An automated evaluation system and method for animal experimental procedures based on spatiotemporal multidimensional fusion.

Abstract

The application discloses an animal experiment operation automatic evaluation system and method based on space-time multi-dimension fusion, and belongs to the technical field of experiment operation automatic evaluation. The system comprises a video collection module, a physiological signal collection module and a multi-dimension data analysis and control module, wherein the multi-dimension data analysis and control module comprises a time sequence analysis unit, a space analysis unit and a comprehensive evaluation unit. The physiological complexity of operation results is analyzed by introducing multi-element multi-scale entropy, the space standardization of operation processes is analyzed by a convolutional neural network, and evaluation parameters are dynamically optimized by a neural network, so that more accurate, objective and comprehensive automatic evaluation is realized.

Description

Technical Field

[0001] This invention relates to the field of automated evaluation technology for experimental procedures, and more specifically to an automated evaluation system and method for animal experimental procedures based on spatiotemporal multidimensional fusion. Background Technology

[0002] The evaluation of animal experiments (such as functional experiments) has long relied on manual observation and scoring, which has problems such as low efficiency, strong subjectivity and difficulty in standardization.

[0003] In existing technologies, although some solutions propose to analyze experimental operation steps through video and analyze physiological signals through sensors, most of them are simple judgments based on independent dimensions and fail to deeply integrate the standardization of operation (spatial dimension) with the complexity of animal physiological responses (temporal dimension).

[0004] Furthermore, the analysis of physiological signals is often limited to traditional time and frequency domain characteristics, neglecting the nonlinear dynamics of physiological systems under pathological or stress states, such as changes in entropy. At the same time, the parameters of evaluation models are usually fixed and cannot be adaptively optimized for different experimental stages or animal states, thus limiting the accuracy of evaluations.

[0005] Therefore, how to provide an automated evaluation system and method for animal experimental operations based on spatiotemporal multi-dimensional fusion is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0006] In view of this, the present invention provides an automatic evaluation system and method for animal experimental operations based on spatiotemporal multidimensional fusion. By introducing multivariate multiscale entropy analysis to determine the physiological complexity of the operation results, by using convolutional neural networks to analyze the spatial regularity of the operation process, and by using neural networks to dynamically optimize the evaluation parameters, a more accurate, objective, and comprehensive automatic evaluation can be achieved.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: An automated evaluation system for animal experimental procedures based on spatiotemporal multi-dimensional fusion includes: The video acquisition module is used to collect video data during the experimental operation process; The physiological signal acquisition module is used to acquire multi-channel physiological signals from animals during experimental procedures. The multi-dimensional data analysis and control module includes a time-series analysis unit, a spatial analysis unit, and a comprehensive evaluation unit; in, The time series analysis unit is used to receive preprocessed physiological signals, construct a physiological signal complexity evaluation model based on a multivariate multiscale entropy algorithm, and output a physiological complexity score. The spatial analysis unit is used to receive the video data, construct an experimental operation standardization evaluation model based on a convolutional neural network, and output an operation standardization score. The comprehensive evaluation unit is used to integrate the physiological complexity score and the operational standardization score to generate a final multi-dimensional evaluation report.

[0008] Furthermore, the video acquisition module includes a macro camera with variable zoom and a fill light.

[0009] Furthermore, the physiological signal acquisition module includes one or more of the following: a blood pressure sensor, a respiration sensor, an electromyography sensor, an electroencephalogram (EEG) sensor, and an electrocardiogram (ECG) sensor.

[0010] Furthermore, the timing analysis unit includes: The entropy calculation model is used to coarsely process and embed multi-channel physiological signals, calculate multi-scale entropy values ​​under different scale factors, and serve as the basis for physiological complexity scoring. A parameter optimization network is used to dynamically correct the key parameters of the entropy calculation model.

[0011] Furthermore, the parameter optimization network is a BP neural network; The BP neural network is used to establish the mapping relationship between the scale factor, embedding dimension, and similarity tolerance of the entropy calculation model and the entropy evaluation effect.

[0012] Furthermore, the spatial analysis unit includes: A stage identification network is used to identify the video data and determine the experimental operation stage to which it belongs; Feature extraction network is used to extract operational features from video data at the corresponding stage; An operation completion assessment model is used to calculate the similarity between the extracted operation features and preset standard operation features, and generate an operation completion score.

[0013] Furthermore, the operation completion evaluation model is built based on a CNN neural network, where the input is video data and the output is an operation completion score that characterizes the extent to which the operator follows the standard operating procedure.

[0014] An automated evaluation method for animal experimental procedures based on spatiotemporal multi-dimensional fusion includes the following steps: Collect video data of the experimental procedure and multi-channel physiological signals of the animals; The physiological signals are preprocessed; The preprocessed physiological signals are analyzed based on the multivariate multiscale entropy algorithm to generate a physiological complexity score; The video data is analyzed using a convolutional neural network to generate an operational standardization score. The physiological complexity score and operational standardization score are combined to generate a final multi-dimensional evaluation report.

[0015] As can be seen from the above technical solution, compared with the prior art, this invention provides an automatic evaluation system and method for animal experimental operations based on spatiotemporal multi-dimensional fusion. It combines spatial features representing process standardization with temporal complexity features representing result stability, solving the problem of the one-sidedness of single-dimensional evaluation. By analyzing the nonlinear dynamic characteristics of physiological signals through multivariate multi-scale entropy analysis, it can detect subtle physiological changes in animals during experiments earlier and more sensitively, improving the depth of experimental result evaluation. By using a BP neural network to dynamically optimize key parameters in the MMSE algorithm, it avoids the subjectivity and instability of manual parameter tuning, ensuring the generalization ability and accuracy of the evaluation model under different experimental scenarios. By calculating the degree of operation completion, the final experimental results are weighted and calibrated to ensure that the evaluation is based on correct operation, improving the logical rigor of the evaluation. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the system structure provided by the present invention; Figure 2 This is a schematic diagram of the method flow provided by the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] See Figure 1 This invention discloses an automatic evaluation system for animal experimental procedures based on spatiotemporal multi-dimensional fusion, comprising: Video acquisition module: Used to acquire video data during the experimental operation process.

[0020] Physiological signal acquisition module: used to acquire multi-channel physiological signals (such as electrocardiogram, blood pressure, electromyography, etc.) of animals during experimental operations.

[0021] Multi-dimensional data analysis and control module: includes time series analysis unit, spatial analysis unit and comprehensive evaluation unit.

[0022] Temporal analysis unit: Based on the multivariate multiscale entropy algorithm, a physiological signal complexity assessment model is constructed to quantify the depth and stability of the impact of experiments on the physiological state of animals.

[0023] Spatial Analysis Unit: Based on convolutional neural networks, an experimental operation standardization evaluation model is constructed to identify operation stages, extract operation features, and determine whether the operation method is correct.

[0024] Comprehensive evaluation unit: Integrates temporal complexity assessment results and spatial normativity assessment results to generate a final multi-dimensional evaluation report.

[0025] In one specific embodiment, the timing analysis unit includes: Entropy calculation model: The preprocessed multi-channel physiological signals are coarsely processed to calculate multivariate multi-scale entropy values ​​under different scale factors. This entropy value reflects the physiological regulatory capacity and complexity of the animal during the experiment. Entropy values ​​that are too high or too low and deviate from the baseline indicate abnormal physiological states.

[0026] The timing analysis unit is used to calculate the complexity of animal physiological signals and reflect the stability of the animal's physiological state during the experiment.

[0027] (1) Input data definition:

[0028] in: The number of channels for physiological signals is [number]. In this embodiment, ECG, blood pressure, and electromyography are collected simultaneously. ; N Number of sampling points per channel (sequence length); For the first k The first channel in the i The sampled value at each time point.

[0029] (2) Coarsening treatment (multi-scale): Coarsening is the first step in constructing multi-scale entropy, used to observe the complexity of a signal at different time scales. For scale factors... t :

[0030] Where: scale factor t A positive integer, representing the degree of coarsening (i.e., window length). For the first k Each channel in scale t The first after coarsening j One point.

[0031] (3) Multi-embedded vector reconstruction: For a given scale τ, define a composite time delay vector (multi-element embedding vector):

[0032] Where m represents the embedding dimension of each channel. d Representing the time delay vector, it is usually taken as d =1.

[0033] At this point, the total dimension of the reconstructed vector is .

[0034] (4) Distance definition and probability calculation: Define two composite time delay vectors and The distance between them is the Chebyshev distance (maximum norm):

[0035] Define similarity tolerance r (Typically, 0.15 times the standard deviation of the original data is used). For each i Vectors, statistics satisfying the conditions j Number of vectors: First probability For all i ,statistics[ ]≤ r (and i = j The number of ) divided by the total number of vector logs gives the matching probability. Then take the logarithmic average:

[0036] Second probability Expanding the embedding dimension M Add 1 dimension (i.e., add an embedding dimension to any channel), repeat the above calculation, and obtain the matching probability. :

[0037] (5) Definition of multivariate multiscale entropy: For a given scale factor t Embedding dimension m and similarity tolerance r :

[0038] This value represents the regularity of a physiological time series at a specific scale. An entropy value that is too low (e.g., tending towards 0) indicates that the signal is too regular (possibly in a suppressed state), while an entropy value that is too high indicates that the signal tends towards random noise (possibly in a disordered state). Deviation from the baseline value indicates an abnormal physiological state.

[0039] Let the preprocessed multi-channel physiological signal be: Parameter optimization network (performance measurement model): This method utilizes existing backpropagation (BP) neural networks to construct a mapping relationship between MMSE parameters such as scale factor, embedding dimension, and similarity tolerance, and the evaluation accuracy. By traversing the parameter space, it selects the parameter combination (correcting the scale factor and embedding dimension) that makes the entropy value evaluation result closest to the standard prognosis, and dynamically corrects the entropy value calculation model to ensure that the model operates in the optimal state.

[0040] In one specific embodiment, the spatial analysis unit includes: Stage identification network: Identifies the experimental stage to which video data belongs.

[0041] Feature extraction network: Extracts the first feature at a specific stage (such as instrument position, animal position, wound morphology).

[0042] Operation completion assessment model: Using a CNN neural network, the extracted first feature is compared with the preset standard operating procedure (expected static / dynamic action data) to calculate the similarity and generate an operation completion score, which is used to measure the degree to which the operator follows the standard operating procedure.

[0043] In one specific embodiment, the comprehensive evaluation unit is integrated in the following manner: Final evaluation score = Operation completion score × Corrected physiological complexity score This product combination ensures that the experiment is considered successful only when the experimental procedure is performed correctly (high degree of completion) and the animal's physiological state is stable (physiological complexity score is within a reasonable range), thus avoiding invalid evaluation of experimental results under the premise of non-standard operation.

[0044] In one specific embodiment, a signal preprocessing module is also included: filtering, denoising, and normalizing the original physiological signal.

[0045] On the other hand, see Figure 2 This invention also discloses an automatic evaluation method for animal experimental procedures based on spatiotemporal multi-dimensional fusion, comprising the following steps: Collect video data of the experimental procedure and multi-channel physiological signals of the animals; The physiological signals are preprocessed; The preprocessed physiological signals are analyzed based on the multivariate multiscale entropy algorithm to generate a physiological complexity score; The video data is analyzed using a convolutional neural network to generate an operational standardization score. The physiological complexity score and operational standardization score are combined to generate a final multi-dimensional evaluation report.

[0046] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0047] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An automated evaluation system for animal experimental procedures based on spatiotemporal multi-dimensional fusion, characterized in that, include: The video acquisition module is used to collect video data during the experimental operation process; The physiological signal acquisition module is used to acquire multi-channel physiological signals from animals during experimental operations. The multi-dimensional data analysis and control module includes a time-series analysis unit, a spatial analysis unit, and a comprehensive evaluation unit; in, The time series analysis unit is used to receive preprocessed physiological signals, construct a physiological signal complexity evaluation model based on a multivariate multiscale entropy algorithm, and output a physiological complexity score. The spatial analysis unit is used to receive the video data, construct an experimental operation standardization evaluation model based on a convolutional neural network, and output an operation standardization score. The comprehensive evaluation unit is used to integrate the physiological complexity score and the operational standardization score to generate a final multi-dimensional evaluation report.

2. The automatic evaluation system for animal experimental procedures based on spatiotemporal multi-dimensional fusion according to claim 1, characterized in that, The video acquisition module includes a macro camera with variable zoom and a fill light.

3. The automatic evaluation system for animal experimental operations based on spatiotemporal multi-dimensional fusion according to claim 1, characterized in that, The physiological signal acquisition module includes one or more of the following: blood pressure sensor, respiratory sensor, electromyography sensor, electroencephalography sensor, and electrocardiogram sensor.

4. The automatic evaluation system for animal experimental procedures based on spatiotemporal multi-dimensional fusion according to claim 1, characterized in that, The time series analysis unit includes: The entropy calculation model is used to coarsely process and embed multi-channel physiological signals, calculate multi-scale entropy values ​​under different scale factors, and serve as the basis for physiological complexity scoring. A parameter optimization network is used to dynamically correct the key parameters of the entropy calculation model.

5. The automatic evaluation system for animal experimental operations based on spatiotemporal multi-dimensional fusion according to claim 4, characterized in that, The parameter optimization network is a BP neural network; The BP neural network is used to establish the mapping relationship between the scale factor, embedding dimension, and similarity tolerance of the entropy calculation model and the entropy evaluation effect.

6. The automatic evaluation system for animal experimental procedures based on spatiotemporal multi-dimensional fusion according to claim 1, characterized in that, The spatial analysis unit includes: A stage identification network is used to identify the video data and determine the experimental operation stage to which it belongs; Feature extraction network is used to extract operational features from video data at the corresponding stage; An operation completion assessment model is used to calculate the similarity between the extracted operation features and preset standard operation features, and generate an operation completion score.

7. The automatic evaluation system for animal experimental procedures based on spatiotemporal multi-dimensional fusion according to claim 6, characterized in that, The operation completion evaluation model is built on a CNN neural network, where the input is video data and the output is an operation completion score that characterizes the degree to which the operator follows the standard operating procedure.

8. An automatic evaluation method for animal experimental procedures based on spatiotemporal multi-dimensional fusion, characterized in that, Includes the following steps: Collect video data of the experimental procedure and multi-channel physiological signals of the animals; The physiological signals are preprocessed; The preprocessed physiological signals are analyzed based on the multivariate multiscale entropy algorithm to generate a physiological complexity score; The video data is analyzed using a convolutional neural network to generate an operational standardization score. The physiological complexity score and operational standardization score are combined to generate a final multi-dimensional evaluation report.

Images