CN115486839A - Gait analysis and diagnosis system and method based on artificial intelligence - Google Patents

Abstract

A gait analysis and diagnosis system and method based on artificial intelligence are based on a three-dimensional gait analysis and acquisition device, a storage device and a computer AI analysis system, and are used for analyzing and diagnosing key parts such as kinematics, dynamics, dynamic electromyography and the like. Compared with manual operation, the AI has strong calculation capability, is more accurate in image analysis, can complete analysis of gait analysis curves in a short time, makes diagnosis and outputs results. The error probability is greatly reduced while the time and labor cost are saved.

Description

Gait analysis and diagnosis system and method based on artificial intelligence

Technical Field

The invention relates to the field of gait analysis, in particular to a gait analysis and diagnosis system and method based on artificial intelligence.

Background

Gait analysis (gait analysis) is an inspection method for studying walking rules, and aims to disclose key links and influencing factors of gait abnormalities through biomechanical and kinematic means so as to guide rehabilitation assessment and treatment and contribute to clinical diagnosis, curative effect assessment, mechanism study and the like. The most common existing gait analysis is three-dimensional gait analysis, and accurate judgment of information such as the range of motion, mechanical feedback, myoelectricity and the like is required for each joint involved in the gait process, including the movement of an ankle joint, a knee joint, a hip joint, a pelvis and a spine in each dimension and each direction, so that missing report or error report of important information often occurs when a gait report is read, errors occur in selection of a diagnosis and treatment scheme, a large amount of time, energy and manpower are consumed when the gait report is analyzed, and certain error probability also exists in manual operation. Artificial intelligence is a new technology capable of subverting the traditional graphical diagnosis system, and can be used for auxiliary diagnosis of an ultrasonic endoscope [ CN202110061525.6, ultrasonic endoscope, artificial intelligence auxiliary identification method, system, terminal, medium ], electrocardiogram diagnosis [ CN202011108883.X, cloud electrocardiogram diagnosis system based on artificial intelligence auxiliary diagnosis ]. However, there is still a gap in the field of gait analysis.

Disclosure of Invention

In view of the background technology, the invention aims to design an artificial intelligence-based gait analysis and diagnosis system, which analyzes key parts such as kinematics, dynamics, dynamic electromyography and the like.

The gait analysis and diagnosis system based on artificial intelligence comprises three-dimensional gait acquisition and analysis equipment, a storage device and a computer AI analysis system;

the three-dimensional gait acquisition and analysis equipment comprises a motion capture system, a three-dimensional force measuring plate, a surface electromyography system and plantar pressure distribution equipment, and acquires and analyzes initial data comprising a moving track, a joint angle, a speed, a period and a time phase, an electromyogram, gravity center displacement and power energy consumption of a testee during walking; storing the collected and analyzed preliminary data into a storage;

the computer AI analysis system uses retrospective gait analysis, uses an AI deep learning method to record gait analysis diagnosis cases into a database for learning and classification, sets multi-classification problems of artificial intelligence modeling, uses a multi-classification algorithm to train and model preliminary data to form an automatic discrimination artificial intelligence model of gait analysis diagnosis, and obtains preliminary data analysis to a three-dimensional gait acquisition and analysis device through the model to obtain an analysis diagnosis result.

Further, in the computer AI analysis system, analysis results including walking cycle, kinematics and dynamics parameters, dynamic electromyogram, center of gravity displacement and energy consumption, standing posture, gait plantar analysis results, and gait analysis results are obtained based on preliminary data obtained by the three-dimensional gait acquisition and analysis device.

Further, for the walking cycle, a database of a normal gait model is established in advance, the average values of the kinematic and dynamic data are recorded, the gait curves of the trunk, the hip joint, the knee joint and the ankle joint are analyzed through the comparison of the preliminary data with the normal parameters and the curves, the abnormal mode is judged, and possible reasons are matched.

Further, for the kinematic and dynamic parameters, the kinematic parameters and the dynamic parameters of each joint and the kinematic parameters of each joint center are calculated; and comparing the flexion angle, the extension angle and the maximum motion range of each joint in each time phase, and judging the reason of gait abnormity.

Further, as for the dynamic electromyogram, the difference between the dominant side and the non-dominant side of the muscle of the lower limbs of the human body when the human body walks freely is analyzed, and the muscle function level is judged according to the changes of the amplitude, the wave crest and the contraction time of the normal electromyogram and the abnormal electromyogram of each time phase.

Further, for gravity center displacement and energy consumption, the space displacement track of the total gravity center of the person in the motion state is obtained through calculation processing based on the force measurement result, and the space velocity is obtained through combination of image measurement, so that the energy consumption during walking is calculated, and the energy consumption is compared with a normal result to make judgment.

Further, for the standing posture analysis, the sizes of the left foot and the right foot, the front and back stress of the feet at two sides and the stability of the pressure centers of the soles in the front and back direction and the left and right direction are compared, and the abnormal state of the standing posture is analyzed; for gait sole analysis, the maximum stress of the front and back feet on both sides, the length of the single supporting line and the position of the pressure center are compared, and abnormal states of the gait sole are analyzed.

Further, analyzing the gait from the space-time parameters, the joint kinematics and the ground reaction force and analyzing the cause of gait abnormity;

the spatio-temporal parameters include: comparing the differences of the gait cycles, step lengths, strides and pace speeds of the bilateral gait frequency, the gait cycles including the support phase and the swing phase with the range of joint dynamics parameters of normal people, judging whether the abnormal conditions exist or not, and simultaneously comparing the bilateral symmetry level;

the joint kinematics comprises: recording bilateral ankle joint plantar flexion, dorsiflexion, inversion and supination; flexion-extension rotation, inversion and eversion of knee joints at two sides; the motion states of adduction, abduction and internal rotation of the bilateral hip joint are compared with the range of joint kinetic parameters of normal people, whether abnormity exists or not is judged, and meanwhile bilateral symmetry levels are compared;

the ground reaction force includes: and (3) comparing the braking force and the driving force of the bilateral feet in the front and back directions, the stability and the repeatability, recording the peak value of the acting force when the heels touch the ground, comparing with the range of the joint dynamic parameters of normal people, judging whether the abnormality exists or not, and comparing the bilateral symmetry level.

Furthermore, in the computer AI analysis system, data are detected through the characteristics of area, point position and numerical value, compared with normal values, the curve is analyzed and the picture is distinguished, and diagnosis is carried out by matching with an AI deep learning algorithm and the image; matching the collected gait data with a data template stored in a database, and further diagnosing according to the height of the correlation and the size of the matched template; finally, the two are complementary to form a final diagnosis, and the diagnosis and analysis results are stored; the diagnostic results will be adjusted according to the underlying parameters of the different patients.

The gait analysis and diagnosis method based on artificial intelligence comprises the following steps:

step 1, acquiring preliminary data including a moving track, a joint angle, a speed, a period and a time phase, an electromyogram, a gravity center displacement and power energy consumption of a subject during walking through three-dimensional gait analysis and acquisition equipment;

step 2, storing the obtained preliminary data into a storage;

step 3, using a computer AI analysis system, modeling, analyzing and diagnosing the preliminary data through the established automatic discrimination artificial intelligence model of gait analysis and diagnosis to obtain an analysis and diagnosis result, wherein the analysis and diagnosis result comprises analysis and diagnosis of walking cycle, kinematics and dynamics parameters, dynamic electromyogram, gravity center displacement and energy consumption, standing posture, gait plantar and gait;

step 4, extracting a data result for auditing, agreeing or modifying the analysis and diagnosis result obtained in the previous step, and confirming that the artificial intelligence gait analysis and diagnosis result is true and effective through signature;

and 5, printing the analysis and diagnosis result and simultaneously keeping the backup.

The invention achieves the following beneficial effects:

1) Compared with manual operation, the AI has strong calculation capability, more accurate analysis on the graph, greatly reduces the error probability and improves the accuracy;

2) Through an AI algorithm and a neural network, the model after deep training can finish analysis of gait analysis curves and make diagnosis in a short time, and the result is output, so that the working efficiency is improved;

3) And an artificial gait judgment is replaced by an AI algorithm and a neural network, so that the time and labor cost are saved.

Drawings

Fig. 1 is a schematic diagram of a gait analysis diagnostic system according to an embodiment of the invention.

Fig. 2 is a flowchart of a gait analysis diagnosis method according to an embodiment of the invention.

Fig. 3 is a schematic diagram of three-dimensional gait analysis acquisition as described in an embodiment of the invention.

Fig. 4 is a front view of a three-dimensional gait analysis acquisition as described in an embodiment of the invention.

Fig. 5 is a side view of a three-dimensional gait analysis acquisition as described in an embodiment of the invention.

Fig. 6 is a diagram illustrating a data analysis result of an exemplary case according to an embodiment of the present invention.

Fig. 7 is a data diagram collected by the three-dimensional gait collecting and analyzing device according to the embodiment of the invention.

Fig. 8 is a schematic diagram illustrating a convolutional neural network for learning and analyzing data according to an embodiment of the present invention.

Fig. 9 is a schematic diagram of an abnormal gait and a normal gait according to an embodiment of the invention.

Fig. 10 is a schematic diagram of comparison between input images and models according to an embodiment of the present invention.

Detailed Description

The technical scheme of the invention is further explained in detail by combining the drawings in the specification.

A gait analysis and diagnosis system and method based on artificial intelligence are disclosed, the system mainly comprises a three-dimensional gait analysis and acquisition device, a storage device and a computer AI analysis system, and the method mainly comprises five steps of acquisition, storage, analysis and diagnosis to obtain results, confirmation of diagnosis by a doctor and printing.

The three-dimensional gait analysis and acquisition equipment comprises a motion capture system, a three-dimensional force measuring plate, a surface myoelectricity system, plantar pressure distribution equipment and the like, and refer to fig. 3-5. Obtaining data includes: the movement track, joint angle, speed, period and time phase, electromyogram, gravity center displacement and power energy consumption of the subject during walking.

The apparatus comprises: the motion capture system, the three-dimensional pressure plate, the plantar pressure distribution device and the surface electromyography system can be connected with the host computer through data lines.

Directly acquiring gait data and storing the gait data into a background server, namely a storage; in addition, the data such as the moving track and the like obtained through the preliminary analysis are subjected to data transmission to a background server, namely a storage through a corresponding interface in a fixed format.

Training and modeling of a computer AI analysis system according to the following steps

The first step is as follows: and collecting the data to form a data set.

The second step is that: and preprocessing the data set, adjusting the format and removing invalid data.

The third step: the data is divided into normal data groups and abnormal data groups.

The criteria for data partitioning are divided into three groups. First, a normal data group having a diagnosis standard is provided on any textbook reference manual, and if data of a subject is within a normal range, the data is classified into the normal data group. The abnormal data group refers to the data of the persons who have gait abnormality to test, and is classified into the abnormal data group.

The fourth step: and selecting a convolutional neural network model as a main body, identifying the collected data picture and graph, and converting the data picture and the graph into a numerical value which can be compared.

The specific implementation method comprises the following steps: the different curves in a graph have different colors, and the boundary of the curve is planned by defining the colors appearing on the image. Then, the relative relationship between the two curves is measured and analyzed.

The fifth step: and comparing the normal data with the test data by using a multi-classification algorithm to obtain a difference value, and further correcting the difference value by comparing the difference value with the abnormal data.

And a sixth step: and defining the final difference value, recording the data, and finally forming a diagnosis database.

The seventh step: and repeatedly evaluating and optimizing until the optimal state is reached.

Finally, an automatic discrimination artificial intelligence model for gait analysis and diagnosis is formed.

The computer AI analysis system mainly performs the following analysis and diagnosis:

(1) A walking cycle: establishing a database of a normal gait model, recording the average values of the kinematics and the dynamics data, analyzing the gait curves of the trunk, the hip joint, the knee joint and the ankle joint by comparing with the normal parameters and the curves, judging the abnormal mode, and matching with possible reasons.

(2) Kinematic and kinetic parameter analysis: and calculating to obtain the kinematic parameters and the dynamic parameters of each joint and the kinematic parameters of each joint center. The flexion angle, the extension angle, the maximum movable range and the like of each joint in each time phase are compared. And judging the cause of gait abnormity.

(3) Dynamic electromyogram: and analyzing the difference between the dominant side and the non-dominant side of the muscle of the lower limbs of the human body when the human body freely walks, and judging the muscle function level according to the changes of the wave amplitude, the wave crest and the contraction time of the normal electromyogram and the abnormal electromyogram of each time phase.

(4) Center of gravity displacement and energy consumption: through the force measurement result, the result data are subjected to operation processing, the space displacement track of the person in the motion state, which is the total gravity center, can be obtained, and the space velocity is obtained by combining image measurement, so that the energy consumption during walking is calculated. And comparing with normal results to make diagnosis.

(5) Analyzing a standing posture: comparing the load bearing of the left foot and the right foot, the front and back stress of the feet on both sides and the stability of the pressure centers of the soles in the front and back direction and the left and right direction, and analyzing the abnormal state of the standing posture.

(6) Gait and plantar analysis: and comparing the maximum stress of the front and rear feet at both sides, the length of the single supporting line and the position of the pressure center, and analyzing the abnormal state of the foot soles of the gait.

(7) Gait analysis: and analyzing the gait from the space-time parameters, the joint kinematics and the ground reaction force, and analyzing the cause of gait abnormity.

Space-time parameters: and comparing the difference of the bilateral step frequency, the gait cycle (support phase and swing phase), the step length, the stride and the pace with the range of the joint dynamics parameters of normal people to judge whether the abnormality exists. While comparing bilateral symmetry levels.

Kinematics of the joint: recording bilateral ankle joint plantar flexion, dorsiflexion, inversion and supination; flexion and extension rotation, varus and valgus of knee joints at two sides; the motion states of adduction and abduction, internal rotation and the like of the bilateral hip joint are compared with the range of joint kinetic parameters of normal people, and whether abnormity exists is judged. While comparing bilateral symmetry levels.

Ground reaction force: and comparing the braking force and the driving force of the two lateral feet in the front and back directions, the stability and the repeatability, recording the peak value of the acting force when the heels touch the ground, comparing the peak value with the range of the joint dynamic parameters of normal crowds, and judging whether the abnormality exists. Bilateral symmetry levels were also compared.

The artificial intelligence gait analysis system detects data through characteristics such as area, point position and numerical value, compares the data with normal values, analyzes curves accurately and distinguishes pictures accurately, and diagnoses by matching with an AI deep learning algorithm and images. And meanwhile, the collected gait data is matched with the data templates stored in the database, and the gait data is further accurately diagnosed according to the height of the correlation and the size of the matched template. And finally, the diagnosis and the analysis result are stored in the background. The diagnosis result can be adjusted according to the basic parameters of different patients.

The doctor can extract the corresponding data result on the doctor interface for auditing, agree or modify the diagnosis and analysis result calculated by the artificial intelligence, and then confirm that the gait analysis result of the artificial intelligence is real and effective through signature.

It is equipped that the printer will print the results of this, while keeping the backup in the cloud.

Referring to fig. 6, an exemplary case is: through analyzing the picture, compare preoperative, postoperative test value and normal scope, find that thighbone external rotation angle right side is 5 degrees, and the left side is 10 degrees. The femoral pronation angle was 75 degrees on both sides, so it was concluded that the patient's pronator splayfoot gait was caused by bilateral increased femoral pronation.

Referring to fig. 7-10, the computer AI analysis system flow is:

the data collected by the three-dimensional gait collecting and analyzing device is composed of data points in a curve form shown in fig. 7, and various parameters mentioned later are summarized to form gait analysis data of a person, and the gait analysis data is recorded into a database. The AI analysis system can automatically complete the step without manual operation.

The collected data comprises a gait abnormal database and data of normal children. The data (as shown in fig. 8) of a normal child is firstly learned and analyzed through a convolutional neural network (common models such as VGG, resNet and the like), so as to form a model of the gait data of the normal child, and the model can simulate the boundary of a normal range of the gait. The abnormal gait analysis database is then imported and the difference between the two curves is compared by analyzing the two images of normal and abnormal. And defining the meaning of the difference between the two images through a multi-classification model. For example, in fig. 9, if the solid line is abnormal gait and the dotted line is normal gait, it is compared whether the difference between the solid line and the dotted line is too much or too little rotation and is within the normal range. Finally, various parameters of each joint are compared once.

Data is continuously accumulated, and diagnosis is summarized and correlated. Finally, a set of model with high accuracy is formed. When an image is input, the image can be compared by the model to obtain whether the result is normal or where the result is abnormal. For example, fig. 10 is introduced into a model, and through intelligent comparison, the rest of images are found to be in a normal range, but the femoral rotation angle (femoral rotation) is not in the normal range, and then comparison is performed until the image is rotated inwards or outwards. The result is a greater inward rotation. That means there is an abnormal gait due to the femoral internal rotation.

The same is true for the remaining various parameter comparison modes.

The above description is only a preferred embodiment of the present invention, and the scope of the present invention is not limited to the above embodiment, but equivalent modifications or changes made by those skilled in the art according to the present disclosure should be included in the scope of the present invention as set forth in the appended claims.

Claims (10)

1. Gait analysis and diagnosis system based on artificial intelligence is characterized in that:

the system comprises three-dimensional gait acquisition and analysis equipment, a storage device and a computer AI analysis system;

the three-dimensional gait acquisition and analysis equipment comprises a motion capture system, a three-dimensional force measuring plate, a surface electromyography system and plantar pressure distribution equipment, and acquires and analyzes primary data comprising a moving track, a joint angle, a speed, a period and a time phase, an electromyogram, gravity center displacement and power energy consumption of a testee during walking; storing the collected and analyzed preliminary data into a storage;

the computer AI analysis system uses retrospective gait analysis, uses an AI deep learning method to record gait analysis diagnosis cases into a database, uses the database to carry out deep neural network training, learns the correlation between relevant data information captured by acquisition equipment and diagnosis results through training, constructs a classification model based on the deep neural network, and is finally used for diagnosis and analysis of actual clinical cases; setting a multi-classification problem of artificial intelligence modeling, training and modeling preliminary data by using a multi-classification algorithm to form an automatic discrimination artificial intelligence model for gait analysis and diagnosis, and analyzing the preliminary data obtained by the three-dimensional gait acquisition and analysis equipment by the model to obtain an analysis and diagnosis result.

2. The artificial intelligence based gait analysis and diagnosis system according to claim 1, characterized in that: in the computer AI analysis system, analysis results including walking period, kinematics and dynamics parameters, dynamic electromyogram, gravity center displacement and energy consumption, standing posture and gait plantar analysis results and gait analysis results are obtained based on the preliminary data obtained by the three-dimensional gait acquisition and analysis equipment.

3. The artificial intelligence based gait analysis diagnosis system according to claim 2, characterized in that: for the walking period, a database of a normal gait model is established in advance, the average values of the kinematics and the dynamics data are recorded, the gait curves of the trunk, the hip joint, the knee joint and the ankle joint are analyzed through the comparison of the primary data and the normal parameters and curves, the abnormal mode is judged, and possible reasons are matched.

4. The artificial intelligence based gait analysis diagnosis system according to claim 2, characterized in that: calculating the kinematic parameters and the kinetic parameters of each joint and the kinematic parameters of each joint center for the kinematic and the kinetic parameters; and comparing the flexion angle, the extension angle and the maximum moving range of each joint in each time phase, and judging the reason of gait abnormity.

5. The artificial intelligence based gait analysis diagnosis system according to claim 2, characterized in that: and for the dynamic electromyogram, analyzing the difference between the dominant side and the non-dominant side of the lower limb muscle of a human body when the human body walks freely, and judging the muscle function level according to the wave amplitude, wave crest and contraction time changes of the normal electromyogram and the abnormal electromyogram at each time phase.

6. The artificial intelligence based gait analysis diagnosis system according to claim 2, characterized in that: for gravity center displacement and energy consumption, the space displacement track of the total gravity center of a person in a motion state is obtained through calculation processing based on the force measurement result, and the space velocity is obtained through combination of image measurement, so that the energy consumption during walking is calculated, and the energy consumption is compared with a normal result to make judgment.

7. The artificial intelligence based gait analysis diagnosis system according to claim 2, characterized in that: for the standing posture analysis, the sizes of the left foot and the right foot, the front and back stress of the feet on two sides and the stability of the pressure centers of the soles in the front and back directions and the left and right directions are compared, and the abnormal state of the standing posture is analyzed; for gait sole analysis, the maximum stress of the front and back feet on both sides, the length of the single supporting line and the position of the pressure center are compared, and abnormal states of the gait sole are analyzed.

8. The artificial intelligence based gait analysis diagnosis system according to claim 2, characterized in that: analyzing the gait from the space-time parameters, the joint kinematics and the ground reaction force and analyzing the cause of gait abnormity;

the spatio-temporal parameters include: comparing the differences of the gait cycles, step lengths, strides and pace speeds of the bilateral gait frequency, the gait cycles including the support phase and the swing phase with the range of joint dynamics parameters of normal people, judging whether the abnormal conditions exist or not, and simultaneously comparing the bilateral symmetry level;

the joint kinematics comprises: recording bilateral ankle joint plantar flexion, dorsiflexion, inversion and supination; flexion-extension rotation, inversion and eversion of knee joints at two sides; the motion states of adduction, abduction and internal rotation of the bilateral hip joint are compared with the range of joint kinetic parameters of normal people, whether abnormity exists or not is judged, and meanwhile bilateral symmetry levels are compared;

the ground reaction force includes: and comparing the braking force and the driving force of the double-side foot in the front and back directions, the stability and the repeatability, recording the peak value of the acting force when the heel touches the ground, comparing with the range of the joint dynamics parameters of normal people, judging whether the abnormality exists or not, and comparing the bilateral symmetry level.

9. The artificial intelligence based gait analysis diagnosis system according to claim 1, characterized in that: in the computer AI analysis system, data are detected through the characteristics of area, point position and numerical value, compared with normal values, the curve is analyzed and the picture is distinguished, and diagnosis is carried out by matching with an AI deep learning algorithm and the image; meanwhile, the collected gait data is matched with a data template stored in a database, and the degree of correlation and the size of the matched template are further diagnosed; finally, the two are complementary to form a final diagnosis, and the diagnosis and analysis results are stored; the diagnostic results will be adjusted according to the underlying parameters of the different patients.

10. An analytical diagnostic method using the artificial intelligence based gait analysis diagnostic system according to claims 1 to 9, characterized in that: the method comprises the following steps:

step 1, acquiring preliminary data including a moving track, a joint angle, a speed, a period and a time phase, an electromyogram, a gravity center displacement and power energy consumption of a subject during walking through three-dimensional gait analysis and acquisition equipment;

step 2, storing the obtained preliminary data into a storage;

step 3, using a computer AI analysis system, modeling, analyzing and diagnosing preliminary data through an established automatic discrimination artificial intelligence model of gait analysis and diagnosis to obtain analysis and diagnosis results, wherein the analysis and diagnosis results comprise analysis and diagnosis of walking cycle, kinematics and kinetic parameters, dynamic electromyography, gravity center displacement and energy consumption, standing posture, gait, foot soles and gait;

step 4, extracting a data result for auditing, agreeing or modifying the analysis and diagnosis result obtained in the previous step, and confirming that the artificial intelligence gait analysis and diagnosis result is true and effective through signature;

and 5, printing the analysis and diagnosis result, and simultaneously saving the backup.

Images