Brain nerve function injury detection method and device, electronic equipment and storage medium

Abstract

The present disclosure relates to a brain nerve function damage detection method and device, an electronic device and a storage medium. The brain nerve function damage detection method comprises: obtaining a brain tissue region of a brain perfusion image; extracting a dynamic image feature of the brain tissue region; obtaining a first feature for a first target and a second feature for a second target based on the dynamic image feature, the first target being whether suffering from a stroke, and the second target being an evaluation index of a brain function damage degree; and determining whether the brain nerve function is damaged based on the first feature and the second feature. The embodiment of the present disclosure can realize automatic detection of brain nerve function damage and improve detection accuracy.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of medical image processing, and in particular to a brain nerve function damage detection method and device, an electronic device and a storage medium. BACKGROUND

[0002] Stroke has become the second leading cause of death worldwide. Most of the surviving patients cannot live independently, and the risk of other neurological sequelae (such as dementia) will also increase. The suddenness, severity and unpredictability of stroke bring serious physical and psychological burden to patients and families. Clinical evidence shows that stroke can cause nerve function damage, and the occupying effect and brain edema produced by severe stroke can lead to brain hernia and death. The National Institutes of Health Stroke Scale (NIHSS) is currently used to assess the degree of brain nerve function damage, with a score range of 0-42 points. A score of 0 indicates normal, and a higher score indicates more severe damage. In clinical practice, the NIHSS score is often assessed by questionnaire, which is time-consuming, and relevant studies have shown that even if the score is 0, it does not mean that the patient does not have a stroke, so there is a misjudgment. If a solution can be proposed to automatically detect brain nerve function damage from multiple angles, not only can the detection time be reduced, but also the detection accuracy can be improved. SUMMARY

[0003] The present disclosure proposes a brain nerve function damage detection method and device, an electronic device and a storage medium, and the embodiments of the present disclosure can realize automatic detection of the degree of nerve function damage and improve the detection accuracy.

[0004] According to an aspect of the present disclosure, a brain nerve function damage detection method is provided, which includes:

[0005] Obtaining a brain tissue region of a brain perfusion image;

[0006] Extracting dynamic image features of the brain tissue region;

[0007] Obtaining a first feature for a first target and a second feature for a second target based on the dynamic image features, the first target being whether suffering from stroke, and the second target being an evaluation index of the degree of brain function damage;

[0008] Determining whether the brain nerve function is damaged based on the first feature and the second feature.

[0009] In some possible implementations, the extracting the dynamic image features of the brain tissue region includes:

[0010] Respectively performing feature extraction processing on the brain tissue region of the brain perfusion image at each time to obtain time image features at each time;

[0011] combining the time point image features of the brain tissue region at each time point respectively to obtain dynamic image features of the brain perfusion image;

[0012] And / or, before extracting the dynamic image features of the brain tissue region, the method further comprises: performing moving smoothing processing on each pixel point of the brain perfusion image in the time dimension.

[0013] In some possible implementation manners, the obtaining, based on the dynamic image features, of a first feature for a first target and a second feature for a second target comprises:

[0014] selecting, from the dynamic image features, a first feature corresponding to a first feature item and a second feature corresponding to a second feature item;

[0015] And / or

[0016] selecting, from the dynamic image features, a first sub-feature corresponding to the first feature item and a second sub-feature corresponding to the second feature item;

[0017] performing dimension reduction processing on the dynamic image features to obtain reduced dimension features;

[0018] obtaining the first feature based on the first sub-feature and the reduced dimension features, and obtaining the second feature based on the second sub-feature and the reduced dimension features.

[0019] In some possible implementation manners, the determining, based on the first feature and the second feature, of whether the brain neural function is damaged comprises:

[0020] in a case where the first feature and the second feature satisfy a first condition, determining that the brain neural function is damaged, and in a case where the first feature and the second feature do not satisfy the first condition, obtaining a third feature by using the first feature and the second feature, and determining whether the brain neural function is damaged by using the third feature.

[0021] In some possible implementation manners, the obtaining, by using the first feature and the second feature, of the third feature, and the determining, by using the third feature, of whether the brain neural function is damaged comprise:

[0022] obtaining a first probability corresponding to the first target by using the first feature;

[0023] obtaining a second probability corresponding to the second target by using the second feature;

[0024] obtaining the third feature based on the first probability and the second probability;

[0025] The presence or absence of brain nerve function damage is determined based on the third feature.

[0026] In some possible implementations, before obtaining the first feature for the first target and the second feature for the second target based on the dynamic image features, the method further includes determining an optimal feature combination strategy for extracting the first feature and the second feature, which includes:

[0027] The dynamic image features and target variables of brain perfusion images are obtained from a set of brain perfusion images, wherein the target variables include a first target and a second target;

[0028] Based on a preset feature selection method and at least two feature dimensionality reduction methods, feature processing is performed on the dynamic image features to obtain a first selected feature and at least two sets of first dimensionality reduction features.

[0029] Based on multiple combination strategies, feature combination processing is performed on the first selected feature and the first dimensionality reduction feature to obtain multiple combined features;

[0030] The combined features are evaluated using a classification model and the target variable, and the combination strategy corresponding to the combined feature with the highest classification score is determined as the optimal feature combination strategy.

[0031] In some possible implementations, the method further includes determining the preset feature selection method, which includes:

[0032] Based on at least two feature selection methods, a first image feature that meets the selection criteria of the feature selection methods is selected from the salient features of the dynamic image features.

[0033] The first image features obtained by each of the aforementioned feature selection methods are evaluated using the classification model, and the feature selection method corresponding to the first image feature with the highest classification score is determined as the preset feature selection method.

[0034] According to a second aspect of this disclosure, a brain nerve function impairment detection device is provided, comprising:

[0035] The acquisition module is used to acquire brain tissue regions from brain perfusion images;

[0036] The extraction module is used to extract dynamic image features of the brain tissue region;

[0037] The feature processing module obtains a first feature for a first target and a second feature for a second target based on the dynamic image features. The first target is whether the patient has suffered a stroke, and the second target is an evaluation index of the degree of brain function impairment.

[0038] The determination module is used to determine whether the brain nerve function is damaged based on the first feature and the second feature.

[0039] According to a third aspect of the present disclosure, an electronic device is provided, comprising:

[0040] a processor;

[0041] a memory for storing processor-executable instructions;

[0042] wherein the processor is configured to invoke the instructions stored in the memory to perform the method of any one of the first aspect.

[0043] According to a fourth aspect of the present disclosure, a computer-readable storage medium is provided, having stored thereon computer program instructions, wherein the computer program instructions, when executed by a processor, implement the method of any one of the first aspect.

[0044] In the embodiments of the present disclosure, the corresponding feature information can be extracted from the dynamic image features of the brain perfusion image according to different targets, and the feature information of different targets is combined to determine whether the brain nerve function is damaged. The dynamic image features of the embodiments of the present disclosure can accurately express blood flow information, and can evaluate the condition of brain function damage through multi-angle feature information, can automatically detect, and can improve the detection accuracy.

[0045] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, but not limiting the present disclosure.

[0046] Other features and aspects of the present disclosure will become apparent from the following detailed description of exemplary embodiments with reference to the drawings. BRIEF DESCRIPTION OF DRAWINGS

[0047] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate embodiments consistent with the present disclosure and serve to explain the technical solutions of the present disclosure together with the specification.

[0048] Figure 1 A flowchart of a brain nerve function damage detection method according to an embodiment of the present disclosure is shown;

[0049] Figure 2 A flowchart of extracting dynamic image features of the brain tissue region according to an embodiment of the present disclosure is shown;

[0050] Figure 3 A flowchart of determining a feature item corresponding to a target according to an embodiment of the present disclosure is shown;

[0051] Figure 4 A flowchart of determining an optimal feature combination strategy according to an embodiment of the present disclosure is shown;

[0052] Figure 5A block diagram of a brain nerve function damage detection apparatus according to an embodiment of the present disclosure is shown.

[0053] Figure 6 A block diagram of an electronic device 800 according to an embodiment of the present disclosure is shown.

[0054] Figure 7 A block diagram of another electronic device 1900 according to an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0055] Various exemplary embodiments, features and aspects of the present disclosure will be explained in detail below with reference to the accompanying drawings. The same reference numerals in the drawings denote the same or similar elements. Although various aspects of the embodiments are illustrated in the drawings, the drawings are not necessarily drawn to scale unless specifically indicated.

[0056] The term "exemplary" is used herein to mean "serving as an example, instance, or illustration." Any implementation described herein as "exemplary" is not necessarily to be construed as preferred or advantageous over other implementations.

[0057] The term "and / or" used herein only means an association relationship of the associated objects, and means that there can be three relationships, for example, A and / or B can mean that A exists alone, A and B exist together, and B exists alone. In addition, the term "at least one" herein means any one of a plurality or any combination of at least two of a plurality, for example, including at least one of A, B, and C can mean including any one or more elements selected from the set consisting of A, B, and C.

[0058] In addition, in order to better illustrate the present disclosure, numerous specific details are given in the specific embodiments below. Those skilled in the art should understand that the present disclosure can also be implemented without some specific details. In some examples, methods, means, elements and circuits that are well known to those skilled in the art are not described in detail, in order to highlight the main idea of the present disclosure.

[0059] The execution subject of the brain nerve function damage detection method can be an image processing apparatus, for example, the image processing method can be executed by a terminal device or a server or other processing device, wherein the terminal device can be a user equipment (User Equipment, UE), a mobile device, a user terminal, a terminal, a cellular phone, a cordless phone, a personal digital assistant (Personal Digital Assistant, PDA), a handheld device, a computing device, a vehicle-mounted device, a wearable device, etc. In some possible implementations, the brain nerve function damage detection method can be realized by a processor calling computer readable instructions stored in a memory.

[0060] It can be understood that the above-mentioned various method embodiments mentioned in the present disclosure can be combined with each other to form combined embodiments without deviating from the principle logic. Due to the limited space, the present disclosure will not be repeated.

[0061] Figure 1 A flowchart of a brain nerve function damage detection method according to an embodiment of the present disclosure is shown as follows. Figure 1 As shown in the figure, the brain nerve function damage detection method comprises:

[0062] S10: Obtain a brain tissue region of a brain perfusion image;

[0063] In some possible implementations, the type of the brain perfusion image can be at least one of a magnetic resonance perfusion weighted imaging (PWI), a computed tomography perfusion imaging (CTP), and an arterial spin labeling perfusion imaging (ASL-MRI). In addition, the brain perfusion image comprises a plurality of groups of brain images, each group of brain images can be a brain image scanned in a time range, and the plurality of groups of brain images in the brain perfusion image can be brain images scanned in continuous time (as a plurality of time points). In addition, the brain perfusion image in the embodiment of the present disclosure can be an image of a patient with a brain disease, such as a perfusion image of an ischemic stroke patient, or can also be an image of a brain glioma patient, which is not specifically limited by the present disclosure.

[0064] S20: Extract a dynamic image feature of the brain tissue region;

[0065] In some possible implementations, the brain perfusion image is a plurality of brain images collected in time sequence at a plurality of time points, so that the image features can be extracted for the brain perfusion image at each time point respectively, and the dynamic image features can be combined based on the time point information.

[0066] S30: Obtain a first feature for a first target and a second feature for a second target based on the dynamic image feature, the first target being whether to have a stroke, and the second target being an evaluation index of a brain function damage degree;

[0067] In some possible implementations, the feature processing can be performed for the first target and the second target of detecting the brain function damage respectively to obtain the corresponding first feature and the second feature. The first feature and the second feature can be respectively used to accurately distinguish whether to have a stroke and whether there is a brain function damage degree.

[0068] S40: Determine whether the brain nerve function is damaged based on the first feature and the second feature.

[0069] In some possible implementations, the first feature and the second feature can be respectively used to detect whether the brain function is damaged from two angles, so as to improve the detection accuracy.

[0070] Based on the above configuration, the embodiment of the present disclosure can extract dynamic image features of the brain perfusion image, extract corresponding feature information according to different targets, and determine whether the brain nerve function is damaged by combining the two types of feature information. The dynamic image features of the embodiment of the present disclosure can accurately express blood flow information, and at the same time, evaluate the brain function damage condition through multi-angle feature information, realize automatic evaluation, and improve the detection accuracy.

[0071] The embodiment of the present disclosure will be described in detail below with reference to the accompanying drawings. First, the embodiment of the present disclosure can obtain a brain perfusion image, wherein the way of obtaining the brain perfusion image can include at least one of the following ways:

[0072] A1) directly using a medical image acquisition device to acquire the brain perfusion image; in the embodiment of the present disclosure, the medical image acquisition device can be a nuclear magnetic resonance device, but it is not specifically limited in the present disclosure.

[0073] A2) transmitting and receiving the brain perfusion image through an electronic device; the embodiment of the present disclosure can receive the brain perfusion image transmitted by other electronic devices through communication, and the communication can include wired communication and / or wireless communication, which is not specifically limited in the present disclosure.

[0074] A3) reading the brain perfusion image stored in the database; the embodiment of the present disclosure can read the brain perfusion image stored in the local or server according to the received data reading instruction, so as to obtain the brain perfusion image, which is not specifically limited in the present disclosure.

[0075] It should be noted that the brain perfusion image in the embodiment of the present disclosure can be a perfusion image acquired by the same device or different devices. The person skilled in the relevant art can select the corresponding device according to the needs, which is not specifically limited herein.

[0076] Further, the embodiment of the present disclosure can perform bone removal processing on the obtained brain perfusion image to determine the brain tissue region in the brain perfusion image. The embodiment of the present disclosure aims to analyze the features of the brain tissue region, so the position of the brain tissue region of the brain perfusion image can be extracted first, and the brain tissue region in the embodiment of the present disclosure can include gray matter, white matter and cerebrospinal fluid. The FSL software can be used to perform bone removal processing on the brain perfusion image and obtain the position information of the brain tissue region. Through the bone removal processing, the skull pixels in the brain perfusion image can be avoided to affect the subsequent processing, and the accuracy of feature extraction can be improved.

[0077] After obtaining the brain tissue region, the brain perfusion image can also be preprocessed, or the subsequent processing can also be directly performed according to the determined brain tissue region in the brain perfusion image.

[0078] The preprocessing of the brain perfusion image includes at least one of the following: time series registration of the brain perfusion image; and smoothing of the brain perfusion image over time.

[0079] The time series registration of the brain perfusion image includes rigid registration of brain images at different times in the brain perfusion image, to eliminate motion bias caused by movement of the subject during image acquisition.

[0080] The smoothing of the brain perfusion image over time includes obtaining grayscale values of the same pixel in the brain tissue at different times to form a time grayscale sequence, and performing smoothing on the time grayscale sequence. The smoothing can include a three-point moving average process, and the embodiment of the present disclosure can use a 1*3 moving window to smooth the time grayscale sequence, which is not limited in the present disclosure. The smoothing can reduce noise in the brain perfusion image and improve image quality.

[0081] After preprocessing or obtaining the brain tissue region, the brain tissue region can be further processed for features. First, dynamic image features of the brain perfusion image can be extracted. In the embodiment of the present disclosure, the brain perfusion image can include brain images at different times, Figure 2 A flowchart of extracting dynamic image features of the brain tissue region according to an embodiment of the present disclosure is shown. As shown in Figure 2 The extraction of the dynamic image features of the brain tissue region includes:

[0082] S21: performing feature extraction processing on the brain tissue region in the brain perfusion image at each time, to obtain time image features at each time;

[0083] S22: combining the time image features of the brain tissue region at each time for the brain perfusion image, to obtain dynamic image features of the brain perfusion image;

[0084] In the embodiment of the present disclosure, the brain perfusion image can include t sets of brain images, each corresponding to a time, and t can be an integer greater than 1 and less than or equal to 50, but is not limited in the present disclosure. The embodiment of the present disclosure can perform feature extraction processing on the brain tissue region in the brain image at t times.

[0085] In one example, the feature extraction process can include: performing at least one image transformation on the brain tissue region, obtaining an augmented set of the brain tissue region based on the brain tissue region and its image transformation results; and extracting at least one of the first order gradient feature, shape feature and texture feature of any image in the augmented set. Wherein the image transformation includes at least one of Fourier transform, Gabor transform, Gaussian-Laplacian transform, wavelet transform, square root filter, exponential function filter. The disclosed embodiments can utilize the original brain tissue region and the results after image transformation to form an augmented set, and perform feature extraction on each brain tissue region in the augmented set to obtain more abundant image features. The extracted first order gradient feature can include features such as mean, maximum and minimum gray values, variance, percentiles (14 and 15) of the brain tissue region, which describe single pixel or single voxel features, skewness and kurtosis features that describe the shape of data intensity distribution, and histogram quotient and energy information, etc. Wherein the skewness reflects the asymmetry of the data distribution curve to the left (negative skew, lower than the mean) or to the right (positive skew, higher than the mean); and the kurtosis reflects the tailing of the data distribution relative to the Gaussian distribution due to outliers. The shape feature can include surface and volume-based features such as compactness and sphericity features. The texture feature can include Absolute Gradient, Gray Level Co-occurrence Matrix (GLCM), Gray Level Run Length Matrix (GLRLM), Gray Level Size Zone Matrix (GLSZM), and Gray Level Dependence Matrix (GLDM).

[0086] In some possible implementation manners, the feature extraction processing described above can be performed by using an extraction radiomics manner to obtain time point image features corresponding to the brain tissue region at each time point. The dynamic image features of the brain tissue region are obtained by combining the time point image features at each time point. The 65800 dynamic image features (3D brain image at 50 time points x 1316 time point image features) can be calculated by the embodiments of the present disclosure. These dynamic image features are divided into 9 groups: (1) shape features x 50 = 700, (2) first-order gradient features: 18 x 50 = 900, (3) gray level co-occurrence matrix GLCM (24 x 50 = 1200), (4) gray level run length matrix GLRLM (16 x 50 = 800), (5) gray level size zone matrix GLSZM (16 x 50 = 800), (6) neighboring gray tone difference matrix NGTDM (5 x 50 = 250), (7) gray level dependence matrix GLDM (14 x 50 = 700), (8) Laplace transform (465 x 50 = 23250), and (9) wavelet transform (744 x 50 = 37200). In the embodiments of the present disclosure, each time point image feature can be defined as the combination of the name of the image feature itself and the time value of the 3D brain image, where t is the time value corresponding to the 3D image. For example, "Log-sigma-1-0-mm-3d_firstorder_skewness_17" represents the time point image feature "Log-sigma-1-0-mm-3d_firstorder_skewness" at the 17th time point in the DSC-PWI brain perfusion image.

[0087] In addition, when performing the feature extraction processing of the brain tissue region, the embodiments of the present disclosure can also optimize the time sequence, reduce the time value, and improve the operation efficiency. Specifically, the embodiments of the present disclosure can divide the time t of the brain perfusion image into three groups, such as a first group for a preparation stage, a second group for a reaction stage, and a third group for a recovery stage. The preparation stage is a stage in which the brain image is not affected by the contrast agent during perfusion imaging, the reaction stage is a stage in which the gray value of the pixel point changes as the contrast agent flows through the blood vessel, and the recovery stage is a process in which the gray value of the pixel point returns to the initial state as the contrast agent leaves. In the embodiments of the present disclosure, t is 50 time points, of which the first group is 1-10 time points, the second group is 11-30 time points, and the third group is 31-50 time points. The above is only an example of the present disclosure and is not a specific limitation. In the case of obtaining three groups of time points, the mean value processing can be performed on the brain image corresponding to the first group of time points, and the mean value processing can be performed on the brain image corresponding to the third group of time points. The brain images obtained by the mean value processing of the first group, the brain images corresponding to the second group of time points, and the brain images obtained by the mean value processing of the third group are used as new brain perfusion images for feature extraction processing. Thus, the amount of calculation is reduced and the feature extraction efficiency is improved under the premise of ensuring comprehensive information.

[0088] The above embodiment can perform image feature extraction on the brain tissue region of the three-dimensional brain image at each time point, to obtain dynamic image features at different time points from the perspective of the three-dimensional image at multiple time points.

[0089] In addition, in some other embodiments of the present disclosure, the features of each layer of brain image in the brain perfusion image at t time point can also be analyzed as a whole to obtain dynamic image features. The dynamic image features of the brain tissue region at different time points can also include: generating a first brain image based on the brain image of the same layer at different time points, respectively, the number of the first brain image being the same as the number of layers of the brain image, the number of layers of the first brain image being the same as the number of time points; performing feature extraction processing on the brain tissue region in the first brain image to obtain layer image features; and obtaining the dynamic image features of the brain perfusion image based on the combination of the layer image features of the brain tissue region.

[0090] The brain image in the brain perfusion image of the embodiment of the present disclosure is a 3D image, and each brain image has the same dimension. The 3D image can include multiple layers of brain images, such as coronal, sagittal and transverse images. Each layer of image in each direction can be used as a feature extraction object of the embodiment of the present disclosure.

[0091] In one example, the dimension of the brain perfusion image can be represented as t*C*W*H, where t represents the number of time points, C represents the number of layers of the brain image, and W and H represent the width and height of the brain image, respectively. In the embodiment of the present disclosure, the brain images at t time points are extracted in the order of the first layer to the Cth layer to form a first brain image. Each first brain image corresponds to the number of layers of the brain image, the number of first brain images is the same as the number of layers of the brain image, and the number of layers of the first brain image is the same as the number of time points. Therefore, the dimension of the first brain image obtained for each brain perfusion image is t*W*H, and the number is C.

[0092] In the case of obtaining the first brain image, the feature extraction processing can be performed on the brain tissue region in the first brain image to obtain layer image features. The feature extraction processing is the same as the configuration of the above embodiment, which includes: performing at least one image transformation on the brain tissue region, obtaining an augmented set of the brain tissue region based on the brain tissue region and the image transformation result of the brain tissue region; and extracting at least one of the first-order gradient feature, the shape feature and the texture feature of any image in the augmented set. After performing the feature extraction processing of the brain tissue region on the first brain image of each layer, the corresponding layer image feature can be obtained, and the combination of the layer image features of each layer can obtain the dynamic image feature. In the case where the number of layers of the brain image is 20, the number of dynamic image features can be 26320 (1316*20), but this is not a specific limitation of the present disclosure.

[0093] Similarly, the embodiment of the present disclosure can optimize the time value before extracting the layer image features, and reduce the time value. For details, refer to the foregoing embodiment.

[0094] Based on the above registration, the embodiment of the present disclosure can construct a three-dimensional image from the perspective of time, extract layer image features, and further enrich the extracted dynamic image features. In some other embodiments, the dynamic image features obtained by the embodiment of the present disclosure can be a combination of the above two types of dynamic image features, but this is not a specific limitation of the present disclosure.

[0095] In the case of obtaining rich dynamic image features, further feature processing in two types of target states can be performed. For example, obtaining a first feature corresponding to a first target, and a second feature corresponding to a second target. The embodiment of the present disclosure can first determine a first feature item of the first feature and a second feature item of the second feature, and then select the corresponding first feature and second feature from the dynamic image features.

[0096] In some embodiments, the first feature item and the second feature item can be selected according to a multi-level feature selection strategy. Figure 3 A flowchart of a method for determining a feature item corresponding to a target according to an embodiment of the present disclosure is shown. As shown in Figure 3 The method for determining a feature item corresponding to a target according to a multi-level feature selection strategy includes:

[0097] S100: selecting a significant feature satisfying significance from dynamic features of a brain perfusion image set;

[0098] S200: screening a first image feature satisfying a selection condition of at least two feature selection methods from the significant feature;

[0099] S300: selecting a second image feature satisfying a classification condition from the first image feature by using at least one classification model, and the feature name of the second image feature is a corresponding feature item.

[0100] In some possible embodiments, a feature item corresponding to a target can be determined through feature analysis of a brain perfusion image set including a plurality of brain perfusion images. The acquisition manner of the brain perfusion images in the brain perfusion image set, the brain tissue region determination manner of the brain perfusion images, and the dynamic feature calculation manner are the same as those described in the foregoing embodiments, and are not repeated here. In the case of obtaining dynamic features of the brain perfusion images, significant features can be selected from the dynamic features.

[0101] Specifically, the obtained dynamic image features can be first subjected to standardization processing to reduce the influence of the numerical span of the features themselves. Each row of the dynamic image features obtained by the embodiments of the present disclosure represents the feature values of different feature items of the same patient, and each table represents the feature values of the same feature of different patients. When performing feature standardization, standardization processing is performed on each column of features of the dynamic image features respectively, and the standardization processing of the embodiments of the present disclosure can be mean-variance standardization, so that the standardized features have a mean of 0 and a variance of 1. In other embodiments, the ratio of each column of features to the maximum value of the column of features can also be used as the standardized feature value. Then, the dynamic image features subjected to the standardization processing can be used to perform significant feature extraction.

[0102] The embodiments of the present disclosure can select significant features from the first image features, calculate the p value (assumed value) between the two groups of state features by performing significance analysis on each dynamic image feature in different states of the first target or the second target, and determine that the feature is a significant feature in the case where the p value is less than a significant threshold. The significant threshold is 0.05, and the p value calculation method includes T test, which is only exemplary and not a specific limitation of the present disclosure. In addition, the embodiments of the present disclosure can also calculate the correlation coefficient between the dynamic image features, and determine that the feature is a significant feature in the case where the correlation coefficient of the feature is higher than a coefficient threshold and the p value is less than the significant threshold. The coefficient threshold can be a value greater than 0.6, such as 0.9.

[0103] In one example, in the case where the first target is whether to have a stroke, the first target can include a first state representing having a stroke and a second state representing not having a stroke. The dynamic image features of the brain perfusion images corresponding to the first state and the dynamic image features of the brain perfusion images corresponding to the second state in the brain perfusion image set are grouped, the significance value p value (assumed value) between the same feature items in the two groups of features is calculated, and the feature is determined to be a significant feature in the case where the p value is less than a significant threshold. The embodiments of the present disclosure can also calculate the correlation coefficient between the two groups of features, and determine that the feature is a significant feature in the case where the correlation coefficient of the feature is higher than a coefficient threshold and the p value is less than the significant threshold. The coefficient threshold can be a value greater than 0.6.

[0104] In another example, the second target represents whether the brain nerve function is impaired, the second target can include representing a first state of impaired nerve function (NIHSS score greater than 0) and a second state of unimpaired nerve function (NIHSS score equal to 0), grouping the dynamic image features of the brain perfusion images corresponding to the first state and the dynamic image features of the brain perfusion images corresponding to the second state in the brain perfusion image set, calculating the p value (hypothetical value) of the significance value of the same feature item between the two groups of features, and determining that the feature is a significant feature if the p value is less than the significant threshold. The embodiments of the present disclosure can also calculate the correlation coefficient between the two groups of features, and determine that the feature is a significant feature if the correlation coefficient of the feature is higher than the coefficient threshold and the p value is less than the significant threshold. The coefficient threshold can be a value greater than 0.6.

[0105] In the case of obtaining significant features, feature selection can be performed using a variety of feature selection methods, which have different selection principles. In one example, the feature selection method can include at least two of the information theory-based method, the similar feature-based method, the statistical feature-based method, and the sparse feature and flow feature-based method. The information theory-based method can include maximum mutual information method (MIM), conditional mutual information maximization method (CMIM), conditional mutual information maximization method (MRMR), best individual feature (BIF), mutual information selection (MIFS), joint mutual information (JMI), and the like, and the similar feature-based method can include distance separability measure (Fisher score algorithm), Laplace score (Lapscore algorithm), feature weight algorithm (ReliefF), the statistical feature-based method can include Tscore algorithm and Fscore algorithm, and the sparse feature and flow feature-based method can include multi-cluster feature selection algorithm (MCFS), least absolute shrinkage and selection operator (Lasso), and Alpha algorithm.

[0106] The embodiments of the present disclosure can perform feature selection on the significant features of the first target and / or the second target using at least two of the above-mentioned feature selection methods, and the selection conditions of the feature selection methods other than the Lasso algorithm can include that the maximum number of features is less than a feature quantity threshold, and the feature score is greater than a score threshold, wherein the feature quantity threshold is greater than 10, such as 20 in the present disclosure, and the score threshold can be greater than 0.6, such as 0.8 in the present disclosure. The selection condition of the Lasso algorithm is to select the feature item with a non-zero feature coefficient. The above is only an exemplary description and is not a specific limitation of the present disclosure.

[0107] Based on the above configuration, each feature selection method can select a group of first image features from the significant features. For example, n groups of first image features are generated corresponding to n feature selection methods.

[0108] In the case of obtaining the first image features, at least one classification model can be further utilized to select second image features meeting classification conditions from the first image features. The embodiments of the present disclosure can adopt two ways to perform the above process.

[0109] In some possible implementations, the embodiments of the present disclosure can combine the first image features obtained by each feature selection method to obtain all the first image features, and utilize at least one classification model to perform classification of the first state and the second state in the first target and / or classification of the first state and the second state in the second target based on all the first image features, and determine the first image features meeting the classification conditions in the first target classification and the second target classification as the second image features.

[0110] Specifically, the embodiments of the present disclosure can utilize the classification model to obtain the importance of each third image feature, and rank the second image features according to the importance. In the process of utilizing the classification model to obtain the importance of each first image feature, each first image feature can be input into the classification model, ten-fold cross-validation can be performed by utilizing the classification model to obtain the index of the classification model, the index includes at least two of AUC (area under the ROC curve), precision, accuracy, Real l, F1, and the average value of each index is utilized as the importance of the feature. In the case of including multiple classification models, the importance corresponding to each classification model can be subjected to mean value processing to obtain the final importance. In the case of obtaining the importance of each first image feature, each first image feature can be ranked from high to low according to the importance, wherein the first image features with the highest importance in a preset number can be selected as the second image features, or the first image features with the importance higher than an importance threshold value can also be selected as the second image features. The preset number can be a value greater than 5, and the importance threshold value can be a value greater than 0.6, but not as a specific limitation of the present disclosure.

[0111] In some possible implementation manners, the second image features can be selected from the first image features obtained by the feature selection methods. Specifically, the first image features can be ranked according to scores of the first image features, and the second image features can be selected from the first image features with the highest scores. The scores of the first image features can be calculated based on performances of the first image features on the classification model. For example, the score of each group of first image features can be calculated based on performances of the first image features in the group on the classification model. The performances of the first image features in the group on the classification model can be determined by performing ten-fold cross validation on the classification model by inputting the first image features in the group into the classification model. The performances of the first image features in the group on the classification model can include at least two of AUC (area under the ROC curve), precision, accuracy, Recall, and F1. The score of each group of first image features can be calculated based on average values of the performances of the first image features in the group. In the case of multiple classification models, the scores of the first image features in the group on the classification models can be averaged to obtain the score of the first image features in the group. In the case of multiple groups of first image features, the scores of the first image features in the groups can be ranked from high to low. The second image features can be selected from the first image features with the highest scores, or the second image features can be selected from the first image features with scores higher than a score threshold. The score threshold can be a value greater than 0.6, but is not limited thereto.

[0112] The classification model can include a machine learning model based on different classification strategies, such as one or more of a support vector machine model (SVM) based on a nonlinear relationship, a decision tree model, a random forest model, an Adaboost model, a neural network model, a nearest neighbor model (KNN), a logistic regression model (LR), a linear discriminant analysis model (DA), a gradient boosting classification model (GBDT), and a Gaussian naive Bayes model (NB).

[0113] In addition, the score of each feature selection method can be obtained based on the scores of the first image features obtained by the feature selection method on the classification model. The feature selection method with the highest score can be determined as the optimal feature selection method.

[0114] Based on the above configuration, the multi-level feature selection strategy can be used to fuse selection methods with different selection principles, so as to select the second image features that can highly distinguish the first state and the second state of the first target or the second target, thereby improving the feature selection accuracy.

[0115] The feature name of the second image is the determined feature item. Through the above configuration, the second image features (first feature item) identifying the first state and the second state of the first target, and the second image features (second feature item) identifying the first state and the second state of the second target can be obtained.

[0116] In some embodiments of the present disclosure, the feature item can also be determined in combination with the feature selection method and the feature dimension reduction method. The embodiments of the present disclosure can determine the optimal feature combination strategy of the feature selection method and the feature dimension reduction method, and obtain the corresponding first feature and second feature based on the optimal feature combination strategy.

[0117] Figure 4 A flowchart for determining the optimal feature combination strategy according to an embodiment of the present disclosure is shown. Before obtaining the first feature for the first target and the second feature for the second target based on the dynamic image features, the optimal feature combination strategy for extracting the first feature and the second feature is determined, which includes:

[0118] S1000: performing feature processing on the dynamic image features of the brain perfusion image set based on a preset feature selection method and at least two feature dimension reduction methods, to obtain first selected features and at least two groups of first dimension reduction features;

[0119] S2000: performing feature combination processing on the first selected features and the first dimension reduction features based on a plurality of combination strategies, to obtain a plurality of combination features;

[0120] S3000: evaluating the combination features using a classification model and the target variable, and determining the combination strategy corresponding to the combination feature with the highest classification score as the optimal feature combination strategy.

[0121] In some embodiments, the preset feature selection method can be a specific information set in advance, or can be an optimal feature selection method determined by comparing a plurality of selection methods. Based on at least two feature selection methods, the first image features satisfying the selection conditions of the feature selection method are selected from the significant features of the dynamic image features; the classification model is used to evaluate the first image features obtained by each feature selection method, and the feature selection method corresponding to the first image feature with the highest classification score is determined as the preset feature selection method. The process of determining the preset feature selection method in the embodiments of the present disclosure is as described in the above embodiments, which is not repeated here.

[0122] In addition, for the feature dimension reduction method, the embodiments of the present disclosure can include various feature dimension reduction methods, including linear dimension reduction methods and nonlinear dimension reduction methods. The linear dimension reduction methods include principal component analysis algorithm (PCA) and independent component analysis (ICA); the nonlinear dimension reduction methods include t-distribution stochastic neighbor embedding algorithm (T-SNE), isometric feature mapping (ISOMAP), and uniform manifold approximation and projection (UMAP). The significant features obtained are processed by using the above dimension reduction methods to obtain corresponding first dimension reduction features. In the embodiments of the present disclosure, the number of features after dimension reduction is set to 10, and in other embodiments, it can also be set to other values, which is not limited in the present disclosure.

[0123] In the case where the first selection features selected by the preset feature selection method and the first dimension reduction features obtained by using the dimension reduction method are obtained, various feature combination strategies can be used to perform feature combination to obtain corresponding combination features, and the specific manner can be at least one of the following manners:

[0124] B1) In some possible embodiments, the first selection features can be combined with each of the first dimension reduction features, respectively, to obtain a plurality of combination features; for example, the first selection feature can be represented as F-select, each of the first dimension reduction features can be represented as a set {F1, F2, …, Fk}, and k represents the number of feature dimension reduction algorithms. The generated combination features are concat(F-select, F1), concat(F-select, F2), …, concat(F-select, Fk), and concat represents combination.

[0125] B2) The first dimension reduction features are subjected to feature selection by using the preset selection method to obtain corresponding second dimension reduction features, and the first selection features and the second dimension reduction features are combined to obtain a plurality of combination features; the embodiments of the present disclosure can further perform feature selection processing on the first dimension reduction features obtained by each dimension reduction algorithm to obtain second dimension reduction features matched with each target. The preset selection method can be used to perform the feature selection operation, and corresponding second dimension reduction features are obtained. The set of second dimension reduction features formed by each dimension reduction method can be represented as {F ’ 1, F ’ 2, …, F ’ k}, and k represents the number of feature dimension reduction algorithms. The generated combination features are concat(F-select, F ’ 1), concat(F-select, F ’ 2), …, concat(F-select, F ’k),concat represents combination. In this way, the feature information in the dimension reduction algorithm can be effectively extracted, the number of features is reduced, the operation precision is improved, and the operation speed is also improved.

[0126] B3) evaluating a classification score of the first dimension reduction feature for the target by using the classification model, and obtaining a plurality of combination features based on a combination of a preset number of first dimension reduction features with the highest scores and the first selection feature. Similarly, the first selection feature obtained by using the classification model to evaluate each dimension reduction algorithm can also be evaluated by the present disclosure. In one example, in the case that the target is whether to have a stroke, the classification model can be used to classify the state of the target based on the first dimension reduction feature, ten-fold cross-validation is performed by using the classification model to obtain the indicators of the classification model, the indicators include at least two of AUC (area under the ROC curve), precision, accuracy, Real l, and F1, and the average value of each indicator is used as the classification score of the group of first image features. In the case of including a plurality of classification models, the classification scores corresponding to each classification model can be processed by using the average value to obtain the final classification score. In the case of obtaining the classification score of each first dimension reduction feature, each group of first dimension reduction features can be ranked from high to low, wherein the first dimension reduction feature with the highest score and the first selection feature can be combined to obtain the combination feature. The preset number can be 1 or a value greater than 1, and the present disclosure does not make specific limitations thereto. In addition, the target can also be the first state (NIHSS score greater than 0) indicating that the neurological function is impaired and the second state (NIHSS score equal to 0) indicating that the neurological function is not impaired.

[0127] In the case of obtaining the combination features corresponding to different feature combination strategies, the classification model can be used to evaluate each combination feature, and the combination features of the embodiments of the present disclosure can be represented as concat1, concat2, …, concatm. Wherein, m is the number of combination features. Similarly, the classification model is used to perform ten-fold cross-validation on each combination feature to obtain the indicators of the classification model, including at least two of AUC (ROC curve area), precision, accuracy, Real l, and F1, and the average value of each indicator is used as the classification score of the first image feature. In the case of including multiple classification models, the classification scores corresponding to each classification model can be processed by mean value to obtain the final classification score. In the case of obtaining the classification scores of each combination feature, each combination feature can be ranked from high to low, wherein the combination feature with the highest classification score is determined as the optimal combination feature, and the corresponding feature combination strategy is the optimal feature combination strategy. For example, the optimal combination strategy obtained by the embodiments of the present disclosure can be Lasso+PCA_Lasso. Wherein, Lasso is used as the preset feature selection method, the dimension reduction algorithm is the PCA algorithm, the first selected feature is obtained by using the preset feature selection method Lasso, the first dimension reduction feature is obtained by using the PCA algorithm, and the second dimension reduction feature PCA_Lasso is further selected from the first dimension reduction feature obtained by using the preset feature selection method. Then, the first selected feature and the second dimension reduction feature are combined to obtain the optimal combination feature, and the corresponding Lasso+PCA_Lasso algorithm is represented as the optimal combination strategy.

[0128] In some embodiments, the classification model of the embodiments of the present disclosure can include a machine learning model based on different classification strategies, such as one or more of a support vector machine model (SVM) based on a nonlinear relationship, a decision tree model, a random forest model, an Adaboost model, a neural network model, a nearest neighbor model (KNN), a logistic regression model (LR), a linear discriminant analysis model (DA), a gradient boosting classification model (GBDT), and a Gaussian naive Bayes model (NB).

[0129] Based on the above configuration, the embodiments of the present disclosure use the feature selection method and the dimension reduction method to process the dynamic signs extracted from the brain perfusion image, obtain the first selected feature and the first dimension reduction feature, and use different combination strategies to combine the first selected feature and the first dimension reduction feature. Through the set classification model, each combination feature is evaluated, and the combination feature that best matches the state target is selected, and then the optimal feature combination strategy corresponding to the state target is selected. The embodiments of the present disclosure can adaptively select the optimal combination strategy of the dimension reduction feature and the selected feature according to different state targets, conduct research on brain function damage, improve the accuracy of analysis, and provide better support for clinical analysis.

[0130] Correspondingly, in the case of determining the optimal feature combination strategy, the dynamic features can be processed according to the feature combination strategy. The embodiments of the present disclosure can directly use the feature items determined by the determined optimal feature selection method to select the corresponding first features and second features from the dynamic image features. As described, the first features for the first target and the second features for the second target are obtained based on the dynamic image features, including: selecting the first features corresponding to the first feature items and the second features corresponding to the second feature items from the dynamic image features. Alternatively, the optimal feature combination strategy can also be used to determine the corresponding first features and second features, and the first features for the first target and the second features for the second target are obtained based on the dynamic image features, including at least one of the following ways:

[0131] In the case of the optimal feature combination strategy being the combination of the first selected features and the first dimensionality reduction features obtained by the preset feature selection method and the dimensionality reduction method, the first sub-features corresponding to the first feature items and the second sub-features corresponding to the second feature items can be directly selected from the dynamic image features; the dimensionality reduction processing is performed on the dynamic image features to obtain dimensionality reduction features; the first features are obtained based on the combination of the first sub-features and the dimensionality reduction features, and the second features are obtained based on the combination of the second sub-features and the dimensionality reduction features.

[0132] In the case of the optimal feature combination strategy being the combination of the corresponding selected features after further using the preset feature selection method for feature selection based on the dimensionality reduction features, the first sub-features corresponding to the first feature items and the second sub-features corresponding to the second feature items can be selected from the dynamic image features; the dimensionality reduction processing is performed on the dynamic image features to obtain dimensionality reduction features; the first sub-dimensionality reduction features corresponding to the first target are selected from the dimensionality reduction features by using the preset feature selection method, and the second sub-dimensionality reduction features corresponding to the second target are selected from the dimensionality reduction features; the first features are obtained based on the combination of the first sub-features and the first sub-dimensionality reduction features, and the second features are obtained based on the combination of the second sub-features and the second sub-dimensionality reduction features.

[0133] In the case of obtaining the first features and the second features, the detection of the brain function damage can be performed based on the first features and the second features. Wherein, the determination of whether the brain neural function is damaged based on the first features and the second features includes: in the case that the first features and the second features satisfy a first condition, determining that the brain neural function is damaged, and in the case that the first features and the second features do not satisfy the first condition, obtaining third features by using the first features and the second features, and judging whether the brain neural function is damaged by using the third features.

[0134] In some possible implementation manners, the first condition can be that the first feature represents that the brain perfusion image has a cerebral stroke lesion, and the second feature represents that the brain perfusion image has a cerebral nerve function damage. The embodiments of the present disclosure can use a classification model to perform state detection on the first feature for the first target, to obtain probability values in two states, and perform state detection on the second feature for the second target, to obtain probability values in two states; and determine a state with a higher probability value as the state of the corresponding target. For example, for the first target, in a case where the probability of the first state of having a cerebral stroke is higher than the probability of the second state of not having a cerebral stroke, it is determined that the first feature represents having a cerebral stroke, and otherwise, it is determined that the first feature represents not having a cerebral stroke, and in a case where the probability of the first state of having a nerve function damage is higher than the probability of the second state of not having a nerve function damage, it is determined that the second feature represents having a nerve function damage, and otherwise, it is determined that the second feature represents not having a nerve function damage. The classification model can be one or multiple, and when multiple classification models are used, the obtained probability value can be the mean of the probability values obtained by the multiple classification models.

[0135] In a case where the first feature represents that the brain perfusion image has a cerebral stroke, and the second feature represents that the brain perfusion image has a nerve function damage, it is determined that the first condition is met, and it is determined that the cerebral nerve function damage exists. Correspondingly, in a case where the first feature represents that the brain perfusion image does not have a cerebral stroke, and the second feature represents that the brain perfusion image does not have a nerve function damage, it is determined that the cerebral nerve function damage does not exist.

[0136] In addition, in a case where the representations of the first feature and the second feature are opposite, a third feature can be obtained by using the first feature and the second feature, and whether the cerebral nerve function damage exists can be determined based on the third feature. That is, in a case where the first feature represents that the brain perfusion image has a cerebral stroke, but the second feature represents that the brain perfusion image does not have a nerve function damage, or the first feature represents that the brain perfusion image does not have a cerebral stroke, but the second feature represents that the brain perfusion image has a nerve function damage, a third feature can be obtained by using the first feature and the second feature, and whether the cerebral nerve function damage exists can be determined based on the third feature.

[0137] In some possible implementation manners, the obtaining of the third feature by using the first feature and the second feature, and the determination of whether the cerebral nerve function damage exists based on the third feature, include: obtaining a first probability corresponding to the first target of the first feature; obtaining a second probability corresponding to the second target of the second feature; obtaining the third feature based on the first probability and the second probability; and determining whether the cerebral nerve function damage exists based on the third feature.

[0138] Specifically, the embodiment of the present disclosure can determine the mean of the first probability represented by the first feature and the second probability represented by the second feature as a third feature, and determine that the brain nerve function is damaged if the third feature is greater than a judgment threshold. Otherwise, it is determined that the brain nerve function is not damaged.

[0139] Based on the above configuration, the embodiment of the present disclosure can extract corresponding feature information from the dynamic image features of the brain perfusion image according to different targets, and determine whether the brain nerve function is damaged in combination with the feature information of different targets. The dynamic image features of the embodiment of the present disclosure can accurately express blood flow information, and can evaluate the condition of brain function damage through multi-angle feature information. In addition, the embodiment of the present disclosure can realize automatic detection and improve detection accuracy.

[0140] Those skilled in the art can understand that in the above method of the specific embodiment, the writing order of each step does not mean a strict execution order and does not constitute any limitation on the implementation process. The specific execution order of each step should be determined by its function and possible internal logic.

[0141] In addition, the present disclosure also provides a brain nerve function damage detection device, an electronic device, a computer readable storage medium, and a program, which can be used to implement any of the brain nerve function damage detection methods provided by the present disclosure. The corresponding technical solutions and descriptions are referred to the corresponding description in the method part, and will not be repeated here.

[0142] Figure 5 A block diagram of a brain nerve function damage detection device according to an embodiment of the present disclosure is shown in FIG. 1. As shown in FIG. 1, the brain nerve function damage device includes: Figure 5

[0143] The acquisition module 10 is configured to acquire a brain tissue region of a brain perfusion image.

[0144] The extraction module 20 is configured to extract dynamic image features of the brain tissue region.

[0145] The feature processing module 30 is configured to obtain a first feature for a first target and a second feature for a second target based on the dynamic image features, wherein the first target is whether suffering from a brain stroke, and the second target is an evaluation index of a brain function damage degree.

[0146] The determination module 40 is configured to determine whether the brain nerve function is damaged based on the first feature and the second feature.

[0147] In some embodiments, the device provided by the embodiment of the present disclosure has functions or contains modules which can be used to execute the methods described in the above method embodiments. The specific implementation can refer to the description of the above method embodiments. For brevity, they will not be repeated here.

[0148] ​The embodiment of the present disclosure further provides a computer readable storage medium, which stores computer program instructions. The computer program instructions are executed by a processor to implement the method described above. The computer readable storage medium can be a non-volatile computer readable storage medium.

[0149] The embodiment of the present disclosure further provides an electronic device, which comprises a processor, and a memory for storing processor-executable instructions. The processor is configured to implement the method described above.

[0150] The electronic device can be provided as a terminal, a server or other forms of devices.

[0151] Figure 6 A block diagram of an electronic device 800 according to an embodiment of the present disclosure is shown. The electronic device 800 can be, for example, a terminal such as a mobile phone, a computer, a digital broadcast terminal, a messaging device, a game console, a tablet device, a medical device, a fitness device, a personal digital assistant, and the like.

[0152] Reference Figure 6 The electronic device 800 can include one or more of the following components: a processing component 802, a memory 804, a power supply component 806, a multimedia component 808, an audio component 810, an input / output (I / O) interface 812, a sensor component 814, and a communication component 816.

[0153] The processing component 802 usually controls overall operations of the electronic device 800, such as operations associated with displaying, making phone calls, data communications, camera operations and recording operations. The processing component 802 can include one or more processors 820 to execute instructions to complete all or part of steps of the methods described above. In addition, the processing component 802 can include one or more modules to facilitate the interaction between the processing component 802 and other components. For example, the processing component 802 can include a multimedia module to facilitate the interaction between the multimedia component 808 and the processing component 802.

[0154] The memory 804 is configured to store various types of data to support operations of the electronic device 800. Examples of these data include instructions for any application or method operating on the electronic device 800, contact data, phonebook data, messages, pictures, videos and the like. The memory 804 can be implemented by any type of volatile or non-volatile storage devices or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0155] The power component 806 provides power to the various components of the electronic device 800. The power component 806 can include a power management system, one or more power sources, and other components associated with generating, managing, and distributing power for the electronic device 800.

[0156] The multimedia component 808 includes a screen providing an output interface between the electronic device 800 and a user. In some embodiments, the screen can include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes the touch panel, the screen can be implemented as a touch screen to receive an input signal from a user. The touch panel includes one or more touch sensors to sense a touch, a slide, and a gesture on the touch panel. The touch sensor can not only sense a boundary of a touching or a sliding action, but also detect duration and pressure related to the touching or sliding action. In some embodiments, the multimedia component 808 includes a front camera and / or a rear camera. When the electronic device 800 is in an operating mode, such as a shooting mode or a video mode, the front camera and / or the rear camera can receive external multimedia data. Each of the front camera and the rear camera can be a fixed optical lens system or have a focal length and optical zoom capability.

[0157] The audio component 810 is configured to output and / or input an audio signal. For example, the audio component 810 includes a microphone (MIC) configured to receive an external audio signal when the electronic device 800 is in an operating mode, such as a call mode, a recording mode, and a voice recognition mode. The received audio signal can be further stored in the memory 804 or transmitted via the communication component 816. In some embodiments, the audio component 810 also includes a speaker for outputting an audio signal.

[0158] The I / O interface 812 provides an interface between the processing component 802 and peripheral interface modules, which can be a keypad, a click wheel, buttons, and the like. The buttons can include, but are not limited to, a home button, a volume button, a start button, and a lock button.

[0159] The sensor component 814 includes one or more sensors for providing status assessments for various aspects of the electronic device 800. For example, the sensor component 814 can detect an open / closed position of the electronic device 800, relative positioning of components of the electronic device 800, such as a display and a keypad of the electronic device 800, a change in position of the electronic device 800 or a component of the electronic device 800, presence or absence of user contact with the electronic device 800, orientation or acceleration / deceleration / g-force and temperature changes of the electronic device 800. The sensor component 814 can include an accelerometer to measure a change in position, acceleration, or deceleration, a gyroscope to measure rotation of the electronic device 800, a magnetometer to measure orientation of the electronic device 800, a pressure sensor to measure air pressure and / or a temperature sensor to measure temperature changes. The sensor component 814 can also include a proximity sensor configured to detect proximity of an object without any physical touch. The sensor component 814 can further include a light sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, the sensor component 814 can further include an acceleration sensor, a gyroscope sensor, a magnetic sensor, a pressure sensor or a temperature sensor.

[0160] The communication component 816 is configured to facilitate wired or wireless communication between the electronic device 800 and other devices. The electronic device 800 can access a wireless network based on a communication standard, such as WiFi, 2G, or 3G, or a combination thereof. In an example embodiment, the communication component 816 receives a broadcast signal or broadcast related information from an external broadcast management system via a broadcast channel. In an example embodiment, the communication component 816 further includes a Near Field Communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on Radio Frequency Identification (RFID) techniques, infrared data association (IrDA) techniques, ultra-wideband (UWB) techniques, Bluetooth (BT) techniques, and other techniques.

[0161] In an example embodiment, the electronic device 800 can be implemented using one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, micro-controllers, microprocessors, other electronic units, or a combination thereof, to perform the above-described methods.

[0162] In an example embodiment, a non-transitory computer-readable storage medium, such as the memory 804 including computer program instructions, is also provided, which can be executed by the processor 820 of the electronic device 800 to complete the above-described methods.

[0163] Figure 7 A block diagram of another electronic device 1900 according to an embodiment of the present disclosure is shown. For example, the electronic device 1900 can be provided as a server. Referring to FIG. 19, the electronic device 1900 includes one or more processors 1910, memory 1920 and 1930, a subscriber identity module (SIM) 1940, an input device 1950, a display 1960, and communication components 1970. The electronic device 1900 can also include a bus 1980. The bus 1980 can be any bus structure(s) such as a memory bus, a peripheral bus, a serial bus, a parallel bus, or a super- or multi- bus structure. Figure 7The electronic device 1900 includes a processing component 1922, which is further composed of one or more processors, and a memory resource represented by the memory 1932 for storing instructions, such as application programs, executable by the processing component 1922. The application programs stored in the memory 1932 can include one or more than one module each corresponding to a set of instructions. In addition, the processing component 1922 is configured to execute the instructions to perform the above method.

[0164] The electronic device 1900 can further include a power supply component 1926 configured to perform power management of the electronic device 1900, a wired or wireless network interface 1950 configured to connect the electronic device 1900 to a network, and an input / output (I / O) interface 1958. The electronic device 1900 can operate based on an operating system stored in the memory 1932, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, or the like.

[0165] In exemplary embodiments, a non-transitory computer readable storage medium, such as the memory 1932 including computer program instructions, is also provided, which can be executed by the processing component 1922 of the electronic device 1900 to complete the above method.

[0166] The present disclosure can be a system, a method, and / or a computer program product. The computer program product can include a computer readable storage medium (or media) having computer readable program instructions thereon for causing a processor to carry out aspects of the present disclosure.

[0167] The computer readable storage medium can be a tangible device that can retain and store instructions for use by an instruction execution device. The computer readable storage medium can be, for example, but is not limited to, an electronic storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium include the following: a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanically encoded device such as punch-cards or punched tape, a

[0168] Computer readable program instructions described herein can be downloaded to respective computing / processing devices from a computer readable storage medium or to an external computer or external storage device via a network, for example, the Internet, a local area network, a wide area network and / or a wireless network. The network can comprise copper transmission cables, optical transmission fibers, wireless transmission, routers, firewalls, switches, gateway computers and / or edge servers. A network adapter card or network interface in each computing / processing device receives computer readable program instructions from the network and forwards the computer readable program instructions for storage in a computer readable storage medium within the respective computing / processing device.

[0169] Computer readable program instructions for carrying out operations of the present disclosure can be assembler instructions, instruction-set-architecture (ISA) instructions, machine instructions, machine dependent instructions, microcode, firmware instructions, state-setting data, or either source code or object code written in any combination of one or more programming languages, including an object oriented programming language such as Smalltalk, C++ or the like, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The computer readable program instructions can execute entirely on the user's computing / processing device, partly on the user's computing / processing device, as a stand-alone software package, partly on the user's computing / processing device and partly on a remote computing / processing device or entirely on the remote computing / processing device or server. In the latter scenario, the remote computing / processing device can be connected to the user's computing / processing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computing / processing device, for example, through the Internet using an Internet Service Provider. In some embodiments, electronic circuitry including, for example, programmable logic circuitry, field-programmable gate arrays (FPGA), or programmable logic arrays (PLA) can execute the computer readable program instructions by utilizing state information of the computer readable program instructions to personalize the electronic circuitry, in order to perform aspects of the present disclosure.

[0170] The computer readable program instructions can also be loaded onto a computing / processing device, other programmable data processing apparatus, or other device to cause a series of operations to be performed on the computing / processing device, other programmable apparatus or other device to produce a computer implemented process, such that the instructions which execute on the computing / processing device, other programmable apparatus, or other device implement the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0171] These computer readable program instructions can be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks. These computer readable program instructions can also be stored in a computer readable storage medium that can include a non-transitory computer readable storage medium that can be a computer- readable storage medium having no data storage cycles that change state. The instructions can be executed by one or more processors of a computer, to cause a series of operational steps to be performed on the computer to produce a computer-implemented process. The instructions can also cause one or more processors of a computer or other programmable data processing apparatus to

[0172] The computer readable program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other device to produce a computer implemented process, such that the instructions which execute on the computer, other programmable data processing apparatus, or other device implement the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0173] The computer readable program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other device to produce a computer implemented process, such that the instructions which execute on the computer, other programmable data processing apparatus, or other device implement the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0174] Embodiments of the present disclosure have been described above, and the description is intended to be illustrative of the embodiments and not restrictive of the disclosed embodiments. Many modifications and variations of the disclosed embodiments are possible in light of the above teachings without departing from the scope and spirit of the described embodiments. It is, therefore, to be understood that there has been described above the principles of the disclosed embodiments in connection with specific embodiments thereof. It is to be understood that this disclosure is in no way limited to the embodiments described above, but includes numerous other embodiments with minor modifications and alterations. It is therefore contemplated to cover any and all modifications and variations of the disclosed embodiments. It is further noted that the use of certain terms in this specification is intended to be interpreted by the reader in light of the discovery and teachings provided herein and by the relevant artisans with the skills in the art, and are not intended to limit the possible meanings of the terms.

Claims

1. A method for detecting impairment of brain nerve function, characterized by, The method comprises the following steps: obtaining a brain tissue region of a brain perfusion image; extracting dynamic image features of the brain tissue region; obtaining a first feature corresponding to whether a brain stroke is suffered and a second feature corresponding to an evaluation index of a brain function damage degree based on the dynamic image features; before obtaining the first feature corresponding to whether a brain stroke is suffered and the second feature corresponding to the evaluation index of the brain function damage degree based on the dynamic image features, determining an optimal feature combination strategy corresponding to the first feature and the second feature comprises: obtaining dynamic image features of a brain perfusion image in a brain perfusion image set and a target variable; wherein the target variable comprises whether a brain stroke is suffered and an evaluation index of a brain function damage degree; performing feature processing on the dynamic image features based on a preset feature selection method and at least two feature dimension reduction methods to obtain first selected features and at least two groups of first dimension reduced features; performing feature combination processing on the first selected features and the two groups of first dimension reduced features based on a plurality of combination strategies to obtain a plurality of combined features; and evaluating the plurality of combined features by using a classification model and the target variable to determine a combination strategy corresponding to a combined feature with the highest classification score as the optimal feature combination strategy corresponding to the first feature and the second feature; determining whether the brain neural function is damaged based on the first feature and the second feature comprises: in the case that the first feature represents that there is a brain stroke lesion in the brain perfusion image and the second feature represents that there is a neural function damage in the brain perfusion image, determining that the brain neural function is damaged; otherwise, in the case that the first feature and the second feature represent the opposite, obtaining a first probability corresponding to the first feature representing that a brain stroke is suffered and a second probability corresponding to the second feature representing that there is a brain neural damage; determining a third feature based on the mean of the first probability and the second probability; if the third feature is greater than a judgment threshold, determining that the brain neural function is damaged; otherwise, determining that there is no brain neural function damage.

2. The brain nerve function impairment detection method according to claim 1, characterized in that, The extraction of the dynamic image features of the brain tissue region comprises: performing feature extraction processing on the brain tissue region of the brain perfusion image at each time to obtain time image features at each time, comprising: performing at least one image transformation on the brain tissue region of the brain perfusion image at each time; obtaining an augmented set corresponding to the brain tissue region based on the brain tissue region and the image transformation; extracting at least one of a first-order gradient feature, a shape feature and a texture feature of any image in the augmented set to obtain time image features at each time; combining the time image features of the brain tissue region at each time for the brain perfusion image to obtain dynamic image features of the brain perfusion image.

3. The brain function impairment detection method according to any one of claims 1 or 2, characterized by, Before extracting the dynamic image features of the brain tissue region, performing moving smoothing processing on each pixel point of the brain perfusion image in the time dimension, comprising: obtaining the gray values of the same pixel points in the brain tissue region of the brain perfusion image at multiple times to form a time gray sequence; performing smoothing processing on the time gray sequence by using a moving window.

4. The brain function impairment detection method according to any one of claims 1 or 2, characterized by, The first feature corresponding to whether the user has a stroke and the second feature corresponding to the evaluation index of the brain function damage degree are obtained based on the dynamic image feature, and the method comprises the following steps: First feature items corresponding to the first feature and second feature items corresponding to the second feature are selected from the dynamic image feature.

5. The brain nerve function impairment detection method according to claim 3, characterized in that, The first feature corresponding to whether the user has a stroke and the second feature corresponding to the evaluation index of the brain function damage degree are obtained based on the dynamic image feature, and the method comprises the following steps: First feature items corresponding to the first feature and second feature items corresponding to the second feature are selected from the dynamic image feature.

6. The brain function impairment detection method according to any one of claims 1 or 2, characterized by, The first feature corresponding to whether the user has a stroke and the second feature corresponding to the evaluation index of the brain function damage degree are obtained based on the dynamic image feature, and the method comprises the following steps: First feature items corresponding to the first feature and second feature items corresponding to the second feature are selected from the dynamic image feature. Dimension reduction processing is performed on the dynamic image feature to obtain a dimension-reduced feature corresponding to the dynamic image feature; The first feature corresponding to whether the user has a stroke is obtained based on the first sub-feature and the dimension-reduced feature; 7. The brain nerve function impairment detection method according to claim 3, characterized by, The second feature corresponding to the evaluation index of the brain function damage degree is obtained based on the second sub-feature and the dimension-reduced feature. The first feature corresponding to whether the user has a stroke and the second feature corresponding to the evaluation index of the brain function damage degree are obtained based on the dynamic image feature, and the method comprises the following steps: First feature items corresponding to the first feature and second feature items corresponding to the second feature are selected from the dynamic image feature. Dimension reduction processing is performed on the dynamic image feature to obtain a dimension-reduced feature corresponding to the dynamic image feature; The first feature corresponding to whether the user has a stroke is obtained based on the first sub-feature and the dimension-reduced feature; 8. The method of claim 1, 2, 5, 7, wherein, The second feature corresponding to the evaluation index of the brain function damage degree is obtained based on the second sub-feature and the dimension-reduced feature. The preset feature selection method is determined, comprising: Based on at least two feature selection methods, a first image feature meeting the selection condition of the feature selection method is selected from the significant features of the dynamic image feature; The first image feature obtained by each feature selection method is evaluated by using the classification model; 9. The brain nerve function impairment detection method according to claim 3, characterized by, The feature selection method corresponding to the first image feature with the highest classification score is determined as the preset feature selection method. The preset feature selection method is determined, comprising: Based on at least two feature selection methods, a first image feature meeting the selection condition of the feature selection method is selected from the significant features of the dynamic image feature; The first image feature obtained by each feature selection method is evaluated by using the classification model; 10. The brain nerve function impairment detection method according to claim 4, characterized in that, The feature selection method corresponding to the first image feature with the highest classification score is determined as the preset feature selection method. The preset feature selection method is determined, comprising: Based on at least two feature selection methods, a first image feature meeting the selection condition of the feature selection method is selected from the significant features of the dynamic image feature; The first image feature obtained by each feature selection method is evaluated by using the classification model; The feature selection method corresponding to the first image feature with the highest classification score is determined as the preset feature selection method.

11. The brain nerve function impairment detection method according to claim 6, characterized in that, The preset feature selection method is determined, including: Based on at least two feature selection methods, the first image feature that meets the selection condition of the feature selection method is screened from the significant features of the dynamic image features; The first image feature obtained by each feature selection method is evaluated by using the classification model; The feature selection method corresponding to the first image feature with the highest classification score is determined as the preset feature selection method.

12. A brain nerve function impairment detection device, characterized by, Including: An acquisition module is configured to acquire a brain tissue region of a brain perfusion image; An extraction module is configured to extract dynamic image features of the brain tissue region; A feature processing module is configured to obtain, based on the dynamic image features, a first feature corresponding to whether a brain stroke is suffered and a second feature corresponding to an evaluation index of a brain function damage degree; before obtaining, based on the dynamic image features, the first feature corresponding to whether the brain stroke is suffered and the second feature corresponding to the evaluation index of the brain function damage degree, an optimal feature combination strategy corresponding to the first feature and the second feature is determined, including: acquiring dynamic image features of a brain perfusion image and a target variable in a brain perfusion image set; wherein the target variable includes whether a brain stroke is suffered and an evaluation index of a brain function damage degree; performing feature processing on the dynamic image features based on a preset feature selection method and at least two feature dimension reduction methods to obtain a first selected feature and at least two groups of first dimension reduction features; performing feature combination processing on the first selected feature and the two groups of first dimension reduction features based on a plurality of combination strategies to obtain a plurality of combined features; evaluating the plurality of combined features by using a classification model and the target variable, and determining a combination strategy corresponding to a combined feature with the highest classification score as the optimal feature combination strategy corresponding to the first feature and the second feature; A determination module is configured to determine, based on the first feature and the second feature, whether the brain neural function is damaged, including: in a case where the first feature represents that the brain perfusion image has a brain stroke lesion and the second feature represents that the brain perfusion image has a neural function damage, it is determined that the brain neural function is damaged; otherwise, in a case where the first feature and the second feature represent opposites, a first probability corresponding to the first feature representing that the brain stroke is suffered and a second probability corresponding to the second feature representing that the brain neural damage exists are acquired; a third feature is determined based on a mean value of the first probability and the second probability; if the third feature is greater than a judgment threshold, it is determined that the brain neural function is damaged; otherwise, it is determined that the brain neural function is not damaged.

13. An electronic device, comprising: Including: A processor; A memory for storing processor-executable instructions; The processor is configured to invoke the instructions stored in the memory to execute the brain neural function damage detection method in any one of claims 1-11.

14. A computer-readable storage medium having stored thereon computer program instructions, wherein, The computer program instructions are executed by the processor to implement the brain neural function damage detection method in any one of claims 1-11.

Images