Patent Application
20140343460
The various embodiments relate to methods for quantifying biomechanical data, and more specifically to biomechanical data related to a mobile gait force associated with a lower limb prosthesis.
American military healthcare system has improved treatment of combat casualties—increase from 76% to 87% survival since Vietnam. As a result of these improvements in survivability, there has been an influx of soldiers dealing with lifelong major injuries attributed to roadside bombs and Improvised Explosive Devises (IEDs). For example, such injuries result in the amputation (transtibial or transfemoral) of a lower limb for a large number of soldiers. In addition to such military amputees, there are also about 100,000 lower limb amputations occurring per year in the general population. As a result, many of these patients are typically fitted with lower limb prostheses. In the case of military personnel, these individuals are typically otherwise healthy and the lower limb prosthesis allows them to engage in an active lifestyle.
However, lower limb prosthetic patients are generally at risk for injury and chronic disease in their unaffected leg, including their healthy joints. A culprit for this increased for injury and chronic disease in unaffected legs and healthy joints is the asymmetric gait generally resulting from the use of some prosthetic lower limb devices. An asymmetric gait can cause increased joint loading and higher energy consumption. Such increased energy consumption can limit the amputee's participation in various activities and work functions. More importantly, if injuries and chronic diseases develop in the unaffected leg or healthy joints, this may permanently limit the amputee's ability to function. Table 1 summarizes peak ground reaction force between intact and prosthetic legs in transtibial amputees at various speeds:
(From L. Nolan et al., “Adjustments in gait symmetry with walking speed in transfemoral and trans-tibial amputees.” Gait and Posture 17 (2003) 142-151.) For at least these reasons, a need exists to limit asymmetric gait by better fitting of prosthetic devices and training or monitoring of patients in the use of said devices.
Typically, fitting is performed by a trained prosthetist or physical therapist during one or more office visits. Prosthetists have training and experience that allows them to adjust the prosthetic based on feedback from the patient and their observation of the patient's gait. This type of traditional fitting is typically an iterative process requiring multiple sessions. During a session, the prosthetist makes observations to determine the required adjustments. First, the patient walks for an interval of time while the prosthetist observes. The prosthetist then makes an adjustment, and the process is repeated until the prosthetist and the patient are satisfied with the operation of the prosthesis. In most prosthetic devices, the adjustments can be made to a stump/device interface, any 6-degrees of freedom (DOF) joints or adapters, a prosthetic foot, or any combination thereof.
In general, the data collected is typically limited to the prosthetist's observations, which are, of course subjective and variable from one prosthetist to another. This subjectivity can lead to inconsistent results across the plurality of patients and also to counterproductive results if one patient visits a plurality of prosthetists.
In some cases, it is possible to obtain quantifiable biomechanical data for a patient's gait in a so-called “gait laboratory.” Unfortunately, the process of obtaining biomechanical data in a gait laboratory is expensive and time-consuming. Accordingly, such methods are not generally available to most prosthetists and therapists. Additionally, a problem exists in translating any quantifiable biomechanical data collected in a gait laboratory into clinically meaningful data which can be used by a prosthetist or a therapist to make actual adjustments to prostheses.
Therefore, a need exists for a mobile gait analysis system which can accurately determine biomechanical data, including joint angles and ground reaction forces, in the lower limb amputee patient and also return clinically meaningful results to a prosthetist. Satisfying this need, would allow for better fitting prosthetic devices, which would in turn limit problems associated with and caused by an asymmetric gait.
Various system embodiments measure the full lower limb kinetics and kinematics of subjects outside of a motion capture gait laboratory. The system can determine the torques and forces at any of the joints in the body or in any of the limb segments. The system can be used for rehabilitation of patients experiencing injury to the lower extremity or amputee subjects. The system can also be used in the analysis and improvement of amputee prosthetics or joint replacement components. This system can also be used for the optimization of biomechanics in an athletic environment. Various embodiments can be employed a variety of commercial applications, including but not limited to rehabilitation, prosthesis evaluation and design, athletic performance, video game development and/or control.
Various embodiments satisfy the aforementioned need for a mobile gait analysis system which can accurately determine biomechanical data, including joint angles and ground reaction forces, in the lower limb amputee patient and also return clinically meaningful results to a prosthetist. Various embodiments, therefore, allow for better fitting prosthetic devices, and limit problems associated with and caused by an asymmetric gait.
In one embodiment, a method of performing gait analysis of a subject is provided. The method includes obtaining a plurality of measurement sets for a subject, each of the plurality of measurement sets including inertial measurements obtained from a sensor device associated with a different one of a plurality of segments of the subject. The segments can include a trunk or torso of the subject as well as limb segments. The method also includes calculating a sensor orientation for the sensor device associated with each of the plurality of segments based at least on a portion of a corresponding one the plurality of measurement sets and computing a segment orientation for each of the plurality of segments based on a data fusion process applied to each of the plurality of segments. The data fusion process includes combining at least a one of the plurality measurement sets and the corresponding sensor orientation to estimate the segment orientation. The method also includes determining joint angles based on the estimate of the segment orientation for each of the plurality of segments. Optionally, ground reaction forces can be obtained based force measurements.
In a second embodiment of the invention, there is provided a system for performing gait analysis. The system includes a processor and a communications interface configured for receiving a plurality of measurement sets, each of the plurality of measurement sets including inertial measurements obtained from a sensor device associated with different one of a plurality of segments of a subject. The system also includes a computer-readable medium having stored thereon a plurality of instructions for causing the processor to perform steps. The steps include calculating a sensor orientation for the sensor device associated with each of the plurality of segments based at least on a portion of a corresponding one the plurality of measurement sets and computing an estimate of a segment orientation for each of the plurality of segments based on a data fusion process applied to each of the plurality of segments, the data fusion process including combining at least a one of the plurality measurement sets and the corresponding sensor orientation to estimate the corresponding segment orientation. The steps also include determining joint angles based on the estimate of the segment orientation for each of the plurality of segments. Optionally, the steps can include computing ground reaction forces can be obtained based force measurements.
In a third embodiment of the invention, there is provided a sensor for analyzing gait and ground reaction forces. The sensor includes a forefoot portion removably attachable to a sole of a subject's forefoot and including at least one force sensor. The sensor also includes a heel portion removably attachable to a sole of the subject's heel and including at least one force sensor. The sensor further includes a processing unit communicatively coupled to the forefoot portion and the heel portion and configured for transmitting sensor signals from the forefoot portion and the heel portion to a remote computing device. In the sensor, at least one of the forefoot portion, the heel portion, and the processing unit includes at least one inertial measurement sensor. The sensor can be utilized with the systems and methods described herein. This embodiment also incorporates a method for calculating the biomechanical forces or moments at any location (prosthetic limb, tibial, or femoral locations for example) or any joint (prosthetic socket, knee, ankle, hip for example) in the biomechanic system for ambulation.
These and other features, aspects, and advantages of the present invention will become better understood with reference to the following description and appended claims, and accompanying drawings where:
Some of the figures illustrate diagrams of the functional blocks of various embodiments. The functional blocks are not necessarily indicative of the division between hardware circuitry. Thus, for example, one or more of the functional blocks (e.g., processors or memories) may be implemented in a single piece of hardware (e.g., a general purpose signal processor or a block or random access memory, hard disk or the like). Similarly, the programs may be standalone programs, may be incorporated as subroutines in an operating system, may be functions in an installed imagining software package, and the like.
It should be understood that the various embodiments are not limited to the arrangements and instrumentality shown in the drawings.
The present invention may be understood more readily by reference to the following detailed description of preferred embodiments of the invention as well as to the examples included therein. All numeric values are herein assumed to be modified by the term “about,” whether or not explicitly indicated. The term “about” generally refers to a range of numbers that one of skill in the art would consider equivalent to the recited value (i.e., having the same function or result). In many instances, the term “about” may include numbers that are rounded to the nearest significant figure.
As noted above, a need exists for a mobile gait analysis system which can accurately determine biomechanical data, including joint angles and ground reaction forces, in the lower limb amputee patient and also return clinically meaningful results to a prosthetist. In view of the limitations of conventional methods, a novel mobile gait analysis system (MGAS) is provided, which can accurately determine biomechanical data, including joint angles and ground reaction forces, in the lower limb amputee patient. Further, the new MGAS can be used to return clinically meaningful results to the prosthetist, who can then determine how to adjust or modify a prosthetic lower limb device.
In particular, the various embodiments are directed to systems and methods for mobile gait analysis, force balancing, and alignment system to determine limb segment positioning, forces, and moments. The systems and methods are not limited solely to analysis of an affected limb, but to analysis of the intact limb. In the various embodiments, in order to mathematically determine joint moments and forces for purposes of adjusting and fitting a prosthesis, knowledge of several kinematic components is required. A first is knowledge of the ground reaction forces. The second is knowledge of limb orientation. Finally, knowledge of velocity and acceleration for the limb components, both linear and angular, is also required. The mobile gait analysis system of the various embodiments such data to be collected.
Although the various embodiments will be described with respect to human subjects with transtibial prostheses, this is solely for illustrative purposes. Rather, the systems and methods described herein can be utilized with either human or non-human subject having transfemoral or transtibial lower limb prostheses. Further, the systems and methods described herein are also not limited solely to use with prosthetic devices. Rather, the systems and methods described herein can also be used with healthy subjects, subjects with orthotic or other rehabilitative devices and conditions, and robotic devices and systems. The system may also be used to assist subjects with neurological or neurodegenerative conditions such as Parkinson's Disease, Multiple Sclerosis, Cerebral Palsy or other condition. This system may be used to measure or improve performance of non-human subjects such as race horses, dogs, or other animals.
In the various embodiments, the mobile gait analysis system allows quality kinetic and kinematic data to be measured without the infrastructure investment of a camera-based motion capture gait analysis facility. Moreover, as a laboratory setting is not needed, the mobile gait analysis system allows measurement of biomechanics data in any environment. That is, it can be utilized in the actual locations and activities that any subject or clinician desires to record or analyze biomechanics data. This allows rehabilitation or examination of the actual activities of daily living (ADL) subjects expect to regularly encounter. Further, a software system can be provided to assist with data interpretation. This software system will help assess component fit and alignment as well as patient biomechanics. However, analyses can also be performed manually in the various embodiments.
Beyond the known alignment systems based on force plates, this system represents a new approach to prosthesis alignment, fitting, and patient rehabilitation and will allow subjects to experience more natural and efficient function from their prosthetic limbs and reduce secondary disabilities. This approach will enhance the maintenance and performance of long-term prosthesis and socket performance/fit by increasing the ease of measurement of prosthesis performance. Mobile gait analysis also represents an evidence-based approach to prosthesis fitting and will allow wider use of evidence-based rehabilitation techniques. More natural function will allow patients to return to a higher level of activity, and the reduction in secondary disabilities includes chronic lower back pain, hip and knee pain, as well as osteoarthritis of the knee(s), hips or lower back.
Turing first to
As noted above, the amputee 100 is fitted with a prosthesis 104. In the exemplary embodiment of
In some embodiments, the foot 112 can be connected to the remainder of the prosthesis 104 via a passive or powered ankle joint. Further, although the various embodiments will be discussed primarily with respect to below-the-knee amputees, the various embodiments are not limited in this regard and are equally applicable to above-the-knee amputees. Thus, in some embodiments, a passive or powered knee joint can also be included between the socket 108 and the remainder of prosthesis 104. Thus, prosthesis 104 can include a knee joint, an ankle joint, or both, each of which can be powered or passive.
The system 100 can also include a footpad 114 that is secured to the amputee's 102 healthy limb. The footpad 114 can comprise a plurality of force/moment (F/M) sensors. That is, sensors that measure loads and torques in one or more axes. For example, the footpad 114 can have an F/M sensor on a forefoot portion and can have an F/M sensor on a heel portion thereof. The F/M sensors can measure ground reaction forces (GRS) for the healthy limb. The F/M sensors can also determine the orientation of the foot or other portions of the healthy limb. In some embodiments, the footpad can include components similar to IMUs 106 to communicate measurements. In other embodiments, the footpad 114 can operate in conjunction with a foot processor unit 115, where the foot processor unit (FPU) 115 is configured to communicate with the footpad 114 and communicate measurements on behalf with footpad 114.
In some embodiments, the prosthetic foot 112 can be configured to operate in a manner substantially similar to footpad 114. Further, the IMU 106 in prosthetic leg 104 can also provide a FPU for the sensors in the prosthetic foot. However, in other embodiments, the prosthetic foot 112 can also be fitted with a footpad and, optionally, a FPU to collect measurements. The footpad 114 and IMU 106 may be integrated into a prosthetic foot with force and orientation measurement or feedback. The footpad may also be integrated into a shoe that can provide propulsive force and torque quantity and direction information. The data from the IMU is combined with the data from the footpad such that propulsive forces and torques, foot orientation, and propulsive force vector directions, with respect to a selected reference frame, are measured.
System 100 also includes a computing device 116 for communicating with various elements of system 100, such as IMUs 106, footpad 114, FPU 115, or any other components for providing sensor information. The communications links between computing device 116 and the various elements of system 100 can be wireless, wired, or a combination of both.
In operation, data from the IMUs 106 or from the footpad 114 can be transmitted as a signal 118 to computing device 116. The computing device 116 can also send a signal 120 to the IMUs 106 and/or footpad 114 (or FPU 115). In the exemplary configuration of
Referring to now to
Referring now to
To provide IMUs 106 for the torso and limb segments, the IMUs 106 can be configured to allow their attachment to the body of amputee 102. For example, the IMUs 106 can be configured to be strapped to a plurality of limb segments at various points on an amputee's body. Referring to
Referring now to
Referring now to
Together, forefoot portion 500 and heel portion 600 can be utilized to define the footpad 114 for foot 602, as shown in
Now referring to now to
Although a specific arrangement of straps and anchors is illustrated in
As noted above, the footpad 114 can be configured in a manner similar to IMU 106 or can include a FPU 115 that operates with footpad 114 and that can also be removably attached to the foot (healthy or prosthetic) of the subject. Regardless of the configuration, the block diagram of
Referring to
Some elements of each of forefoot portion 500 and heel portion 600 have already been described above. However, to allow operation with FPU 115, additional elements can be provided. For example, as show in
It should be noted that in some embodiments, controllers 922 and 932 are not utilized. In such embodiments, the forefoot portion 500 and heel portion can be controlled via microcontroller 902. Additionally, the arrangement in
In the various embodiments, it can be advantageous to provide inertial measurement sensors at the foot. For example, for purposes of gait analysis, it is not only use useful to obtain the orientation of a foot (healthy or prosthetic), but also necessary, as direction and orientation are relatively significant for any type of gait analysis. Thus, in some embodiments the FPU 115 at the foot can be configured in substantially the same way as an IMU 106. That is, the FPU 115 can include inertial measurement sensors 914. However, in other embodiments, inertial measurement sensors, or any other type of sensor, can be provided within the footpad 114 as well. For example, inertial measurement sensors 929 and 939 can be configured for measuring orientation for at least one of the forefoot portion 500 and the heel portion 600, respectively.
Now referring to
Now turning to
Although exemplary embodiments below will be described with respect to an algorithm based on an extended Kalman filter process, the various embodiments are not limited in this regard. Rather, in the various embodiments, any other type of data fusion techniques or processes can be utilized.
Generally, pitch and roll angles can be calculated by estimating the direction of gravity using the accelerometer signals. Alternatively, the gyroscope signals can be integrated to determine pitch, roll and yaw (heading). However, these calculations are subject to drift and noise which cause increasing error as the signal is integrated over time. Individually, these respective angle calculations are inaccurate. Accordingly, an extended Kalman filter process was developed in the various embodiments to fuse the accelerometer and gyroscope data to estimate segment orientation while accounting for noise and drift.
In one particular embodiment, the filter uses a 14-element state vector (1)
where vint and aint are velocity and acceleration in three axes transformed to an intermediate reference frame, ωbody and bgyr are the gyroscope signals and the gyroscope bias in three axes, and r and p are roll and pitch of the segment. The intermediate reference frame can be initially aligned with a laboratory reference frame but rotates about the gravity vector and is propagated outside of the roll and pitch EKF. The rotations from the laboratory reference frame to the IMU frames can be represented using direction cosine matrices so pitch and roll rotations can be isolated while rotations about the gravity vector are ignored.
The orientation about the gravity vector (internal/external rotation, heading or yaw) can be calculated through a second EKF that assumes the yaw rotation is minimal and about 0 degrees. This is a valid assumption when the relative rotation between segments (joint rotations) is being examined as opposed to the absolute rotation in global space.
The velocity at step k+1, in the intermediate frame are found by numerically integrating aint (2) over timestep Δt.
v
int,k+1
=v
int,k
+a
int,k
Δt. (2)
Accelerations, angular rates and angular biases are modeled by using the value at the previous time step, adding noise, wa, wω, wgyr to acceleration, gyroscopes and bias and, for the acceleration model subtracting a factor multiplied by velocity, yvp, to stabilize the velocity calculation (3)
a
int,k+1
=a
int,k
+w
k
α
−yv
p,k
w
body,k+1=ωbody,k+wkα
b
gyr,k+1
=b
gyr,k
+w
k
gyr (3)
Angles of the segments in the laboratory reference frame can then be calculated by transforming the gyroscope signals to the laboratory reference frame (4)
{dot over (r)}=ω
x+(ωx cos r+ωy sin r)tan p
p=(ωy cos r−ωx sin r)
{dot over (y)}=(ωx cos r+ωy sin r)sec p (4)
representing the time derivative of roll, pitch and yaw, {dot over (r)}, {dot over (p)}, and ÿ, then numerically integrating the angular velocities (5)
r
k+1
=r
k
+{dot over (r)}
k
Δt
p
k+1
=p
k
+{dot over (p)}
k
Δt
y
k+1
=y
k
+{dot over (y)}
k
Δt (5)
Here, y is yaw, which is not included in the pitch and roll EKF state equations and represents the rotation of the intermediate reference frame about the gravity vector.
The measurement vector (6) consists of the three signals from the accelerometer, aimu which include the gravity vector in the IMU frame, three signals from the gyroscope in the IMU frame and any drift associated with the gyroscope, wIMU and bgyr. An estimate of roll and pitch, rest and pest, respectively, is calculated using the direction of gravity from the accelerometer signals, basic trigonometry and sequential rotations. The filtered measurement vector, vk can then be defined by:
where wk is the measurement noise at time k.
The process covariances were calculated using the ideal signals calculated with a kinematic model. Only covariances for aint, ωbody, and bgyr were used. All other covariances were set to zero. The angle calculated using the gravity signal from the accelerometers is only accurate during low linear acceleration (when the IMU is not in motion). Therefore, the measurement covariances were optimized so that the algorithm weights the gyroscope measurements more heavily than the accelerometer and estimated angle measurements. The covariances may also be adjusted in real time based on the IMU data or by using data from the force sensors on the feet. During times of low linear acceleration the covariances can be adjusted so as to weight the accelerometer measurements and the pitch and yaw calculations from gravity more heavily than the gyroscope measurements.
In the various embodiments, the gravity vector can be obtained in a variety of methods. One method is to use accelerometers, such as those in the IMUs. Accelerometers use a “proof mass” which experience forces when the sensor is subjected to acceleration. This can cause a movement of the mass and a strain in the members that support the mass. Typically MEMS devices, like those described above, use capacitive components supported on flexures with integrated capacitive elements to measure the deflection of the proof mass. The changes in capacitance can be measured as a change in acceleration or a feedback voltage can be used to maintain the proof mass at a constant location. The amount of voltage required to hold the proof mass in place is proportional to the acceleration.
If a single axis accelerometer is stationary and oriented such that the measurement axis is perfectly vertical, the mass will deflect due to gravity and the accelerometer will detect an acceleration that equals 9.81 m/s2 or 1 g. The three axis accelerometers used herein consist of three single axis accelerometers placed orthogonal to each other. Therefore, when the sensor is not moving or experiencing no inertial acceleration, the only signal the accelerometer senses is that of gravity deflecting the proof masses. If one of the three axes is perfectly aligned to vertical it will sense 1 g of acceleration and the other axes will read 0 g. If the accelerometer is randomly oriented and no single axis is aligned with gravity, then the proof masses in the other axes will also deflect and the gravity component in each of the axes will be detected. Assuming that the sensor is not moving, the orientation of the sensor can therefore be calculated using the known magnitude of gravity, the gravity component in each axes, and trigonometric functions arcsine and arccosine.
Depending on the amount of motion, the covariance values can be adjusted for the gyroscope signals and the accelerometer signals. That is, the angle values are either calculated primarily based on a direction of gravity calculated by the accelerometers or by using the gyroscope signals. During motion, an algorithm can be weighted to consider the gyroscope signals more and the angle calculated from gravity less, because using the accelerometers is less accurate when an IMU is in motion. When the IMU is not moving, the covariance values can be adjusted to consider the angle calculated from gravity using the accelerometers more, because that calculation should be of higher accuracy. This provides a correcting measurement which mitigates the effect of gyroscope drift over time.
As noted above data fusion methods, such as the EKF methods described above, provide only a portion of an overall algorithm in accordance with the various embodiments. For example, the overall algorithm provides parameters that are input to the data fusion portion and also determines when motion is occurring and when motion has ceased.
In operation, the parameters that are input to the data fusion portion of the overall algorithm can be divided into recursive-type and non-recursive-type parameters. The overall algorithm is configured to provide initial values for both types of parameters. New values for the recursive-type parameters are then calculated by the data fusion portion during a sampling/calculation period and fed back into the data fusion portion as an input for the next sampling/calculation period. The non-recursive-type parameters, such as the measurement covariance values, are also fed into the data fusion portion but remain unchanged by the data fusion portion. However, the overall algorithm may change these non-recursive-type parameter values over time.
The overall algorithm determines when motion starts by using a threshold value for the gyroscope signal of each respective IMU. Also, depending on what portion of the gait activity (stance phase or swing phase) the particular limb is experiencing, the overall algorithm can change the measurement covariance and can determine how the data fusion algorithm for each segment will weigh the various measurement signals. For covariances for the thigh, shank, and trunk segments, only the IMU data is used as a measure to adjust the measurement covariances. For the foot segments, the force measurements are used to determine when the foot is on the ground and in a stance phase. However, these force measurements are only used to adjust the measurement covariances outside of the data fusion segment and are not used in the orientation calculation.
Other parameters that the overall algorithm calculates for input to the data fusion segment include an initial gyroscope bias which is taken from an average of a static data collection from the respective IMUs. The standard deviations of the gyroscope signals from these static trials are also used as the basis for some of the measurement covariance values.
The operation of the overall algorithm is described below in greater detail with respect to
Loop 1308 first begins with a determination that motion has started. In particular, at 1310, a determination is made that motion was started if gyroscope measurements exceed a threshold value for motion start. If motion start is determined at 1310, initial angles for the segments are generated at 1312. In particular, IMU data before motion, specifically acceleration data (Anorm), is used at 1312 to calculate initial angles using yaw=0. The loop 1308 can then utilize EKFs to get data for each of the segments.
In particular, at 1314. the current IMU data, force data, and old EKF data (if available) is received. The IMU data can then be organized by segment at 1314. Further, a determination of whether a current phase is a stance or a swing phase can be made at 1314. Once the IMU data is organized and a current phase (swing or stance) is determined, the EKF process can be performed by an EKF engine 1316.
The EFK engine 1316 receives as inputs current IMU data, old or previous EKF data (if available), and the current phase information (i.e., swing or stance). The old or previous EKF data can include x, the 14 state vector of equation (1), and P, prior variance estimates for the states. Initially, P can be initialized as a square (14×14) matrix of zeroes. In the data fusion portion, the following is then performed at a configuration stage 1318:
Thereafter, the updated EKF parameters from EKFs 1320 and 1322 are passed back to loop 1308 at 1314 and the updated EKF parameters and states of the segments (i.e., the updated segment orientations) are then provided. The updated data can then be stored at 1324. Further, the updated data can be passed back into loop 1308 to provide the old or current EKF data for a next iteration performing 1314 and 1316. The stored data can then be plotted or otherwise presented to the user at 1326. Finally, all EKF data generated by loop can be stored at 1328.
The filtering process described above can be used to carry out the methods of
The shank orientation can be computed in a substantially similar way. In particular, steps 1410, 1412, 1414, and 1416 can be performed for the data from an IMU 106 associated with a shank, for which the knee angle is to be calculated, in substantially a same way as steps 1402, 1404, 1406, and 1408, respectively. Steps 1410-1416 can be performed at a same or different time as steps 1402-1408. However, for purposes of reducing errors, it is preferable to perform these steps concurrently so that the calculated orientation is based on corresponding measurements of the thigh and shank. Once the orientation of each of the thigh and shank is obtained at steps 1406, 1408, 1414, and 1416, the associated knee angle can be calculated at step 1418. The method 1400 can then be repeated to capture development of the knee angle over time. Further, the method 1400 can be performed for both legs, concurrently. Additionally, a similar process can also be performed for the individual segments of the footpad (i.e., the forefoot and heel). Thus, in addition to capturing knee angle, angles for the ankle and angle associated with the foot can also be captured. In such configurations, the ground reaction forces can be considered.
As a result, the kinematics of the legs of the amputee can be captured without the need for a conventional gait laboratory setup. Therefore, together with other information regarding the prosthesis and the amputee and without the need of a gait laboratory, the prosthetist can perform adjustments to improve the gait of the amputee until such kinematics are within acceptable tolerance limits. For example, a method of carrying out such adjustments is presented below with respect to
Once the initial configuration of the prosthesis is completed, the method 1500 can proceed to step 1506 to collect additional patient data using the MGAS systems described herein. For example, the amputee/patient can be outfitted as described above with respect to
The prosthetist, manually or programmatically, can then evaluate the kinematic data. In particular, at step 1510, the kinematic data for the patient can be evaluated to determine whether or not if falls within acceptable limits. For example, to provide a proper gait for the user, it may be preferred that the development of knee angles on both the healthy and affected legs be symmetric. Further, it may also be preferred that the overall stride of the healthy and affected legs also be similar as possible. Thus, a determination is made whether there are any unbalanced or irregular aspects of the patient's gait for the current prosthesis configuration.
If the kinematic data is within acceptable limits at step 1510, the method 1500 can end at step 1512. If the kinematic data is not within acceptable limits at step 1510, the method 1500 can proceed to step 1514. At step 1514, further adjustments can be made to the prosthesis, based on the kinematic data and other data. Such adjustments can be determined manually or automatically based on the kinematic and other data. Thereafter, the method can repeat steps 1506, 1508, 1510, and 1514 until the kinematic data is within acceptable limits.
Although the various embodiments have been described with respect to evaluating gait with respect to the development of the knee angle over time, the various embodiments are not limited in this regard. Substantially similar methods can be utilized for purposes of evaluating the kinematics of any of the other joints of the amputee and/or the prosthesis. For example, similar methods can be applied to adjust the gait of below-the-knee amputees by using measurements of ankle angle to determine how to adjust the prosthesis. In another example, similar methods can be applied to adjust a prosthesis including both knee joints and ankle joints.
Although examples of various aspects of the present invention are discussed in detail below, the various embodiments are not limited in this regard. Rather, these examples are provided solely for illustrating or clarifying the various embodiments of the present invention.
In the following examples, the method of extended Kalman filtering of the various embodiments was evaluated conducting experiments to confirm whether the data collected by the IMUs and the kinematic data obtained using the extending Kalman filtering corresponded to actual kinematic data. In particular, these experiments employed using a Mitsubishi Heavy Industry (Tokyo, Japan) PA-107C robot arm with IMUs and a data processing system, as described above. The difference between applying the EKF algorithm to data from a robot and data from a human subject is adjusting the measurement covariance values, or adjusting how much the EKF weighs one signal over another. The robot was controlled with a personal computer which moved the arm through a repeatable motion sequence while simultaneously recording the encoder data from the robot. A separate personal computer recorded the gyroscope and accelerometer data from the IMUs.
The overall process for the experiment is shown in
In the experiments, an IMU was attached to the “thigh” and “shank” segments of the robot. The goal was to determine the orientation of the segments and the angle between them, or the “knee” angle. These IMUs were attached to the robot “thigh” and “shank” segments using custom holders designed to put the IMUs in the same place for each trial.
A simplified, hardwired data collection system was used for these experiments as telemetry was not required. For this study the IMU data was collected using a MSP430 (Texas Instruments, Inc., Dallas, Tex.) microcontroller and stored on a computer for post processing using MATLAB (The Mathworks, Inc. Natick, Mass.). Three trials were performed for all three activities. The root mean-squared-error (RMSE) between the MGAS orientation angles and the robot orientation angles were calculated.
A mathematical model of the robot was created using the Matlab (The Mathworks, Natick, Mass.) programming environment. The position and orientation of the IMUs relative to the robot segments, the robot joint angle data, and the robot segment lengths were the inputs for the model. The outputs were the position and orientation of the IMUs as well as calculated accelerometer and gyroscope “signals” used as the ground truth when determining the accuracy of the IMU signals. The calculated IMU data was used to synchronize the robot and IMU data, evaluate IMU performance and develop the algorithm used to calculate joint angles from IMU data.
Referring back to
Although specific IMUs were utilized in these experiments, a range of IMU sensors were evaluated using these robotic procedures. The IMU chip used herein, that consists of a three-axis accelerometer and a three-axis gyroscope, was selected based on performance, cost, size, form factor, communication interface and ease of implementation. This chip was incorporated to an expansion board for a commercially available computer-on-module device that uses an ARM reduced instruction set processor that runs the Linux operating system, runs compiled. This configuration is substantially similar to that described above with respect to
Average results of six trials of Walk A were computed. The thigh segment pitch RMSE was 0.2 degrees (standard deviation=0.1 degrees), shank segment pitch RMSE was 0.5 degrees (standard deviation=0.0 degrees), and the knee flexion calculation RMSE was 0.5 degrees (standard deviation=0.1 degrees) with a max error of 1.5 degrees (standard deviation=0.2 degrees) of knee flexion. The RMSE of the out of sagittal plane angles were not calculated but by inspection are within one or two degrees throughout the trials.
The average results of the three trials of Walk B were the thigh segment pitch RMSE was 0.1 degrees (standard deviation=0.0 degrees), the shank segment pitch RMSE was 0.3 degrees (standard deviation=0.1 degrees) and the knee flexion RMSE was 0.8 degrees (standard deviation=0.2 degrees). Similar to the Walk A data, the out of sagittal plane orientation data appeared to be within one or two degrees by visual inspection of the plots.
The average slow Stair Climb activity segment pitch RMSE for the thigh segment was 0.3 degrees (standard deviation=0.1 degrees), for the shank segment was 0.1 degrees (standard deviation=0.1 degrees) and for knee flexion 0.4 degrees (standard deviation=0.0 degrees). Similar to the previous two activities the error of out of sagittal plane motion appeared to be within a few degrees by visual inspection of the data.
The average maximum knee flexion errors per trial were 1.5 degrees (standard deviation=0.2 degrees), 3.1 degrees (standard deviation=0.2 degrees) and 1.2 degrees (standard deviation=0.3 degrees) for Walk A, Walk B and Stair Climb activities, respectively. The flexion error from the different robot motions are shown in
As noted above, the results from these kinematic tests show good agreement and are within the accuracy requirements for gait analysis. The RMSE values for sagittal plane orientations are within 1 degree RMSE, which was the goal set for the kinematic portion of the system. The out of sagittal plane motions are also accurate to within a few degrees, however sagittal plane motions are of secondary importance for gait analysis purposes There is no additional reference for the yaw component of the limb orientation; therefore this calculation is dependent purely on the gyroscope signal and vulnerable to drift errors. This may be a factor in real world testing of the system when soft tissue artifact and inconsistent motion comes into play. However, with additional processing, a stable estimate of yaw orientation, or heading, can be made without the use of additional sensors such as magnetometers. By avoiding using magnetometers, the concern over ferrous magnetic perturbations is avoided and this system will work in any setting.
These trials demonstrate that portable, inexpensive inertial sensors can be used to accurately track complicated repeated biomechanical motion. Incorporating this into the proposed MGAS will result in a tool that will give prosthetists, clinicians and researchers more information to improve the performance of lower leg prosthesis and the overall quality of life of amputee.
For testing in a gait laboratory environment, an expansion board for these IMUs was designed for two applications. The first application is to control and manage data from a portable force/moment (F/M) foot sensor which is strapped to the bottom of the shoe, as discussed above with respect to
The second application of the expansion board is as a limb segment (e.g. thigh, shank) IMU sensor, as discussed above with respect to
To further evaluate the effectiveness of the analysis system of the various embodiments, a gait lab study was conducted in which measurements were recorded from the system in accordance with the various embodiments (labeled “MGAS” in the following figures). Video gait lab data was recorded simultaneously and is presented for comparison purposes (labeled “GL” in the following figures). The measurements were obtained for a single, healthy subject walking at normal speed. Preliminary data from this study is presented in
Video gait data is show for comparison in these plots, however video gait data is an indirect measurement and should not be considered ground truth. The markers used for video gait analysis were not mounted in the same location as the inertial measurement sensors, and some variation in results is to be expected. Both methods will be subject to skin, clothing, and muscle motion artifacts, however these artifacts will be different for the two measurement techniques. Both sets of data are provided to illustrate that the results from the MGAS system are physiologically meaningful.
Thigh and shank angle measurements are compared in
As shown in
Knee angle measurements are compared in
As shown in
Heel and toe orientation measurements of the foot are compared in
The standard gait laboratory measures the orientation of the foot as one body whereas the MGAS system measures the orientation of the heel and forefoot. The data are presented together for illustration only, and comparison should not be made due to these differences in measurement technique. As expected, the gait laboratory foot results are more similar to the MGAS heel measurements as compared to the forefoot.
Various embodiments of the present technology are carried out using one or more computing devices. With reference to
The system bus 3310 may be any of several types of bus structures including a memory bus or memory controller, a peripheral bus, and a local bus using any of a variety of bus architectures. A basic input/output (BIOS) stored in ROM 3340 or the like, may provide the basic routine that helps to transfer information between elements within the computing device 3300, such as during start-up. The computing device 3300 further includes storage devices 3360 such as a solid state hard disk drive, a magnetic disk drive, an optical disk drive, tape drive or the like. The storage device 3360 can include software modules 3362, 3364, 3366 for controlling the processor 3320. Other hardware or software modules are contemplated. The storage device 3360 is connected to the system bus 3310 by a drive interface. The drives and the associated computer readable storage media provide nonvolatile storage of computer readable instructions, data structures, program modules and other data for the computing device 3300. In one aspect, a hardware module that performs a particular function includes the software component stored in a non-transitory computer-readable medium in connection with the necessary hardware components, such as the processor 3320, bus 3310, display 3370, and so forth, to carry out the function. The basic components are known to those of skill in the art and appropriate variations are contemplated depending on the type of device, such as whether the device 3300 is a small, handheld computing device, a desktop computer, or a computer server.
Although the exemplary embodiment described herein employs the hard disk 3360, it should be appreciated by those skilled in the art that other types of computer readable media which can store data that are accessible by a computer, such as magnetic cassettes, solid state memory devices, digital versatile disks, cartridges, random access memories (RAMs) 3350, read only memory (ROM) 3340, a cable or wireless signal containing a bit stream and the like, may also be used in the exemplary operating environment. Non-transitory computer-readable storage media expressly exclude media such as energy, carrier signals, electromagnetic waves, and signals per se.
To enable user interaction with the computing device 3300, an input device 3390 represents any number of input mechanisms, such as a microphone for speech, a touch-sensitive screen for gesture or graphical input, keyboard, mouse, motion input, speech and so forth. An output device 3370 can also be one or more of a number of output mechanisms known to those of skill in the art. In some instances, multimodal systems enable a user to provide multiple types of input to communicate with the computing device 3300. The communications interface 3380 generally governs and manages the user input and system output. There is no restriction on operating on any particular hardware arrangement and therefore the basic features here may easily be substituted for improved hardware or firmware arrangements as they are developed.
For clarity of explanation, the illustrative system embodiment is presented as including individual functional blocks including functional blocks labeled as a “processor” or processor 3320. The functions these blocks represent may be provided through the use of either shared or dedicated hardware, including, but not limited to, hardware capable of executing software and hardware, such as a processor 3320, that is purpose-built to operate as an equivalent to software executing on a general purpose processor. For example, the functions of one or more processors presented in
The logical operations of the various embodiments are implemented as: (1) a sequence of computer implemented steps, operations, or procedures running on a programmable circuit within a general use computer, (2) a sequence of computer implemented steps, operations, or procedures running on a specific-use programmable circuit; and/or (3) interconnected machine modules or program engines within the programmable circuits. The system 3300 shown in
Although the present invention has been described in considerable detail with reference to certain preferred versions thereof, other versions are possible. Therefore, the spirit and scope of the appended claims should not be limited to the description of the preferred versions contained herein.
The reader's attention is directed to all papers and documents which are filed concurrently with this specification and which are open to public inspection with this specification, and the contents of all such papers and documents are incorporated herein by reference.
All the features disclosed in this specification (including any accompanying claims, abstract, and drawings) may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise. Thus, unless expressly stated otherwise, each feature disclosed is one example only of a generic series of equivalent or similar features.
Any element in a claim that does not explicitly state “means for” performing a specified function, or “step for” performing a specific function, is not to be interpreted as a “means” or “step” clause as specified in 35 U.S.C §112, sixth paragraph. In particular, the use of “step of” in the claims herein is not intended to invoke the provisions of 35 U.S.C §112, sixth paragraph.
This invention was made with government support under contract No. 2095-V215-10 awarded by the U.S. Army Medical Material Command and contract No. DE-AC05-00OR22725 awarded by the Department of Energy. The government has certain rights in the invention.
