CN115089112B

CN115089112B - Method, device and electronic device for establishing risk assessment model of post-stroke cognitive impairment

Info

Publication number: CN115089112B
Application number: CN202210487286.5A
Authority: CN
Inventors: 徐枫; 马靓笛; 郭雨晨
Original assignee: Tsinghua University
Current assignee: Tsinghua University
Priority date: 2022-05-06
Filing date: 2022-05-06
Publication date: 2024-04-30
Anticipated expiration: 2042-05-06
Also published as: CN115089112A

Abstract

The present application relates to a method, device and electronic device for establishing a risk assessment model for post-stroke cognitive impairment, wherein the method includes: obtaining relevant data sets of multiple modalities of a target population, and inputting them into a pre-built stroke segmentation model to obtain the size and location characteristics of the stroke lesions of the target population, and analyzing the clinical report data set based on a pre-built PSCI correlation analysis model for post-stroke cognitive impairment to obtain the PSCI risk probability of the target population, inputting the voice data set into a pre-built post-stroke language dysfunction assessment model to obtain the language dysfunction rating of the target population, inputting the video data set into a post-stroke motor dysfunction assessment model to obtain the motor dysfunction rating of the target population, and based on the obtained relevant rating results, training a prediction model composed of a multi-layer neural network to obtain a risk assessment model for post-stroke cognitive impairment. Thus, a more accurate assessment of post-stroke cognitive impairment is achieved.

Description

Post-stroke cognitive impairment risk assessment model building method and device and electronic equipment

Technical Field

The application relates to the technical field of risk assessment, in particular to a method and a device for establishing a risk assessment model of cognitive impairment after stroke and electronic equipment.

Background

Cerebral apoplexy is the primary cause of fatal disability of urban residents, partial apoplexy patients can continuously experience Post-apoplexy cognitive dysfunction (Post-stroke Cognitive Impairment, PSCI), the life quality and survival time of the patients are seriously affected, and the cerebral apoplexy is one of the important causes of serious cerebral apoplexy diseases in China.

In the related art, the prevention and treatment of PSCI is mainly performed by early evaluation and intervention to prevent PSCI and delay the development of cognitive impairment. The PSCI diagnosis and evaluation method widely used at present mainly utilizes a screening scale, a doctor performs scoring on each function of a patient by requesting the patient to perform a series of demonstration and controls the screening scale, and the cognitive disorder development condition of the patient is comprehensively evaluated by combining the clinical diagnosis result of stroke.

However, manually analyzing clinical diagnosis results of stroke patients and calculating relevant diagnosis indexes is a very time-consuming task, and the PSCI is evaluated by screening a screening table, so that the contradiction of short time consumption and complete coverage exists, and the evaluation results have strong subjective dependence on doctors, are easy to generate larger errors, and are in need of being solved.

Disclosure of Invention

The application provides a method, a device, electronic equipment and a storage medium for establishing a risk assessment model of cognitive impairment after stroke, which are used for solving the problems that the time consumption is short, the coverage is complete, the subjective dependence of an assessment result on doctors is strong, large errors are easy to generate and the like when PSCI is assessed through a screening scale in the related technology.

An embodiment of a first aspect of the present application provides a method for establishing a risk assessment model for cognitive impairment after stroke, including the following steps:

acquiring a multi-modal image dataset, a clinical report dataset, a voice dataset and a video dataset of a target crowd;

Inputting the image data sets of the multiple modes into a pre-constructed cerebral apoplexy segmentation model to obtain the size and position characteristics of a cerebral apoplexy focus of the target crowd, and analyzing the clinical report data set based on a pre-constructed PSCI correlation analysis model of cognitive impairment after the cerebral apoplexy to obtain PSCI risk probability of the target crowd;

Inputting the voice data set into a pre-constructed post-stroke language dysfunction assessment model to obtain a language dysfunction rating of the target crowd, and inputting the video data set into a post-stroke movement dysfunction assessment model to obtain a movement dysfunction rating of the target crowd; and

And training a prediction model formed by a multi-layer neural network based on the size and position characteristics of the stroke focus, the PSCI risk probability, the language dysfunction rating and the movement dysfunction rating to obtain a post-stroke cognitive dysfunction risk assessment model.

Optionally, before inputting the image data sets of the multiple modalities into a pre-constructed brain stroke segmentation model, the method further comprises:

Registering the image data sets of the multiple modes by using a medical image registration method to obtain registered image data sets of the multiple modes;

training a preset convolutional neural network based on the registered image data sets of the multiple modes and corresponding diagnosis results to generate the cerebral apoplexy segmentation model.

Optionally, before analyzing the clinical report data set based on the post-stroke cognitive impairment PSCI correlation analysis model, further comprising:

learning, using natural language processing algorithms, from the clinical report dataset, the association of demographic characteristics, clinical factors, and post-stroke cognitive impairment;

and constructing the PSCI relevance analysis model according to the relevance.

Optionally, before inputting the speech data set into the post-stroke language dysfunction assessment model, further comprising:

Extracting features related to language functions from the speech dataset;

training the recurrent neural network based on the language function related features, and constructing the post-stroke language dysfunction assessment model.

Optionally, before inputting the video dataset into the post-stroke motor dysfunction assessment model, further comprising:

Extracting human body joint motion from the video dataset using neural networks and spectral analysis techniques;

calculating an envelope surface of joint point coordinates based on the human body joint movement to obtain human body movement amplitude, and fitting the human body movement amplitude by using wavelet transformation to obtain the frequency of joint tremors;

the post-stroke motor dysfunction assessment model is constructed based on the amplitude of the human motion and the frequency of the joint tremors.

An embodiment of a second aspect of the present application provides a post-stroke cognitive impairment risk assessment model building apparatus, including:

The acquisition module is used for acquiring a multi-modal image data set, a clinical report data set, a voice data set and a video data set of a target crowd;

The risk analysis module is used for inputting the image data sets of the multiple modes into a pre-constructed cerebral apoplexy segmentation model to obtain the size and position characteristics of a cerebral apoplexy focus of the target crowd, and analyzing the clinical report data set based on a pre-constructed post-apoplexy cognitive impairment PSCI correlation analysis model to obtain PSCI risk probability of the target crowd;

The grading module is used for inputting the voice data set into a pre-constructed post-stroke language dysfunction evaluation model to obtain the language dysfunction grading of the target crowd, and inputting the video data set into the post-stroke movement dysfunction evaluation model to obtain the movement dysfunction grading of the target crowd; and

The generation module is used for training a prediction model formed by a multi-layer neural network based on the size and position characteristics of the stroke focus, the PSCI risk probability, the language dysfunction rating and the movement dysfunction rating to obtain a post-stroke cognitive dysfunction risk assessment model.

Optionally, before inputting the image data sets of the multiple modalities into a pre-constructed brain stroke segmentation model, the risk analysis module is specifically configured to:

Optionally, before analyzing the clinical report data set based on the post-stroke cognitive impairment PSCI correlation analysis model, the risk analysis module is further configured to:

and constructing the PSCI relevance analysis model according to the relevance.

Optionally, before inputting the speech data set into the post-stroke language dysfunction assessment model, the assessment module is specifically configured to:

Extracting features related to language functions from the speech dataset;

Optionally, before inputting the video dataset into the post-stroke motor dysfunction assessment model, the assessment module is further to:

An embodiment of a third aspect of the present application provides an electronic device, including: the system comprises a memory, a processor and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the program to realize the post-stroke cognitive impairment risk assessment model building method according to the embodiment.

An embodiment of a fourth aspect of the present application provides a computer readable storage medium having stored thereon a computer program for execution by a processor for implementing a post-stroke cognitive impairment risk assessment model building method as described in the above embodiment.

Therefore, the method for establishing the risk assessment model for post-stroke cognitive impairment has the following advantages.

(1) The embodiment of the application utilizes data in a plurality of different modes, and the different data can mutually supplement and mutually promote.

(2) The embodiment of the application can extract more abundant characteristic representations by utilizing data of a plurality of different modes, and improves the accuracy and the robustness of the model.

(3) According to the embodiment of the application, the grading and risk prediction of cognitive impairment after stroke are automatically realized by utilizing different modal data and multiple neural network technologies, so that subjectivity of manual diagnosis is avoided.

Additional aspects and advantages of the application will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the application.

Drawings

The foregoing and/or additional aspects and advantages of the application will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings, in which:

FIG. 1 is a flowchart of a method for establishing a risk assessment model for cognitive impairment following a stroke according to an embodiment of the present application;

FIG. 2 is a block diagram illustrating a method and apparatus for establishing a risk assessment model for cognitive impairment following a stroke according to an embodiment of the present application;

Fig. 3 is an exemplary diagram of an electronic device according to an embodiment of the present application.

Detailed Description

Embodiments of the present application are described in detail below, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to like or similar elements or elements having like or similar functions throughout. The embodiments described below by referring to the drawings are illustrative and intended to explain the present application and should not be construed as limiting the application.

The following describes a post-stroke cognitive impairment risk assessment model establishment method, a post-stroke cognitive impairment risk assessment model establishment device, electronic equipment and a storage medium according to an embodiment of the application with reference to the accompanying drawings. Aiming at the problems that PSCI is evaluated through a screening table in the related technology mentioned in the background technology center, the time consumption is short, the coverage is complete, the subjective dependence of an evaluation result on doctors is strong, and large errors are easy to generate, the application provides a post-stroke cognitive impairment risk evaluation model building method.

Before describing the method for establishing the risk assessment model of the cognitive impairment after stroke, the application introduces the cognitive impairment after stroke and related technologies.

Post-stroke cognitive impairment (PSCI) is a persistent cognitive impairment closely related to the pathological nature of stroke, and has a high incidence, with about 30% to 80% of stroke patients experiencing post-stroke cognitive impairment within 3-6 months after stroke, resulting in severe impact on quality of life and survival.

Diagnosis of stroke and calculation of various quantitative indicators related to stroke (such as ischemic volume, hemorrhagic volume, etc.) are not separated from various medical image assistance, including CT (Computed Tomography, electronic computer tomography), MR (Magnetic Resonance, magnetic resonance examination), etc. Most of evaluation contents in the screening scale need to record and analyze the voice and the behaviors of the patient, so that if the images, the clinical reports, the voice records and the video records of the patient can be simultaneously utilized, the data of the multiple modes can be comprehensively analyzed by an automatic method, and an automatic post-stroke cognitive impairment risk evaluation system based on the multimode data can be developed, so that powerful assistance can be provided for the modern stroke diagnosis and treatment process.

In addition, deep learning techniques have been rapidly developed in recent years, and have been widely used in various fields. The natural language processing method based on the Transformer can effectively analyze large-scale text data, is suitable for analyzing clinical text data and extracting key information; the medical image processing method based on the convolutional neural network can carry out high-efficiency feature extraction on the brain neural image and assist in stroke diagnosis and calculation of related quantitative indexes; the voice recognition technology and the frequency spectrum analysis method based on the cyclic neural network can accurately analyze and extract key information in voice and video data, and accurately evaluate various functions of a patient, such as language, movement and the like. Therefore, by integrating and fusing the technologies and adopting a multi-task deep learning strategy, a post-stroke cognitive impairment risk assessment model based on multi-mode data can be constructed.

Specifically, fig. 1 is a schematic flow chart of a method for establishing a risk assessment model of cognitive impairment after stroke according to an embodiment of the present application.

As shown in fig. 1, the method for establishing the risk assessment model of cognitive impairment after stroke comprises the following steps:

In step S101, a multi-modality image dataset, a clinical report dataset, a voice dataset, and a video dataset of a target population are acquired.

In particular, the present examples were trained using image datasets of 3 different modalities (CT, MR, CTA (CT contrast examination of arterial blood vessels)), clinical report datasets, speech datasets containing language dysfunction, and video datasets containing movement dysfunction.

In step S102, the image dataset of multiple modes is input to a pre-constructed brain stroke segmentation model to obtain the size and position characteristics of the stroke focus of the target crowd, and the clinical report dataset is analyzed based on the pre-constructed post-stroke cognitive impairment PSCI relevance analysis model to obtain the PSCI risk probability of the target crowd.

Optionally, in some embodiments, before inputting the image dataset of the plurality of modalities into the pre-constructed brain stroke segmentation model, further comprising: registering the image data sets of the multiple modes by using a medical image registration method to obtain registered image data sets of the multiple modes; based on the registered image data sets of multiple modes and corresponding diagnosis results, training a preset convolutional neural network to generate a cerebral apoplexy segmentation model.

The cerebral stroke segmentation model can be constructed based on a large number of medical images of different modes by using a convolutional neural network. Optionally, in some embodiments, prior to analyzing the clinical report dataset based on the post-stroke cognitive impairment PSCI correlation analysis model, further comprising: learning, using natural language processing algorithms, from the clinical report dataset, the association of demographic characteristics, clinical factors, and post-stroke cognitive impairment; and constructing a PSCI relevance analysis model according to the relevance.

The PSCI correlation analysis model can learn the correlation of demographic characteristics, clinical factors and post-stroke cognitive impairment from a large number of clinical diagnosis reports by using a BERT (Bidirectional Encoder Representation from Transformers, pre-trained language characterization model) algorithm, and constructs the PSCI correlation analysis model based on clinical factor analysis.

Specifically, the embodiment of the application needs to use a medical image registration method to align the medical images of different input modes to the template so as to obtain the segmentation results of different brain regions. After registration, the image data of all the data sets are used together with the corresponding diagnosis results to train an image segmentation model formed by a convolutional neural network, so as to obtain the size and position characteristics of the focus of the stroke. The image segmentation model comprises an encoder and a decoder module, wherein the encoder comprises 3 paths, and the images are mapped to low-dimensional image features in the same space corresponding to 3 different image modes respectively; the decoder decodes the low-dimensional image features into an original image. In order to map medical images of different modes to the same low-dimensional feature space, an countermeasure learning strategy is used, a discriminator is arranged to judge which mode the low-dimensional feature belongs to, and an encoder is caused to encode images of different modes to the same feature space.

The embodiment of the application also needs to use a natural language processing method, train a PSCI relevance analysis model by using text data in a clinical report data set, and acquire the characteristics with high PSCI relevance in the report. The PSCI relevance analysis model comprises an encoder module consisting of a bidirectional transducer structure and a classifier consisting of a multi-layer neural network, wherein the encoder module divides input text data into a group of word vectors, maps the word vectors into a high-dimensional feature space and generates a deep bidirectional language representation capable of fusing left and right context information in the text. The classifier takes the generated language representation as input through a supervised learning mode, learns the mapping relation between the representation and PSCI classification results, and finally obtains PSCI risk probability predicted according to text data.

In step S103, the voice data set is input to a pre-constructed post-stroke language dysfunction assessment model to obtain a language dysfunction rating of the target crowd, and the video data set is input to a post-stroke movement dysfunction assessment model to obtain a movement dysfunction rating of the target crowd. Optionally, in some embodiments, before inputting the speech dataset into the post-stroke language dysfunction assessment model, further comprising: extracting features related to language functions from the voice dataset; and training the recurrent neural network based on the characteristics related to the language functions, and constructing a post-stroke language dysfunction assessment model.

The post-stroke language dysfunction assessment model can be constructed by learning language function related features from voice recordings of a large number of stroke patients by using an LSTM (Long Short-Term Memory) algorithm.

Optionally, in some embodiments, before inputting the video data set into the post-stroke motor dysfunction assessment model, further comprising: extracting human body joint motions from the video data set by using a neural network and a spectrum analysis technology; based on the joint movement of the human body, calculating an envelope surface of the joint point coordinates to obtain the movement amplitude of the human body, and fitting the movement amplitude of the human body by using wavelet transformation to obtain the frequency of joint tremors; based on the amplitude of human motion and the frequency of joint tremors, a post-stroke motor dysfunction assessment model is constructed.

The post-stroke motor dysfunction assessment model can be realized by extracting relevant indexes in a PSCI screening scale from videos of patients through a spectrum analysis technology. Specifically, the embodiment of the application trains a speech feature analysis and language dysfunction assessment model formed by a cyclic neural network by using a language dysfunction data set, the model inputs a section of audio data recorded by a normal person or a patient with language dysfunction, the cyclic neural network is utilized to perform feature extraction, a section of audio data is mapped into an abstract feature space, and the neural network is further utilized to learn a mapping relation between abstract feature representation and whether the patient suffers from language dysfunction or not, so that a post-stroke language dysfunction assessment result generated by the audio data analysis is obtained.

The embodiment of the application also utilizes a neural network and a spectrum analysis technology, uses a dyskinesia data set to design a human body posture estimation model, extracts human body joint motion, then calculates an envelope surface of joint point coordinates to obtain human body motion amplitude, and uses wavelet transformation fitting waveforms to obtain the frequency of joint tremors. Finally, according to the tremor amplitude and frequency, the motor dysfunction of the patient is automatically graded by referring to a scale.

In step S104, a prediction model composed of a multi-layer neural network is trained based on the size and position characteristics of the stroke focus, PSCI risk probability, language dysfunction rating and movement dysfunction rating, and a post-stroke cognitive dysfunction risk assessment model is obtained.

The post-stroke cognitive impairment risk assessment model can utilize the multi-layer feature extraction capability of the neural network, and extract complete post-stroke cognitive impairment feature representation to achieve accurate construction according to the acquired multiple features and indexes.

Specifically, the embodiment of the application synthesizes the multi-task results and is used for training a PSCI prediction model formed by a multi-layer neural network. And (3) inputting the adopted stroke position and size characteristics, PSCI risk probability predicted based on clinical reports, language dysfunction rating generated based on audio data and movement dysfunction rating generated based on video data by using a PSCI prediction model, and learning the mapping relation between the multiple indexes and the final PSCI diagnosis grade by using the multi-layer characteristic extraction capability of a neural network to obtain a final post-stroke cognitive dysfunction risk assessment model based on multi-mode data.

According to the post-stroke cognitive impairment risk assessment model building method provided by the embodiment of the application, the demographics of a patient, the relevant characteristics of the stroke and the relevant risk factors of PSCI are extracted by using a transducer, the diagnosis of the stroke and the calculation of the quantitative index of the stroke are realized by using a convolutional neural network, the language function of the patient is assessed by using the convolutional neural network, the abnormal actions in the video of the patient are identified and quantified by using spectrum analysis, and finally, the accurate rating of the cognitive impairment is realized by using a multi-task learning strategy. Therefore, the problems that the time consumption is short, the coverage is complete, the subjective dependence of an evaluation result on doctors is strong, and large errors are easy to generate and the like in the related technology are solved by evaluating PSCI through a screening table.

Next, a post-stroke cognitive impairment risk assessment model building device according to an embodiment of the present application will be described with reference to the accompanying drawings.

Fig. 2 is a block diagram of a post-stroke cognitive impairment risk assessment model building apparatus according to an embodiment of the present application.

As shown in fig. 2, the post-stroke cognitive impairment risk assessment model building apparatus 10 includes: the risk analysis module 200, the rating module 300 and the generation module 400.

The acquisition module 100 is configured to acquire a multi-modal image dataset, a clinical report dataset, a voice dataset and a video dataset of a target population;

The risk analysis module 200 is configured to input a multi-modal image dataset into a pre-constructed brain stroke segmentation model to obtain the size and position characteristics of a stroke focus of a target crowd, and analyze a clinical report dataset based on the pre-constructed post-stroke cognitive impairment PSCI relevance analysis model to obtain PSCI risk probability of the target crowd;

The rating module 300 is configured to input a voice data set to a pre-constructed post-stroke language dysfunction assessment model to obtain a language dysfunction rating of a target crowd, and input a video data set to a post-stroke movement dysfunction assessment model to obtain a movement dysfunction rating of the target crowd; and

The generating module 400 is configured to train a prediction model formed by a multi-layer neural network based on the size and position characteristics of the stroke focus, PSCI risk probability, language dysfunction rating and movement dysfunction rating, and obtain a post-stroke cognitive dysfunction risk assessment model.

Optionally, in some embodiments, before inputting the image dataset of the plurality of modalities into the pre-constructed brain stroke segmentation model, the rating module 300 is specifically configured to:

based on the registered image data sets of multiple modes and corresponding diagnosis results, training a preset convolutional neural network to generate a cerebral apoplexy segmentation model.

Optionally, in some embodiments, the risk analysis module 200 is specifically configured to, prior to analyzing the clinical report data set based on the post-stroke cognitive impairment PSCI correlation analysis model:

and constructing a PSCI relevance analysis model according to the relevance.

Optionally, in some embodiments, before inputting the speech dataset into the post-stroke language dysfunction assessment model, the rating module 300 is specifically configured to:

extracting features related to language functions from the voice dataset;

And training the recurrent neural network based on the characteristics related to the language functions, and constructing a post-stroke language dysfunction assessment model.

Optionally, in some embodiments, before inputting the video dataset into the post-stroke motor dysfunction assessment model, the rating module 300 is further to:

extracting human body joint motions from the video data set by using a neural network and a spectrum analysis technology;

Based on the joint movement of the human body, calculating an envelope surface of the joint point coordinates to obtain the movement amplitude of the human body, and fitting the movement amplitude of the human body by using wavelet transformation to obtain the frequency of joint tremors;

Based on the amplitude of human motion and the frequency of joint tremors, a post-stroke motor dysfunction assessment model is constructed.

It should be noted that the foregoing explanation of the embodiment of the method for establishing a risk assessment model of post-stroke cognitive impairment is also applicable to the device for establishing a risk assessment model of post-stroke cognitive impairment in this embodiment, and is not described herein again.

According to the post-stroke cognitive impairment risk assessment model building device provided by the embodiment of the application, the demographics of a patient, the relevant characteristics of the stroke and the relevant risk factors of PSCI are extracted by using a transducer, the diagnosis of the stroke and the calculation of the quantitative index of the stroke are realized by using a convolutional neural network, the language function of the patient is assessed by using the convolutional neural network, the abnormal actions in the video of the patient are identified and quantified by using spectrum analysis, and finally, the accurate rating of the cognitive impairment is realized by using a multi-task learning strategy. Therefore, the problems that the time consumption is short, the coverage is complete, the subjective dependence of an evaluation result on doctors is strong, and large errors are easy to generate and the like in the related technology are solved by evaluating PSCI through a screening table.

Fig. 3 is a schematic structural diagram of an electronic device according to an embodiment of the present application. The electronic device may include:

Memory 301, processor 302, and a computer program stored on memory 301 and executable on processor 302.

The processor 302 implements the post-stroke cognitive impairment risk assessment model building method provided in the above embodiments when executing a program.

Further, the electronic device further includes:

a communication interface 303 for communication between the memory 301 and the processor 302.

A memory 301 for storing a computer program executable on the processor 302.

The memory 301 may comprise a high-speed RAM memory or may further comprise a non-volatile memory (non-volatile memory), such as at least one disk memory.

If the memory 301, the processor 302, and the communication interface 303 are implemented independently, the communication interface 303, the memory 301, and the processor 302 may be connected to each other through a bus and perform communication with each other. The bus may be an industry standard architecture (Industry Standard Architecture, abbreviated ISA) bus, an external device interconnect (PERIPHERAL COMPONENT, abbreviated PCI) bus, or an extended industry standard architecture (Extended Industry Standard Architecture, abbreviated EISA) bus, among others. The buses may be divided into address buses, data buses, control buses, etc. For ease of illustration, only one thick line is shown in fig. 3, but not only one bus or one type of bus.

Alternatively, in a specific implementation, if the memory 301, the processor 302, and the communication interface 303 are integrated on a chip, the memory 301, the processor 302, and the communication interface 303 may perform communication with each other through internal interfaces.

Processor 302 may be a central processing unit (Central Processing Unit, abbreviated as CPU), or an Application SPECIFIC INTEGRATED Circuit, abbreviated as ASIC, or one or more integrated circuits configured to implement embodiments of the present application.

The embodiment of the application also provides a computer readable storage medium, on which a computer program is stored, which when being executed by a processor, realizes the method for establishing the cognitive impairment risk assessment model after stroke.

In the description of the present specification, a description referring to terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present application. In this specification, schematic representations of the above terms are not necessarily directed to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or N embodiments or examples. Furthermore, the different embodiments or examples described in this specification and the features of the different embodiments or examples may be combined and combined by those skilled in the art without contradiction.

Furthermore, the terms "first," "second," and the like, are used for descriptive purposes only and are not to be construed as indicating or implying a relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include at least one such feature. In the description of the present application, "N" means at least two, for example, two, three, etc., unless specifically defined otherwise.

Any process or method descriptions in flow charts or otherwise described herein may be understood as representing modules, segments, or portions of code which include one or more N executable instructions for implementing specific logical functions or steps of the process, and further implementations are included within the scope of the preferred embodiment of the present application in which functions may be executed out of order from that shown or discussed, including substantially concurrently or in reverse order, depending on the functionality involved, as would be understood by those reasonably skilled in the art of the embodiments of the present application.

Logic and/or steps represented in the flowcharts or otherwise described herein, e.g., a ordered listing of executable instructions for implementing logical functions, can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For the purposes of this description, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer-readable medium would include the following: an electrical connection (electronic device) having one or N wires, a portable computer cartridge (magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer readable medium may even be paper or other suitable medium on which the program is printed, as the program may be electronically captured, via, for instance, optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner, if necessary, and then stored in a computer memory.

It is to be understood that portions of the present application may be implemented in hardware, software, firmware, or a combination thereof. In the above-described embodiments, the N steps or methods may be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. As with the other embodiments, if implemented in hardware, may be implemented using any one or combination of the following techniques, as is well known in the art: discrete logic circuits having logic gates for implementing logic functions on data signals, application specific integrated circuits having suitable combinational logic gates, programmable Gate Arrays (PGAs), field Programmable Gate Arrays (FPGAs), and the like.

Those of ordinary skill in the art will appreciate that all or a portion of the steps carried out in the method of the above-described embodiments may be implemented by a program to instruct related hardware, where the program may be stored in a computer readable storage medium, and where the program, when executed, includes one or a combination of the steps of the method embodiments.

In addition, each functional unit in the embodiments of the present application may be integrated in one processing module, or each unit may exist alone physically, or two or more units may be integrated in one module. The integrated modules may be implemented in hardware or in software functional modules. The integrated modules may also be stored in a computer readable storage medium if implemented in the form of software functional modules and sold or used as a stand-alone product.

The above-mentioned storage medium may be a read-only memory, a magnetic disk or an optical disk, or the like. While embodiments of the present application have been shown and described above, it will be understood that the above embodiments are illustrative and not to be construed as limiting the application, and that variations, modifications, alternatives and variations may be made to the above embodiments by one of ordinary skill in the art within the scope of the application.

Claims

1. A post-stroke cognitive impairment risk assessment model building device, comprising:

The generation module is used for training a prediction model formed by a multi-layer neural network based on the size and position characteristics of the stroke focus, the PSCI risk probability, the language dysfunction rating and the movement dysfunction rating to obtain a post-stroke cognitive dysfunction risk assessment model;

before the image data sets of the multiple modes are input into a pre-constructed cerebral apoplexy segmentation model, the risk analysis module is specifically configured to:

training a preset convolutional neural network based on the registered image data sets of the multiple modes and corresponding diagnosis results to generate the cerebral apoplexy segmentation model;

the risk analysis module is further configured to, prior to analyzing the clinical report dataset based on the post-stroke cognitive impairment PSCI relevance analysis model:

constructing the PSCI relevance analysis model according to the relevance;

before inputting the video dataset into the post-stroke motor dysfunction assessment model, the assessment module is further to: