A processing method and device for a hematoma aspiration decision system for cerebral hemorrhage

Abstract

The present application provides a processing method and device for a hematoma aspiration decision system for cerebral hemorrhage, and constructs a two-layer deep neural network for the processing of hematoma aspiration schemes. The timing-based perception-decision-control method implemented thereby overcomes the unknowability of the results of the hematoma aspiration process in the prior art, provides direct and convenient information transmission for doctors' decision-making, can effectively improve the accuracy of cerebral hematoma aspiration treatment schemes, and provide strong digital and intelligent support for clinical hematoma aspiration treatment.

Description

Technical Field

[0001] The present application relates to the field of medical technology, and in particular to a processing method and device for a hematoma aspiration decision system for cerebral hemorrhage. Background Art

[0002] Minimally invasive surgery is a new surgical treatment method for cerebral hemorrhage. It has certain advantages, such as small trauma and high hematoma clearance rate. Among the minimally invasive surgical treatment methods, the most important thing is hematoma aspiration. Hematoma aspiration can be understood as inserting a drainage tube after drilling through the skull, and performing operations such as liquefaction, aspiration and drainage of the hematoma to reduce the hematoma and reduce intracranial pressure.

[0003] In current clinical treatment, the method of hematoma aspiration mainly depends on the experience of clinicians. First, the aspiration plan is determined based on the patient's preoperative physical signs, including age, hematoma location, hematoma size, and preoperative imaging. During the aspiration process, the doctor relies on experience to control the aspiration speed, ultimately achieving the goal of reducing the hematoma and lowering intracranial pressure.

[0004] However, the hematoma aspiration method takes a long time, and the control of the aspiration process depends entirely on the doctor's personal experience. During a long operation, the aspiration effect cannot be guaranteed. Problems that may occur include: (1) incomplete aspiration. During the aspiration process, the hematoma is deformed due to aspiration, and the residual hematoma cannot be absorbed, which worsens the cerebral hemorrhage. (2) excessive aspiration. Since the changes in the hematoma cannot be predicted during the aspiration process, excessive aspiration may occur, causing damage to normal brain tissue and affecting normal function. (3) unstable control of the aspiration process. During the aspiration process, the intracranial pressure changes dramatically, affecting the pressure balance and possibly causing blood vessel rupture, causing secondary damage to the brain tissue during the aspiration process.

[0005] Therefore, there is a demand for using digital technology to assist in hematoma aspiration. The inventors of the present application found that only some functions were realized in the existing solutions. For example, real-time monitoring of key information in hematoma aspiration was ignored, and treatment plans were only recommended based on preoperative information. However, digital control and intelligent real-time decision-making in the hematoma aspiration process were not achieved. For example, only real-time situations were focused on during the hematoma aspiration process, and there was a lack of real-time predictions, making it difficult to control the abnormal situations that would be faced next.

[0006] In other words, current technology is still in a relatively primitive stage in terms of how to use digital technology to assist in hematoma aspiration, and has not yet reached a deep level of intelligent processing effects, and its actual value is still lacking. Summary of the invention

[0007] The present application provides a processing method and device for a hematoma aspiration decision system for cerebral hemorrhage, and constructs a two-layer deep neural network for the processing of hematoma aspiration schemes. The timing-based perception-decision-control method implemented thereby overcomes the unknowability of the results of the hematoma aspiration process in the prior art, provides direct and convenient information transmission for doctors' decision-making, can effectively improve the accuracy of cerebral hematoma aspiration treatment schemes, and provide strong digital and intelligent support for clinical hematoma aspiration treatment.

[0008] In a first aspect, the present application provides a processing method for a hematoma aspiration decision system for cerebral hemorrhage, the method comprising:

[0009] Acquiring pre-operative data of hematoma aspiration, wherein the pre-operative data of hematoma aspiration involves a first indicator;

[0010] The preoperative data of hematoma aspiration is input into the first deep neural network to determine the adapted initial hematoma aspiration scheme before the operation. In fact, the first deep neural network is used to perform multi-center clustering on the preoperative data of hematoma aspiration input into the network, and output the hematoma aspiration scheme pre-adapted by the most adapted cluster center;

[0011] Acquiring real-time intraoperative data of hematoma aspiration, wherein the real-time intraoperative data of hematoma aspiration involves a second indicator;

[0012] The real-time hematoma aspiration intraoperative data is input into a second deep neural network constructed by an LSTM network and pre-adapted with the most suitable clustering center to determine the hematoma aspiration plan decision result and the hematoma aspiration plan prediction result during the operation, wherein the second deep neural network is used to determine the timing control plan corresponding to the real-time hematoma aspiration process based on the real-time hematoma aspiration intraoperative data input by the network, and the timing control plan includes the decision-making and prediction plan contents of the hematoma aspiration plan.

[0013] In a second aspect, the present application provides a processing device for a hematoma aspiration decision system for cerebral hemorrhage, the device comprising:

[0014] A first acquisition unit is used to acquire pre-operative data of hematoma aspiration, wherein the pre-operative data of hematoma aspiration involves a first indicator;

[0015] The first processing unit is used to input the preoperative data of hematoma aspiration into the first deep neural network to determine the adapted initial hematoma aspiration scheme before the operation. In fact, the first deep neural network is used to perform multi-center clustering on the preoperative data of hematoma aspiration input into the network, and output the hematoma aspiration scheme pre-adapted by the most adapted cluster center;

[0016] A second acquisition unit is used to acquire real-time data during hematoma aspiration surgery, wherein the real-time data during hematoma aspiration surgery involves a second indicator;

[0017] The second processing unit is used to input the real-time hematoma aspiration intraoperative data into a second deep neural network constructed by an LSTM network and pre-adapted with the most suitable clustering center, so as to determine the hematoma aspiration plan decision result and the hematoma aspiration plan prediction result during the operation, wherein the second deep neural network is used to determine the timing control plan corresponding to the real-time hematoma aspiration process based on the real-time hematoma aspiration intraoperative data input by the network, and the timing control plan includes the decision-making and prediction plan contents of the hematoma aspiration plan.

[0018] In a third aspect, the present application provides a hematoma aspiration decision system, including a processor and a memory, wherein a computer program is stored in the memory, and when the processor calls the computer program in the memory, the method provided in the first aspect of the present application or any possible implementation method of the first aspect of the present application is executed.

[0019] In a fourth aspect, the present application provides a computer-readable storage medium, which stores multiple instructions, and the instructions are suitable for a processor to load to execute the method provided by the first aspect of the present application or any possible implementation of the first aspect of the present application.

[0020] From the above content, it can be concluded that the present application has the following beneficial effects:

[0021] Aiming at the auxiliary goal of hematoma aspiration treatment, the present application constructs a two-layer deep neural network for the processing of hematoma aspiration scheme. The timing-based perception-decision-control method implemented thereby overcomes the unknowability of the results of the hematoma aspiration process existing in the prior art, and provides direct and convenient information transmission for doctors' decision-making. It can effectively improve the accuracy of cerebral hematoma aspiration treatment schemes and provide strong digital and intelligent support for clinical hematoma aspiration treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0023] Figure 1 A flowchart of a processing method of a hematoma aspiration decision system for cerebral hemorrhage in this application;

[0024] Figure 2 A schematic diagram of a scenario for clustering processing in this application;

[0025] Figure 3 A schematic diagram of a scenario for the working logic of this application scheme;

[0026] Figure 4 A structural schematic diagram of a processing device of a hematoma aspiration decision system for cerebral hemorrhage in the present application;

[0027] Figure 5 A structural schematic diagram of the hematoma aspiration decision system of this application. DETAILED DESCRIPTION

[0028] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of this application.

[0029] The terms "first", "second", etc. in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described here can be implemented in an order other than that illustrated or described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or modules is not necessarily limited to those steps or modules clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products or devices. The naming or numbering of steps in this application does not mean that the steps in the method flow must be executed in the time / logical sequence indicated by the naming or numbering. The process steps that have been named or numbered can change the execution order according to the technical purpose to be achieved, as long as the same or similar technical effects can be achieved.

[0030] The division of modules in this application is a logical division. There may be other division methods when it is implemented in actual applications. For example, multiple modules can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, and the indirect coupling or communication connection between modules can be electrical or other similar forms, which are not limited in this application. In addition, the modules or submodules described as separate components may or may not be physically separated, may or may not be physical modules, or may be distributed in multiple circuit modules, and some or all of the modules may be selected according to actual needs to achieve the purpose of the present application.

[0031] Before introducing the processing method of the hematoma aspiration decision system for cerebral hemorrhage provided by the present application, the background content involved in the present application is first introduced.

[0032] The processing method, device and computer-readable storage medium of the hematoma aspiration decision system for cerebral hemorrhage provided in the present application can be applied to the hematoma aspiration decision system. A two-layer deep neural network is constructed for the processing of the hematoma aspiration plan. The time-series-based perception-decision-control method implemented thereby overcomes the unknowability of the results of the hematoma aspiration process existing in the prior art, provides direct and convenient information transmission for doctors' decision-making, can effectively improve the accuracy of the cerebral hematoma aspiration treatment plan, and provide strong digital and intelligent support for clinical hematoma aspiration treatment.

[0033] The processing method of the hematoma aspiration decision system for cerebral hemorrhage mentioned in the present application can be executed by a processing device of the hematoma aspiration decision system for cerebral hemorrhage, or a server, physical host or user equipment (UE) and other different types of hematoma aspiration decision systems that integrate the processing device of the hematoma aspiration decision system for cerebral hemorrhage. Among them, the processing device of the hematoma aspiration decision system for cerebral hemorrhage can be implemented in hardware or software, and the UE can be a terminal device such as a smart phone, tablet computer, laptop computer, desktop computer or personal digital assistant (PDA), and the hematoma aspiration decision system can be set up in the form of a device cluster.

[0034] It can be understood that the hematoma aspiration decision system that executes the processing method of the hematoma aspiration decision system for cerebral hemorrhage of the present application, or the hematoma aspiration decision system equipped with the application service corresponding to the processing method of the hematoma aspiration decision system for cerebral hemorrhage of the present application, can be deployed on a device that is specifically responsible for auxiliary work in hematoma aspiration treatment in clinical application scenarios, or it can be deployed on a device that is responsible for other work (such as other clinical treatments or even routine office work) in addition to auxiliary work in hematoma aspiration treatment. It is obviously more flexible, and it can be either an on-site device or a device that provides remote services to the site through remote services such as web, which is also possible in actual situations. In this way, the specific device type and device deployment form of the hematoma aspiration decision system can be flexibly adjusted according to actual needs, and the present application does not make specific limitations.

[0035] Next, we will begin to introduce the processing method of the hematoma aspiration decision system for cerebral hemorrhage provided by this application.

[0036] First, see Figure 1 , Figure 1A flowchart of a processing method of a hematoma aspiration decision system for cerebral hemorrhage provided by the present application is shown. The processing method of a hematoma aspiration decision system for cerebral hemorrhage provided by the present application may specifically include the following steps S101 to S104:

[0037] Step S101, obtaining pre-operative data of hematoma aspiration, wherein the pre-operative data of hematoma aspiration involves a first indicator;

[0038] It can be understood that the data support for hematoma aspiration decision in this application involves two stages: preoperative and intraoperative. In this regard, data will be collected according to corresponding indicators before and during the operation to provide a basis for the two-layer deep neural network to process the hematoma aspiration plan.

[0039] For the convenience of explanation, the data obtained before the operation is recorded as hematoma aspiration preoperative data, which involves the first indicator and corresponds to the first deep neural network; the data obtained in real time during the operation is recorded as real-time hematoma aspiration intraoperative data, which involves the second indicator and corresponds to the second deep neural network.

[0040] As an exemplary embodiment here, the first indicator involved in the pre-hematoma aspiration data may specifically include:

[0041] Indicators include age, NIHSS score (a common stroke rating scale in clinical practice that quantifies severity by scoring), medical history, blood pressure, blood oxygen, heart rate, intracranial pressure, hematoma location, hematoma volume, CT images and MRI images (both can be called preoperative images).

[0042] It can be understood that a specific implementation plan is provided here for the content of pre-operative data of hematoma aspiration.

[0043] Step S101 is for the acquisition and processing of preoperative data of hematoma aspiration. Generally speaking, it is the acquisition of ready-made data, for example, it can be directly retrieved from the relevant system, or it can be manually entered. Of course, the possibility of real-time data collection is not ruled out, but this means that the system needs to be modified in terms of software and hardware. Considering that these data themselves are relatively common data in clinical practice, how to collect data and the corresponding data collection equipment / devices involved are not specifically explained here.

[0044] After obtaining the preoperative data of hematoma aspiration, preprocessing may also be involved. For example, digital methods can be used to digitize data of multiple modalities, and then normalization or standardization methods can be used to unify the digital features under the same metric, so as to form data that is more convenient for the model to process. Among them, preprocessing in specific operations can be handed over to corresponding automated processing tools or even AI products such as deep neural networks. This is relatively flexible and can achieve the goal of data preprocessing.

[0045] It is worth adding that step S101 is about the acquisition and processing of the pre-operative data of hematoma aspiration, which may also directly involve the pre-processing mentioned above, which is also possible in actual situations.

[0046] Step S102, inputting the preoperative data of hematoma aspiration into the first deep neural network to determine an adapted initial hematoma aspiration scheme before the operation. In fact, the first deep neural network is used to perform multi-center clustering on the preoperative data of hematoma aspiration input into the network, and outputting a hematoma aspiration scheme pre-adapted by the most suitable cluster center;

[0047] refer to Figure 2 A schematic diagram of a scenario of clustering processing of the present application is shown. It can be understood that before hematoma aspiration, the present application will determine the hematoma aspiration scheme suitable for the current patient (user) based on the acquired pre-operative hematoma aspiration data. In this process, multi-center clustering is performed specifically through the first deep neural network. It can be understood that the present application pre-sets adapted hematoma aspiration schemes for different clustering centers (clustering clusters). Therefore, after completing the clustering processing of multi-center clustering, the most suitable clustering center for the current patient situation can be determined (the most suitable clustering center is determined by the shortest distance principle). The hematoma aspiration scheme pre-adapted by the most suitable clustering center is the initial hematoma aspiration scheme that the network can output, and the initial hematoma aspiration scheme is the hematoma aspiration scheme determined by digital and intelligent means in the preoperative stage. The subsequent decision-making and prediction of the hematoma aspiration scheme during the operation are based on the initial hematoma aspiration scheme.

[0048] Among them, for the first deep neural network, its main purpose is to realize multi-center clustering processing. The specific model architecture adopted is obviously more flexible and can be configured according to actual conditions. For example, the existing model architecture can be adopted, or it can be optimized and improved on the basis of the existing model architecture. For example, a completely self-developed and more novel model architecture can be adopted.

[0049] As for the different cluster centers (cluster clusters) involved in the clustering process, it can be understood that this application is pre-configured based on different sample data (including real patient data, data transformed on the basis of real patient data, and directly configured virtual patient data), corresponding to the subsequent clustering processing that may be involved. After the different cluster centers are determined, the corresponding adapted hematoma aspiration schemes can be configured one by one, so that the clustering processing can be carried out through the model to obtain the initial hematoma aspiration scheme pre-adapted to the most suitable cluster center.

[0050] As an example, the application may have the following processing in advance:

[0051] (1), obtain the existing case set and record it as ,in, Represents a single case;

[0052] (2), randomly select k points from D As a starting point;

[0053] (3), calculate all points arrive The distance function is calculated using the Euler distance, and the corresponding Euler distance calculation formula is as follows:

[0054] ,

[0055] Finally, the distance matrix of the points is obtained:

[0056] ,

[0057] (4), each point Divide into the category of the nearest center point, obtain k classified point clusters, and recalculate the center point of the point cluster, also using the Euler distance;

[0058] (5) Repeat (3) and (4) until the center point of the cluster no longer changes. The final centers of the multiple clusters are the center points of the k classes.

[0059] (6), finally, according to the final point cluster center , divided into k classes , these classes are classifications of patient characteristics, commonly known as patient categories.

[0060] Since each predetermined class is configured with a corresponding and adapted hematoma aspiration scheme (scheme template), after being put into actual use, when new patient data x is input, x and the point cluster center are calculated Distance , select the center point c with the smallest distance as the center class (cluster center) of x, so that the patient's category can be determined, and the corresponding initial hematoma aspiration plan can be determined, the initial treatment plan can be obtained, and the initial decision can be completed.

[0061] Specifically, as an exemplary embodiment herein, the content of the initial hematoma aspiration program may include:

[0062] The tolerance range of intracranial pressure changes, single aspiration time, hematoma status, whether other drug injections are needed (including hematoma liquefaction agents, etc.) and maximum aspiration speed, etc.

[0063] Step S103, obtaining real-time hematoma aspiration intraoperative data, wherein the real-time hematoma aspiration intraoperative data involves a second indicator;

[0064] It can be understood that after obtaining the initial hematoma aspiration plan, the corresponding hematoma aspiration treatment can be carried out. During the hematoma aspiration treatment process or during the operation, the real-time decision-making and prediction of the hematoma aspiration plan of this application can be continued to be promoted to achieve real-time and accurate hematoma aspiration plan control effect.

[0065] To this end, it is necessary to obtain real-time hematoma aspiration intraoperative data corresponding to the second indicator in real time during the operation to provide a reference for the subsequent processing of the second deep neural network.

[0066] As an exemplary embodiment here, the second indicator involved in the data during hematoma aspiration may specifically include:

[0067] Suction speed, fluid inflow speed, blood pressure, blood oxygen, heart rate, intracranial pressure, hematoma location and hematoma size, etc.

[0068] The suction speed and liquid inlet speed can be measured by a micro flow meter, which is used to directly detect the liquid flow rate in the inlet and outlet channels. A gravimeter is used to calculate the increase and decrease in mass per unit time to correct the inlet and outlet flow rates. The final flow rate can be expressed as follows:

[0069] ,

[0070] ,

[0071] in, represents the directly measured liquid inlet velocity, It represents the mass of medicine or drug input per unit time t. Indicates the density of the input drug or medicine, represents the directly measured liquid suction velocity, It represents the mass of hematoma sucked out per unit time t, Indicates the density of the aspirated hematoma;

[0072] For blood pressure, it can be measured with a sphygmomanometer;

[0073] Blood oxygen can be measured using a blood oximeter;

[0074] For heart rate, it can be measured using a heart rate monitor;

[0075] Intracranial pressure can be measured using an intracranial pressure monitor;

[0076] The location of the hematoma can be measured using a near-infrared measuring instrument;

[0077] As for the size of hematoma, it cannot be measured directly. The position of hematoma can be measured by near-infrared probe, and the current size of hematoma can be calculated based on the mass of hematoma aspirated.

[0078] Step S104, inputting the real-time hematoma aspiration intraoperative data into a second deep neural network constructed by an LSTM network and pre-adapted with the most suitable cluster center, so as to determine the hematoma aspiration plan decision result and the hematoma aspiration plan prediction result during the operation, wherein the second deep neural network is used to determine the timing control plan corresponding to the real-time hematoma aspiration process based on the real-time hematoma aspiration intraoperative data input by the network, and the timing control plan includes the decision-making and prediction plan contents of the hematoma aspiration plan.

[0079] It can be seen that for the second deep neural network, this application specifically uses LSTM to create it. LSTM, that is, long short-term memory network, Long Short-Term Memory, is aimed at the prediction of time series data. That is to say, this application regards the real-time hematoma aspiration plan during surgery as a kind of time series data, and makes decisions and predictions based on this to achieve the control goal of real-time and high-precision control of the hematoma aspiration plan.

[0080] In addition, it should be noted that, for the second deep neural network of the present application, there is a matching mechanism with the cluster centers involved in the previous clustering processing, that is, different cluster centers are pre-adapted with different second deep neural networks. In this way, after determining the most suitable cluster center for the current patient condition, the second deep neural network pre-adapted to the most suitable cluster center can be used as a basis. Compared with the second deep neural networks pre-adapted to other cluster centers, the real-time decision and prediction of the hematoma aspiration plan can have better matching, thereby promoting better treatment and suction effects.

[0081] In specific operations, the first deep neural network and the second deep neural network can form a multi-center deep neural network. In this case, the first deep neural network, that is, the first layer, can be understood as a multi-center clustering layer, and the second deep neural network, that is, the second layer, can be understood as a multi-center real-time hematoma aspiration network.

[0082] During the operation, since the second deep neural network is built for the most suitable center in front, at the beginning, it will be based on the initial hematoma aspiration plan determined before the operation or adopted at the beginning, combined with the real-time hematoma aspiration intraoperative data, to analyze and process the hematoma aspiration plans at different time sequences. The analysis result can be called a timing control plan. In the timing control plan, the hematoma aspiration plan for the current time sequence is recommended for direct use, that is, the decision result of the hematoma aspiration plan. For the hematoma aspiration plan for the subsequent time sequence, it is the predicted result of the hematoma aspiration plan. When the subsequent time sequence is reached in actual circumstances, the new real-time hematoma aspiration intraoperative data of the subsequent time sequence can be combined to carry out a new round of analysis and processing of the hematoma aspiration plans at different time sequences, so as to continuously iterate and update, and continuously adjust the adopted / applied hematoma aspiration plan in real time, so as to achieve real-time and high-precision hematoma aspiration plan control effect.

[0083] As an exemplary embodiment herein, the decision result of the hematoma aspiration scheme output by the second deep neural network may specifically include the control amount of the micro-metering pump for both the fluid inlet and fluid outlet.

[0084] The prediction result of the hematoma aspiration plan output by the second deep neural network can specifically include two parts: the prediction result quantity and the prediction decision quantity. The former, i.e., the prediction result quantity, can include blood pressure, blood oxygen, heart rate, intracranial pressure, hematoma location and hematoma size; the latter, i.e., the prediction decision quantity, can include fluid intake and fluid output.

[0085] In addition, for the second deep neural network, as an exemplary embodiment, the timing control scheme involved is expressed as:

[0086] ,

[0087] in, is the control plan at time t (hematoma aspiration plan), T is the maximum value of t (i.e. t is the time variable, the maximum value is T), T corresponds to the total time (the total duration of hematoma aspiration treatment),

[0088] ,

[0089] Includes several states at time t, including control state , static state and dynamic state , specifically, there are:

[0090] for , which is used to control the action at the next moment, including the type of liquid inlet, liquid inlet speed and suction speed,

[0091] for , is not a specific value, but a descriptive value. As a status monitoring, including heart rate, blood pressure and blood oxygen, it can include specific descriptive values ​​such as "normal", "high", "low", etc.

[0092] for , used to compare with the actual value to correct the control plan at the next moment , including intracranial pressure, hematoma size, and hematoma location,

[0093] Control plan for the next moment Calculated by the following formula:

[0094] ,

[0095] in, is the actual status at time t, including the actually measured intracranial pressure, hematoma size and hematoma location, , and are the different network parameters of the second deep neural network (i.e., the network parameters involved in the LSTM model itself), and It is used to calculate the decision error, and use the decision error to correct the decision value at the next moment to obtain the actual decision.

[0096] For a more vivid understanding, you can also refer to Figure 3 A schematic diagram of a scenario of the working logic of the present application scheme is shown. Most of the above processing is handled by the decision module. (the actual state at time t) is obtained by the perception module, (Control state) is executed by the control module, (Dynamic state) is predicted by the prediction module.

[0097] In addition, the present application scheme may also be configured with a visualization mechanism to show doctors or medical staff the relevant circumstances during the hematoma aspiration treatment process assisted by the present application scheme.

[0098] Correspondingly, the present application method may further include:

[0099] Through the visual interface, display , , , , and decision errors.

[0100] It can be understood that the display screen involved in the visualization interface can be either a structure of the system itself or an external device (including a dedicated display device or other device with a display screen).

[0101] On the basis of the visual interface, the present application scheme can also introduce a human-machine fusion mechanism, so that doctors can make empirical judgments based on their work experience, so as to make more suitable hematoma aspiration plan control that is difficult for machines to learn in the first time in some special cases. At the same time, this also plays a bottom-line role and provides an entry point for manual intervention.

[0102] In this regard, as an exemplary embodiment, the method of the present application may further include:

[0103] Receive control actions entered by decision-makers (usually doctors) ,in, It includes at least one of changing the type of liquid inlet, adjusting the liquid inlet speed and adjusting the suction speed;

[0104] implement and , and Replace with ,in, In , and the rest same, The calculation method is changed to:

[0105] ,

[0106] ,

[0107] .

[0108] In this way, at a practical level, the hematoma aspiration decision plan adjusted by the doctor's decision can ensure better stability and safety in specific operations.

[0109] In addition to being able to adjust decisions based on a visual interface, doctors or medical staff can also configure an emergency stop control button to limit the occurrence of dangerous situations and issue an alarm at the same time.

[0110] At the same time, with the assistance of the decision-making system of the present application, the process of hematoma aspiration treatment is easy to understand and may also involve a data recording mechanism.

[0111] For this purpose, a time interval (e.g. 50ms) can be set, and the data collected after each time interval (dynamic state) and (Dynamic state), record (static state) and dynamic state (Dynamic state), if the doctor has decision control, record If the doctor does not make a decision, then Record it as 0. For these data, create an independent document and record them at regular time intervals until the treatment is completed, recording the entire process of hematoma aspiration treatment.

[0112] In addition, in order to facilitate data recording and promote better data utilization, the present application can also be configured in the form of a decision matrix.

[0113] For this purpose, you can also , , , , and The digitized eigenvalues ​​are standardized and recorded as a decision matrix as shown below:

[0114] ,

[0115] In the decision matrix, each column represents all states and decision feature values ​​at a certain moment, and each row represents the value of a certain feature at all moments.

[0116] In addition, the present application can also optimize the previous second neural network (LSTM network) based on the actual situation of the current patient's hematoma aspiration treatment / control plan, so as to achieve the effect of iteratively updating the second neural network during actual use.

[0117] In this regard, it can also be combined with the processing settings of the above decision matrix. As an exemplary embodiment, the method of the present application can also include:

[0118] Will , , , , and The digitized eigenvalues ​​are standardized and recorded as a decision matrix as shown below:

[0119] ,

[0120] In the decision matrix, each column represents all states and decision feature values ​​at a certain moment, and each row represents the value of a certain feature at all moments;

[0121] The error between the decision matrix and the actual perception value is calculated by the following formula:

[0122] ,

[0123] in, is the decision output value of the decision model, is the actual output value;

[0124] The error is used as the gradient direction for the second deep neural network involved , and Gradient optimization is performed, and each moment (t value) is gradually optimized. The final optimized model is used as the second deep neural network pre-adapted with the most suitable clustering center, and the current patient's information is classified into the point cluster of the most suitable clustering center for subsequent clustering of the first deep neural network.

[0125] In addition, it is understandable that for the first deep neural network and the second deep neural network, the solution of the present application can also be handed over to the cloud server for processing (including storage maintenance and updating, etc.), so that remote data support for the model can be provided for different solution deployment objects (in actual cases, it may involve objects of different granularity / levels such as hospitals, departments or doctors), and even remote data support for the entire solution of the present application can be directly provided.

[0126] In this case, in the above embodiment, all the parameters involved can be uploaded to the cloud server, which uses its powerful data processing capabilities to perform complex optimization on the second deep neural network. The optimized model parameters are then sent down for the solution deployment object that originally uploaded the data to update the model. In this way, for the local solution deployment object, the individuality of the treatment of patients in different regions and with different characteristics is improved, and it is more targeted and matched, further improving the effectiveness of hematoma aspiration treatment. In general, a multi-center intelligent optimization effect can be achieved (multi-center means multi-solution deployment objects).

[0127] If the second neural network is updated locally autonomously, it is obvious that the model is updated locally autonomously, which can also improve the individuality of treatment for patients in local areas and with local characteristics, and has better autonomy.

[0128] Regarding the above solution contents, in general, it can be seen that, for the auxiliary goal of hematoma aspiration treatment, the present application constructs a two-layer deep neural network for the processing of hematoma aspiration scheme. The timing-based perception-decision-control method thus implemented overcomes the unknowability of the results of the hematoma aspiration process existing in the prior art, provides direct and convenient information transmission for doctors' decision-making, can effectively improve the accuracy of the cerebral hematoma aspiration treatment plan, and provide strong digital and intelligent support for clinical hematoma aspiration treatment.

[0129] The above is an introduction to the processing method of the hematoma aspiration decision system for cerebral hemorrhage provided by the present application. In order to facilitate better implementation of the processing method of the hematoma aspiration decision system for cerebral hemorrhage provided by the present application, the present application also provides a processing device for the hematoma aspiration decision system for cerebral hemorrhage from the perspective of functional modules.

[0130] See also Figure 4 , Figure 4 This is a structural schematic diagram of a processing device of a hematoma aspiration decision system for cerebral hemorrhage in this application. In this application, the processing device 400 of the hematoma aspiration decision system for cerebral hemorrhage may specifically include the following structure:

[0131] A first acquisition unit 401 is used to acquire pre-operative data of hematoma aspiration, wherein the pre-operative data of hematoma aspiration involves a first indicator;

[0132] The first processing unit 402 is used to input the preoperative data of hematoma aspiration into the first deep neural network to determine an adapted initial hematoma aspiration scheme before the operation. In fact, the first deep neural network is used to perform multi-center clustering on the preoperative data of hematoma aspiration input into the network, and output a hematoma aspiration scheme pre-adapted by the most adapted cluster center;

[0133] A second acquisition unit 403 is used to acquire real-time hematoma aspiration intraoperative data, wherein the real-time hematoma aspiration intraoperative data involves a second indicator;

[0134] The second processing unit 404 is used to input the real-time hematoma aspiration intraoperative data into a second deep neural network constructed by an LSTM network and pre-adapted with the most suitable cluster center, so as to determine the hematoma aspiration plan decision result and the hematoma aspiration plan prediction result during the operation, wherein the second deep neural network is used to determine the timing control plan corresponding to the real-time hematoma aspiration process based on the real-time hematoma aspiration intraoperative data input by the network, and the timing control plan includes the decision-making and prediction plan contents of the hematoma aspiration plan.

[0135] In an exemplary embodiment, the first indicator involved in the pre-hematoma aspiration data specifically includes:

[0136] Age, NIHSS score, medical history, blood pressure, blood oxygen, heart rate, intracranial pressure, hematoma location, hematoma volume, CT images, and MRI images;

[0137] The second indicator involved in the data of hematoma aspiration includes:

[0138] Aspiration rate, fluid inflow rate, blood pressure, blood oxygen, heart rate, intracranial pressure, hematoma location and hematoma size.

[0139] In an exemplary embodiment, the initial hematoma aspiration protocol includes:

[0140] The tolerance range of intracranial pressure changes, single aspiration time, hematoma status, whether other drug injections are needed and maximum aspiration speed.

[0141] In yet another exemplary embodiment, the hematoma aspiration scheme decision result includes the control amount of the micro-metering pump for both the inflow and outflow amounts;

[0142] The predicted results of the hematoma aspiration plan include predicted result quantities and predicted decision quantities, wherein the predicted result quantities include blood pressure, blood oxygen, heart rate, intracranial pressure, hematoma location and hematoma size, and the predicted decision quantities include fluid intake and fluid output.

[0143] In yet another exemplary embodiment, for the second deep neural network, the timing control scheme is expressed as:

[0144] ,

[0145] in, is the control scheme at time t, T is the maximum value of t, T corresponds to the total time,

[0146] ,

[0147] Includes several states at time t, including control state , static state and dynamic state , specifically, there are:

[0148] for , which is used to control the action at the next moment, including the type of liquid inlet, liquid inlet speed and suction speed,

[0149] for , is not a specific value, but a descriptive value, used as status monitoring, including heart rate, blood pressure and blood oxygen,

[0150] for , used to compare with the actual value to correct the control plan at the next moment , including intracranial pressure, hematoma size, and hematoma location,

[0151] Control plan for the next moment Calculated by the following formula:

[0152] ,

[0153] in, is the actual status at time t, including the actually measured intracranial pressure, hematoma size and hematoma location, , and are different network parameters of the second deep neural network, and It is used to calculate the decision error, and use the decision error to correct the decision value at the next moment to obtain the actual decision.

[0154] In yet another exemplary embodiment, the apparatus further comprises:

[0155] The display unit 405 is used to display , , , , and decision errors;

[0156] Receiving unit 406, used to receive the control action entered by the decision-making participant ,in, It includes at least one of changing the type of liquid inlet, adjusting the liquid inlet speed and adjusting the suction speed;

[0157] Execution unit 407, used to execute and , and Replace with ,in, In , and the rest same, The calculation method is changed to:

[0158] ,

[0159] ,

[0160] .

[0161] In yet another exemplary embodiment, the apparatus further comprises:

[0162] The recording unit 408 is used to record , , , , and The digitized eigenvalues ​​are standardized and recorded as a decision matrix as shown below:

[0163] ,

[0164] In the decision matrix, each column represents all states and decision feature values ​​at a certain moment, and each row represents the value of a certain feature at all moments;

[0165] The calculation unit 409 is used to calculate the error between the decision matrix and the actual perception value by the following formula:

[0166] ,

[0167] in, is the decision output value of the decision model, is the actual output value;

[0168] An updating unit 410 is used to update the second deep neural network involved in the error as a gradient direction , and Gradient optimization is performed, and each moment is gradually optimized. The final optimized model is used as the second deep neural network pre-adapted with the most suitable clustering center, and the current patient's information is classified into the point cluster of the most suitable clustering center for subsequent clustering of the first deep neural network.

[0169] This application also provides a hematoma aspiration decision system from the perspective of hardware structure, which is regarded as a processing device, see Figure 5 , Figure 5 5 shows a structural schematic diagram of the hematoma suction decision system of the present application. Specifically, the hematoma suction decision system of the present application may include a processor 501, a memory 502, and an input / output device 503. The processor 501 is used to execute the computer program stored in the memory 502 to implement the following Figure 1 The steps of the processing method of the hematoma aspiration decision system for cerebral hemorrhage in the corresponding embodiment; or, the processor 501 is used to execute the computer program stored in the memory 502 to implement the following Figure 4 Corresponding to the functions of each unit in the embodiment, the memory 502 is used to store the processor 501 executing the above Figure 1 The computer program required for the processing method of the hematoma aspiration decision system for cerebral hemorrhage in the corresponding embodiment.

[0170] Exemplarily, the computer program may be divided into one or more modules / units, one or more modules / units are stored in the memory 502, and executed by the processor 501 to complete the present application. One or more modules / units may be a series of computer program instruction segments capable of completing specific functions, and the instruction segments are used to describe the execution process of the computer program in the computer device.

[0171] The hematoma suction decision system may include, but is not limited to, a processor 501, a memory 502, and an input-output device 503. Those skilled in the art will appreciate that the illustration is merely an example of a hematoma suction decision system and does not constitute a limitation on the hematoma suction decision system. The hematoma suction decision system may include more or fewer components than shown in the illustration, or a combination of certain components, or different components. For example, the hematoma suction decision system may also include a network access device, a bus, etc., and the processor 501, the memory 502, the input-output device 503, etc. are connected via a bus.

[0172] The processor 501 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the hematoma aspiration decision system, and uses various interfaces and lines to connect various parts of the entire device.

[0173] The memory 502 can be used to store computer programs and / or modules. The processor 501 implements various functions of the computer device by running or executing the computer programs and / or modules stored in the memory 502 and calling the data stored in the memory 502. The memory 502 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function, etc.; the data storage area can store data created according to the use of the hematoma aspiration decision system, etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0174] When the processor 501 is used to execute the computer program stored in the memory 502, the following functions can be implemented:

[0175] Acquiring pre-operative data of hematoma aspiration, wherein the pre-operative data of hematoma aspiration involves a first indicator;

[0176] The preoperative data of hematoma aspiration is input into the first deep neural network to determine the adapted initial hematoma aspiration scheme before the operation. In fact, the first deep neural network is used to perform multi-center clustering on the preoperative data of hematoma aspiration input into the network, and output the hematoma aspiration scheme pre-adapted by the most adapted cluster center;

[0177] Acquiring real-time intraoperative data of hematoma aspiration, wherein the real-time intraoperative data of hematoma aspiration involves a second indicator;

[0178] The real-time hematoma aspiration intraoperative data is input into a second deep neural network constructed by an LSTM network and pre-adapted with the most suitable clustering center to determine the hematoma aspiration plan decision result and the hematoma aspiration plan prediction result during the operation, wherein the second deep neural network is used to determine the timing control plan corresponding to the real-time hematoma aspiration process based on the real-time hematoma aspiration intraoperative data input by the network, and the timing control plan includes the decision-making and prediction plan contents of the hematoma aspiration plan.

[0179] A person skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process of the processing device of the hematoma aspiration decision system for cerebral hemorrhage, the hematoma aspiration decision system and its corresponding units described above can refer to the following. Figure 1 The description of the processing method of the hematoma aspiration decision system for cerebral hemorrhage in the corresponding embodiment will not be repeated here.

[0180] A person of ordinary skill in the art will appreciate that all or part of the steps in the various methods of the above embodiments may be completed by instructions, or by controlling related hardware through instructions. The instructions may be stored in a computer-readable storage medium and loaded and executed by a processor.

[0181] To this end, the present application provides a computer-readable storage medium, in which a plurality of instructions are stored, and the instructions can be loaded by a processor to execute the present application as follows: Figure 1 The steps of the processing method of the hematoma aspiration decision system for cerebral hemorrhage in the corresponding embodiment, the specific operation can refer to the following Figure 1 The description of the processing method of the hematoma aspiration decision system for cerebral hemorrhage in the corresponding embodiment will not be repeated here.

[0182] The computer-readable storage medium may include: a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0183] Due to the instructions stored in the computer-readable storage medium, the present application can be executed. Figure 1The steps of the processing method of the hematoma aspiration decision system for cerebral hemorrhage in the corresponding embodiment, therefore, the present application can be implemented as follows Figure 1 The beneficial effects that can be achieved by the processing method of the hematoma aspiration decision system for cerebral hemorrhage in the corresponding embodiment are detailed in the previous description and will not be repeated here.

[0184] The above is a detailed introduction to the processing method, device, hematoma aspiration decision system and computer-readable storage medium for the hematoma aspiration decision system for cerebral hemorrhage provided by the present application. This article uses specific examples to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea; at the same time, for technical personnel in this field, according to the ideas of the present application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A processing method for a hematoma aspiration decision system for cerebral hemorrhage, characterized in that: The method comprises: Acquiring pre-operative data of hematoma aspiration, wherein the pre-operative data of hematoma aspiration relates to a first indicator; Inputting the preoperative data of hematoma aspiration into a first deep neural network to determine an adapted initial hematoma aspiration scheme before surgery, wherein the first deep neural network is used to perform multi-center clustering on the preoperative data of hematoma aspiration input into the network, and outputting a hematoma aspiration scheme pre-adapted by the most adapted cluster center; Acquiring real-time intraoperative data of hematoma aspiration, wherein the real-time intraoperative data of hematoma aspiration relates to a second indicator; Inputting the real-time hematoma aspiration intraoperative data into a second deep neural network constructed by an LSTM network and pre-adapted by the most suitable cluster center, so as to determine the decision result of the hematoma aspiration scheme and the prediction result of the hematoma aspiration scheme during the operation, wherein the second deep neural network is used to determine a timing control scheme corresponding to the real-time hematoma aspiration process based on the real-time hematoma aspiration intraoperative data input by the network, and the timing control scheme includes the decision-making and prediction scheme contents of the hematoma aspiration scheme; For the second deep neural network, the timing control scheme is expressed as: ， in, is the control scheme at time t, T is the maximum value of t, T corresponds to the total time, ， Including several states at time t, including control state , static state and dynamic state , specifically, there are: for , which is used to control the action at the next moment, including the type of liquid inlet, liquid inlet speed and suction speed, for , not a specific value, but a descriptive value, as a status monitoring, including heart rate, blood pressure and blood oxygen, for , used to compare with the actual value to correct the control plan at the next moment , including intracranial pressure, hematoma size, and hematoma location, The control scheme at the next moment Calculated by the following formula: ， in, is the actual state at time t, including the actually measured intracranial pressure, hematoma size and hematoma location, , and are different network parameters of the second deep neural network, and It is used to calculate the decision error, and use the decision error to correct the decision value at the next moment to obtain the actual decision.

2. The method according to claim 1, characterized in that The first indicator involved in the pre-hematoma aspiration data specifically includes: Age, NIHSS score, medical history, blood pressure, blood oxygen, heart rate, intracranial pressure, hematoma location, hematoma volume, CT images, and MRI images; The second indicator involved in the data during the hematoma aspiration operation specifically includes: Aspiration rate, fluid inflow rate, blood pressure, blood oxygen, heart rate, intracranial pressure, hematoma location and hematoma size.

3. The method according to claim 1, characterized in that The initial hematoma aspiration protocol includes: The tolerance range of intracranial pressure changes, single aspiration time, hematoma status, whether other drug injections are needed and maximum aspiration speed.

4. The method according to claim 1, characterized in that: The hematoma aspiration scheme decision result includes the control amount of the micro-metering pump for both the inflow and outflow of liquid; The predicted result of the hematoma aspiration plan includes a predicted result quantity and a predicted decision quantity, wherein the predicted result quantity includes blood pressure, blood oxygen, heart rate, intracranial pressure, hematoma location and hematoma size, and the predicted decision quantity includes the fluid intake and the fluid output.

5. The method according to claim 1, characterized in that The method further comprises: Through the visual interface, display , , , , and said decision error; Receive control actions entered by decision-makers ,in, It includes at least one of changing the type of liquid inlet, adjusting the liquid inlet speed and adjusting the suction speed; implement and , and Replace with ,in, In , and the rest same, The calculation method is changed to: ， ， 。 6. The method according to claim 5, characterized in that The method further comprises: Will , , , , and The digitized eigenvalues ​​are standardized and recorded as a decision matrix as shown below: ， Wherein, in the decision matrix, each column represents all states and decision feature values ​​at a certain moment, and each row represents the value of a certain feature at all moments; The error between the decision matrix and the actual perception value is calculated by the following formula: ， in, is the decision output value of the decision model, is the actual output value; The error is used as the gradient direction for the second deep neural network involved , and Gradient optimization is performed, and each moment is gradually optimized. The final optimized model is used as the second deep neural network pre-adapted to the most suitable clustering center, and the information of the current patient is classified into the point cluster of the most suitable clustering center for subsequent clustering of the first deep neural network.

7. A processing device for a hematoma aspiration decision system for cerebral hemorrhage, characterized in that: The device comprises: A first acquisition unit is used to acquire pre-operative data of hematoma aspiration, wherein the pre-operative data of hematoma aspiration relates to a first indicator; A first processing unit is used to input the preoperative data of hematoma aspiration into a first deep neural network to determine an adapted initial hematoma aspiration scheme before surgery, wherein the first deep neural network is used to perform multi-center clustering on the preoperative data of hematoma aspiration input into the network, and output a hematoma aspiration scheme pre-adapted by the most adapted cluster center; A second acquisition unit is used to acquire real-time hematoma aspiration intraoperative data, wherein the real-time hematoma aspiration intraoperative data relates to a second indicator; A second processing unit is used to input the real-time hematoma aspiration intraoperative data into a second deep neural network constructed by an LSTM network and pre-adapted by the most suitable cluster center, so as to determine the decision result of the hematoma aspiration scheme and the prediction result of the hematoma aspiration scheme during the operation, wherein the second deep neural network is used to determine a timing control scheme corresponding to the real-time hematoma aspiration process based on the real-time hematoma aspiration intraoperative data input by the network, and the timing control scheme includes the decision-making and prediction scheme contents of the hematoma aspiration scheme; For the second deep neural network, the timing control scheme is expressed as: ， in, is the control scheme at time t, T is the maximum value of t, T corresponds to the total time, ， Including several states at time t, including control state , static state and dynamic state , specifically, there are: for , which is used to control the action at the next moment, including the type of liquid inlet, liquid inlet speed and suction speed, for , not a specific value, but a descriptive value, as a status monitoring, including heart rate, blood pressure and blood oxygen, for , used to compare with the actual value to correct the control plan at the next moment , including intracranial pressure, hematoma size, and hematoma location, The control scheme at the next moment Calculated by the following formula: ， in, is the actual state at time t, including the actually measured intracranial pressure, hematoma size and hematoma location, , and are different network parameters of the second deep neural network, and It is used to calculate the decision error, and use the decision error to correct the decision value at the next moment to obtain the actual decision.

8. A hematoma aspiration decision system, characterized in that: The method comprises a processor and a memory, wherein the memory stores a computer program, and the processor executes the method according to any one of claims 1 to 6 when calling the computer program in the memory.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a plurality of instructions, and the instructions are suitable for being loaded by a processor to execute the method according to any one of claims 1 to 6.

Images