by the embodiment, so that the same technical problems can be solved, and the same technical effects can be achieved.

The embodiment of the application further provides an electronic device, and specifically, the electronic device comprises a processor and a storage device; the storage means has stored thereon a computer program which, when executed by the processor, performs the method of any of the above described embodiments.

Fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present application, where the electronic device 400 includes: a processor 40, a memory 41, a bus 42 and a communication interface 43, wherein the processor 40, the communication interface 43 and the memory 41 are connected through the bus 42; the processor 40 is arranged to execute executable modules, such as computer programs, stored in the memory 41.

The Memory 41 may include a high-speed Random Access Memory (RAM) and may also include a non-volatile Memory (non-volatile Memory), such as at least one disk Memory. The communication connection between the network element of the system and at least one other network element is realized through at least one communication interface 43 (which may be wired or wireless), and the internet, a wide area network, a local network, a metropolitan area network, etc. may be used.

The bus 42 may be an ISA bus, PCI bus, EISA bus, or the like. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one double-headed arrow is shown in FIG. 4, but this does not indicate only one bus or one type of bus.

The memory 41 is used for storing a program, the processor 40 executes the program after receiving an execution instruction, and the method executed by the apparatus defined by the flow process disclosed in any of the foregoing embodiments of the present invention may be applied to the processor 40, or implemented by the processor 40.

The processor 40 may be an integrated circuit chip having signal processing capabilities. In implementation, the steps of the above method may be performed by instructions in the form of hardware integrated logic circuits or software in the processor 40. The processor 40 described above may be a general purpose processor, including a central processing unit

A Central Processing Unit (CPU), a Network Processor (NP, abbreviated as 5), and the like; the device can also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other Programmable logic devices, discrete Gate or transistor logic devices, discrete hardware components. Can implement or execute the invention

Methods, steps, and logic diagrams of the disclosure in the illustrative embodiments are described. The general purpose processor may be microprocessor 0 or the processor may be any conventional processor or the like. The steps of the method disclosed in connection with the embodiments of the present invention may be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software module can be located in RAM, flash memory, ROM, PROM, EPROM, register, etc

Domain matured storage media. The storage medium is located in the memory 41, and the processor 40 reads the information in the memory 415, and completes the steps of the above method in combination with the hardware thereof.

Corresponding to the method, the embodiment of the application also provides a computer readable storage medium, and the computer readable storage medium stores machine executable instructions, when the computer executable instructions are called and executed by a processor, the computer executable instructions cause the processor to execute the steps of the method.

0 in the embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be used in general

But in other ways. The above-described embodiments of the apparatus are merely illustrative, and for example, the division of the units into only one type of logical function may be implemented in other ways, and for example, a plurality of units or components may be combined or integrated into another system, or some features may be integrated

May be omitted, or not performed. In other words, the shown or discussed mutual coupling or direct coupling 5 or communication connection may be an indirect coupling or communication connection via some communication interfaces, devices or units,

and may be electrical, mechanical or otherwise.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, functional units in the embodiments provided in the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit.

The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, an electronic device, or a network device) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.

It should be noted that: like reference numbers and letters indicate like items in the figures, and thus once an item is defined in a figure, it need not be further defined or explained in subsequent figures, and moreover, the terms "first," "second," "third," etc. are used merely to distinguish one description from another and are not to be construed as indicating or implying relative importance.

Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and these modifications or substitutions do not depart from the spirit of the corresponding technical solutions of the embodiments of the present invention.

Claims (10)

1. A data processing method for medical images is characterized by comprising the following steps:

training the data analysis decision model by using a training database; the training database is generated based on a medical image data processing knowledge base; the medical image data processing knowledge base comprises: a data processing 5 processes a target and a data analysis link;

generating a target data analysis execution flow by using the trained data analysis decision model according to the data analysis requirement; the data analysis requirements include: a processing target for medical image data; the target data analysis execution stream comprises a plurality of medical image data analysis links, relevant parameters of all links and corresponding execution sequences;

and 0, analyzing and executing the flow based on the target data, processing the acquired actual medical image data, and generating a final processing result so as to realize automatic analysis and processing of the medical image data.

2. The medical image-oriented data processing method according to claim 1,

the method further comprises the following steps:

constructing a medical image data processing knowledge base; the framework 5 of the medical image data processing knowledge base comprises a plurality of top-level categories and a plurality of sub-level categories; each current sub-level category is obtained by subdividing a previous level category of the current sub-level category; the top level categories include: data processing target, data analysis link, object data, data storage, server, file and task.

3. The medical image-oriented data processing method according to claim 2,

the method further comprises the following steps:

0, constructing a training database based on the medical image data processing knowledge base; the training database includes: data analysis requirements and a data analysis execution flow; the data analysis requirement is a coded text sequence expression of a processing target of the medical image data; the data analysis execution flow includes: a text sequence expression of a component corresponding to at least one medical image data analysis link; each of the components includes at least one module for representing a specific algorithm of a corresponding link.

4. The medical image-oriented data processing method according to claim 2, further comprising: constructing a computing facility decision model; the computing facility decision model includes: a correspondence rule and a machine learning model;

generating corresponding rules among user data processing requirements, data storage and a server according to the medical image data processing knowledge base;

training and generating a machine learning model based on relevant data in the server operation process; the machine learning model includes: a data storage decision model and a server decision model; the input of the data storage decision model is the user data processing requirement, and the output is a data storage target; the input of the server decision model is the user data processing requirement, and the output is a server target; determining a target computing facility based on the computing facility decision model; the target computing facility is used for executing a data analysis link of the target data analysis execution flow.

5. The medical image-oriented data processing method according to claim 4, wherein the step of determining a target computing facility based on the computing facility decision model comprises:

determining a data storage goal based on the data storage decision model; the data storage target includes: storage type, capacity, and target storage device; the target storage equipment is used for storing initial data to be processed and data generated by each data analysis link;

determining a server objective based on the server decision model; the server object includes: a target server type and a target server; the target server is used for executing the calculation of each data analysis link.

6. The medical image-oriented data processing method according to claim 5, further comprising:

generating a processing strategy related to the target data analysis execution flow by utilizing a trained data analysis decision model according to the data analysis requirement; the processing strategy comprises a data calculation strategy and a data analysis strategy;

the data analysis strategy is used for expressing a mode of executing each analysis link in the target data analysis execution flow; the data analysis strategy comprises: single analysis, voting analysis and multidimensional analysis;

the data calculation strategy comprises a data calculation mode in the analysis link; the data calculation mode comprises the following steps: CPU calculation, GPU calculation and mixed calculation.

7. The medical image-oriented data processing method according to claim 6, further comprising:

and respectively filtering the data calculation strategy and the data analysis strategy based on predefined rules among the data storage target, the server target and the processing strategy to generate the filtered data calculation strategy and the filtered data analysis strategy.

8. The medical image-oriented data processing method according to claim 1, wherein the medical image data analysis step comprises: any of preprocessing, feature extraction, statistical analysis, machine learning modeling, deep learning modeling, visualization, result annotation and processing strategies;

processing the acquired actual medical image data based on the target data analysis execution flow, including:

sequentially executing the medical image data analysis links on the acquired actual medical image data according to the corresponding sequence, and generating a processing result of each link;

calculating a data quality control index according to the processing result of the data analysis link; the data control index is used for evaluating the availability of data;

and generating result evaluation parameters aiming at the final processing result.

9. A medical image-oriented data processing system, the system comprising:

the model training module is used for training the data analysis decision model by utilizing a training database; the training database is generated based on a medical image data processing knowledge base; the medical image data processing knowledge base comprises: a data processing target and a data analysis link;

the execution flow generation module is used for generating a target data analysis execution flow by utilizing the trained data analysis decision model according to the data analysis requirement; the data analysis requirements include: a processing target for medical image data; the target data analysis execution stream comprises a plurality of medical image data analysis links, relevant parameters of all links and a corresponding execution sequence;

and the processing result generation module is used for analyzing the execution flow based on the target data, processing the acquired actual medical image data and generating a final processing result so as to realize automatic analysis and processing of the medical image data.

10. An electronic device comprising a memory and a processor, wherein the memory stores a computer program operable on the processor, and wherein the processor implements the steps of the method of any of claims 1 to 8 when executing the computer program.

Images