Gait Analysis

Quantitative analysis of animal motion can provide unique insight into normal and abnormal limb function. This information is useful clinically for the diagnosis of disease and as a measure of outcome following treatment intervention. In addition, gait analysis is commonly used in the development and subsequent use of animal models of disease. Historically, gait analysis has been extremely cumbersome and time-consuming. Routine clinical use was impractical because of the time required for set-up, data collection, and evaluation as well as using bulky and/or awkward equipment. Dramatic computational and equipment improvements over the past 10 years, however, have made these techniques readily available and now allow for routine assessment of gait in a modern-day laboratory or hospital. Perhaps the most attractive draw to gait analysis is that it produces objective, reproducible data of limb function.

Several techniques are available for the analysis of gait. The most common method in veterinary medicine is visual observation of gait. Objective measures of ground reaction forces (GRF) using a force platform, however, is an attractive alternative that is also commonly utilized. Publications of analysis of joint and body movements using two- and three-dimensional stereometric methods and visible markers also exist. These techniques attach markers to the skin on rigid segments of the body structure and track their motion using imaging equipment. Less frequently used methods of analyzing gait in animals include accelerometers, electromyography, electrogoniometry, and roentgenographic and magnetic coupling. Simultaneous use of a combination of evaluation methods should be considered.

Visual Observation of Gait

For most species the entire gait cycle lasts for only one second; thus, a systematic and disciplined approach must be used to clinically evaluate a patient’s gait. The animal should be viewed from the front, side, and back. The observation of gait begins with a general assessment, noting symmetry and smoothness of movements of the various body parts. During stance, the clinician should take note of the position of the patient’s center of mass, as the center of mass will shift toward a normal or less affected side to reduce the force applied to more painful joints or muscles. In addition, it is important to look at the base width. In general, if the forelimbs are apart and the rear limbs near each other, the animal is shifting its weight to the front; the opposite is also true. Finally, as the patient stands, one should observe for excessive internal or external rotation of a limb. In my experience, external rotation of a limb is associated with an abnormality in that limb, and internal rotation of a limb is associated with an abnormality in a different limb. Of course, disorders that have caused a permanent rotation of the skeleton make this point moot. Watching a patient sit and move from a sitting to standing position is also helpful. The "sit-test" is sometimes advocated. When a dog sits, it is considered abnormal for a foot to be pointed outward. In fact, some consider that position pathognomonic for knee pain. Although this concept frequently holds true, many normal dogs also sit with a foot or feet externally rotated. When a dog rises from a sitting position, the focus should be on the shift of weight onto the front limbs. With experience, one will notice excessive shifts.

As the animal walks, attention to the stance phase of gait is helpful. If lameness is present because of pain, the animal will avoid placing weight on the affected limb. When this occurs, the stance phase of gait will be shorter. For example, if a dog is suspected of having a unilateral forelimb lameness, the observer should watch how much time the dog spends with each of the front feet in contact with ground. The foot that is on the ground for the shortest period of time may be more painful. Some clinicians prefer to observe the position of the head when observing for a forelimb lameness. The concept is similar; the patient will place more weight on the less painful side; as the foot on this side is positioned firmly on the ground, the shoulder and head will drop. Alternatively, the foot on the more painful side will not be in contact with the ground as much and the shoulder and head will be elevated. Circumduction of a limb is abnormal. It is most likely present because of a painful joint; the patient resists flexion because of this pain and swings the limb instead. A preference for observing a patient’s gait at several different velocities is common. In addition, some gait abnormalities may be observable only after the patient has been exercised.

The largest concern with the use of visual observation of gait is its subjective nature. Clinical and personal experiences can influence interpretation, thus introducing opinion, which, in the author’s point of view, reduces precision, accuracy, and reproducibility. For example, in one publication when the sensitivity of visual observation of gait was compared with that of force platform gait analysis, visual observation was found to be vastly inferior [1]. In this comparison, they evaluated 148 adult Labrador retrievers; 17 were free of orthopedic and neurologic abnormalities, and 131 were 6 months after surgery for unilateral CCL injury. The observer was blinded to the dog’s group assignment. Of the 17 normal dogs, the observer correctly identified all as having no gait abnormality, as did the force platform. However, the observer identified only 15 of the 131 dogs that were 6 months after knee surgery as not being normal. Using ground reaction forces from force platform gait analysis, 75% of the 131 dogs failed to achieve GRFs consistent with sound Labradors. In effect, if a dog looked lame it was lame; but if it looked normal to the observer, it may in fact have been abnormal. It should not be surprising that a computational gait analysis is more sensitive than our powers of observation. This fact should be remembered when we inform an owner that a dog "has returned to normal" after a clinical examination.

Force Platforms

A force platform measures the ground reaction forces (GRFs) exerted when it is stepped on during the stance phase. It consists of a metal plate that is mounted level with the surrounding floor or walkway that is separated from a bottom frame by force transducers near each corner. Forces in an X, Y, or Z direction that are exerted on the top surface are transmitted through the force transducers. Commercially available force plates are frequently grouped into either piezoelectric or strain gauge. Although subtle differences exist in the capabilities and cost of the two types, for clinical gait applications, the type used probably has little influence on the data generated. Piezoelectric force plates utilize quartz transducers, which generate an electric charge when stressed. Strain-gauge force plates utilize strain gauges to measure the stress in load cells when a force is applied. Piezoelectric force plates are a bit more sensitive, have a built-in amplifier, and allow for a greater force range than do strain-gauge types. However, they may have some drift, which requires resetting of the charge amplifiers just prior to data acquisition, and they are usually more expensive. There are several different manufacturers to choose from and aside from cost, the potential for customization of a system is something that should be factored into the decision-making process. The size of the force platform, transparency (if photography under the platform is desired), portability, and use with a treadmill (only z direction forces can be measured) are all potential options. Although all of the manufacturers offer compatible software, for veterinary use most clinicians and researchers use commercially available veterinary-specific software. A final consideration is professional installation in an area that encourages frequent clinical and research use but avoids abuse from general hallway traffic.

Pressure Platforms and Walkways

Pressure platforms are comparatively new systems that allow investigators to measure both temporal and spatial gait parameters. In general, pressure platforms are a bit larger (2 square feet) than a traditional force platform. Pressure walkways, which consist of multiple platforms in series, can be customized to virtually any length. These platforms function using a sensor pad that is an ultra-thin (∼0.1 mm), flexible printed circuit. Contained within the circuit are thousands of pressure-sensing locations, or elements, arranged in rows and columns along the length and width of the platform or walkway. The sensing elements act as variable resistors within an electrical circuit. When the sensors are loaded, their resistance is low; when the force is reduced, resistance increases. The output resistance created is then converted to a raw sum for use in analysis. In effect, this allows one to measure ground reaction forces in the z-force direction. Sensors can be produced with pressure ranges from as low as 0-5 psi to as high as 0-25,000 psi. Perhaps it is important to note that ground reaction forces generated from normal dogs on pressure walkways have been shown to be nearly identical to those of force platforms [2].

Several advantages exist to using a pressure walkway. First, because of its extended length, multiple readings and simultaneous, consecutive, and contralateral foot-strikes can be recorded with a single pass over the walkway. Some of this advantage can be attenuated if a laboratory is equipped with multiple force platforms. A single force platform, however, can only measure on half of a gait cycle in a single pass or trial. Thus, by increasing the length of the measuring field, one not only gathers data that are more reflective of the patient’s gait but fewer trials are needed to generate an adequate amount of data for statistical comparison. Reducing trial repetition is important because it saves time, because some weak or lame patients may not be able to physically perform many trials, and because the majority of variance in force platform data is attributable to trial repetition [3]. Second, because the geometry of the mat is known, spatial parameters of gait can be calculated. Patient stride length and width for each limb can be measured for consecutive steps, allowing the clinician to look for inconsistencies, improvements, or disease progression. Limb velocity and acceleration can be calculated using these data or they can be measured using a speed gun or photoelectric cells. Third, the distribution of pressure from the entire foot can be investigated [4]. Both types of platforms measure cumulative force(s) over the cell. For pressure platforms the cell is usually 0.25-square centimeters and for a traditional force platform the cell is 3-square feet. This difference allows for clinicians to estimate changes in load over the patient’s foot (e.g. is load being shifted to the left side of the foot because of an injury on the right). Fourth, because of the small size of the cell in pressure platforms, there are few limitations to the size of the animal for which data can be accurately collected. Using a traditional force platform, a dog’s stride length must be long enough that it place only one foot on the platform at a time. In general, dogs weighing less than 20 kg have a difficult time accomplishing this. Likewise, an extremely tall dog with a long stride may step over a traditional force platform. Increasing the animal’s velocity through the gaited area will increase its stride length but this methodology creates other limitations. Because pressure platforms allow for data collection in animals with a very short stride length they provide the opportunity to measure limb function in small dogs and cats [5,6] or in large dogs that have a short stride length because of disease or because they are recovering from surgery [7]. An added benefit to this advantage is that pressure platforms allow users to create, view, and save data in the form of a "continuous framed movie". Movies are recorded over a specified time period, at a specified frame per second speed. These gait movies allow for clinicians to measure load-bearing while the animal stands still [7]. Finally, pressure platforms are easy to set up, break down, and move to a different location. This portability provides opportunities to measure limb function that would otherwise be impossible with a traditional force platform.

Unfortunately, pressure platforms have some disadvantages. One drawback is that they can only measure total ground reaction force. Forces are not deciphered between the x-, y-, and z-direction. Second, our experiences with the software are that it is easy to work with but requires the user to invest much more time to extrude the data from each trial. For example, when using a traditional force platform, the software immediately provides peak vertical force and vertical impulse numbers. The pressure platform would require approximately one minute of additional work to gather that data. In effect, data are not automatically tabulated and summarized; this must be done manually. Furthermore, although some platform systems have components that incorporate velocity and acceleration data for each trial, most pressure platform systems do not. Again, these values, to date, must be manually calculated and added to the database. Pressure platforms are also not nearly as durable as a metal force platform and would not last long if used to measure limb function in a horse. Finally, the many advantages of pressure platforms come with a significant cost.

Technical Aspects of Force Platform Gait Analysis

Documentation of patient kinetics generates an enormous amount of data. Forces in the x-, y-, and z-direction are expressed in both peak and impulse forms, and average rising and falling slope are just a few of the points available. One obvious question is what data the clinician or investigator should focus on. Forces in the z-direction are generated by a vertical compression of the platform, and these forces are dramatically larger than those of the other two directions. Forces in the y-direction (second largest) are generated by cranial-caudal tipping of the platform and are closely related to acceleration or deceleration, which can be measured or controlled in other ways. Forces in the x-direction are generated by medial-lateral tipping and are extremely small in the quadruped. Peak vertical force (PVF) is the single largest force during the stance phase and represents only a single data point. Vertical impulse (VI) is the total area under the stance-phase curve. The results of these facts are that most veterinary publications have focused on z-direction PVF and VI forces. Although this method of expressing the data is reasonable and common, it may not be ideal. Evaluation of the force-time curves that are provided by the software is likely the best representation of limb function. Unfortunately, the statistical methods associated with this can be complicated. A simple method to mathematically explore gait analysis is to use a multivariate, as opposed to a univariate, statistical approach. In one report, the optimal set of GRFs was selected using logistic regression, with normal or not normal as the binary dependent variable and GRFs as candidate explanatory variables [8]. This method built a descriptive yet economical model that evaluated most candidate combinations of GRFs that best discriminated between the GRFs of normal Labrador retrievers and those of Labrador retrievers that had unilateral lameness from a torn cranial cruciate ligament. A cutoff that maximized Youden’s index (sensitivity + specificity -1) was established and then a receiver operating curve (an ROC is a plot that represents the relationship between sensitivity and specificity of a diagnostic test) was used to assess the diagnostic properties of GRFs obtained from the force platform. These values were then applied to a group of Labrador retrievers that were 6 months after unilateral CCL-rupture surgery. The probability that an individual Labrador could be discriminated from the normal population of Labradors was calculated from data collected from visual observation of gait and from GRFs generated by force platform gait analysis. There are several advantages to this methodology. Limb function is determined by evaluating all GRFs. In comparison, if clinicians use z-direction PVF, they truly are looking at only 1/1200 (this assumes a sampling frequency of 1200 samples per second and a stance phase of 1 second) of the data. In addition, this technique allows clinicians to estimate the probability of normal function after gait analysis and communicate that to an owner. For example, we now know that 6 months after some surgical procedures the probability of normal limb function is only 20% [9]. This does not mean that function after surgery is not improved or that surgery wasn’t successful, only that function is not normal. This is an important step toward differentiating outcomes between treatment options.

Subject velocity influences most GFR measurements. For example, as patient velocity increases, z-force PVF increases because the patient hits the platform harder, and z-force VI decreases because the patient spends less time on the platform. Historically, for force platform gait analysis, patient velocity has been recorded via the use of photoelectric cells that project a beam of light across the runway and are set up at a known distance apart from one another with the center photoelectric cell centered at the center of the platform. When a subject interrupts the first photoelectric cell’s beam of light, the system is triggered to begin recording time as the subject passes across the force platform. When the last cell’s beam of light has been interrupted, the recording system stops. The patient’s average velocity through the runway is then calculated. If three or more photoelectric cells are used in the system, patient acceleration can also be calculated. Velocity can also be measured using a speed gun. One limitation of the photoelectric cell system is that the cells are often aligned so that the subject’s torso is the trigger of the system. Thus, photoelectric cells measure average torso velocity rather than limb velocity. The significance of this can be envisioned if one were to compare velocities of a Great Dane to that of toy poodle. If both dogs have similar torso velocities, the limb velocity of the smaller dog would have to proportionally increase relative to its shortened stride length. This limitation can be averted by comparing data from dogs of similar stature. The other, perhaps easier, method for controlling for this problem is to standardize stance time on the platform [10].

Given the importance of subject velocity, it is common for investigators to control this parameter when evaluating and comparing a group of dogs. For walking trials, 1.0 to 1.3 meters per second, and for trotting, 1.7 to 2.0 meters per second, are used. It has been suggested by some that a trotting velocity should be used because it is more of a challenge to the dog and thus more lameness will be detected. This argument may have merit if the patient population being studied has only a subtle limp. However, in one study the effect of patient velocity on GRFs was studied and a walking velocity was clearly superior [8]. Several important messages came from that study. First, GRFs collected at a walk were linearly related and highly correlated to those collected at a trot. Second, in this investigation of dogs with lameness from a torn cranial cruciate ligament, data could be collected from all dogs at a walking velocity but only 62% of the dogs at a trotting velocity. The large proportion of dogs that failed to obtain acceptable trotting trials would dramatically impact a study. Ground reaction forces at a walk were significantly different among dogs that could and could not successfully trot; this occurred because the dogs that were the most severely lame failed to trot (most likely because as velocity increased, force increased, and thus pain increased). These dogs would be eliminated from a clinical study, and the study would be biased toward dogs that were less lame. This may make a study group that received a certain treatment appear less lame than if they were studied at a walking velocity. Use of trotting velocity would affect subject accrual because more candidate dogs would be necessary to reach the sample size required for the study. For example, for a study in which 50 dogs are required, 81 candidate dogs would be needed to perform the study by use of trot velocity (assuming the 38% failure rate reported). In contrast, a similar study to evaluate 50 dogs at a walk would likely require only 50 candidate dogs. Third, among dogs that could walk and trot, fewer trials were required to obtain acceptable walking data than trotting data and walking data had a smaller coefficient of variation, suggesting that walking has less inherent variation than trotting.

Interestingly, the ideal velocity should probably not be determined by the investigator but the subject. If one dog is most comfortable walking through the runway at 1.1 meters per second then that is arguably the speed that you will gather data that is most reflective of that dog’s normal gait. If a second dog in the study wants to go 1.8 meters per second that also should be allowed. The velocity of these dogs for subsequent trials should remain the same. The velocity, and/or stance time of these dogs can vary and their influence on GRFs can be calculated and adjusted mathematically.

Kinematic Analysis of Gait

A trained observer may be able to make critical judgments about a patient’s gait, but by viewing a video recording, especially if viewed in slow motion, more subtle abnormalities may be detected. Thus, one of the simplest pieces of gait instrumentation, a picture video system, also is one of the most useful. It is also useful to document gait prior to applying any instrumentation so that differences can be resolved if, after equipment has been attached to the animal, motion data do not correspond to the clinician’s initial visual image of the subject. In addition to motion photography, automated stereometric systems are commonly used. Other technologies, such as electromechanical linkage and electrogoniometry, which use an exoskeleton that is applied to the patient, are too cumbersome for most animals.

The stereometric method employs visible markers that are attached to the skin on rigid segments of the body (e.g., joints, centers of rotation, or bony prominences) and tracks their motion using imaging equipment. The markers are either infrared light-emitting diodes for active marker systems or solid shapes covered with retroreflective tape for passive marker systems. Digital image analysis allows the horizontal and vertical coordinates of each marker to be computed as the subject moves within the field of view. Using triangulation of the views from an array of cameras and the known location of each camera, computer software computes the coordinates for each marker. This technique has minimal impact on the natural motion of the subject and two- or three-dimensional methods can be employed. Two-dimensional methods have a diminished time and financial investment for the laboratory or clinic that will only perform this method occasionally, but joint motions can only be determined in one plane. Three-dimensional methods generally employ the use of 3 to 8 cameras that are mounted to the ceiling or wall and create a semi-permanent field of view. It is important to note that, in order to document three-dimensional kinematics, each body segment must be defined by at least three markers, each joint center must be defined, and Euler angles calculated. Although these systems allow for collection of state-of-the-art motion data, they are relatively expensive and, if the methods are not well understood, they can be intimidating.

Motion data allows for calculation of time/distance parameters (velocity, cadence, stance, and swing times, etc.) and the angular position of the joints (hips, knees, and ankles) during the different phases of gait. These methods have been well described and demonstrated in normal dogs at a trot, in dogs with orthopedic disease, and in dogs that are swimming [11-13]. Although historically these methods have been used more in the field of veterinary orthopedics, applications for patients with a neurologic disorder that quantify deviation from normal may offer the greatest clinical treatment potential for this technology. Three-dimensional kinematics, when collected simultaneously with force (kinetic) data can provide useful information about the force around a joint.

Electromyography (EMG)

EMG provides a representation of the contribution from muscles during gait. EMG can be useful to the clinician, but detailed attention to the instrumentation and techniques must be made to provide high-quality EMG signals. Surface electrodes have gained widespread use owing to their ease of application and because skin penetration is not required. However, results for deep muscles can be reliably obtained only with intramuscular wire electrodes. As with many of the other techniques described in this chapter, collaboration with an experienced group early on will pay long-term dividends.