KR20230050067A - System and method for automatically evaluating complexity of infant movement - Google Patents

Abstract

A method and a system for automatically evaluating the complexity of infant movement are provided. According to one embodiment of the present invention, the method for automatically evaluating the complexity of infant movement provides the steps of: i) providing movement images photographing movement of a plurality of infants lying on their backs; ii) inputting the image into a pose prediction model to provide position coordinates of each joint of the plurality of infants; iii) providing a joint angle and a joint angular velocity of the joint using each position coordinate; and iv) calculating the complexity of each joint angle and each joint angular velocity with sample entropy (SE). In the step of providing the position coordinates of each joint, each joint is one or more joints selected from the group consisting of left shoulder, right shoulder, left elbow, right elbow, left wrist, right wrist, left hip, right hip, left knee, right knee, left ankle, and right ankle.

Description

Method and system for automatic evaluation of movement complexity of infants

The present invention relates to a method and system for automatically evaluating motion complexity. More particularly, the present invention relates to systems and methods for automatically assessing the complexity of movements of infants and toddlers.

A developmental disability refers to a state in which mental or physical development has not been achieved according to age. Intellectual disability, chromosomal disorders such as cerebral palsy, Down syndrome, and attention deficit hyperactivity disorder (ADHD) are classified as developmental disabilities. Among them, cerebral palsy refers to motor dysfunction caused by brain lesions before birth, at birth, or when the brain is immature after birth. Cerebral palsy is not a single disease, but a clinical syndrome involving multiple causes and lesions, and is non-progressive.

Identification of high-risk markers is important for early detection of cerebral palsy in infants and young children. Early detection of detectable risk in infants and young children enables early intervention to improve neurodevelopmental outcomes in high-risk infants. Beginning in early fetal life and continuing through the first half of term, infants' voluntary movements may reflect the integrity of the developing nervous system.

Korea Patent No. 1,440,098

It is intended to provide a method for automatically evaluating the complexity of movements of infants and toddlers. In addition, it is intended to provide a system that automatically evaluates the complexity of movements of infants and toddlers.

A method for automatically evaluating motion complexity of infants according to an embodiment of the present invention includes i) providing an image obtained by photographing motion images of a plurality of infants lying supine, ii) inputting the image to a pose prediction model to determine the Providing the position coordinates of each joint, iii) providing the joint angle and joint angular velocity of the joint using each position coordinate, and iv) calculating the complexity of each joint angle and each joint angular velocity by sample entropy (SE ). In the step of providing position coordinates of each joint, each joint is divided into left shoulder, right shoulder, left elbow, right elbow, left wrist, right wrist, left hip joint, right hip joint, left knee, right knee, left ankle, and right ankle. It is one or more joints selected from the group consisting of

A method for automatically evaluating motion complexity of infants according to an embodiment of the present invention includes: i) loading pre-stored examination data of a plurality of infants and young children and classifying a plurality of infants and young children into a first group and a second group; ii) a first group The method may further include extracting complexities from each group and the second group, and iii) calculating a correlation coefficient between the first group and the second group from the complexities.

In the method for automatically evaluating motion complexity of infants and young children according to an embodiment of the present invention, after the step of providing position coordinates, a reliability of 0 to 1 is assigned to each position coordinate and deleting the position coordinates assigned a reliability of less than a preset value. may further include. The preset value may be 0.25 to 0.75.

The method for automatically evaluating the motion complexity of infants and young children according to an embodiment of the present invention may further include, after the step of deleting the position coordinates, a step of smoothing missing data among the position coordinates.

In the step of calculating the complexity, each joint angle is one or more joint angles selected from the group consisting of a right shoulder, a right elbow, a left hip joint, a left knee, and a right knee, and each joint angular velocity is a left shoulder, a right shoulder, a right elbow, and a left knee. It may be the angular velocity of one or more joints selected from the group consisting of a hip joint, a right hip joint, a left knee, and a right knee.

A method for automatically evaluating motion complexity of infants according to an embodiment of the present invention includes the steps of i) providing images of motion images of a plurality of infants lying supine, ii) inputting the images into a pose prediction model to determine each of the plurality of infants and young children Providing position coordinates of one or more joints selected from the group consisting of left shoulder, right shoulder, left elbow, right elbow, left wrist, right wrist, left hip joint, right hip joint, left knee, right knee, left ankle, and right ankle , iii) providing joint angles or angular velocities of joints using each position coordinate, iv) classifying a plurality of infants into a first group and a second group by loading pre-stored examination data of a plurality of infants and young children, and v) calculating correlation coefficients from joint angles or joint angular velocities of different combinations of joints in the first group and the second group, respectively.

In the step of calculating the correlation coefficient, the combination of each joint of the joint angle is a combination of a left hip joint and a right hip joint, a combination of a left knee and a right knee, a combination of a left ankle and a right ankle, a combination of a left hip joint and a left shoulder, and a right It may be one or more combinations selected from the group consisting of a combination of a hip joint and a right shoulder, a combination of a left shoulder and a left elbow, a combination of a right shoulder and a right elbow, a combination of a left hip joint and a left knee, and a combination of a right hip joint and a right knee. . More preferably, the combination of each joint may be one or more combinations selected from the group consisting of a combination of a left hip joint and a right hip joint, a combination of a left hip joint and a left shoulder, and a combination of a left shoulder and a left elbow.

The system for automatically evaluating the motion complexity of infants according to an embodiment of the present invention includes i) an infant photographing unit for capturing motion images of a plurality of infants lying supine, ii) connected to the infant photographing unit, and the position of each joint of a plurality of infants and young children An arithmetic unit that provides joint angles and joint angular velocities of each joint using coordinates and angular positional coordinates, and computes the complexity of each joint angle and each joint angular velocity as sample entropy (SE), and iii) Connected to a calculating unit, A database unit for storing examination data of a plurality of infants and young children is included. Each joint is one or more joints selected from the group consisting of a left shoulder, a right shoulder, a left elbow, a right elbow, a left hip joint, a right hip joint, a left knee, a right knee, a left wrist, a right wrist, a left ankle, and a right ankle.

The calculation unit may load the data, classify the plurality of infants into a first group and a second group, extract complexities from each of the first group and the second group, and calculate a correlation coefficient between the first group and the second group. The system for automatically evaluating motion complexity of infants and young children according to an embodiment of the present invention may further include an output unit that is connected to the operation unit and outputs a correlation coefficient.

The system for automatically evaluating the motion complexity of infants according to an embodiment of the present invention includes: i) an infant photographing unit for photographing motion images of a plurality of infants lying supine, ii) each left shoulder of a plurality of infants and young children connected to the infant photographing unit, The joint angle and A calculation unit for calculating joint angular velocity, and iii) a database unit connected to the calculation unit and storing examination data of a plurality of infants and toddlers. The calculation unit loads the examination data, classifies a plurality of infants into a first group and a second group, and calculates a correlation coefficient from joint angles or joint angular velocities of different combinations of joints in the first group and the second group.

The combinations of joints in the joint angle are the combination of left hip joint and right hip joint, the combination of left knee and right knee, the combination of left ankle and right ankle, the combination of left hip and left shoulder, the combination of right hip and right shoulder, the combination of left shoulder and It may be one or more combinations selected from the group consisting of a combination of a left elbow, a combination of a right shoulder and a right elbow, a combination of a left hip and a left knee, and a combination of a right hip and a right knee. More preferably, the combination of joints may be one or more combinations selected from the group consisting of a combination of a left hip joint and a right hip joint, a combination of a left hip joint and a left shoulder, and a combination of a left shoulder and a left elbow.

Based on the quantitative evaluation established through the automatic standardization method for quantitatively analyzing the spontaneous movements of premature infants, the spontaneous movements according to early neural development can be analyzed. That is, the complexity of the infant's voluntary movements can be automatically evaluated. As a result, it is easier for clinicians to interpret based on quantitative assessments of important spatiotemporal variables including limb complexity and interlimb synchronization in the clinical setting. In addition, since it is possible to quantitatively evaluate the voluntary movements of ultra-premature infants using a deep learning algorithm, it is possible to identify high-risk indicators by detecting developmental disorders, such as cerebral palsy, at an early stage. As a result, early intervention may improve neurodevelopmental outcomes in high-risk infants.

1 is a schematic conceptual diagram of a method for automatically evaluating motion complexity of infants and toddlers according to an embodiment of the present invention.
2 is a schematic flowchart of a method for automatically evaluating motion complexity of infants and young children according to an embodiment of the present invention.
3 is a schematic flowchart of a method for automatically evaluating motion complexity of infants and young children according to another embodiment of the present invention.
FIG. 4 is a schematic conceptual diagram of the method for automatically evaluating motion complexity of infants and young children of FIG. 3 .
5 is a schematic diagram of an automatic evaluation system for motion complexity of infants and toddlers according to an embodiment of the present invention.

The terminology used herein is intended only to refer to specific embodiments and is not intended to limit the present invention. As used herein, the singular forms also include the plural forms unless the phrases clearly indicate the opposite. As used herein, the meaning of "comprising" specifies specific characteristics, regions, integers, steps, operations, elements, and/or components, and other specific characteristics, regions, integers, steps, operations, elements, elements, and/or groups. does not exclude the presence or addition of

Although not defined differently, all terms including technical terms and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the art to which the present invention belongs. Terms defined in commonly used dictionaries are additionally interpreted as having meanings consistent with related technical literature and currently disclosed content, and are not interpreted in ideal or very formal meanings unless defined.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily carry out the present invention. However, the present invention may be embodied in many different forms and is not limited to the embodiments described herein.

1 schematically shows a conceptual diagram of a method for automatically evaluating motion complexity of infants and young children according to an embodiment of the present invention. The method for automatically evaluating motion complexity of infants and young children of FIG. 1 is only for exemplifying the present invention, and the present invention is not limited thereto. Therefore, the automatic evaluation method for infants' movement complexity can be modified in other forms.

As shown in FIG. 1, in one embodiment of the present invention, first, an image obtained by capturing an image of an infant or toddler is provided. Then, position coordinates are obtained from the image using a pose prediction model. Next, joint angles and joint angular velocities are acquired by pre-processing the positional coordinates. That is, the joint angles and joint angular velocities of the shoulder (S), elbow (E), wrist (W), hip joint (H), knee (K), and ankle (A) in FIG. 1 are acquired.

The joint angle is defined as an angle formed by positional coordinates of the corresponding joint and two adjacent joints. Shoulder joint angle, angle between corresponding shoulder joints, opposite shoulder joint, one elbow joint, elbow joint angle, angle between corresponding elbow joint and two adjacent shoulder and wrist joints, hip joint angle, corresponding hip joint, opposite hip joint and one side The angle between the knee joints, the knee joint angle, and the angle between the corresponding knee joint and two adjacent hip and ankle joints correspond to this.

For example, the right shoulder angle means an angle formed by the center of the right shoulder, the left shoulder, and the right elbow. The left shoulder angle means an angle formed by the center of the left shoulder, the right shoulder, and the left elbow. The right elbow angle means an angle formed by the center of the right elbow, the right shoulder, and the right wrist. The left elbow angle means an angle formed by the center of the left elbow, the left shoulder, and the left wrist. The right hip joint angle refers to an angle formed by the center of the right hip joint, the left hip joint, and the right knee. The left hip joint angle refers to an angle formed by the center of the left hip joint, the right hip joint, and the left knee. The right knee angle means an angle formed by the center of the right knee, the right hip joint, and the right ankle. The left knee angle refers to the angle formed by the center of the left knee, the left hip joint, and the left ankle.

In particular, among various joints, the joint angles of these joints may be less affected by the height, tilt/roll angle, and direction of the camera than position coordinates. In addition, the joint angle has the advantage that clinicians can easily interpret it, so it is good to use.

Meanwhile, infants and young children may be divided into a first group and a second group using examination data of infants and young children. For example, the previously validated Hammersmith Infant Neurological Examinations (HINE) score can be used to verify the reliability of experimental data. HINE is a score evaluated by experienced physical therapists or physical therapists, and is used to evaluate the neurological outcome of infants and toddlers at the corrected age of 4 months. The HINE consists of five test sections including cranial nerve function, posture, movement, tone, reflexes and reactions, and has a score range of 0 to 78. That is, the total score is provided after the five sections are individually evaluated and then summed. In relation to one embodiment of the present invention, the HINE scores of infants of 3 to 4 months of corrected age are about 62.5 to 69. On the other hand, if the HINE score of a 3-month-old infant is 40 to 60, more specifically less than 56, the possibility of cerebral palsy is high. Considering these HINE scores, they are classified into two groups based on the preset values. For example, divide by the HINE score of 60. Therefore, infants are classified into a first group with a score of less than 60 and a second group with a score of 60 or more according to the HINE score.

Then, parameters such as correlation coefficient or complexity are extracted from the joint angle and joint angular velocity. In addition, the maximum value, minimum value, average value or standard deviation may be used as parameters. Parameters are extracted from each of the first group and the second group to calculate a correlation coefficient. When the correlation coefficient is smaller than the preset value and closer to 0, it means that there is no relationship between the first group and the second group. That is, since the difference between the first group and the first group is greater than that of the second group, which is normal infants and toddlers, it means that the first group has a high possibility of an abnormal condition, for example, cerebral palsy. Through this method, the possibility of cerebral palsy in infants is diagnosed. The basic concept according to an embodiment of the present invention will be described in more detail with reference to FIG. 2 .

2 schematically shows a method for automatically evaluating motion complexity of infants and toddlers according to an embodiment of the present invention. The method of automatically evaluating the motion complexity of infants and young children of FIG. 2 is performed by automatically evaluating the complexity of voluntary movements of infants and young children. Meanwhile, the method for automatically evaluating motion complexity of infants and young children of FIG. 2 is only for exemplifying the present invention, and the present invention is not limited thereto. Therefore, the method of automatically evaluating motion complexity of infants and young children of FIG. 2 may be modified differently.

As shown in FIG. 2, the method for automatically evaluating the motion complexity of infants includes providing an image obtained by taking motion images of infants and young children (S10), inputting the image into a pose prediction model, and providing position coordinates of each joint of infants and young children. (S20), providing joint angles and joint angular velocities using positional coordinates (S30), and calculating the complexity of each joint angle and each joint angular velocity as sample entropy (SE). In addition, the method for automatically evaluating motion complexity of infants and young children may further include other steps.

First, in step S10, an image of motion images of infants and young children is provided. That is, as shown on the left side of FIG. 1, an image of spontaneous movement of infants who lie down and have no movement restrictions is taken. Since cerebral palsy is diagnosed, infants and young children may be premature or premature. The imaging time may be 3 to 5 minutes. If the shooting time is too short, sufficient time required for position coordinate extraction cannot be secured. In addition, considering the analysis efficiency, it is preferable that the shooting time does not exceed 5 minutes.

In step S20, the image in step S10 is input to the pose prediction model to provide positional coordinates of each joint of infants and toddlers. AlphaPose, a posture estimation algorithm, can be used as a pose prediction model. AlphaPose is a pre-trained pose estimation model based on CNN (Convolutional Neural Network) architecture. Alphapose can estimate the postures of multiple infants in real time and extracts the positional coordinates. The spatio-temporal data for the head, torso, arms, and legs are automatically acquired through Alphapose. More specifically, since six joints of the shoulder, elbow, wrist, hip, knee, and ankle are located left and right, respectively, position coordinates of a total of 12 joints are automatically extracted from both sides.

The pose prediction model assigns a confidence score from 0 to 1 in each frame of the image to generate position coordinates of each joint. Position coordinates whose confidence level is less than the preset value are regarded as measurement errors and are removed from the position coordinates. That is, the overall reliability can be improved by removing noise through deletion of the location coordinates assigned with a reliability of less than the preset value. Here, the preset value may be 0.25 to 0.75. If the preset value is too small, a lot of noise may be included in the position coordinates. In addition, if the preset value is too large, the location coordinate data is too small, which may cause a problem in reliability. Therefore, the preset value is adjusted within the above range.

Meanwhile, after deleting the location coordinates, missing data among the location coordinates may be smoothed. Through this method, position coordinates are processed close to perfection.

Next, in step S30, joint angles and joint angular velocities are provided using the positional coordinates. Joint angles and joint angular velocities are less sensitive to changes in camera distance or field of view for infants than the position coordinates themselves or the lengths between adjacent joints. For this reason, it is desirable to use joint angles and joint angular velocities as basic data in terms of kinematic analysis. Joint angles are provided on a frame-by-frame basis. Joint angles are interpolated with a locally weighted smoothing method for missing data and noise reduction processing. The joint angular velocity is approximated using a symmetric difference exponent as a sequence of joint angles and finite differences.

In step S40, the complexity of each joint angle and each joint angular velocity is calculated as sample entropy (SE).

Complexity is a parameter used to quantify limb movement in infants and young children. Complexity could be a potential candidate for detecting infants at high risk for cerebral palsy. The complexity of movements of the full-term fetus is related to early neural development. Therefore, cerebral palsy in infants and young children can be detected through the index of joint movement complexity in both upper and lower limbs. In one embodiment of the present invention, sample entropy (SE) is used as complexity. Sample entropy is the degree of signal regularity for time series data, and is used to quantify the complexity of limb movement in infants and young children. Sample entropy can be calculated in the following way.

First, x: vi = [x _i , x _i+1 , … , x _i+m1 ] to construct an embedding vector with m consecutive data points. For example, the joint angle and joint angular velocity of each joint correspond to this. where m is the embedding dimension. Second, the complexity C _i ^m for each i (1≤i≤Nm) is defined by Equation 1 below.

[Equation 1]

Here, r is the tolerance value, and r=εσ _X. ε is the scaling parameter, and σ _X is the standard deviation of x. and Θ and || v _i - v _j || ₁ is defined by Equation 2 and Equation 3 below, respectively.

[Equation 2]

[Equation 3]

Thirdly, U ^m is obtained in Equation 4 below by averaging all embedding vectors, and U ^m+1 is defined by Equation 5 below.

[Equation 4]

[Equation 5]

As a result, the sample entropy (SampEn) of x is calculated as shown in Equation 6 below.

[Equation 6]

SampEn = - ln (U ^m+1 /U ^m )

Complexity can be calculated and used according to the above method.

Although not shown in FIG. 1 , a plurality of infants and young children may be classified into a first group and a second group according to the examination data. For example, HINE data of infants and young children are loaded and infants are classified into a first group and a second group based on HINE. If the HINE score is less than a preset value, it is classified as a first group, and if it is more than a preset value, it is classified as a second group. For example, the preset value may be 60. If the HINE score is low, the likelihood of cerebral palsy in infants and young children is high. In addition, when the HINE score is high, the possibility of cerebral palsy in infants and young children is low. After dividing the infants into the first group and the second group according to the HINE scores, the joint angles and joint angular velocities of each group are obtained and compared with each other. If there is a large difference in joint angle and joint angular velocity between the first group and the second group, it means that the first group is not normal to the second group, so whether or not cerebral palsy can be diagnosed through this.

In addition to the correlation coefficient or complexity, the maximum value, minimum value, average value, and standard deviation of the joint angle and joint angular velocity may be extracted as parameters in the first group and the second group, respectively. Interlimb synchronization and complexity during voluntary movements in infants and young children are hallmarks of brain dysfunction and are associated with developmental disabilities. Therefore, these parameters are utilized for kinematic analysis.

A correlation coefficient between the first group and the second group may be calculated from parameters such as the maximum value, minimum value, average value, standard deviation, correlation coefficient, or complexity. The correlation coefficient may be a Pearson correlation coefficient and is represented by a P value.

A correlation coefficient is a coefficient that numerically represents the degree of a specific correlation in order to express a statistical relationship between two variables. All values of the correlation coefficient range from -1 to +1, with ±1 representing the strongest potential agreement and 0 representing the strongest potential disagreement. The correlation coefficient is obtained by Equation 7 below using the variance.

[Equation 7]

In joint angles and joint angular velocities, the possibility of cerebral palsy can be diagnosed by obtaining correlation coefficients of the first group and the second group of combinations (x, y) of the shoulder, elbow, wrist, hip, knee, and ankle.

If the P value between the first group and the second group is less than a preset value, for example, if the P value is less than 0.05, it means that the infants of the first group are abnormal compared to the normal infants of the second group. For example, it can be diagnosed that the infants and young children of the first group are highly likely to have cerebral palsy. Conversely, when the P value between the first group and the second group is greater than or equal to the preset value, for example, when the P value is greater than or equal to 0.05, the possibility of cerebral palsy in the infants and toddlers in the first group may be regarded as low. In this case, it is possible to determine whether or not the complexity of the first group is low by classifying the infants into the first group and the second group differently, and then repeating the above process to derive a correlation coefficient.

3 schematically illustrates a method for automatically evaluating motion complexity of infants and toddlers according to another embodiment of the present invention. Since the automatic motion complexity estimation method of FIG. 3 is similar to the motion complexity automatic estimation method of FIG. 2 , like reference numerals are used for the same parts, and detailed description thereof is omitted.

As shown in FIG. 3, the method for automatically evaluating motion complexity includes providing an image obtained by capturing motion images of infants (S10), inputting the image into a pose prediction model, and providing position coordinates of each joint of infants and young children. (S20), providing joint angles and joint angular velocities using positional coordinates (S30), classifying infants into first and second groups by loading examination data of infants (S42), and first group and calculating a correlation between joint angles or joint angular velocities of different combinations of joints in the second group (S52). In addition, the method for automatically evaluating motion complexity of infants and young children may further include other steps.

In step S42, a plurality of infants and toddlers may be classified into a first group and a second group by loading examination data. For example, HINE data of infants and young children are loaded and infants are classified into a first group and a second group based on HINE. If the HINE score is less than a preset value, it is classified as a first group, and if it is more than a preset value, it is classified as a second group. For example, the preset value may be 60.

Next, in step S52, a correlation between joint angles or joint angular velocities of different combinations of joints in the first group and the second group is calculated. That is, two different joints among the 12 joints of the left shoulder, right shoulder, left elbow, right elbow, left wrist, right wrist, left hip joint, right hip joint, left knee, right knee, left ankle, and right ankle are each controlled. Select and combine from Group 1 and Group 2. Then, the correlation coefficient of the joint angle or joint angular velocity of the combination is obtained. In this case, the total number of combinations is 56. The correlation coefficient represents a P value as a Pearson correlation coefficient.

Although not shown in FIG. 3 , it is possible to estimate the possibility of cerebral palsy in infants and toddlers of the first group according to the correlation coefficient. That is, when the P value is less than the predetermined value, for example, when the P value is less than 0.05, it means that the infants of the first group are in an abnormal condition compared to the infants of the second group who are normal. As a result, it can be estimated that the possibility of cerebral palsy in infants and young children of the first group is high. Conversely, when the P value between the first group and the second group is greater than or equal to the preset value, for example, when the P value is greater than or equal to 0.05, the possibility of cerebral palsy in the infants and toddlers in the first group may be regarded as low. Through this method, it is possible to determine the degree of synchronization of the limbs of infants, which can be said to be the criterion for determining cerebral palsy.

FIG. 4 schematically illustrates the concept of the method for automatically evaluating motion complexity of infants and young children of FIG. 3 . The method for automatically evaluating motion complexity of infants and young children of FIG. 4 is merely for exemplifying the present invention, and the present invention is not limited thereto.

As shown in FIG. 4 , correlation coefficients of joint angles and joint angular velocities are derived by combining different joints in a pair in the first group and the second group, respectively, for joint angles and joint angular velocities, respectively. For example, in the first group and the second group, a combination of a left hip joint and a right hip joint, a combination of a left knee and a right knee, a combination of a left ankle and a right ankle, a combination of a left hip joint and a left shoulder, a combination of a right hip joint and a right shoulder, respectively. A combination of a left shoulder and a left elbow, a combination of a right shoulder and a right elbow, a combination of a left hip and a left knee, or a combination of a right hip and a right knee may be considered. Since these combinations have low correlation coefficients, they can be used as indicators necessary for determining developmental disabilities. More preferably, among these combinations, a combination of a left hip joint and a right hip joint, a combination of a left hip joint and a left shoulder, or a combination of a left shoulder and a left elbow may be used. As a result, the possibility of cerebral palsy in infants and young children can be diagnosed considering the degree of synchronization of the limbs of infants and young children.

Currently, general movement assessment (GMA) to assess voluntary infant movement is mainly used for early detection of cerebral palsy, along with neonatal magnetic resonance imaging and Hammersmith infant neurological examination (HINE). The GMA is a standardized motor assessment using video recordings of spontaneous infant movements. Abnormal writhing and restless movements in the GMA represent an increased risk of negative neurodevelopmental outcomes. However, GMA of voluntary movements in infants and young children is performed in a qualitative or semi-quantitative manner and depends on the skill and experience of the rater. In addition, performing GMA requires a quality training course, which has limited its use as routine practice in a clinical setting.

In one embodiment of the present invention, the problem of subjective GMA is improved to accurately estimate whether an infant is abnormal in a more objective way. This is in contrast to GMA, where the results differ depending on the evaluator.

5 schematically shows a system 100 for automatically evaluating motion complexity of infants and toddlers according to an embodiment of the present invention. The automatic evaluation system for motion complexity of infants and young children of FIG. 1 is only for exemplifying the present invention, and the present invention is not limited thereto. Accordingly, the automatic evaluation system 100 for motion complexity of infants and young children may be modified in other forms.

The system 100 for automatically evaluating motion complexity of infants and young children of FIG. 5 includes an infant photographing unit 10 , a calculation unit 20 , a database unit 30 and an output unit 40 . In addition, the system 100 for automatically evaluating motion complexity of infants and toddlers may further include other units. The automatic infant motion complexity evaluation system 100 of FIG. 5 is for performing the automatic infant motion complexity evaluation method of FIG. 2 or 3 .

The infant photographing unit 10 automatically captures motion images of infants and toddlers lying down. The calculation unit 20 is electrically connected to the infant photographing unit 10 . The calculation unit calculates the position coordinates of each joint of infants and young children and the joint angle and joint angular velocity of each joint. Meanwhile, the database unit 30 is electrically connected to the calculation unit 20 . The database unit 30 stores examination data of infants and toddlers. Examination data may include HINE data.

The calculation unit 20 loads the examination data of infants and young children from the database unit 30 and classifies the infants into a first group and a second group. For example, the calculation unit 20 loads HINE data and classifies infants into a first group and a second group based on HINE. In the first group, the HINE of the infants is less than the first preset value, and in the second group, the HINE is equal to or greater than the first preset value. Then, parameters such as complexity of each joint angle and each joint angular velocity are extracted from the first group and the second group, respectively. In addition, the parameter may further include a maximum value, a minimum value, an average value, or a standard deviation. A correlation coefficient between the first group and the second group is calculated from the parameters. When the correlation coefficient is smaller than the second preset value, the calculation unit 20 estimates that the possibility of cerebral palsy is high among the plurality of infants and toddlers in the first group. Conversely, when the correlation coefficient is greater than or equal to the second predetermined value, the calculation unit 20 estimates that the possibility of cerebral palsy in the plurality of infants and toddlers in the first group is low.

In another embodiment of the present invention, the calculation unit 20 may perform step S52 of FIG. 3 . In this case, correlation coefficients of joint angles or joint angular velocities of different combinations of joints in the first group and the second group are calculated.

Meanwhile, the obtained correlation coefficient is output through the output unit 40 . The output unit 40 may be a display. Since the output unit 40 is connected to the infant photographing unit 10, the calculation unit 20, and the database unit 30, respectively, it can output images of infants and young children, calculation processes, and examination data together.

The following experimental examples are only for exemplifying the present invention, and the present invention is not limited thereto. Accordingly, the present invention may be modified in other forms as well.

Experimental example

Subject infants and toddlers

Premature infants admitted to the neonatal intensive care unit of two tertiary hospitals with a gestational age of less than 32 weeks or a birth weight of less than 1500 g were enrolled. Gestational age was determined by the date of the last menstrual period and confirmed by ultrasound in the first or second trimester of pregnancy. Infants and young children with genetic syndromes, major congenital anomalies or medically unstable conditions defined as requiring cardiovascular support, active sepsis, or surgical pathology at the time of image acquisition of spontaneous movement were excluded from the study.

clinical information

Perinatal and postnatal clinical trials of infants and young children, including infant birth data including sex, gestational age, birth weight, 1/5 minute Apgar score, prenatal steroid use, history of preeclampsia, ventilation use, bronchopulmonary dysplasia severity13, and history of sepsis. got the information Seizure history, presence of retinopathy of prematurity, and brain ultrasonography were performed weekly. Intracranial hemorrhage was also classified. Brain magnetic resonance imaging was performed for infants with a birth weight of 1000 g and a gestational age of less than 29 weeks or full-term infants with severe ventricular hemorrhage on brain ultrasonography.

video recording

Spontaneous movement images of preterm infants were automatically recorded according to the standard protocol of the GMA at the equivalent age of a period considered 40±1 weeks postmenstrual age. As a condition for video recording of spontaneous movements, infants are supine, naked, fully awake but not agitated, such as crying or scolding. It should also be comfortable, quiet and undisturbed at neutral temperatures. Move the body and all limbs freely, including fingers and toes included in the recorded scene. For video recording, a smartphone camera can be used if necessary, and it is recorded for 3 to 5 minutes for at least 2 minutes in a row. If it was determined that the recorded video did not meet the above conditions, it was repeatedly recorded within the age range. The recorded footage was automatically edited and put together into a three to five minute sequence.

Movement recognition and kinematic analysis of infants and toddlers

Using AlphaPose, the location coordinates of 12 joints, including the shoulder, elbow, wrist, hip, knee, and ankle, were automatically extracted from both sides of the image taken of the infant in voluntary movement. Position coordinates of each joint were generated at each frame with a confidence level of 0 to 1, and position coordinates with a confidence level of less than 0.5 were considered measurement errors and automatically removed. Joint angles were acquired frame by frame and interpolated with a locally weighted smoothing method for missing data handling and noise reduction. Joint angular velocities were approximated using symmetric difference exponents as sequences of finite differences.

After preprocessing the time series data of joint angles and joint angular velocities, maximum, minimum, average, and standard deviation values were calculated for each infant. Interlimb synchronization and complexity during voluntary movements are hallmarks of brain dysfunction and are associated with developmental disabilities, so they were investigated for kinematic analysis. To quantify the complexity of limb movement in preterm infants, we adopted sample entropy, which is the degree of signal regularity for series data. Interlimb synchronization of joint angles and joint angular velocities was quantified using a similarity index including Pearson's correlation coefficient, which is a measure of dependence between two pairs of data.

Hammersmith Infant Neurological Screening

The Hammersmith Infant Neurological Examination (HINE) was performed. HINE was used to evaluate neurological outcomes in infants at 4 months of corrected age. Five test sections, including infants' cranial nerve function, posture, movement, tension, reflexes, and reactions, were evaluated individually, and then a total score was assigned and the HINE data was saved. Based on this information, 65 infants were divided into two groups: the first group with HINE score less than 60 and the second group with HINE score of 60 or more.

Experiment result

Measurement results of joint angles, joint angular velocities, and correction coefficients between HINE and infant complexity

Table 1 below shows the correction coefficients and P values of joint angles and joint angular velocities for each joint for 65 infants and toddlers. The correction factor is the correction value between HINE and infant complexity in joint angle and joint angular velocity. Since the asterisk (*) means that the P value is less than 0.05, it means that the difference between the first group and the second group is large, so that the first group is highly likely to have cerebral palsy.

measurement target part correction factor P value joint angle right shoulder 0.288 0.020* left shoulder 0.277 0.026* right elbow 0.206 0.099 left elbow 0.234 0.060 right hip joint 0.227 0.069 left hip joint 0.312 0.012* right knee 0.246 0.048* left knee 0.315 0.011* joint angular velocity right shoulder 0.267 0.032* left shoulder 0.279 0.024* right elbow 0.244 0.051 left elbow 0.264 0.034* right hip joint 0.281 0.023* left hip joint 0.370 0.002* right knee 0.228 0.068 left knee 0.386 0.001*

Complexity measurement results

As shown in Table 2, as a result of measuring the complexity of infants by dividing them into the first group and the second group, when the P value is less than 0.05, the right shoulder, right elbow, left hip joint, left knee, and right knee in the joint angle it was applicable In addition, in joint angular velocity, the P value was less than 0.05 in the left shoulder, right shoulder, right elbow, left hip joint, right hip joint, left knee, and right knee. Therefore, when complexity was used as a parameter, it was possible to diagnose the possibility of cerebral palsy in infants at these joints.

measurement target part HINE<60
(16 people) HINE≥60
(44 people) P value joint angle right shoulder 0.050±0.031 0.074±0.038 0.025* left shoulder 0.056±0.036 0.074±0.038 0.083 right elbow 0.042±0.028 0.064±0.038 0.048* left elbow 0.049±0.035 0.065±0.040 0.106 right hip joint 0.049±0.030 0.068±0.039 0.053 left hip joint 0.046±0.025 0.069±0.030 0.009* right knee 0.043±0.030 0.062±0.033 0.014* left knee 0.043±0.022 0.064±0.030 0.008* joint angular velocity right shoulder 0.122±0.068 0.166±0.072 0.041* left shoulder 0.121±0.069 0.168±0.069 0.031* right elbow 0.100±0.068 0.148±0.078 0.027* left elbow 0.113±0.078 0.154±0.081 0.057 right hip joint 0.122±0.063 0.160±0.063 0.029* left hip joint 0.112±0.045 0.163±0.060 0.003* right knee 0.108±0.065 0.140±0.057 0.023* left knee 0.095±0.052 0.146±0.062 0.003*

Maximum value, minimum value, average value, standard deviation measurement result of joint angle

The maximum, average, minimum, and standard deviation of the measured joint angles are shown in Table 3 below for each upper and lower limb on both sides. Maximum value of right elbow joint angle (156.85±26.97 vs 174.88±14.42, P=0.009), mean value of right elbow joint angle (70.01±29.66 vs 90.91±27.1, P=0.014), standard deviation of right shoulder joint angle (13.31 There were significant differences between infants with HINE <60 and HINE ≥60 in ±2.96 vs 15.38±3.19, P=0.021) and standard deviation of left shoulder joint angle (14.06±8.63 vs 16.57±4.25, P=0.009).

measurement target part HINE<60
(16 people) HINE≥60
(49 people) P value right shoulder max value 172.16±8.03 176.33±4.35 0.097 medium 135.22±19.35 140.59±10.01 0.301 minimum 97.01±22.31 93.24±29.08 0.976 Standard Deviation 13.31±2.96 15.38±3.19 0.021* left shoulder max value 170.22±10.57 176.49±2.56 0.074 medium 131.6±15.43 139.5±10.35 0.070 minimum 93.95±34.58 90.43±31.42 0.738 Standard Deviation 14.06±8.63 16.57±4.25 0.009* right elbow max value 156.85±26.97 174.88±14.42 0.009* medium 70.01±29.66 90.91±27.1 0.011* minimum 6.72±11.9 10.6±13.85 0.411 Standard Deviation 34.22±11.15 38.41±7.83 0.179 left elbow max value 164.31±27.13 172.4±13.93 0.217 medium 74.94±31.19 90.81±27.39 0.056 minimum 4.97±10.65 9.86±17.71 0.394 Standard Deviation 37.22±12.04 38.46±9.07 0.663 right hip joint max value 173.85±8.57 174.71±6.95 0.522 medium 125.86±20.58 128.87±15.28 0.534 minimum 77.25±23.9 73.86±21.48 0.605 Standard Deviation 21.3±7.94 21.72±5.13 0.803 left hip joint max value 171.91±9.43 174.04±10.77 0.385 medium 131.2±31.8 130.92±15.1 0.322 minimum 72.13±37.24 77.37±23.95 0.843 Standard Deviation 18.14±4.88 20.13±6.39 0.223 right knee max value 172.48±12.92 178.05±5.53 0.161 medium 118.42±25.09 116.11±23.44 0.737 minimum 44.87±22.67 40.47±20.26 0.467 Standard Deviation 29.24±8.36 32.23±7.01 0.162 left knee max value 173.02±23.2 177.13±6.53 0.648 medium 100.77±33.74 114.63±22.88 0.141 minimum 37.63±25.64 47.98±21.2 0.113 Standard Deviation 25.98±9.61 30.35±6.85 0.050

Maximum, minimum, average, and standard deviation measurement results of joint angular velocity

The maximum, average, minimum, and standard deviation of the measured joint angular velocity are shown in Table 4 below for each upper and lower extremity on both sides. The maximum value of the right shoulder joint angular velocity was significantly different between the two groups. (34.55±16.41 vs 48.2±24.26, P=0.027) And there was a significant difference between the two groups in the minimum angular velocity of the right shoulder joint. (-35.08±16.39 vs -46.72±22.79, P=0.048). The average value of the joint angular velocity of the right hip joint showed a significant difference between the two groups. (-0.14±0.15 vs 0.03±0.17, P=0.001) Also, the average value of joint angular velocity of the right knee showed a significant difference between the two groups. (0.16±0.31 vs -0.05±0.26, P=0.018). And the standard deviation of the joint angular velocity of the right elbow showed a significant difference between the two groups. (17.53±8.94 vs 24.62±9.31, P=0.017) The standard deviation of joint angular velocity of the right knee also showed a significant difference between the two groups. (14.97±8.1 vs 20.79±8.62, P=0.022)

measurement target part HINE<60
(16 people) HINE≥60
(49 people) P value right shoulder max value 34.55±16.41 48.2±24.26 0.027* medium -0.03±0.1 -0.01±0.18 0.710 minimum -35.08±16.39 -46.72±22.79 0.048* Standard Deviation 7.81±3.65 10.83±4.87 0.017* left shoulder max value 41.13±32.17 47.71±26.04 0.152 medium -0.03±0.22 0.02±0.15 0.796 minimum -39.9±30.06 -48.42±25.06 0.120 Standard Deviation 9.06±6.69 11.32±5.36 0.106 right elbow max value 74.25±30.79 96.11±31.82 0.051 medium 0.03±0.28 0.02±0.42 0.941 minimum -74.4±35.05 -93.49±33.4 0.054 Standard Deviation 17.53±8.94 24.62±9.31 0.017* left elbow max value 79.59±37.56 93.79±32.72 0.151 medium 0.17±0.28 0.0±0.36 0.078 minimum -76.55±35.89 -94.63±31.19 0.057 Standard Deviation 19.55±10.76 23.95±9.77 0.133 right hip joint max value 47.1±26.33 57.43±22.36 0.129 medium -0.14±0.15 0.03±0.17 0.001* minimum -46.95±27.95 -55.75±21.94 0.198 Standard Deviation 10.9±5.68 14.03±5.6 0.080 left hip joint max value 53.81±35.37 54.85±21.09 0.912 medium 0.02±0.12 0.05±0.17 0.161 minimum -48.59±28.45 -49.2±18.71 0.937 Standard Deviation 10.47±5.68 12.85±5.23 0.196 right knee max value 63.82±30.12 78.72±29.13 0.083 medium 0.16±0.31 -0.05±0.26 0.010* minimum -66.07±30.16 -80.37±29.69 0.100 Standard Deviation 14.97±8.1 20.79±8.62 0.022* left knee max value 66.31±35.5 72.53±25.96 0.451 medium 0.04±0.19 -0.06±0.24 0.153 minimum -67.7±35.3 -75.87±25.39 0.316 Standard Deviation 14.98±7.78 19.46±7.75 0.049*

Comparison result of correlation coefficients of the first group and the second group of joint angles

In Table 5, parentheses mean ± deviation, and rows and columns indicate combinations of joints selected from the first group and the second group, respectively. Different joints are selected and combined in the first group and the second group, and mutually identical joints are not selected and displayed as N/A (not available). Table 5 shows correlation coefficients of the first group and the second group of joint angles for combinations of these joints.

As shown in Table 5, the correlation coefficient of joint angle showed a significant positive correlation with the HINE score in most upper and lower extremity joints within the range of 0.2-0.4. In addition, it was found that the P value was less than 0.05 in the combination of the left hip joint and the right hip joint, the combination of the left hip joint and the left shoulder joint, and the combination of the left elbow and left shoulder joint indicated in bold in Table 5. Therefore, it was confirmed that there was a significant difference between the first group and the second group in this combination.

right
shoulder right
elbow right
hip joint right
knee left side
shoulder left side
elbow left side
hip joint left side
knee right
shoulder NA 0.14 (0.43)
vs
0.06 (0.34), 0.469 0.01 (0.23)
vs
-0.05 (0.21),
0.342 0.03 (0.22)
vs
0.05 (0.19),
0.831 0.22 (0.29)
vs
0.20 (0.21),
0.839 0.06 (0.29)
vs
0.02 (0.23),
0.581 -0.1 (0.21)
vs
-0.07 (0.21),
0.551 0.07 (0.24)
vs
0.10 (0.19),
0.682 right
elbow 0.14 (0.43)
vs
0.06 (0.34), 0.469 NA 0.1 (0.25)
vs
0.03 (0.21),
0.220 -0.09 (0.24)
vs
-0.01 (0.18),
0.150 0.09 (0.20)
vs
0.03 (0.19),
0.314 0.21 (0.27)
vs
0.15 (0.25),
0.373 0.05 (0.21)
vs
0.03 (0.21),
0.777 -0.02 (0.22)
vs
-0.04 (0.19),
0.707 right
hip joint 0.01 (0.23)
vs
-0.05 (0.21),
0.342 0.1 (0.25)
vs
0.03 (0.21),
0.220 NA 0.77 (0.28)
vs
-0.83 (0.11),
0.831 0.06 (0.19)
vs
-0.01 (0.18),
0.205 0.14 (0.19)
vs
0.11 (0.24),
0.663 0.06 (0.35)
vs
0.27 (0.26),
0.016 -0.22 (0.28)
vs
-0.3 (0.21),
0.238 right
knee 0.03 (0.22)
vs
0.05 (0.19),
0.831 -0.09 (0.24)
vs
-0.01 (0.18),
0.150 -0.77 (0.28)
vs
-0.83 (0.11),
0.831 NA 0.02 (0.23)
vs
0.02 (0.19),
0.446 -0.15 (0.16)
vs
-0.13 (0.2),
0.769 -0.21 (0.32)
vs
-0.30 (0.23),
0.484 0.29 (0.27)
vs
0.29 (0.22),
0.615 left side
shoulder 0.22 (0.29)
vs
0.20 (0.21),
0.839 0.09 (0.2)
vs
0.03 (0.19),
0.314 0.06 (0.19)
vs
-0.01 (0.18),
0.205 0.02 (0.23)
vs
0.02 (0.19),
0.446 NA 0.23 (0.27)
vs
0.08 (0.34),
0.029 0.16 (0.23)
vs
0.00 (0.19),
0.010 0.08 (0.24) vs.
0.02 (0.16),
0.257 left side
elbow 0.06 (0.29)
vs
0.02 (0.23),
0.581 0.21 (0.27)
vs
0.15 (0.25),
0.373 0.14 (0.19)
vs
0.11 (0.24),
0.663 -0.15 (0.16)
vs
-0.13 (0.2),
0.769 0.23 (0.27)
vs
0.08 (0.34),
0.029 NA -0.04 (0.22)
vs
0.06 (0.22),
0.116 0.00 (0.2)
vs
-0.04 (0.22),
0.507 left side
hip joint -0.1 (0.21)
vs
-0.07 (0.21),
0.551 0.05 (0.21)
vs
0.03 (0.21),
0.777 0.06 (0.35)
vs
0.27 (0.26),
0.016 -0.21 (0.32)
vs
-0.3 (0.23),
0.484 0.16 (0.23)
vs
0.0 (0.19),
0.010 -0.04 (0.22)
vs
0.06 (0.22),
0.116 NA 0.66 (0.24)
vs
-0.75 (0.24),
0.117 left side
knee 0.07 (0.24)
vs
0.1 (0.19),
0.682 0.02 (0.22)
vs
-0.04 (0.19),
0.707 -0.22 (0.28)
vs
-0.3 (0.21),
0.238 0.29 (0.27)
vs
0.29 (0.22),
0.615 0.08 (0.24) vs.
0.02 (0.16),
0.257 0.00 (0.20)
vs
-0.04 (0.22),
0.507 -0.66 (0.24)
vs
-0.75 (0.24),
0.117 NA

Comparison result of correlation coefficients of the first group and the second group of joint angular velocities

In Table 6, parentheses mean ± deviation, and rows and columns indicate combinations of joints selected from the first group and the second group, respectively. Different joints are selected and combined in the first group and the second group, and mutually identical joints are not selected and displayed as N/A (not available). Table 6 shows correlation coefficients of the first group and the second group of joint angular velocities for combinations of these joints.

As listed in Table 6, the correlation coefficient of joint angular velocity showed a significant positive correlation with the HINE score within the range of 0.2–0.4 for some data. Parentheses mean ± deviation. No P values less than 0.05 were found between the first and second groups in this combination.

right
shoulder right
elbow right
hip joint right
knee left side
shoulder left side
elbow left side
hip joint left side
knee right
shoulder N/A 0.13 (0.38)
vs
0.12 (0.24),
0.896 -0.04 (0.11)
vs
-0.07 (0.1),
0.278 0.05 (0.12)
vs
0.08 (0.1),
0.349 0.19 (0.24)
vs
0.17 (0.14),
0.726 0.09 (0.17)
vs
0.04 (0.16),
0.361 -0.05 (0.14)
vs
-0.05 (0.12),
0.920 -0.03 (0.19)
vs
0.04 (0.11),
0.139 right
elbow 0.13 (0.38)
vs
0.12 (0.24),
0.896 N/A 0.05 (0.15)
vs
0.02 (0.14),
0.472 -0.02 (0.18)
vs
0.00 (0.13),
0.575 0.10 (0.12)
vs
0.07 (0.14),
0.273 0.19 (0.18)
vs
0.10 (0.15),
0.128 -0.03 (0.16)
vs
-0.02 (0.12),
0.916 0.04 (0.16) vs.
0.00 (0.14),
0.275 right
hip joint -0.04 (0.11)
vs
-0.07 (0.1),
0.278 0.05 (0.15)
vs
0.02 (0.14),
0.472 N/A -0.76 (0.24)
vs
-0.81 (0.13),
0.670 0.01 (0.12)
vs
0.00 (0.13),
0.686 0.03 (0.13)
vs
0.02 (0.14),
0.819 0.18 (0.23)
vs
0.26 (0.22),
0.179 -0.26 (0.20)
vs
-0.25 (0.2),
0.841 right
knee 0.05 (0.12)
vs
0.08 (0.10),
0.349 -0.02 (0.18)
vs
0.0 (0.13),
0.575 -0.76 (0.24)
vs
-0.81 (0.13),
0.670 N/A 0.01 (0.11)
vs
0.00 (0.11),
0.779 -0.03 (0.12)
vs
-0.04 (0.15),
0.835 -0.25 (0.20)
vs
-0.26 (0.21),
0.891 0.31 (0.17)
vs
0.22 (0.20),
0.114 left side
shoulder 0.19 (0.24)
vs
0.17 (0.14),
0.726 0.10 (0.12)
vs
0.07 (0.14),
0.273 0.01 (0.12)
vs
0.00 (0.13),
0.686 0.01 (0.11)
vs
0.00 (0.11),
0.779 N/A 0.26 (0.26)
vs
0.12 (0.26),
0.060 -0.04 (0.15)
vs
-0.02 (0.13),
0.657 0.03 (0.13)
vs
0.03 (0.11),
0.901 left side
elbow 0.09 (0.17)
vs
0.04 (0.16),
0.361 0.19 (0.18)
vs
0.10 (0.15),
0.128 0.03 (0.13)
vs
0.02 (0.14),
0.819 -0.03 (0.12)
vs
-0.04 (0.15),
0.835 0.26 (0.26)
vs
0.12 (0.26),
0.060 N/A -0.03 (0.13)
vs
0.04 (0.15),
0.110 0.06 (0.18)
vs
-0.03 (0.12),
0.057 left side
hip joint 0.05 (0.14)
vs
-0.05 (0.12),
0.920 -0.03 (0.16)
vs
-0.02 (0.12),
0.916 0.18 (0.23)
vs
0.26 (0.22),
0.179 -0.25 (0.20)
vs
-0.26 (0.21),
0.891 -0.04 (0.15)
vs
-0.02 (0.13),
0.657 -0.03 (0.13)
vs
0.04 (0.15),
0.110 N/A -0.66 (0.25)
vs
-0.73 (0.26),
0.196 left side
knee -0.03 (0.19)
vs
0.04 (0.11),
0.139 0.04 (0.16) vs.
0.0 (0.14),
0.275 -0.26 (0.20)
vs
-0.25 (0.20),
0.841 0.31 (0.17)
vs
0.22 (0.20),
0.114 0.03 (0.13) vs.
0.03 (0.11),
0.901 0.06 (0.18)
vs
-0.03 (0.12),
0.057 -0.66 (0.25)
vs
-0.73 (0.26),
0.196 N/A

Voluntary movements in infants and young children are endogenously generated by central pattern generator networks in the spinal cord and brainstem. This network can result in simple movements of the body, but during brain development it can be controlled by movements of spinal superstructures such as the subplate and cortical plate, which leads to motor complexity. However, abnormal patterns of voluntary movement are characterized by deficiencies, resulting from damage or dysfunction of the sublamina, cortical lamina, and/or connective fibers.

A decrease in complexity is closely related to cerebral palsy. Therefore, complexity, which means spatial and temporal variability, can be used as a key index for quantitative measurement in order to objectively measure the voluntary movements of premature infants as in the above experimental example. Decreased complexity indicates fetal or neonatal impairment and is considered one of the important characteristics distinguishing between normal and abnormal spontaneous movements during the moving period. In particular, bilateral lower extremity movement is associated with a risk of developmental delay.

The similarity of Pearson's correlation coefficient does not match the result of complexity. It can be hypothesized that the similarity of inter-limb or intra-limb movement is characteristic of asynchronous movement. However, the similarity index based on Pearson's correlation coefficient is not a measure for excluding normal motion.

Although the present invention has been described as described above, those working in the art to which the present invention belongs will readily understand that various modifications and variations are possible without departing from the concept and scope of the claims described below.

10. Infant & Toddler Photography Department
20. Calculation unit
40. Output
30. Database section
100. Automatic evaluation system for movement complexity in infants and young children

Claims (14)

Providing an image obtained by photographing motion images of a plurality of infants lying supine;
Providing positional coordinates of each joint of the plurality of infants by inputting the image to a pose prediction model;
Providing joint angles and joint angular velocities of the joints using the angular position coordinates; and
Calculating complexity of each joint angle and each joint angular velocity as sample entropy (SE)
including,
In the step of providing the positional coordinates of each joint, the respective joints are left shoulder, right shoulder, left elbow, right elbow, left wrist, right wrist, left hip joint, right hip joint, left knee, right knee, left ankle, and right A method for automatic evaluation of motion complexity in infants, which is one or more joints selected from the group consisting of the ankle. In paragraph 1,
Classifying the plurality of infants into a first group and a second group by loading previously stored examination data of the plurality of infants and young children;
extracting the complexity from each of the first group and the second group; and
Calculating a correlation coefficient between the first group and the second group from the complexity
Method for automatically assessing movement complexity of infants and toddlers further comprising a. In paragraph 1,
After the step of providing the location coordinates, a reliability of 0 to 1 is assigned to each location coordinate and deleting the location coordinates to which the reliability is less than a preset value, wherein the preset value is 0.25 to 0.75. A method for automatic assessment of movement complexity in infants and young children. In paragraph 3,
After the step of deleting the position coordinates, missing data among the position coordinates is smoothed. In paragraph 1,
In the step of calculating the complexity, each joint angle is one or more joint angles selected from the group consisting of a right shoulder, a right elbow, a left hip joint, a left knee, and a right knee, and each joint angular velocity is a left shoulder, a right shoulder, and a right knee. A method for automatically evaluating motion complexity in infants, which is the angular velocity of one or more joints selected from the group consisting of an elbow, a left hip joint, a right hip joint, a left knee, and a right knee. Providing an image obtained by photographing motion images of a plurality of infants lying supine;
By inputting the image into a pose prediction model, the left shoulder, right shoulder, left elbow, right elbow, left wrist, right wrist, left hip joint, right hip joint, left knee, right knee, left ankle, and right ankle of each of the plurality of infants and toddlers Providing positional coordinates of one or more joints selected from the group consisting of;
Providing a joint angle or joint angular velocity of a joint using the angular position coordinates;
classifying the plurality of infants into a first group and a second group by loading previously stored examination data of the plurality of infants and young children; and
Calculating a correlation coefficient from joint angles or joint angular velocities of different combinations of joints in the first group and the second group, respectively.
A method for automatically assessing movement complexity in infants and toddlers, including a. In paragraph 6,
In the step of calculating the correlation coefficient,
The combination of the joints of the joint angle is a combination of a left hip joint and a right hip joint, a combination of a left knee and a right knee, a combination of a left ankle and a right ankle, a combination of a left hip joint and a left shoulder, a combination of a right hip joint and a right shoulder, A method for automatically evaluating movement complexity in infants, which is one or more combinations selected from the group consisting of a combination of the left shoulder and left elbow, a combination of the right shoulder and right elbow, a combination of the left hip joint and the left knee, and a combination of the right hip joint and the right knee. In paragraph 7,
The combination of the joints is one or more combinations selected from the group consisting of a combination of a left hip joint and a right hip joint, a combination of a left hip joint and a left shoulder, and a combination of a left shoulder and a left elbow. An infant filming unit that photographs motion images of a plurality of infants lying supine;
It is connected to the infant photographing unit to provide joint angles and joint angular velocities of each joint using position coordinates of each joint of the plurality of infants and young children and each position coordinate, and the complexity of each joint angle and each joint angular velocity is calculated as sample entropy. A calculation unit that calculates with (sample entropy, SE), and
A database unit connected to the calculation unit and storing examination data of the plurality of infants and young children
including,
Each of the above joints is one or more joints selected from the group consisting of a left shoulder, a right shoulder, a left elbow, a right elbow, a left wrist, a right wrist, a left hip joint, a right hip joint, a left knee, a right knee, a left ankle, and a right ankle. Movement Complexity Automatic Rating System. In paragraph 9,
The calculation unit loads the data, classifies the plurality of infants into a first group and a second group, extracts the complexity from each of the first group and the second group, and calculates the distance between the first group and the second group. An automatic assessment system for motion complexity in infants and young children that calculates correlation coefficients. In paragraph 10,
An automatic evaluation system for motion complexity of infants and young children further comprising an output unit connected to the operation unit and outputting the correlation coefficient. An infant filming unit that photographs motion images of a plurality of infants lying supine;
It is connected to the infant photographing unit and consists of a left shoulder, a right shoulder, a left elbow, a right elbow, a left wrist, a right wrist, a left hip joint, a right hip joint, a left knee, a right knee, a left ankle, and a right ankle of each of the plurality of infants. A calculation unit for calculating joint angles and joint angular velocities of each joint using the positional coordinates of the joints and the respective positional coordinates; and
A database unit connected to the calculation unit and storing examination data of the plurality of infants and young children
including,
The calculation unit loads the examination data, classifies the plurality of infants into a first group and a second group, and correlates joint angles or joint angular velocities of different combinations of joints in the first group and the second group. An automatic evaluation system for infants' movement complexity that calculates coefficients. In paragraph 12,
The combination of the joints in the joint angle is a combination of a left hip joint and a right hip joint, a combination of a left knee and a right knee, a combination of a left ankle and a right ankle, a combination of a left hip joint and a left shoulder, a combination of a right hip joint and a right shoulder, a left An automatic evaluation system for infants' movement complexity that is one or more combinations selected from the group consisting of a combination of a shoulder and a left elbow, a combination of a right shoulder and a right elbow, a combination of a left hip and a left knee, and a combination of a right hip and a right knee. In paragraph 13,
The combination of the joints is one or more combinations selected from the group consisting of a combination of a left hip joint and a right hip joint, a combination of a left hip joint and a left shoulder, and a combination of a left shoulder and a left elbow.

Images