CN116798635B - Movement dysfunction degree evaluation model, evaluation device and evaluation system - Google Patents
Movement dysfunction degree evaluation model, evaluation device and evaluation system Download PDFInfo
- Publication number
- CN116798635B CN116798635B CN202311076913.7A CN202311076913A CN116798635B CN 116798635 B CN116798635 B CN 116798635B CN 202311076913 A CN202311076913 A CN 202311076913A CN 116798635 B CN116798635 B CN 116798635B
- Authority
- CN
- China
- Prior art keywords
- mode
- evaluation
- degree evaluation
- joint
- movement
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000011156 evaluation Methods 0.000 title claims abstract description 137
- 230000033001 locomotion Effects 0.000 title claims abstract description 137
- 230000004064 dysfunction Effects 0.000 title claims abstract description 109
- 238000013210 evaluation model Methods 0.000 title claims abstract description 54
- 238000002001 electrophysiology Methods 0.000 claims abstract description 29
- 230000007831 electrophysiology Effects 0.000 claims abstract description 29
- 230000004927 fusion Effects 0.000 claims abstract description 27
- 239000013598 vector Substances 0.000 claims description 41
- 208000012661 Dyskinesia Diseases 0.000 claims description 33
- 238000011176 pooling Methods 0.000 claims description 30
- 230000007246 mechanism Effects 0.000 claims description 27
- 238000010606 normalization Methods 0.000 claims description 14
- 208000037170 Delayed Emergence from Anesthesia Diseases 0.000 claims description 11
- 206010061296 Motor dysfunction Diseases 0.000 claims description 11
- 238000012545 processing Methods 0.000 claims description 10
- 208000016285 Movement disease Diseases 0.000 claims description 6
- 238000005259 measurement Methods 0.000 claims description 6
- 108010001267 Protein Subunits Proteins 0.000 claims description 4
- 208000012659 Joint disease Diseases 0.000 abstract description 23
- 238000004458 analytical method Methods 0.000 abstract description 9
- 230000008407 joint function Effects 0.000 abstract description 5
- 238000011084 recovery Methods 0.000 abstract description 3
- 238000000034 method Methods 0.000 description 21
- 230000006870 function Effects 0.000 description 10
- 230000004913 activation Effects 0.000 description 9
- 238000004364 calculation method Methods 0.000 description 9
- 210000001503 joint Anatomy 0.000 description 9
- 230000008569 process Effects 0.000 description 9
- 238000010586 diagram Methods 0.000 description 7
- 201000010099 disease Diseases 0.000 description 7
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 7
- 208000024891 symptom Diseases 0.000 description 5
- 238000013528 artificial neural network Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 239000000203 mixture Substances 0.000 description 4
- 230000003287 optical effect Effects 0.000 description 4
- 238000012549 training Methods 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 238000010276 construction Methods 0.000 description 3
- 238000009472 formulation Methods 0.000 description 3
- 241001465754 Metazoa Species 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000003745 diagnosis Methods 0.000 description 2
- 230000005284 excitation Effects 0.000 description 2
- 210000003414 extremity Anatomy 0.000 description 2
- 210000003205 muscle Anatomy 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000004393 prognosis Methods 0.000 description 2
- 238000011158 quantitative evaluation Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 241000282326 Felis catus Species 0.000 description 1
- 201000005569 Gout Diseases 0.000 description 1
- 206010018634 Gouty Arthritis Diseases 0.000 description 1
- 206010019468 Hemiplegia Diseases 0.000 description 1
- 206010020772 Hypertension Diseases 0.000 description 1
- 206010061218 Inflammation Diseases 0.000 description 1
- 206010060820 Joint injury Diseases 0.000 description 1
- 206010023230 Joint stiffness Diseases 0.000 description 1
- 241000489861 Maximus Species 0.000 description 1
- 206010028980 Neoplasm Diseases 0.000 description 1
- 208000027089 Parkinsonian disease Diseases 0.000 description 1
- 206010034010 Parkinsonism Diseases 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 230000006793 arrhythmia Effects 0.000 description 1
- 206010003119 arrhythmia Diseases 0.000 description 1
- 210000000544 articulatio talocruralis Anatomy 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 210000000988 bone and bone Anatomy 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000006735 deficit Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 210000002310 elbow joint Anatomy 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 210000002216 heart Anatomy 0.000 description 1
- 208000019622 heart disease Diseases 0.000 description 1
- 210000001624 hip Anatomy 0.000 description 1
- 210000004394 hip joint Anatomy 0.000 description 1
- 230000004054 inflammatory process Effects 0.000 description 1
- 230000009916 joint effect Effects 0.000 description 1
- 210000003734 kidney Anatomy 0.000 description 1
- 210000003127 knee Anatomy 0.000 description 1
- 210000000629 knee joint Anatomy 0.000 description 1
- 210000004185 liver Anatomy 0.000 description 1
- 210000004072 lung Anatomy 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000007659 motor function Effects 0.000 description 1
- 201000008482 osteoarthritis Diseases 0.000 description 1
- 230000001575 pathological effect Effects 0.000 description 1
- 210000003314 quadriceps muscle Anatomy 0.000 description 1
- 206010039073 rheumatoid arthritis Diseases 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 210000002832 shoulder Anatomy 0.000 description 1
- 210000000323 shoulder joint Anatomy 0.000 description 1
- 210000002027 skeletal muscle Anatomy 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 210000001835 viscera Anatomy 0.000 description 1
- 210000003857 wrist joint Anatomy 0.000 description 1
Abstract
The invention relates to the technical field of medical equipment, and provides a movement dysfunction degree evaluation model, an evaluation device and an evaluation system, wherein the model comprises the following components: a plurality of single-mode evaluation branches and a cross-mode fusion branch; the plurality of single-mode evaluation branches respectively correspond to a kinematic mode, an electrophysiology mode and a biomechanical mode; each single-mode evaluation branch comprises an input unit, an attention network unit and a residual error network unit which are sequentially connected, and the residual error network unit in each single-mode evaluation branch is connected with the cross-mode fusion branch. The model can realize multidimensional evaluation of the movement dysfunction degree of the diseased joint, further evaluate the movement dysfunction degree of the joint disease object timely, accurately, finely and comprehensively, and provide a convenient and efficient analysis method for the joint function recovery progress of the joint disease object.
Description
Technical Field
The invention relates to the technical field of medical equipment, in particular to a movement dysfunction degree evaluation model, an evaluation device and an evaluation system.
Background
Joint diseases can relate to joints of the whole body of a human body, wherein the joints of the knee, hip, shoulder and hand are the most common, the early pathological characteristics of the joint diseases are inflammation and pain, the end stage of the joint diseases can cause limb disability due to daily activity limitation, the daily life quality of a patient is seriously reduced, and meanwhile, a heavy burden is brought to families of the patient.
In order to master the movement dysfunction degree of a patient with joint diseases in clinic at present, a targeted treatment scheme is formulated for the patient, the treatment effect is evaluated and analyzed, and a doctor is usually assisted to master the range and degree of the body function damage of the patient by means of scale scoring.
Although the method for evaluating the clinical joint function can comprehensively and reliably evaluate the motor function impairment of the patient, the whole process is too dependent on the professional experience of doctors and is easily influenced by subjective factors of the doctors, and a great deal of medical resources are needed to be input. Moreover, such a method focuses on specific movement characteristic indexes such as joint movement degree, joint stiffness state and pain relief condition in a certain aspect, and has the limitations of different emphasis points and poor universality, and the grade standard in the quantitative table is large, so that the minute change of the functional state of a patient is difficult to discover in time, and therefore, the efficient, timely and accurate evaluation and analysis of the movement dysfunction of the patient with the joint disease are difficult to realize.
Disclosure of Invention
The invention provides a movement dysfunction degree evaluation model, an evaluation device and an evaluation system, which are used for solving the defects in the prior art.
The invention provides a movement dysfunction degree evaluation model, which comprises the following steps: a plurality of single-mode evaluation branches and a cross-mode fusion branch; the plurality of single-mode evaluation branches respectively correspond to a kinematic mode, an electrophysiology mode and a biomechanical mode;
each single-mode evaluation branch comprises an input unit, an attention network unit and a residual error network unit which are sequentially connected, and the residual error network unit in each single-mode evaluation branch is connected with the cross-mode fusion branch;
the input units are all used for receiving index data of the diseased joint in the corresponding mode;
the attention network unit is used for extracting target feature vectors of index data of corresponding modes based on a multi-head attention mechanism and a channel attention mechanism;
the residual network unit is used for analyzing and calculating target feature vectors of corresponding modes based on a plurality of residual blocks to obtain a movement dysfunction degree evaluation result of the corresponding modes of the diseased joint;
the cross-modal fusion branch is used for fusing the movement dysfunction degree evaluation results of all modes to obtain the cross-modal movement dysfunction degree evaluation result of the diseased joint.
According to the invention, the attention network unit specifically comprises:
the multi-head attention network subunit is used for extracting the reconstruction feature vector of the index data of the corresponding mode based on a multi-head attention mechanism;
a channel attention subunit, configured to determine, based on a channel attention mechanism, a feature attention score of the reconstructed feature vector in a channel dimension;
and the multiplication subunit is used for multiplying the characteristic attention score with the reconstructed characteristic vector along the channel dimension to obtain the target characteristic vector.
According to the invention, there is provided a movement dysfunction degree evaluation model, the attention network unit further includes:
and the layer normalization subunit is used for carrying out normalization processing on the reconstructed feature vector and index data of the corresponding mode.
According to the movement dysfunction degree evaluation model provided by the invention, the attention network unit further comprises a bottleneck structure formed by two fully-connected layers connected in sequence and an activation layer connected after each fully-connected layer;
the bottleneck structure is used to optimize the feature attention score.
According to the movement dysfunction degree evaluation model provided by the invention, the residual error network unit comprises a convolution layer, a maximum pooling layer, a residual error subunit formed by a plurality of residual error blocks, an average pooling layer and a full connection layer which are connected in sequence;
and the target feature vectors of the corresponding modes sequentially pass through a convolution layer, a maximum pooling layer, a residual sub-unit, an average pooling layer and a full-connection layer in the residual network unit to obtain a movement dysfunction degree evaluation result of the corresponding modes of the diseased joint.
According to the movement dysfunction degree evaluation model provided by the invention, the cross-modal fusion branch comprises a first CBR unit, a second CBR unit and an average pooling unit which are sequentially connected;
and the movement dysfunction degree evaluation results of all modes are taken as a whole and sequentially pass through the first CBR unit, the second CBR unit and the average pooling unit to obtain the cross-mode movement dysfunction degree evaluation results.
According to the movement dysfunction degree evaluation model provided by the invention, the movement dysfunction degree evaluation model is obtained by training based on index data of a sample object in a kinematic mode, an electrophysiology mode and a biomechanical mode and movement dysfunction degree evaluation results of the sample object.
The invention also provides a movement dysfunction degree evaluation device, which comprises:
the data acquisition module is used for acquiring index data of the diseased joint in a kinematic mode, an electrophysiology mode and a biomechanical mode;
the degree evaluation module is used for inputting index data under each mode into the movement dysfunction degree evaluation model to obtain movement dysfunction degree evaluation results of each mode of the diseased joint and cross-mode movement dysfunction degree evaluation results output by the movement dysfunction degree evaluation model.
The invention also provides a movement dysfunction degree evaluation system which comprises a processor, and a motion capture device, an electric measuring instrument and a mechanical sensor which are connected with the processor;
the motion capture device is used for collecting three-dimensional position information of a diseased joint and calculating kinematic data of the diseased joint based on the three-dimensional position information;
the electrical measurement instrument is used for acquiring electrophysiology data of the diseased joint;
the mechanical sensor is used for collecting biomechanical data of the diseased joint;
the processor is used for receiving the kinematic data, the electrophysiology data and the biomechanical data, and obtaining the movement dysfunction degree evaluation results of all modes of the diseased joint and the cross-mode movement dysfunction degree evaluation results output by the movement dysfunction degree evaluation model by adopting the movement dysfunction degree evaluation model.
According to the motor dysfunction degree evaluation system provided by the invention, the mechanical sensor comprises a plantar pressure sensor or a joint moment sensor.
The invention provides a movement dysfunction degree evaluation model, an evaluation device and an evaluation system, wherein the model comprises: a plurality of single-mode evaluation branches and a cross-mode fusion branch; the plurality of single-mode evaluation branches respectively correspond to a kinematic mode, an electrophysiology mode and a biomechanical mode; each single-mode evaluation branch comprises an input unit, an attention network unit and a residual error network unit which are sequentially connected, and the residual error network unit in each single-mode evaluation branch is connected with the cross-mode fusion branch. The dyskinesia degree evaluation model can realize multidimensional evaluation of dyskinesia degree of a diseased joint. The method comprises the steps of respectively extracting target feature vectors with high clinical relevance of diseased joints in index data of multiple modes such as kinematics, electrophysiology and biomechanics through an attention network unit and combining multiple attention mechanisms, outputting quantized movement dysfunction degree evaluation results of the modes closely related to joint movement through a residual network unit, quantitatively representing physiological mode change information of a joint disease object under the condition of clinical symptoms by fully utilizing relevance and complementarity among joint movement physiological information of different modes through a cross-mode fusion branch, outputting cross-mode movement dysfunction degree evaluation results, more comprehensively and accurately quantizing movement dysfunction degree of the diseased joints, timely, accurately, finely and comprehensively evaluating the movement dysfunction degree of the joint disease object, and providing a convenient and efficient analysis method for joint function recovery progress of the joint disease object.
Drawings
In order to more clearly illustrate the invention or the technical solutions of the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious to those skilled in the art that other drawings can be obtained according to these drawings without inventive effort.
FIG. 1 is a schematic diagram of a motor dysfunction level assessment model according to the present invention;
FIG. 2 is a schematic diagram of the structure of a multi-head attention network subunit in the motor dysfunction level assessment model provided by the invention;
FIG. 3 is a schematic diagram of bottleneck structure in the movement disorder degree evaluation model provided by the invention;
FIG. 4 is a second schematic diagram of the movement disorder degree evaluation model according to the present invention;
fig. 5 is a schematic structural view of the movement disorder degree evaluation device provided by the invention;
fig. 6 is a schematic diagram of the movement disorder degree evaluation system according to the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is apparent that the described embodiments are some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The features of the invention "first", "second" and the like in the description and in the claims may be used for the explicit or implicit inclusion of one or more such features. In the description of the invention, unless otherwise indicated, the meaning of "a plurality" is two or more. Furthermore, in the description and claims, "and/or" means at least one of the connected objects, and the character "/", generally means that the associated object is an "or" relationship.
In recent years, a large number of quantitative evaluation methods for movement dysfunction of joint diseases appear, and most of the quantitative evaluation methods are to collect movement physiological data of the trunk and limbs of a patient by using one or two of an optical motion capturing device, a surface electromyographic signal collector and a plantar pressure sensor, and then to identify abnormal movement patterns and quantitatively evaluate the abnormal movement patterns by using a machine learning method based on data driving. However, most of the methods cannot fully cover the physiological data of multiple modes such as kinematics, electrophysiology and biomechanics in the human body joint movement, lack of feature extraction processes highly related to clinical symptoms of patients, further cannot fully utilize complementarity in the physiological information of the multiple modes of human body movement, and the constructed joint movement function evaluation model is relatively one-sided and poor in interpretation, so that efficient and reliable information support is difficult to provide for the establishment of subsequent treatment schemes and rehabilitation effect evaluation of patients.
Based on the above, the embodiment of the invention provides a comprehensive and accurate movement dysfunction degree evaluation model which can evaluate the movement dysfunction degree of the diseased joint of the patient facing the joint disease object, wherein the joint disease object can be a human body or an animal body. Accordingly, the diseased joint may be a human diseased joint or an animal diseased joint, and is not particularly limited herein.
The movement dysfunction degree evaluation model provided in the embodiment of the invention can comprise: a plurality of single-mode evaluation branches and a cross-mode fusion branch; the plurality of single-mode evaluation branches respectively correspond to a kinematic mode, an electrophysiology mode and a biomechanical mode;
each single-mode evaluation branch comprises an input unit, an attention network unit and a residual error network unit which are sequentially connected, and the residual error network unit in each single-mode evaluation branch is connected with the cross-mode fusion branch;
the input units are all used for receiving index data of the diseased joint in the corresponding mode;
the attention network unit is used for extracting target feature vectors of index data of corresponding modes based on a multi-head attention mechanism and a channel attention mechanism;
the residual network unit is used for analyzing and calculating target feature vectors of corresponding modes based on a plurality of residual blocks to obtain a movement dysfunction degree evaluation result of the corresponding modes of the diseased joint;
the cross-modal fusion branch is used for fusing the movement dysfunction degree evaluation results of all modes to obtain the cross-modal movement dysfunction degree evaluation result of the diseased joint.
Specifically, fig. 1 is a schematic structural diagram of a dyskinesia degree evaluation model provided in an embodiment of the present invention, and as shown in fig. 1, the dyskinesia degree evaluation model may include a single-mode evaluation branch 11 of a kinematic mode, a single-mode evaluation branch 12 of an electrophysiology mode, a single-mode evaluation branch 13 of a biomechanical mode, and a cross-mode fusion branch 2. Each single-mode evaluation branch comprises an input unit, an attention network unit and a residual error network unit which are sequentially connected, and the residual error network unit in each single-mode evaluation branch is connected with the cross-mode fusion branch 2.
The input unit of each single-mode evaluation branch is used for receiving index data of the diseased joint in the corresponding mode. In order to ensure the normal application of the dyskinesia degree evaluation model, the joint disease object to which the diseased joint belongs needs to have only joint diseases, namely the following conditions need to be met: a) The composition meets the requirements that the shoulder joint, the elbow joint, the wrist joint, the hip joint, the knee joint and the ankle joint have one of the diseases such as osteoarthritis, rheumatoid arthritis and gouty arthritis; b) The hearing comprehension capability is basically normal, and the test can be matched with the examination and completion; c) No other basic diseases which have been diagnosed and influence the joint function, such as hemiplegia, parkinsonism and the like; d) Patients with severe heart disease, arrhythmia and hypertension, and severe internal diseases such as heart, lung, liver, kidney, etc. and important viscera dysfunction and tumor, etc.
The index data of the diseased joint in the kinematic mode can comprise kinematic data such as joint activity, joint angular velocity, joint angular acceleration and the like, and can be determined through three-dimensional position information of key bone nodes of the diseased joint, which is acquired by the three-dimensional motion capture equipment. The three-dimensional motion capture device may be an optical three-dimensional motion capture device.
The index data of the diseased joint in the electrophysiology mode can comprise the electric activation data of the related muscles of the diseased joint, and can be acquired by a multichannel surface electromyographic signal acquisition instrument. Wherein the muscle related to the affected joint can be biceps brachii, triceps brachii, deltoid anterior, deltoid middle, deltoid posterior, pectoral major, trapezius superior, brachioradial, tricuspid anterior, gluteus maximus, rectus femoris, medial femoral, lateral femoral, biceps femoris, tibialis anterior, gastrocnemius, soleus muscle, etc.
Index data of the diseased joint in a biomechanical mode can comprise joint moment signals, and the index data can be acquired through a plantar pressure sensor or a joint moment sensor.
The attention network unit of each single-mode evaluation branch is used for extracting the target feature vector of index data of the corresponding mode based on a multi-head attention mechanism and a channel attention mechanism.
It can be understood that the multi-head attention mechanism can be used for building complete global dependence based on the correlation between index data of the same mode of the attention mechanics of a plurality of scaling dot products, and the channel attention mechanism can be used for determining the importance degree of the index data of the corresponding mode on the motion dysfunction degree evaluation result of the mode in each channel dimension.
If the attention network element is，/>For the structural parameters of the attention network element, the index data of a certain mode is +.>C is the index data dimension of the mode, < ->C-th dimension index data representing the mode, the target feature vector of the mode is +.>The following steps are: />。
The residual network element can be a network element formed by a residual neural network (Residual Neural Network, resNet), and the movement dysfunction degree of the diseased joint can be evaluated from each mode level through the residual network element.
The residual neural network may be composed of a plurality of residual blocks, each consisting of two convolutional layers and one jump connection, defining the input characteristics of the residual blocks asThe output characteristic of the residual block is +.>The analytical calculation process for each residual block can be described as:
；
。
wherein,and->Respectively two convolution layers, ">Express lot standardization, ++>Representing the ReLU activation function.
When inputting featuresAnd output characteristics->When the channel dimensions of (a) do not match, the calculation process of both changes slightly: />。
Wherein,representing a linear projection, used as input feature only +.>Is realized using convolution with a convolution kernel size of 1 x 1 at the time of the actual construction.
In the embodiment of the invention, the residual network unit can be composed of four different residual layers, each residual layer is provided with a plurality of residual blocks, and the target feature vectors of the corresponding modes are obtained through the residual blocks of the four residual layersAnd (5) performing analytic calculation to obtain a movement dysfunction degree evaluation result of the corresponding mode of the diseased joint. The evaluation result of the dyskinesia degree of each mode is a score of the dyskinesia degree under a single mode, and the value range of the score is related to the category of the diseased joint and can be related to the clinical situationThe same joints in the scale commonly used on beds have the same range of scores, with higher scores indicating a greater degree of motor dysfunction in the affected joint.
In the embodiment of the invention, the expandability and generalization capability of the movement dysfunction degree evaluation model can be improved through each single-mode evaluation branch, and the accurate and reliable joint movement dysfunction evaluation analysis can be realized, so that the individual difference and joint injury characteristics of a joint disease object can be better comprehensively considered by the evaluation model construction mode based on the deep neural network, a more comprehensive and accurate evaluation result is provided, and a doctor is assisted to make more reliable treatment effect analysis and treatment scheme formulation.
In order to fully exploit and utilize the nonlinear complementary relation of the motion function evaluation results of the human joint motion in the aspects of kinematics, electrophysiology and biomechanics, a cross-mode fusion branch is also introduced in the embodiment of the invention, and the cross-mode motion dysfunction degree evaluation results of all modes are fused through the cross-mode fusion branch to obtain the cross-mode motion dysfunction degree evaluation results of the diseased joint. The cross-modal fusion branch can use one-dimensional convolution to perform fusion calculation, the obtained cross-modal movement dysfunction degree evaluation result is a comprehensive score of movement dysfunction degree under each mode, the value range of the comprehensive score is related to the category of the diseased joint, the value range of the comprehensive score can be the same as the score range of the same joint in a clinically common scale, and the higher the comprehensive score is, the more serious the movement dysfunction degree of the diseased joint is indicated.
The dyskinesia degree evaluation model can output four types of evaluation results, namely a dyskinesia degree evaluation result of each mode and a trans-mode dyskinesia degree evaluation result.
The movement dysfunction degree evaluation model provided by the embodiment of the invention comprises the following steps: a plurality of single-mode evaluation branches and a cross-mode fusion branch; the plurality of single-mode evaluation branches respectively correspond to a kinematic mode, an electrophysiology mode and a biomechanical mode; each single-mode evaluation branch comprises an input unit, an attention network unit and a residual error network unit which are sequentially connected, and the residual error network unit in each single-mode evaluation branch is connected with the cross-mode fusion branch. The dyskinesia degree evaluation model can realize multidimensional evaluation of dyskinesia degree of a diseased joint. The method comprises the steps of respectively extracting target feature vectors with high clinical relevance of diseased joints in index data of multiple modes such as kinematics, electrophysiology and biomechanics through an attention network unit and combining multiple attention mechanisms, outputting quantized movement dysfunction degree evaluation results of the modes closely related to joint movement through a residual network unit, quantitatively representing physiological mode change information of a joint disease object under the condition of clinical symptoms by fully utilizing relevance and complementarity among joint movement physiological information of different modes through a cross-mode fusion branch, outputting cross-mode movement dysfunction degree evaluation results, more comprehensively and accurately quantizing movement dysfunction degree of the diseased joints, timely, accurately, finely and comprehensively evaluating the movement dysfunction degree of the joint disease object, and providing a convenient and efficient analysis method for joint function recovery progress of the joint disease object.
On the basis of the above embodiment, the attention network unit specifically includes:
the multi-head attention network subunit is used for extracting the reconstruction feature vector of the index data of the corresponding mode based on a multi-head attention mechanism;
a channel attention subunit, configured to determine, based on a channel attention mechanism, a feature attention score of the reconstructed feature vector in a channel dimension;
and the multiplication subunit is used for multiplying the characteristic attention score with the reconstructed characteristic vector along the channel dimension to obtain the target characteristic vector.
Specifically, the multi-head attention network subunit may extract the reconstructed feature vectors of the index data of the corresponding modality by using a multi-head attention mechanism. Wherein, the global dependence of the multi-head attention mechanism construction can be expressed as:
；
where Q, K and V represent a query vector, a key vector, and a value vector, respectively,representing the length, i.e. the dimension, of K.
As shown in fig. 2, the multi-head attention mechanism divides key value pairs into h subspaces through linear transformation of a Full Connection (FC), performs scaling dot product attention calculation in parallel, and then splices output vectors of the subspaces together, outputs final reconstructed feature vectors through the Full Connection layer, and the calculation process is as follows:
；
；
；
Q=K=V=X。
wherein H represents the number of attention heads, H i Indicating that the i-th attention head is to be paid,the structural parameters corresponding to Q, K and V of the ith attention head are respectively, MHA (X) is a reconstruction feature vector, cat is a splicing function, < + >>For the structural parameters of the full connection layer in the multi-head attention network subunit, < >>The dimensions Q, K and V are represented respectively, and c represents the dimension of index data X of the corresponding modality.
In order to solve the problem that the model obtains nonlinearity by means of a Softmax function by a multi-head attention mechanism, but gradient disappearance easily occurs, in the embodiment of the invention, a layer normalization subunit is further connected behind the multi-head attention network subunit, and the reconstructed feature vector and index data of a corresponding mode are normalized through the layer normalization subunit. The method comprises the following steps:
；
where M is the normalization result and Layer Norm represents the Layer normalization operation (Layer Normalization).
Thereafter, the channel attention subunit may utilize the channel attention mechanism to determine a feature attention score of the reconstructed feature vector in the channel dimension. Wherein after normalization processing is performed on the reconstructed feature vector, the normalization processing result can be regarded as a plurality of channels, namelyThe number of channels is the same as the dimension of index data of the corresponding mode, and is c, m c The normalization processing result of the c-th channel.
And then, determining the importance degree of index data of the corresponding mode on the evaluation result of the dyskinesia degree of the mode in each channel dimension by using a channel attention mechanism.
The channel attention mechanism may be performed using a compression excitation (Squeeze and Excitation, SE) module, with the global averaging pooling layer yielding a characteristic attention score of the normalized processing result in the channel dimension:
。
wherein z is c For the feature attention score on the c-th channel,representing global average pooling, H, W represents the height and width of the normalized processing results, respectively.
As shown in fig. 3, to control the computational complexity of the model and to increase generalization capability, feature attention scores may be fed into two fully connected layers, a first fully connected layer 31 and a second fully connected layer 32, respectively, after each of which an active layer is connected, i.e. a first active layer 33 is connected after the first fully connected layer 31 and a second active layer 34 is connected after the second fully connected layer 32.
The first fully connected layer 31 may project features to a lower dimension and the second fully connected layer 32 may map features of the lower dimension back to the pre-projection feature dimension. Therefore, the two full-connection layers and the connected activation layers can form a bottleneck structure, the feature attention score is optimized to determine the interdependencies among different channel dimensions, and the feature which is more important to the evaluation result is given a larger weight by continuously adjusting in the training process, so that the sensitivity of the important feature is enhanced by the model, and the calculation process is as follows:
；
wherein s is c Representing z c The corresponding result of the optimization is that,representing the ReLU activation function, implemented by the activation layer connected after the first fully connected layer,/->Representing a Sigmoid activation function, implemented by the activation layer connected after the second fully connected layer,、/>the structural parameters of the two fully connected layers in the channel attention subunit, respectively.
Finally, the multiplication subunit 35 multiplies the optimization result of each channel dimension by the normalization processing result along the channel dimension to obtain a target feature vector:
wherein,，/>representing scalar +.>Andis a multiplication of (a) by (b).
In the embodiment of the invention, the multi-head attention mechanism is utilized to realize the cross-modal reconstruction of the feature vector closely related to the clinical symptom manifestation of the patient, and the feature vector with different clinical symptom manifestation correlations is given different weights by combining the channel attention mechanism.
On the basis of the above embodiment, the residual network unit includes a convolution layer, a maximum pooling layer, a residual subunit composed of a plurality of residual blocks, an average pooling layer, and a full connection layer that are sequentially connected;
and the target feature vectors of the corresponding modes sequentially pass through a convolution layer, a maximum pooling layer, a residual sub-unit, an average pooling layer and a full-connection layer in the residual network unit to obtain the movement dysfunction degree evaluation result of the corresponding modes of the diseased joint.
Specifically, in the embodiment of the invention, the convolution layer is used for performing convolution operation, the maximum pooling layer is used for performing maximum pooling operation, the residual subunit is used for performing analysis calculation operation of the features, the average pooling layer is used for performing global average pooling operation, and the full connection layer is used for outputting.
The calculation process of the residual network element can be expressed as:
；
。
wherein,representing global average pooling,/->Representing four residual layers in residual sub-units, respectively, ">Representing maximum pooling, ++>And representing structural parameters of the full-connection layer in the residual network unit, wherein T is an output result of the average pooling layer, and Y is an output result of the full-connection layer, namely a motion dysfunction degree evaluation result of the corresponding mode.
On the basis of the above embodiment, the cross-modal fusion branch includes a first CBR unit, a second CBR unit, and an average pooling unit connected in sequence;
and the movement dysfunction degree evaluation results of all modes are taken as a whole and sequentially pass through the first CBR unit, the second CBR unit and the average pooling unit to obtain the cross-mode movement dysfunction degree evaluation results.
Specifically, the first CBR unit and the second CBR unit each include a convolution layer (Conv), a batch normalization layer (BN), and a ReLU activation layer connected in sequence.
Before the motion dysfunction degree evaluation results of all modes are input into the first CBR unit, splicing can be performed first to obtain splicing results. The method comprises the following steps:。
wherein,for the splice result, ->Respectively representThe degree of dyskinesia of the kinematic mode, the electrophysiological mode and the biomechanical mode is evaluated.
Thereafter, there are:；
wherein output is the evaluation result of the cross-modal movement dysfunction degree, H 1 H is the output result of the first CBR unit 2 For the output of the second CBR unit Avg represents the average pooling operation, implemented by the average pooling layer.And->The weight matrix and bias coefficients representing the ith CBR cell.
As shown in fig. 4, a complete structural diagram of the movement dysfunction level evaluation model provided in the embodiment of the present invention is shown. The single-mode evaluation branch 11, the single-mode evaluation branch 12 and the single-mode evaluation branch 13 of the kinematic model comprise an input unit, an attention network unit and a residual network unit which are sequentially connected, and index data X of a diseased joint in the kinematic model are respectively received through the input units of the corresponding modes 1 Index data X of diseased joint in electrophysiological mode 2 Index data X of diseased joint in biological and biological mode 3 The motion dysfunction degree evaluation results Y of the kinematics modes of the diseased joints are respectively output through residual network units of the corresponding modes 1 Evaluation result Y of degree of dyskinesia in electrophysiological modality of diseased Joint 2 Evaluation result Y of degree of dyskinesia in biomechanical modality of diseased Joint 3 。
Y 1 、Y 2 、Y 3 And obtaining a cross-modal movement dysfunction degree evaluation result output after sequentially passing through the first CBR unit, the second CBR unit and the average pooling unit of the cross-modal fusion branch 2.
On the basis of the above embodiment, the movement dysfunction degree evaluation model is obtained by training based on index data of a sample object in a kinematic mode, an electrophysiology mode, a biomechanical mode, and movement dysfunction degree evaluation results of the sample object.
Specifically, when the movement dysfunction degree evaluation model is applied, the movement dysfunction degree evaluation model can be obtained by training an initial model by taking index data of a sample object in a kinematic mode, an electrophysiology mode and a biomechanical mode as input data and taking a movement dysfunction degree evaluation result of the sample object as a data tag.
The initial model is identical to the above-described dyskinesia degree evaluation model in structure, except that the structural parameters of the initial model are initialized or pre-trained.
As shown in fig. 5, on the basis of the above embodiment, there is provided a movement dysfunction degree evaluation device according to an embodiment of the present invention, including:
a data acquisition module 51, configured to acquire index data of the diseased joint in a kinematic mode, an electrophysiology mode, and a biomechanical mode;
the degree evaluation module 52 is configured to input the index data in each mode into the movement dysfunction degree evaluation model provided in the above embodiments, and obtain a cross-mode movement dysfunction degree evaluation result of the diseased joint output by the movement dysfunction degree evaluation model.
Specifically, in the motor dysfunction level evaluation device, the data acquisition module 51 may be connected to the level evaluation module 52.
The data acquisition module 51 is configured to acquire index data of the diseased joint in a kinematic mode, an electrophysiology mode and a biomechanical mode, where the data acquisition module 51 may be connected to the motion capture device, the electrical measurement device and the mechanical sensor, respectively, and configured to acquire index data of the diseased joint in the kinematic mode, the electrophysiology mode and the biomechanical mode.
The degree evaluation module 52 applies the movement dysfunction degree evaluation model provided in the above embodiments, and is configured to input index data in each mode into the movement dysfunction degree evaluation model provided in the above embodiments, and output a movement dysfunction degree evaluation result of each mode of the diseased joint and a cross-mode movement dysfunction degree evaluation result through the movement dysfunction degree evaluation model.
Here, the movement dysfunction level evaluation model may be placed in the movement dysfunction level evaluation device or may be placed in a third party device and called by the movement dysfunction level evaluation device. The processing procedure of the movement dysfunction level evaluation model on the index data under each mode is specifically referred to each embodiment of the movement dysfunction level evaluation model, and will not be described herein.
The dyskinesia degree evaluation device provided by the embodiment of the invention firstly utilizes a data acquisition module to acquire index data of a diseased joint in a kinematic mode, an electrophysiology mode and a biomechanics mode; and then, the index data under each mode is input into the movement dysfunction degree evaluation model provided in the above embodiments by using the degree evaluation module, so as to obtain movement dysfunction degree evaluation results of each mode and cross-mode movement dysfunction degree evaluation results of the diseased joints output by the movement dysfunction degree evaluation model. The device can realize multidimensional evaluation on the movement dysfunction degree of the diseased joint, further evaluate the movement function of the joint disease object timely, accurately, finely and comprehensively, and has important significance for disease diagnosis, prescription formulation and prognosis analysis in the disease development and treatment process of the joint disease object.
As shown in fig. 6, on the basis of the above-described embodiment, there is provided a movement disorder degree evaluation system in an embodiment of the present invention, which includes a processor 61, and a motion capture device 62, an electrical measuring instrument 63, and a mechanical sensor 64 connected to the processor 61;
the motion capture device 62 is used for acquiring three-dimensional position information of the diseased joint and calculating kinematic data of the diseased joint based on the three-dimensional position information;
the electrical measurement instrument 63 is used for acquiring electrophysiological data of the diseased joint;
the mechanical sensor 64 is used for acquiring biomechanical data of the diseased joint;
the processor 61 is configured to receive the kinematic data, the electrophysiological data, and the biomechanical data, and obtain the dyskinesia degree evaluation result of each modality of the diseased joint and the trans-modality dyskinesia degree evaluation result by using the dyskinesia degree evaluation model provided in each embodiment.
Specifically, in the embodiment of the present invention, the transmission of the data stream between the processor 61 and the motion capture device 62, the electrical measurement instrument 63 and the mechanical sensor 64 may be realized through a USB port.
The motion capture device 62 may be an optical motion three-dimensional capture device for acquiring three-dimensional positional information of the diseased joint and calculating kinematic data of the diseased joint based on the three-dimensional positional information. After the kinematic data is obtained, the kinematic data may be sent to the processor 61. It can be understood that the kinematic data is index data of the diseased joint in a kinematic mode.
After the electrophysiology data of the diseased joint is acquired, the electrical measurement 63 may send the electrophysiology data to the processor 61. The electrophysiology data is index data of the diseased joint in the electrophysiology mode.
The mechanical sensor 64, after acquiring biomechanical data of the diseased joint, may send the biomechanical data to the processor 61. The biomechanical data is index data of the diseased joint in biomechanical mode.
The processor 61 is configured to receive the kinematic data, the electrophysiological data, and the biomechanical data, and obtain the dyskinesia degree evaluation result of each modality of the diseased joint and the trans-modality dyskinesia degree evaluation result by using the dyskinesia degree evaluation model provided in each embodiment.
Here, the movement dysfunction degree evaluation model may be configured in the processor, or may be configured in a third party device, and invoked by the processor. The processing procedure of the movement dysfunction level evaluation model on the index data under each mode is specifically referred to each embodiment of the movement dysfunction level evaluation model, and will not be described herein.
The system for evaluating the degree of dyskinesia provided by the embodiment of the invention comprises a processor, and a motion capture device, an electric measuring instrument and a mechanical sensor which are connected with the processor. The system can realize multidimensional evaluation of the movement dysfunction degree of the diseased joint, further evaluate the movement function of the joint disease object timely, accurately, finely and comprehensively, and has important significance for disease diagnosis, prescription formulation and prognosis analysis in the disease development and treatment process of the joint disease object.
On the basis of the above embodiments, in the embodiments of the present invention, there is provided a motor dysfunction degree evaluation system, wherein the mechanical sensor includes a plantar pressure sensor or a joint moment sensor.
The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.
From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on this understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the respective embodiments or some parts of the embodiments.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (9)
1. A motor dysfunction level assessment model comprising: a plurality of single-mode evaluation branches and a cross-mode fusion branch; the plurality of single-mode evaluation branches respectively correspond to a kinematic mode, an electrophysiology mode and a biomechanical mode;
each single-mode evaluation branch comprises an input unit, an attention network unit and a residual error network unit which are sequentially connected, and the residual error network unit in each single-mode evaluation branch is connected with the cross-mode fusion branch;
the input units are all used for receiving index data of the diseased joint in the corresponding mode;
the attention network unit is used for extracting target feature vectors of index data of corresponding modes based on a multi-head attention mechanism and a channel attention mechanism;
the residual network unit is used for analyzing and calculating target feature vectors of corresponding modes based on a plurality of residual blocks to obtain a movement dysfunction degree evaluation result of the corresponding modes of the diseased joint;
the cross-modal fusion branch is used for fusing the movement dysfunction degree evaluation results of all modes to obtain the cross-modal movement dysfunction degree evaluation result of the diseased joint;
the cross-modal fusion branch comprises a first CBR unit, a second CBR unit and an average pooling unit which are sequentially connected;
and the movement dysfunction degree evaluation results of all modes are taken as a whole and sequentially pass through the first CBR unit, the second CBR unit and the average pooling unit to obtain the cross-mode movement dysfunction degree evaluation results.
2. The motor dysfunction level assessment model according to claim 1, wherein the attention network element specifically comprises:
the multi-head attention network subunit is used for extracting the reconstruction feature vector of the index data of the corresponding mode based on a multi-head attention mechanism;
a channel attention subunit, configured to determine, based on a channel attention mechanism, a feature attention score of the reconstructed feature vector in a channel dimension;
and the multiplication subunit is used for multiplying the characteristic attention score with the reconstructed characteristic vector along the channel dimension to obtain the target characteristic vector.
3. The motor dysfunction level assessment model according to claim 2, wherein the attention network element further comprises:
and the layer normalization subunit is used for carrying out normalization processing on the reconstructed feature vector and index data of the corresponding mode.
4. The movement dysfunction level assessment model according to claim 2, wherein the attention network unit further comprises a bottleneck structure based on two fully connected layers connected in sequence and an active layer connected after each fully connected layer;
the bottleneck structure is used to optimize the feature attention score.
5. The motor dysfunction level assessment model according to any one of claims 1-4, wherein the residual network element comprises a convolutional layer, a max pooling layer, a residual subunit consisting of a plurality of residual blocks, an average pooling layer, and a full connection layer connected in sequence;
and the target feature vectors of the corresponding modes sequentially pass through a convolution layer, a maximum pooling layer, a residual sub-unit, an average pooling layer and a full-connection layer in the residual network unit to obtain a movement dysfunction degree evaluation result of the corresponding modes of the diseased joint.
6. The model for evaluating the degree of dyskinesia according to any one of claims 1 to 4, wherein the model for evaluating the degree of dyskinesia is trained based on index data of a sample object in a kinematic mode, an electrophysiology mode, a biomechanical mode, and a result of evaluating the degree of dyskinesia of the sample object.
7. A movement disorder degree evaluation device, comprising:
the data acquisition module is used for acquiring index data of the diseased joint in a kinematic mode, an electrophysiology mode and a biomechanical mode;
the degree evaluation module is used for inputting index data under each mode into the movement dysfunction degree evaluation model according to any one of claims 1-6 to obtain movement dysfunction degree evaluation results of each mode and cross-mode movement dysfunction degree evaluation results of the diseased joint output by the movement dysfunction degree evaluation model.
8. The system is characterized by comprising a processor, and a motion capture device, an electric measuring instrument and a mechanical sensor which are connected with the processor;
the motion capture device is used for collecting three-dimensional position information of a diseased joint and calculating kinematic data of the diseased joint based on the three-dimensional position information;
the electrical measurement instrument is used for acquiring electrophysiology data of the diseased joint;
the mechanical sensor is used for collecting biomechanical data of the diseased joint;
the processor is configured to receive the kinematic data, the electrophysiological data, and the biomechanical data, and obtain a dyskinesia degree evaluation result of each modality of the diseased joint and a trans-modal dyskinesia degree evaluation result output by using the dyskinesia degree evaluation model according to any one of claims 1 to 6.
9. The motor dysfunction level evaluation system according to claim 8, wherein the mechanical sensor includes a plantar pressure sensor or a joint moment sensor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311076913.7A CN116798635B (en) | 2023-08-25 | 2023-08-25 | Movement dysfunction degree evaluation model, evaluation device and evaluation system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311076913.7A CN116798635B (en) | 2023-08-25 | 2023-08-25 | Movement dysfunction degree evaluation model, evaluation device and evaluation system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116798635A CN116798635A (en) | 2023-09-22 |
| CN116798635B true CN116798635B (en) | 2023-11-17 |
Family
ID=88045035
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311076913.7A Active CN116798635B (en) | 2023-08-25 | 2023-08-25 | Movement dysfunction degree evaluation model, evaluation device and evaluation system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116798635B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114795831A (en) * | 2022-04-19 | 2022-07-29 | 中国科学技术大学 | Exercise training equipment based on multi-source information fusion, and evaluation system and method |
| CN114944012A (en) * | 2022-05-19 | 2022-08-26 | 常州大学 | Motion behavior recognition method based on multi-mode collaborative self-attention network |
| CN115376694A (en) * | 2022-10-24 | 2022-11-22 | 佛山科学技术学院 | Multi-mode self-adaptive assessment system and method for limb dysfunction |
| CN115424108A (en) * | 2022-11-08 | 2022-12-02 | 四川大学 | Cognitive dysfunction evaluation method based on audio-visual fusion perception |
| CN115662623A (en) * | 2022-09-08 | 2023-01-31 | 华南师范大学 | Human engineering risk assessment method based on multi-modal input and early fusion GCN |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP6289313B2 (en) * | 2014-08-29 | 2018-03-07 | マクセル株式会社 | Brain dysfunction evaluation system, cerebral dysfunction evaluation method and program |
-
2023
- 2023-08-25 CN CN202311076913.7A patent/CN116798635B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114795831A (en) * | 2022-04-19 | 2022-07-29 | 中国科学技术大学 | Exercise training equipment based on multi-source information fusion, and evaluation system and method |
| CN114944012A (en) * | 2022-05-19 | 2022-08-26 | 常州大学 | Motion behavior recognition method based on multi-mode collaborative self-attention network |
| CN115662623A (en) * | 2022-09-08 | 2023-01-31 | 华南师范大学 | Human engineering risk assessment method based on multi-modal input and early fusion GCN |
| CN115376694A (en) * | 2022-10-24 | 2022-11-22 | 佛山科学技术学院 | Multi-mode self-adaptive assessment system and method for limb dysfunction |
| CN115424108A (en) * | 2022-11-08 | 2022-12-02 | 四川大学 | Cognitive dysfunction evaluation method based on audio-visual fusion perception |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116798635A (en) | 2023-09-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111728609B (en) | Electroencephalogram signal classification method, classification model training method, device and medium | |
| Mokri et al. | Muscle force estimation from lower limb EMG signals using novel optimised machine learning techniques | |
| Zhang et al. | Fuzzy inference system based automatic Brunnstrom stage classification for upper-extremity rehabilitation | |
| Hajian et al. | Deep multi-scale fusion of convolutional neural networks for EMG-based movement estimation | |
| Rosa et al. | Machine learning to extract muscle fascicle length changes from dynamic ultrasound images in real-time | |
| Herrera-González et al. | Knee functional state classification using surface electromyographic and goniometric signals by means of artificial neural networks | |
| CN112806977A (en) | Physiological parameter measuring method based on multi-scale fusion network | |
| CN115054243A (en) | Closed-loop design method and system for upper limb rehabilitation training system | |
| Su et al. | Measurement of upper limb muscle fatigue using deep belief networks | |
| Prabakaran et al. | FPUS23: an ultrasound fetus phantom dataset with deep neural network evaluations for fetus orientations, fetal planes, and anatomical features | |
| CN116798635B (en) | Movement dysfunction degree evaluation model, evaluation device and evaluation system | |
| Jahanandish et al. | Gait phase identification during level, incline and decline ambulation tasks using portable sonomyographic sensing | |
| CN103690192B (en) | Based on the system and method for the heel string morphological analysis of ultrasonic image | |
| CN110801227B (en) | Method and system for testing three-dimensional color block obstacle based on wearable equipment | |
| CN116645346A (en) | Processing method of rotator cuff scanning image, electronic equipment and storage medium | |
| Veer et al. | Analysis and recognition of operations using SEMG from upper arm muscles | |
| CN115736920A (en) | Depression state identification method and system based on bimodal fusion | |
| CN106913342A (en) | A kind of gait test system and method based on fiber grating and pressure sensor | |
| Pogadadanda et al. | Disease diagnosis using ayurvedic pulse and treatment recommendation engine | |
| Theofilou et al. | Automatic assessment of Parkinson's patients' dyskinesia using non-invasive machine learning methods | |
| Yee et al. | Data-driven model for upper limb spasticity detection | |
| Martinez et al. | Strategic attention learning for modality translation | |
| CN115177291B (en) | Method and device for identifying myasthenia gravis acuminata in intensive care unit | |
| Mohamed et al. | Classification of ECG signals using deep neural networks | |
| Goswami et al. | Intra-Class Classification of Psoriasis Using Deep Learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |