Joint evaluation device, method, and program

Abstract

A joint evaluating device (10) is provided with: an inertial sensor unit (30) which is mounted in the vicinity of a joint linking bones on both sides thereof, with the joint axis of the joint and a detection axis parallel to one another, and which detects, as a waveform signal, the movement of the bones of the joint axis of which the range of motion of joint movement is limited, among the joint axes; a load detecting unit (34) which detects a load applied to the joint; a data acquiring unit (21) which, when generation of a load is detected, acquires the waveform signal detected by the inertial sensor unit (30), in the time direction and the intensity direction; and a feature quantity calculating unit (22) which analyzes the waveform signal to calculate a feature quantity for evaluating the quality of movement of the joint. As a result, the behavior of the joint in the direction in which the range of motion is limited is detected in an anatomically meaningful way, and an injury, disability or the like is easily and accurately evaluated and predicted.

Description

[Technical Field]

[0001] The present invention relates to a technique for easily and unconstrainedly evaluating joint movements that may lead to injuries or disorders by measuring the translational acceleration and angular velocity of each axis of an inertial sensor attached along the joint axis. [Background technology]

[0002] For example, the valgus of the elbow seen during the acceleration phase of pitching, and the valgus and rotation of the knee seen during a turn are movements that cause these joints to deviate from their normal range of motion, placing stress on tissues such as the ligaments and joint capsules that resist this. Furthermore, repeated impact stress in the longitudinal direction of the lower leg during running, etc., is a source of stress that causes the accumulation of microdamage in the bones and periosteum. In the field of sports medicine, there has been interest in establishing measurement and evaluation methods to easily and accurately assess poor kinematic and kinetic characteristics observed during these movements that could pose a risk for injury or disorders.

[0003] Conventional evaluation methods have evaluated the magnitude of joint angles, joint moments, and joint contact forces (Non-Patent Documents 1 and 2). Recently, a method has been proposed in which knee movements during landing are measured using a small inertial sensor, and the correlation between the peak value of the acceleration data and the peak value of the knee moment is evaluated (Non-Patent Document 3). Patent Document 1 also proposes a walking motion evaluation system that uses motion capture to estimate principal component scores, including changes in knee joint angles and joint moments associated with walking, based on floor reaction force data detected during walking, and evaluates walking motion. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent No. 5315504 [Non-patent literature]

[0005] [Non-Patent Document 1] Hewett, T., Myer, G., Ford, K., Heidt, R., Colosimo, A., McLean, S., Bogert, A., Paterno, M., Succop, P. (2005). Biomechanical measures of neuromuscular control and valgus loading of the knee predict anterior cruciate ligament injury risk in female athletes: AProspective Study The American Journal of Sports Medicine 33(4), 492-501. https: / / dx.doi.org / 10.1177 / 0363546504269591 [Non-patent document 2] Kristianslund, E., Faul, O., Bahr,R., Myklebust, G., Krosshaug, T. (2014). Sidestep cutting technique and kneeabduction loading: implications for ACL prevention exercises. British Journal of Sports Medicine 48(9), 779 783. https: / / dx.doi.org / 10.1136 / bjsports-2012-091370 [Non-patent document 3] Morgan, A., O'Connor, K. (2019).Evaluation of an accelerometer to assess knee mechanics during a drop landingJournal of Biomechanics 86, 125-131. https: / / dx.doi.org / 10.1016 / j.jbiomech.2019.01.055 Summary of the Invention [Problem to be solved by the invention]

[0006] In Patent Document 1 and Non-Patent Documents 1 and 2, an expensive measurement environment, such as a 3D motion capture system or a ground reaction force sensor, was required to calculate variables such as the magnitude of joint angles. In addition, motion capture systems are limited in the measurement space, which significantly limits the types and range of motion that can be measured. Furthermore, measuring movements with these systems requires attaching multiple (numerous) reflective markers to the body surface, obtaining their 3D position coordinates, and then calculating their acceleration and posture matrix, an analytical process that requires extensive offline processing. Due to these limitations, these methods are not suitable for evaluation and feedback immediately after movements. The method described in Non-Patent Document 3 evaluates only peak values ​​that appear transiently in the time series of acceleration data, which is insufficient for detailed evaluation of time-varying behavior of lower limb joints.

[0007] The present invention has been made in consideration of the above, and focuses on the relationship between anterior cruciate ligament (ACL) injuries, which are common sports injuries, and postural instability characteristics, as well as injuries to the elbow joint, etc., and proposes a joint evaluation device, method, and program that can easily and accurately evaluate and predict joint damage, etc., based on the behavior of the joint in directions where the range of motion is limited. [Means for solving the problem]

[0008] The joint evaluation device of the present invention is equipped with an inertial sensor unit that is attached near a joint connecting bones on both sides with its joint axis and detection axis parallel, and that detects, as a waveform signal, the movement of the bones of a joint axis that has a limited range of motion of the joint movement among the joint axes; a load detection unit that detects a load applied to the joint; data acquisition means that, when the generation of the load is detected, acquires the waveform signal detected by the inertial sensor unit in the time direction and the intensity direction; and feature calculation means that analyzes the waveform signal and calculates feature values ​​for evaluating the movement quality of the joint.

[0009] Furthermore, the joint evaluation method of the present invention uses an inertial sensor unit that is attached near a joint connecting bones on both sides with its joint axis and detection axis parallel, and that detects, as a waveform signal, the movement of the bones of a joint axis that has a limited range of motion of joint movement, and a load detection unit that detects the load applied to the joint, and when a data acquisition means detects the occurrence of the load, it acquires the waveform signal detected by the inertial sensor unit in the time direction and the intensity direction, and a feature calculation means analyzes the waveform signal and calculates a feature for evaluating the movement quality of the joint.

[0010] In addition, the program of the present invention uses an inertial sensor unit that is attached near a joint connecting bones on both sides with the joint axis and detection axis parallel to each other and detects, as a waveform signal, the movement of the bones of a joint axis that has a limited range of motion of the joint movement, and a load detection unit that detects the load applied to the joint, and causes a computer to function as data acquisition means that, when it detects the occurrence of the load, acquires the waveform signal detected by the inertial sensor unit in the time direction and the intensity direction, and feature calculation means that analyzes the waveform signal and calculates feature values ​​for evaluating the movement quality of the joint.

[0011] According to these inventions, a wearable inertial sensor detects the bone movement of a joint axis with a limited range of motion as a waveform signal based on anatomical evidence. When a data acquisition means detects the occurrence of a load, for example, when landing on the floor during a one-leg drop, it acquires the waveform signal detected by the inertial sensor in the time and intensity directions. The feature calculation means analyzes the acquired waveform signal to calculate a feature for evaluating the quality of joint movement. In this way, by using a wearable inertial sensor to detect bone movement along a joint axis with a limited range of motion and bone rotation around a joint axis with a limited range of motion and obtaining these feature values, it becomes possible to easily and accurately evaluate and predict damage or injury to such joints based on the correlation between postural fluctuations, such as left-right or forward-backward deviations, twisting, imbalance, and other instability during exercise, such as a drop landing, and the risk of injury or injury. The inertial sensor is not limited to detecting the movement of all or a plurality of joint axes; it may also detect a signal from, for example, a single joint axis of interest. Furthermore, the joint axis to be detected may be set appropriately depending on the joint part to be inspected and the way in which a load is applied. [Effects of the Invention]

[0012] According to the present invention, it is possible to detect the behavior of a joint in a direction in which the range of motion is limited by assigning anatomical significance, and to easily and accurately evaluate or predict damage, disability, etc. [Brief explanation of the drawings]

[0013] [Figure 1] 1 is a block diagram showing an embodiment of a joint evaluation device according to the present invention. [Figure 2] 10 is a diagram showing the correspondence between joint axes in multiple directions of a joint and detection axes of an inertial sensor. FIG. [Figure 3] Figures explaining the movements of a subject performing joint movement quality evaluation, where (A) shows the state at the time of one-leg drop landing, and (B) shows the posture immediately afterwards. [Figure 4]FIG. 1A is a block diagram of a control unit showing a first embodiment of a joint evaluation device, and FIG. 1B is a block diagram of a control unit showing a second embodiment of a joint evaluation device. [Figure 5] This figure illustrates the detection results of a subject using the first evaluation method. (A) shows the detection signals from three sensors, (B) is a histogram of signal strength in the movable direction (around the Y-axis), (C) is a histogram of signal strength in the non-movable direction (around the Z-axis), and (D) is a histogram of signal strength in the non-movable direction (around the X-axis). [Figure 6] Figures showing the detection results of another subject, illustrating the first evaluation method, where (A) shows the detection signals from three sensors, (B) is a histogram of signal strength in the movable direction (around the Y-axis), (C) is a histogram of signal strength in the non-movable direction (around the Z-axis), and (D) is a histogram of signal strength in the non-movable direction (around the X-axis). [Figure 7] 10 is a flowchart illustrating the procedure of feature amount calculation processing I. [Figure 8] This figure explains the second evaluation method. (A) is a diagram showing the relationship between the knee and the sensor unit. (B) is a diagram showing each detection signal (acceleration signal) from multiple subjects. (C) is a diagram showing the steps for creating a dataset (Z-scoring, variance-covariance matrix (X) values). (D) is the step for calculating a coordinate transformation matrix from accumulated data. (E) is the step for projecting the detection signal of a new evaluation subject onto the principal component analysis space. (F) is an example of the principal component space obtained in the data learning process. (G) is a diagram showing the state of the test process for two new evaluation subjects, with the results projected into the inside and outside of the ellipse. [Figure 9] (A) corresponds to FIG. 8(F), and (B) corresponds to FIG. 8(G). [Figure 10] 10 is a flowchart illustrating the procedure of feature amount calculation processing II. [Figure 11] FIG. 10 is a diagram showing evaluation thresholds (2SD, 3SD) set in the principal component space (PC1, PC2). DETAILED DESCRIPTION OF THE INVENTION

[0014] FIG. 1 is a block diagram showing one embodiment of a joint evaluation device according to the present invention, and FIG. 2 is a diagram showing the correspondence between joint axes in multiple directions of a joint and the detection axes of an inertial sensor unit. In FIG. 1, a joint evaluation device 10 includes a control unit 20 and an inertial sensor unit 30. The inertial sensor unit 30 detects the movement of each joint axis of a joint to be evaluated, for example, a knee joint, and is attached near the joint to be evaluated. The control unit 20 is typically composed of a computer (processor) and acquires detection signals from the inertial sensor unit 30 and executes a predetermined joint evaluation process, as described below.

[0015] The inertial sensor unit 30 here includes a first sensor 31 to a third sensor 33, a fourth sensor 34, a communication unit 35, and an alarm unit 36 ​​that is provided as needed. The communication unit 35 exchanges signals with the communication unit 24 on the control unit 20 side via wired or wireless communication. The alarm unit 36 ​​has, for example, an emission unit that emits a beep sound, and is controlled to emit a beep when, for example, the evaluation result in the control unit 20 is poor.

[0016] The inertial sensor unit 30 is a miniaturized wearable sensor, and is, for example, a disk-shaped member as shown in Figure 2, and is fixed to the knee joint with a fastener. The fastener may be a string, a band, a hook-and-loop fastener, or an adhesive.

[0017] The posture of attachment to the knee joint is shown in Figure 2. In Figure 2, the knee joint connects the femur 1 and tibia (lower leg) 2 in the vertical direction (w direction). The patella 3 is located in front of the knee joint, and the front-to-back direction of the knee joint is the u direction, and the left-to-right direction is the v direction. Here, the u, v, and w directions are perpendicular to each other.

[0018] Within the knee joint, the lower leg 2 has a range of motion (flexion / extension) around the V axis relative to the femur 1, but the range of motion in other directions and around axes (internal rotation / external rotation, inversion / eversion) is limited (non-range of motion).

[0019] Assuming that the X, Y, and Z axes are perpendicular to each other, the first sensor 31 built into the inertial sensor unit 30 detects angular velocity around the Z axis, and the second sensor 32 detects angular velocity around the X axis. The third sensor 33 detects acceleration in the Y axis direction, and the fourth sensor 34 detects acceleration in the Z axis direction. The inertial sensor unit 30 is preferably attached to the tibial tuberosity, which is the anterior upper portion of the tibia 2, in order to accurately detect the movement of the tibia 2 relative to the femur 1. Furthermore, the inertial sensor unit 30 is attached to the knee joint in an orientation such that the X axis is parallel to the u axis, the Y axis is parallel to the v axis, and the Z axis is parallel to the w axis. This allows anatomical meaning to be assigned to the detection data from the first to third sensors 31 to 33.

[0020] The inertial sensor unit 30 may be a general-purpose type in which all four sensors, the first sensor 31 to the fourth sensor 34, are integrally built in, or, as will be described later, may be a dedicated type in which only the sensors required in the first and second embodiments are installed.

[0021] Returning to FIG. 1, the control unit 20 is connected to a storage unit 201, a display unit 202, and an operation unit 203. The storage unit 201 has a memory area for storing control programs and various data required for processing, and a work area for temporarily storing the operation of acquiring detected data, data processing, and data in the middle of processing. The display unit 202 displays confirmation of operation details and evaluation results. The operation unit 203 is used to input various instructions for processing, and may be a touch panel made of transparent pressure-sensitive elements laminated on the surface of the display unit 202.

[0022] The control unit 20 executes a control program to function as a data acquisition unit 21, a feature calculation unit 22, an evaluation unit 23, and a communication unit 24.

[0023] The data acquisition unit 21 samples the detection signals from the first sensor 31 to the fourth sensor 34 at a predetermined period and acquires them as waveform signals for a predetermined time. As shown in FIG. 3, the data acquisition unit 21 starts detecting signals when the subject Hu drops from a platform St of a predetermined height, e.g., 20 cm, onto the floor FL and lands on the leg Le on the side being evaluated. The landing timing of the drop jump is determined from the change in acceleration detected by the fourth sensor 34. That is, when the data acquisition unit 21 detects from the fourth sensor 34 that the acceleration in the Z-axis direction exceeds a predetermined threshold, e.g., 7 G (see FIG. 3(A)), it determines that the subject Hu has landed and starts acquiring detection signals. FIG. 3(B) shows the posture of the subject Hu immediately thereafter, and movement of the subject's knee joint is detected from the time of landing. The subject is expected to maintain the posture at the time of landing for several seconds, e.g., about 5 seconds, after landing, and data from that predetermined time is used for evaluation.

[0024] The feature amount calculation unit 22 calculates feature amounts from the detection signals from the first sensor 31 to the third sensor 33. There are a first embodiment and a second embodiment as methods for calculating feature amounts, and the data acquisition process and feature amount calculation process differ accordingly. Each embodiment will be described later with reference to the drawings.

[0025] The joint evaluation can be performed automatically or by using the feature calculated by the feature calculation unit 22 for manual evaluation. The evaluation unit 23 is provided as needed, and displays the feature calculated by the feature calculation unit 22 in correspondence with a threshold for evaluating the quality of the joint movement (manual evaluation), or notifies the user on the screen whether the feature calculated by the feature calculation unit 22 is within or outside the threshold, or outputs an instruction to cause the notification unit 36 ​​to notify the user.

[0026] Next, the first embodiment will be described with reference to Fig. 4(A) and Figs. 5 to 7. The control unit 20A includes a data acquisition unit 21A, a feature calculation unit 22A, an evaluation unit 23A, a communication unit 24, and a display processing unit 25A. The first embodiment uses detection signals from a first sensor 31 and a second sensor 32. That is, the data acquisition unit 21A acquires detection signals about joint axes (w-axis (Z-axis) and u-axis (X-axis)) with limited ranges of motion among the joint axes.

[0027] The feature calculation unit 22A includes a histogram creation unit 221A. The display processing unit 25A displays the creation results on the display unit 202. The histogram creation unit 221A takes in detection signals from the first sensor 31 and the second sensor 32, which are sampled at a predetermined cycle (e.g., 200 Hz), for a predetermined time (e.g., 100 samples: 0.5 seconds). The histogram creation unit 221A extracts each detection sampling data from the first sensor 31 and the second sensor 32 for each predetermined intensity range to create a histogram.

[0028] Figure 5(A) shows a mixture of detection signals from the first sensor 31 and the second sensor 32, with time on the horizontal axis and intensity (here, angular velocity (degrees / second)) on the vertical axis. As can be seen in Figure 5(A), the output of the second sensor 32 shows a low level of vibration throughout (the dark part on the low level side), while the output of the first sensor 31 is high immediately after landing (causing large vibrations) and gradually decreases over time (the light part on the high level side). The histograms in Figures 5(C) and (D) illustrate this state. The histograms show angular velocity (degrees / second) on the horizontal axis and the number of occurrences on the vertical axis.

[0029] From the data of a large number of subjects, it was found that the distribution around the joint axis with a limited range of motion (w-axis (Z-axis)) was close to a normal distribution with a small variance, and that the distribution around the joint axis with a limited range of motion (u-axis (X-axis)) was close to a normal distribution with a small level and variance.

[0030] The risk areas shown in Figures 5(C) and (D) are those above a threshold of 420 degrees / second around the joint axis (w-axis (Z-axis)) and above a threshold of 150 degrees / second around the joint axis (u-axis (X-axis)). The threshold determines whether the quality of the joint movement is good or bad, and is evaluated based on the number of parts exceeding the threshold or the level of exceedance. The subject in Figure 5 has a relatively high proportion of high angular velocity components exceeding the threshold in Figure 5(C), which predicts a high risk of injury or disability, and the quality of the knee joint movement is evaluated as poor. On the other hand, the subject in Figure 6 has a relatively low proportion of high angular velocity components exceeding the threshold, and almost no high angular velocity components are observed, as shown in Figure 6(C), so the quality of the knee joint movement is evaluated as good or fair. In the time period immediately after landing, the frequency of high-level waveforms from the first sensor 31 is low, and there is significant inter-individual variability, which can be used as a feature.

[0031] For reference, Figures 5(B) and 6(B) show histograms based on the detection results from a sensor (not shown) that is integrally provided in the inertial sensor unit 30 and detects the angular velocity around the joint axis (v-axis (Y-axis)), i.e., around the joint axis having a movable axis. By presenting these as needed, it becomes possible to evaluate the speed, smoothness, and quality of reproducibility of the movement of the joint axis having a range of motion.

[0032] 5(C) and (D) on the display unit 202 by the display processing unit 25A, the subject's histogram and risk area are displayed side by side, allowing the quality of joint movement to be easily confirmed from the displayed content. Furthermore, the evaluation unit 23A can quickly perform quality evaluation based on, for example, how much of the histogram exceeds a threshold. Furthermore, by immediately feeding back the evaluation result to the notification unit 36 ​​via the communication units 24 and 35, the subject can also know the evaluation result in approximately real time.

[0033] 7 is a flowchart illustrating the procedure of feature calculation process I. First, the joint evaluation device 10 is started, the inertial sensor unit 30 is put into an operating state, and time signals detected by the first, second, and fourth sensors 31, 32, and 34 are transmitted to the control unit 20.

[0034] In this state, the data acquisition unit 21 determines whether the acceleration detected by the fourth sensor 34 is 7G or more (step S1), and if it does not reach 7G, it returns, and if it does reach 7G, it determines that the subject has landed.

[0035] Next, waveform signals detected by the first sensor 31 and the second sensor 32 at a predetermined sampling period for 0.5 seconds from the time of landing are acquired by the data acquisition unit 21 as new data (step S3) and stored in the storage unit 201. After the 0.5-second signals are acquired, the histogram creation unit 221A creates a histogram for each level from the acquired waveform data (step S5). Next, risk area information, which is accumulated data, is read from the storage unit 201 and associated with the created histogram, and displayed on the display unit 202, for example, as shown in FIGS. 5(C) and 5(D) (step S7). Next, the quality of the created histogram, i.e., the quality of the joint movement, is evaluated based on the degree of high levels and the ratio of high-level areas, etc., from the risk areas and the created histogram (step S9). The evaluation process further includes, for example, immediately transmitting the evaluation results to the inertial sensor unit 30 and notifying the notification unit 36.

[0036] Next, a second embodiment will be described with reference to Fig. 4(B) and Figs. 8 to 10. The control unit 20B includes a data acquisition unit 21B, a feature calculation unit 22B, an evaluation unit 23B, a communication unit 24, and a display processing unit 25B. It also includes a storage unit 2011 that stores a principal component load matrix U. The second embodiment uses an acceleration signal detected by a third sensor 33. That is, the data acquisition unit 21B acquires an acceleration signal in the direction of the joint axis (v-axis (Y-axis)) whose range of motion is limited among the joint axes.

[0037] The feature calculation unit 22B includes a principal component analysis unit 221B. The display processing unit 25B displays the analysis results on the display unit 202. The principal component analysis unit 221B captures the detection signal from the third sensor 33, sampled at a predetermined frequency (e.g., 200 Hz), as a waveform signal for a predetermined period (e.g., 20 samples: 0.1 seconds). The principal component analysis unit 221B applies a previously obtained principal component load matrix U to the captured waveform signal to convert it into feature quantities in the principal component space, as described below. The display processing unit 25B also plots the feature quantities of the current subject on an accumulated data distribution diagram (see FIG. 9), making it easy to recognize the detailed state of the quality of the joint movement. The evaluation unit 23B also sets a threshold value in the principal component space to determine whether the quality of the joint movement is good or bad, and the display processing unit 25B also displays a figure indicating the threshold value. The threshold value is displayed as a 95% confidence ellipse, which is a range that includes 95% of multiple subjects.

[0038] Next, the procedures for creating and applying the principal component load matrix U will be explained using Figure 8. After detecting drop landing, the acceleration of the tibial tuberosity of the landing leg was measured when the subject maintained a stationary standing position for 5 seconds. Two types of acceleration at that time (a predetermined time after landing, for example, 0.1 seconds) were analyzed using PCA (principal component analysis) as data with high variability: a sudden inward acceleration immediately after landing, and repeated small accelerations and decelerations that occurred in the time period after landing. Furthermore, the individual characteristics of knee joint instability were considered from a plot of these principal components.

[0039] First, waveform data (see FIG. 8(B)) consisting of M samples obtained from subject i out of N subjects (for example, 300 landings) is expressed as in equation (1).

[0040]

number

[0041] The data matrix in which the waveform data for N subjects are arranged in the row direction is expressed as in equation (2) (see FIG. 8(C)).

[0042]

number

[0043] Next, the data matrix X is standardized for each column (time axis direction) to obtain the standardized waveform data matrix shown in equation (3).

[0044]

number

[0045] In addition, bar x j is the mean value of column j, and σ j is the standard deviation of column j. In the principal component analysis, the variance-covariance matrix of the standardized waveform data matrix is ​​subjected to singular value decomposition as shown in equation (4), and the eigenvalues ​​and the corresponding eigenvectors are calculated.

[0046]

number

[0047] The superscript T denotes the transposition of a matrix. The eigenvector matrix consisting of eigenvectors is expressed as in equation (5).

[0048]

number

[0049] Furthermore, a principal component loading matrix U is obtained, which serves to linearly transform the vector X, which is the standardized waveform data matrix, into a representation Z in the feature space, as shown in equation (6) (see FIG. 8(D)).

[0050]

number

[0051] In the biplot diagram of the principal component scores for N subjects, consisting of the first principal component (PC1) and the second principal component (PC2), the data distribution centered on the origin is obtained (Fig. 8(F), Fig. 9(A)). Note that the inside of the ellipses in these figures indicates the 95% confidence ellipse.

[0052] Here, the closer the subject's data is to the origin, the more characteristic the lateral knee movement is that everyone exhibits. On the other hand, the further away from the origin the pattern of lateral knee movement (acceleration on the Y-axis) deviates from normal, and data outside the 95% confidence ellipse is defined here as movement that increases knee strain.

[0053] Next, the linear projection of new waveform data onto the feature space and risk detection (test process) after the principal component loading matrix U has been acquired will be described with reference to Figures 8(E) to 8(G) and 9. The extent to which the newly acquired waveform data from the third sensor 33 of the subject deviates from the origin in the feature space is confirmed using equation (7).

[0054]

number

[0055] The mean value and standard deviation of the learned waveform data matrix X are used to standardize it using equation (8).

[0056]

number

[0057] Then, using the principal component loading matrix U obtained (accumulated) in the learning process, a linear mapping to the feature space is performed using equation (9).

[0058]

number

[0059] When the first and second principal components of the newly obtained feature vector Z are plotted on the biplot of FIG. 9(A), a diagram like that of FIG. 9(B) is obtained. In FIG. 9, the horizontal axis is PC1 and the vertical axis is PC2. PC1 is the first principal component, with a contribution rate of, for example, 72.7%, and is characterized by the appearance of a sudden inward peak value immediately after landing. PC2 is the second principal component, with a contribution rate of, for example, 24.3%, and is characterized by the appearance of a slow inward peak value and acceleration / deceleration. The number of principal components is not limited to two, and may be three or more.

[0060] In Figure 9(B), data for two new subjects has been added. New data 1 is outside the 95% confidence ellipse of the distribution of the original data (outlier region). For this reason, it exhibits knee movement that deviates from the knee pattern common to many subjects, exhibiting left-right movement that increases knee strain, and is therefore subject to an alert by the evaluation unit 23. On the other hand, new data 2 is present within the 95% confidence ellipse of the original data, and is therefore determined by the evaluation unit 23 to be knee movement in the left-right direction within a normal range, and is not subject to an alert.

[0061] 10 is a flowchart illustrating the procedure of feature calculation process II. Step S11 is the same as step S1, so its description will be omitted. Next, if the acceleration signal detected by the fourth sensor 34 reaches 7G, ​​it is determined that the subject has landed.

[0062] Next, waveform signals detected by the third sensor 33 at a predetermined sampling period for 0.1 seconds from the time of landing are acquired as new data by the data acquisition unit 21 (step S13) and stored in the storage unit 201. Then, once the waveform signals for 0.1 seconds have been acquired, the acquired waveform signals are multiplied by the principal component load matrix U, which is accumulated data read from the principal component load matrix U storage unit 2011, to calculate feature quantities projected onto the principal component space (step S15).

[0063] Next, 95% confidence ellipse information is read from the storage unit 201 (or storage unit 2011) and displayed on the display unit 202 in association with the calculated feature amount (step S17). Next, it is determined whether the position of the calculated feature point is inside or outside the 95% confidence ellipse, which is a threshold value, that is, the quality of the joint movement is evaluated (step S19). The evaluation process may further include, for example, immediately transmitting the evaluation result to the inertial sensor unit 30 and notifying the notification unit 36.

[0064] The threshold region displayed in the principal component space may be an ellipse or a rectangle set for each principal component as shown in Fig. 11. In the figure, the threshold is displayed as a 95% confidence frame, which is a range that includes 95% of multiple subjects, set at 2SD, and an outer confidence frame that includes 99% is set at 3SD.

[0065] Furthermore, although the present embodiment assumes landing on one foot (using an impact load), evaluation may also be performed using both feet, for example, by knee bending and extension using the load of a squat exercise. Furthermore, a jump landing may be included instead of a drop landing. Even in such cases, the principal component load matrix U corresponding to the load generation mechanism can be obtained from the detection information of the movement of a joint with a limited range of motion, making it possible to evaluate the quality of joint movement with that load generation mechanism. Furthermore, the fourth sensor 34 may be configured to be located not only within the inertial sensor unit 30, but also, for example, on the floor surface FL side.

[0066] In addition to the knee, it is also possible to evaluate the movement of bones in joint axes of elbows, wrists, shoulders, and ankles, where the range of motion of the joint movement is limited, using waveform signals.

[0067] Although the sensors 31 and 32 are angular velocity sensors and the sensors 33 and 34 are acceleration sensors, the present invention is not limited to this and any type of sensor may be used. The mode using angular velocity has the advantage that it is less susceptible to the influence of gravity, and therefore, improved accuracy can be expected.

[0068] The present invention can be applied to various aspects in addition to predicting the risk of developing joint disorders, etc. For example, it can be used to develop athletes by teaching coaches and trainers how to use joints. It can also be used to evaluate movements and monitor risks in rehabilitation using exercise equipment. Furthermore, it can be used as a medical diagnostic expert system for quantifying joint function.

[0069] As described above, a joint evaluation device according to the present invention preferably comprises an inertial sensor unit that is attached near a joint connecting bones on both sides with its joint axis and detection axis parallel, and that detects, as a waveform signal, the movement of the bones of a joint axis that has a limited range of joint movement among the joint axes; a load detection unit that detects a load applied to the joint; data acquisition means that, when the generation of the load is detected, acquires the waveform signal detected by the inertial sensor unit in the time direction and the intensity direction; and feature calculation means that analyzes the waveform signal and calculates feature values ​​for evaluating the movement quality of the joint.

[0070] Furthermore, the joint evaluation method according to the present invention preferably uses an inertial sensor unit that is attached near a joint connecting bones on both sides with its joint axis and detection axis parallel, and that detects, as a waveform signal, the movement of the bones of a joint axis that has a limited range of motion of joint movement, and a load detection unit that detects the load applied to the joint, wherein, when a data acquisition means detects the occurrence of the load, the waveform signal detected by the inertial sensor unit is acquired in the time direction and the intensity direction, and a feature calculation means analyzes the waveform signal and calculates a feature for evaluating the movement quality of the joint.

[0071] Furthermore, the program of the present invention preferably uses an inertial sensor unit that is attached near a joint connecting bones on both sides with the joint axis and detection axis parallel to each other and detects, as a waveform signal, the movement of the bones of a joint axis that has a limited range of motion of the joint movement, and a load detection unit that detects the load applied to the joint, and causes a computer to function as data acquisition means that, when the generation of the load is detected, acquires the waveform signal detected by the inertial sensor unit in the time direction and in the intensity direction, and feature calculation means that analyzes the waveform signal and calculates feature values ​​for evaluating the movement quality of the joint.

[0072] According to these inventions, a wearable inertial sensor detects the bone movement of a joint axis with a limited range of motion as a waveform signal based on anatomical evidence. When a data acquisition means detects the occurrence of a load, for example, when landing on the floor during a one-leg drop, it acquires the waveform signal detected by the inertial sensor in the time and intensity directions. The feature calculation means analyzes the acquired waveform signal to calculate a feature for evaluating the quality of joint movement. In this way, by using a wearable inertial sensor to detect bone movement along a joint axis with a limited range of motion and bone rotation around a joint axis with a limited range of motion and obtaining these feature values, it becomes possible to easily and accurately evaluate and predict damage or injury to such joints based on the correlation between postural fluctuations, such as left-right or forward-backward deviations, twisting, imbalance, and other instability during exercise, such as a drop landing, and the risk of injury or injury. The inertial sensor is not limited to detecting the movement of all or a plurality of joint axes; it may also detect a signal from, for example, a single joint axis of interest. Furthermore, the joint axis to be detected may be set appropriately depending on the joint part to be inspected and the way in which a load is applied.

[0073] Preferably, the inertial sensor unit is integrally provided with the load detection unit, the load detection unit being an inertial sensor having a detection axis parallel to the joint direction of the bones on both sides of the joint and detecting the time point at which the load is generated, and the data acquisition means starts acquiring the waveform signal from the time point at which the load is generated. With this configuration, it is possible to detect the load generation signal on the inertial sensor unit side.

[0074] Furthermore, it is preferable that the inertial sensor unit is a first and a second sensor that detect angular velocities around two axes that are mutually orthogonal to the movable joint axis having the range of motion of the joint motion, and the feature calculation means is a histogram creation means that creates a histogram of signal intensity from the waveform signal as the feature. With this configuration, the movement status of the joint can be detected while minimizing the influence of gravity as much as possible. Furthermore, by using the histogram for each detection level for evaluation, the behavior of the joint with a limited range of motion when a load is applied can be obtained in a form suitable for judgment.

[0075] The inertial sensor unit is preferably two angular velocity sensors with detection axes around the vertical axis of the lower leg and around the sagittal axis of the lower leg. With this configuration, highly reliable behavior information with anatomical meaning can be obtained.

[0076] Furthermore, it is preferable that the inertial sensor unit is a third sensor that detects acceleration with a detection axis parallel to the movable joint axis having the range of motion of the joint movement, and the feature calculation means is principal component analysis means that calculates the feature by subjecting the waveform signal to principal component transformation. With this configuration, evaluation can be easily performed by detecting movement in a direction parallel to the movable joint axis, i.e., a direction in which the range of motion is limited, and further subjecting the detection signal to principal component transformation for the required number of types.

[0077] Preferably, the principal component analysis means analyzes the data into first and second principal components, the first principal component indicating a rapid early swing toward the inner thigh in the time period immediately after the load is applied, and the second principal component indicating the presence or absence of a peak value in the time period after the load is applied. With this configuration, a principal component space can be created that focuses on information that varies greatly between individuals, and discrimination accuracy can be maintained.

[0078] Furthermore, the present invention is preferably a joint evaluation method characterized in that the joint is a knee joint, and the load is a reaction force from the landing surface that the lower leg receives when jumping from a predetermined height. This simplifies the examination work. [Explanation of symbols]

[0079] 10 Joint evaluation device 20, 20A, 20B control section 21, 21A, 21B Data acquisition section 22, 22A, 22B Feature calculation unit (feature calculation means) 221A histogram creation unit (feature calculation means) 221B Principal component analysis unit (feature calculation means) 23, 23A, 23B Evaluation section 30 Inertial sensor unit 31 First sensor 32 Second sensor 33 Third Sensor 34 4th sensor (load detection section)

Claims

1. an inertial sensor unit that is attached near the joint connecting the bones on both sides with its detection axis parallel to the joint axis of the joint, and that detects, as a waveform signal, the movement of the bone of the joint axis having a limited range of motion; a load detection unit that detects a load applied to the joint; a data acquisition means for acquiring a waveform signal detected by the inertial sensor unit in a time direction and an intensity direction when the occurrence of the load is detected; a feature value calculation means for analyzing the waveform signal and calculating a feature value for evaluating the movement quality of the joint; the inertial sensor unit is a sensor that has a detection axis parallel to a movable joint axis having a range of motion of the joint movement and detects acceleration as the waveform signal, the feature calculation means is a principal component analysis means for calculating the feature by subjecting the waveform signal to principal component transformation; The principal component analysis means analyzes the joint into first and second principal components, the first principal component indicating a rapid early swing toward the inner thigh in the time period immediately after the load is applied, and the second principal component indicating the presence or absence of a peak value in the time period after the load is applied.

2. the inertial sensor unit is integrally provided with the load detection unit, the load detection unit is an inertial sensor having a detection axis parallel to a connection direction of bones on both sides of the joint, and detecting a time point when the load is generated; The joint evaluation device according to claim 1 , wherein the data acquisition means starts acquiring the waveform signal from the time when the load is generated.

3. An inertial sensor unit is mounted near a joint connecting bones on both sides with its joint axis and detection axis parallel, and detects the movement of the bones of a joint axis having a limited range of motion of the joint movement as a waveform signal, the inertial sensor unit having a detection axis parallel to a movable joint axis having a limited range of motion of the joint movement and detecting acceleration as the waveform signal, and a load detection unit is used to detect the load applied to the joint, When the data acquisition means of the computer detects the occurrence of the load, it acquires the waveform signal detected by the inertial sensor unit in the time direction and the intensity direction; a feature calculation means of the computer analyzing the waveform signal to calculate a feature for evaluating the movement quality of the joint; and the feature calculation means of the computer performs principal component analysis to convert the waveform signal into principal components and calculate the feature; The principal component analysis is performed into first and second principal components, the first principal component indicating a rapid early swing toward the inner thigh in the time period immediately after the load is applied, and the second principal component indicating the presence or absence of a peak value in the time period after the load is applied.

4. The joint is a knee joint, 4. The joint evaluation method according to claim 3, wherein the load is a reaction force from the landing surface that is applied to the lower leg when jumping down from a predetermined height.

5. An inertial sensor unit is mounted near a joint connecting bones on both sides with its joint axis and detection axis parallel, and detects the movement of the bones of a joint axis having a limited range of motion of the joint movement as a waveform signal, the inertial sensor unit having a detection axis parallel to the movable joint axis having the range of motion of the joint movement and detecting acceleration as the waveform signal, and a load detection unit is used to detect the load applied to the joint, a data acquisition means for acquiring a waveform signal detected by the inertial sensor unit in a time direction and an intensity direction when the occurrence of the load is detected; causing a computer to function as feature amount calculation means that analyzes the waveform signal and calculates feature amounts for evaluating the movement quality of the joint; the feature calculation means performs principal component analysis to convert the waveform signal into principal components and calculate the feature; The principal component analysis is performed into first and second principal components, the first principal component indicating a rapid early swing toward the inner thigh in the time period immediately after the load is applied, and the second principal component indicating the presence or absence of a peak value in the time period after the load is applied.

Images