Abstract
Background and Purpose Gait training is an important component of rehabilitation after lower-extremity amputation. Abnormal gait performance often persists even for individuals who reacquire a high level of function. This case report describes the use of a virtual reality (VR)–based gait training program that provides real-time feedback in order to improve biomechanical and physiological performance. The aim of this case report is to describe the effects of the training in a person with a transfemoral amputation.
Case Description A 24-year-old man with a transfemoral amputation completed a 3-week gait training program. The intervention consisted of 12 sessions of treadmill walking with real-time visual feedback on full-body gait kinematics. A treating therapist directed the patient's attention to specific gait deviations as a means to normalize gait biomechanics.
Outcomes The patient completed overground biomechanical gait analyses and multiple-speed treadmill tests 3 weeks apart prior to and following the training program. Biomechanical gait analyses indicated the training produced improved frontal-plane hip, pelvis, and trunk motion during overground walking. Improvement in trunk motion was observed at the posttraining test, and improvements in pelvis and hip motion were observed at the 3-week follow-up test. Decreases of up to 23% in oxygen consumption also were demonstrated.
Discussion Although the exact contribution of the visual feedback could not be isolated, the training was effective in improving the patient's walking performance. Biomechanical data suggest correcting trunk motion and increasing hip abductor strength (force-generating capacity) may be important in facilitating improvements at the pelvis and hip. Observed improvements in oxygen consumption were significantly larger than achieved through previously reported interventions.
Approximately 60,000 transtibial and transfemoral amputations are performed each year in the United States alone.1 Gait training is a key aspect of the rehabilitation that follows amputation. A majority of patients regain the ability to walk functional distances,2,3 yet gait deviations are common beyond the completion of conventional rehabilitation. The impairments contribute to significant increases in physiological energy requirements for walking,4,5 and the development of debilitating long-term musculoskeletal comorbidities including osteoarthritis and low back pain.6,7 Clinicians have long sought improved methods to restore more desirable gait biomechanics with the belief that doing so would improve efficiency and overall function.
During level-ground walking, people with transfemoral amputation demonstrate significant deviations in frontal-plane kinematics of the pelvis and trunk.8–11 People with a transfemoral amputation commonly display a contralateral rise of the pelvis during mid-stance on the prosthetic limb instead of maintaining a more neutral position.8,10,11 An excessive lateral lean over the prosthetic limb and larger total trunk excursions also are common.8 Normalizing frontal-plane motion has been suggested as important in restoring an efficient, aesthetic, and stable gait pattern following a transfemoral amputation.10,12 Unfortunately, improving the interrelated frontal-plane motion of the pelvis and trunk using available treatment approaches often is difficult.12 For the pelvis in particular, treatment effectiveness can be hindered by soft tissue coverage and a complex multiplanar motion, which decrease the reliability of observation gait analysis.13 Furthermore, there can be difficulties associated with effectively communicating desired kinematic changes in a manner that can be acted on by the patient.
The use of virtual reality (VR), defined as a simulation of a real-world environment that is generated through computer software and is experienced by the user through a human-machine interface,14 has increased during rehabilitation as researchers and clinicians search for new innovative methods to augment conventional clinical care and optimize the physical abilities of their patients. Proponents highlight the capability of VR systems to enhance skill acquisition and retention by providing task specificity (ecological validity), repetition, and external real-time feedback (knowledge of results).15 For several clinical populations, VR-based gait training with feedback was beneficial in improving parameters such as symmetry and increased walking speed.14,16–18 In people with an amputation, feedback on symmetry of limb push-off forces produced a reduction in the metabolic cost of nearly 6%.19 However, it remains to be seen whether the complex motions of the pelvis and trunk can be effectively modified using VR-based feedback and whether the changes would benefit the overall gait performance of people with an amputation. The purpose of this case report is to present the effects of a novel VR-based gait training program using real-time feedback to restore normal walking kinematics, with a particular emphasis on frontal-plane pelvis and trunk motion, and consequently decrease the physiological effort required while walking in a person with a transfemoral amputation.
Patient History and Review of Systems
The patient was a 24-year-old man who experienced a traumatic blast resulting in a left transfemoral amputation and a right tibia fracture with soft tissue injury requiring skin grafting. Following the injury, he was transferred to a specialized care facility for people with amputations where he was walking independently on a transfemoral prosthetic limb with no assistive devices within 3 months. Overall, he completed approximately 20 months of extensive physical rehabilitation. At the completion of his rehabilitation, he was highly functional and living independently in the local community.
Approximately 6 months after completing his rehabilitation program, the patient volunteered for the VR-based gait training program described in this case report. The patient provided written consent to participate as part of a larger clinical study approved by the Institutional Review Board at Brooke Army Medical Center, Fort Sam Houston, Texas. At the time of enrollment, the patient was 174 cm tall, weighed 86.8 kg, and walked using a prosthetic limb containing an ischial containment socket, a Total Knee,* and a Reflex VSP foot.* He remained functionally independent in the community and met the following minimum criteria for participation in the clinical study: (1) 18 to 45 years of age; (2) independent ambulation without an assistive device for a minimum of 3 months; (3) ability to ambulate continuously for a minimum of 15 minutes; (4) verbal pain rating of less than 4/10 on the involved side; and (5) sagittal-plane ankle, knee, and hip strength (force-generating ability) of 4 or greater on the uninvolved side, as determined by manual muscle test. No conditions other than the amputation that would affect walking ability were identified in the patient's health history. He was not actively engaged in a treadmill walking or other physical fitness program at the time of enrollment in the study. For the duration of the study, the patient continued with his normal activity profile outside of the training and did not begin a new exercise program or make changes to his prosthetic limb.
Examination
A total of 4 tests were completed over the course of 12 weeks. All tests were identical and involved collecting biomechanical and physiological measurements of walking ability. Two pretraining tests were completed 3 weeks apart in order to create a baseline status and define the patient's normal fluctuation in performance over time. A posttest was completed after the 3 weeks of training to evaluate the effect of the intervention. Finally, a 3-week follow-up was used to determine the retention and potential continued emergence of benefit from the training. A group of 30 individuals without amputations enrolled in the larger clinical study completed a single test session to establish normative values.
Biomechanical
Full-body gait kinematic and kinetic measurements were collected as the patient walked over a level surface at a self-selected walking speed and a controlled walking speed (1.22 m/s). The controlled walking speed was derived using the patient's measured leg length,20 and software-generated auditory cues† were used to ensure he remained within 5% of the predetermined speed. Walking speed was controlled to account for its potential influence on the measures of interest.21 Kinematic data were collected as the patient walked across a 5-m-long walkway using a 24-camera optoelectronic motion capture system† recording at 120 Hz. Kinetic data were collected at 1,200 Hz using 8 forceplates.‡ The local coordinate system for each segment was defined using International Society of Biomechanics standards and tracked using a 6-degrees-of-freedom marker set.22 Five strides from each limb normalized to 100% of the gait cycle were used to characterize patient performance at each testing session.
The training was primarily assessed using frontal-plane motions of the hip, pelvis, and trunk. Mid-stance joint angles (20% of the gait cycle) were of particular interest in determining whether the training established a more desirable posture during single-leg stance on the prosthetic limb. To fully characterize the patient's gait performance at each session, temporal-spatial and other kinematic and kinetic parameters for the lower extremities, pelvis, and trunk also were assessed. Changes in kinematic variables greater than 4 degrees between tests were identified as a true change in performance, based on minimal detectable change values23 calculated within the laboratory.
Physiological
Breath-by-breath oxygen consumption (mL O2/kg/min) was monitored continuously during a multiple-speed treadmill walking test. The test involved walking at a speed matching the leg length–controlled speed, as well as speeds 20% and 40% slower and 20% faster (ranging from 0.73 to 1.47 m/s). Data were collected by a computer-interfaced portable metabolic unit§ consisting of a face mask worn by the patient connected to the device secured in a harness around the chest. The device was calibrated for gas concentration, turbine flow, and delay before each test according to manufacturer guidelines. A safety harness that provided no weight-bearing support was worn by the patient during testing to eliminate the risk of falling. Each stage lasted 5 minutes in order to allow the patient to achieve physiological steady state.
Averaged steady-state data from the final 2 minutes of each stage were used to represent the oxygen consumption. A change in oxygen consumption greater than 7% was used to evaluate differences between tests based on minimal detectable change values calculated from collections in the laboratory.
Clinical Impression
Temporal and spatial data displayed in the Table demonstrate asymmetries in step length and stance time and an increased step width. Self-selected walking speed exceeded values typically reported for people with a transfemoral amputation, approximating the walking speeds demonstrated by people without amputations.24 Clinically, the greatest deficits during pretraining test 1 were observed in frontal-plane motion. Frontal-plane kinematic data for the left hip, pelvis, and trunk are presented in Figure 1 and indicate a compensated gluteus medius muscle gait pattern, as noted by the increased hip abduction, contralateral pelvic rise, and excessive ipsilateral trunk motion. The primary kinetic change related to the altered frontal-plane motion was a decrease in peak hip abductor torque (Fig. 2). In addition, the range of reciprocal transverse-plane motion between the pelvis and trunk was 7 degrees less than in the reference group of individuals without amputations, and the onset of right ankle plantar flexion occurred at 25% of the gait cycle (vaulting) instead of at approximately 50% of the gait cycle. Similar gait kinematics and kinetics were observed during pretraining test 2, with most differences attributed to normal variation in gait performance. Ensemble-averaged figures for lower-extremity kinematic and kinetic parameters, normalized to percentage of the gait cycle, are shown in eFigures 1, 2, and 3.
Temporal-Spatial Gait Parameters for a Controlled Overground Walking Speeda
Mean (±SD) frontal-plane joint angles for the hip, pelvis, and trunk at mid-stance for the patient from each test with reference to a group of individuals without amputations.
Peak hip torque for the patient in the frontal plane of the prosthetic limb.
Kinematic gait deviations were accompanied by a 66% to 102% increase in oxygen consumption compared with the reference group of individuals without amputations for walking speeds ranging from 0.71 to 1.43 m/s (Fig. 3). A comparison of oxygen consumption was not made for the 2 slowest speeds from the first test, as data were omitted due to the patient's use of the treadmill handrails. Use of the handrails was not evident for the higher speeds or any of the speeds in subsequent tests. At the higher speeds (1.22 and 1.47 m/s), there was an increase of 2% to 3% between pretraining tests 1 and 2.
Oxygen consumption for the patient expressed as (A) values from each test with reference to a group of individuals without amputations and (B) percentage of change between tests. Data from pretraining test 1 (0.73 and 0.97 m/s) are excluded due to weight bearing through handrails.
Intervention
The training was conducted within a domed Computer Assisted Rehabilitation Environment (CAREN).‖ The one-of-a-kind VR system includes 8 projectors, which project a VR environment and visual feedback on the inside surface of the dome. A 24-camera motion capture system# within the dome was used to track 3-dimensional, full-body kinematics. Figure 4 displays the CAREN system as well as the virtual environment developed for the training. The training environment consisted of a straight walking path through a forested area and was designed to minimize distractions from the visual feedback displayed on the screen directly in front of the individual. The feedback consisted of a real-time, full-body virtual representation (marker cloud) of the person and a trace of the frontal-plane trunk motion (based on a marker at the level of C7). In addition, to assist the treating therapist in identifying specific deviations in pelvis motion, graphs of the frontal-, transverse-, and sagittal-plane pelvic motion relative to normative data were displayed behind the patient. The normative kinematic data were collected from the 30 individuals without amputations walking in the same virtual scene and presented as time normalized to 100% of the gait cycle.
Computer-Assisted Rehabilitation Environment (CAREN) system and virtual training environment.
Training consisted of twelve 30-minute treadmill walking sessions over approximately 3 weeks. The walking bouts were divided into three 10-minute blocks with time for rest and instruction between bouts. Walking speed was adjusted by the treating therapist to avoid excessive fatigue or limit the patient's ability to modify gait kinematics. The treating therapist used both observational assessment and real-time kinematic data displayed on the CAREN screen to identify gait deviations and guide patient feedback. Verbal cues from the therapist were used to communicate corrective strategies by directingthe patient's attention to specific aspects of the visual feedback. The patient also was encouraged to independently develop strategies to achieve a normalized gait pattern.
Overall, the goal of the training program was to correct the patient's compensated gluteus medius muscle gait pattern. To meet this goal, verbal cueing was primarily directed toward achieving single-leg support on the prosthetic limb with a more vertical trunk position and a level pelvis. Specifically, cues addressing frontal plane motion of the trunk used the C7 marker trace. The patient was asked to generate a more symmetrical and narrower trace in order to correct the flattened and elongated trace displayed with stance on the prosthetic limb. A focus on frontal-plane pelvic rotation was achieved by directing the patient's attention to the relative vertical motion of the left and right anterior superior iliac spine (ASIS) markers. Cues addressing hip abduction involved positioning the left foot, as noted by a marker on the posterior aspect of the shoe, more medially in reference to the left ASIS marker. Additional, less frequent cues were provided using a combination of markers from multiple segments to increase reciprocal transverse-plane pelvis-trunk motion and decrease vaulting caused by the early heel rise in the intact lower extremity. Overall, the frequency of the cueing was consistent until the final 3 sessions. During those sessions, an intentional progressive decrease in cueing was provided.
Outcomes: Pretraining and Posttraining Comparisons
Over the course of the 12 CAREN-based training sessions, the most significant changes occurred in the frontal-plane kinematics. Figure 5 highlights fluctuations in gait kinematic parameters over the course of treatment, depending on the focus of the cueing during each session. A general trend toward improved kinematics, however, was observed, with the final sessions demonstrating gait kinematics that resembled the subsequent overground gait performance. This more desirable strategy included decreased total frontal-plane trunk excursion and improved positioning of the pelvis over the prosthetic limb. Temporal spatial parameters, reciprocal pelvis and trunk motion, and early heel rise did not exhibit any identifiable trends or overall improvement within the CAREN system.
Training trends for the patient during Computer-Assisted Rehabilitation Environment (CAREN) training for (A) total peak-to-peak trunk excursion in the frontal plane (in degrees) and (B) minimum medial-lateral distance between the foot and mid-pelvis segments during stance on the prosthetic limb.
Overground testing following the training program demonstrated transfer of the frontal-plane improvements noted in the VR system (Fig. 1). The patient's lateral trunk lean at mid-stance decreased by 4 degrees, and the total excursion over the gait cycle decreased by half from 16.6 degrees to 8.5 degrees. Mid-stance hip abduction decreased by approximately 4 degrees, but the leg remained abducted almost 5 degrees. The position of the pelvis over the foot (Fig. 6) and peak hip torque (Fig. 2) also improved following the training program. No changes were observed at the pelvis, but reciprocal transverse-plane motion with the trunk did increase by 2.5 degrees to approximately 12 degrees. Although the increase was closer to the 15 degrees exhibited in the walkers without amputations, it resulted from increased trunk motion and not improvements in pelvis rotation, as would have been desired. The patient continued to vault on the intact side with the onset of right ankle plantar flexion at 26% of the gait cycle.
Minimum medial-lateral distance between the foot and mid-pelvis segments during stance on the prosthetic limb while walking overground.
At the final test session, 3 weeks after the completion of training, a clinically relevant improvement in frontal-plane pelvis position was the primary biomechanical change. The pelvis was shown to have a more normal 2.4-degree drop on the contralateral side instead of the previously observed contralateral hip rise. The modified pelvis position likely contributed to the further normalization in frontal-plane hip motion that produced a 9-degree improvement overall compared with before training. In addition, the modified frontal-plane pelvis and hip angles coincided with a 19% reduction in step width to 0.17 m (Table). Other parameters, including trunk position, hip torque, reciprocal transverse-plane motion between the pelvis and trunk, and the early heel rise did not change or demonstrate further improvement.
Pretraining and posttraining comparisons demonstrated minimal change in oxygen consumption at the slowest walking speed (0.73 m/s), but improvements of 13% to 16% were found for the 3 highest speeds (ie, 0.97–1.47 m/s) (Fig. 2). During the final testing session, additional improvements of 3% to 9% were observed at the 3 lowest speeds, whereas a 3% increase was observed at 1.47 m/s. Overall, an 8% to 23% improvement in oxygen consumption was observed between the pretraining session and the final test session.
Discussion
This case report describes the use of a VR-based gait training program in a person with a transfemoral amputation. The goal of the training was to improve gait performance by providing real-time feedback during gait. The results show 12 sessions of VR-based gait training were effective for this patient in producing more normalized overground gait biomechanics and a decrease in oxygen consumption.
The patient had several gait deviations that are commonly observed in individuals with a lower-extremity amputation. Deviations included increased hip abduction on the involved limb, elevation of the contralateral hip, and exaggerated trunk motion over the prosthetic limb. After completing the training program, the patient demonstrated improvements in frontal-plane trunk, pelvis, and hip kinematics (Fig. 1). However, the kinematic changes did not manifest simultaneously. The nearly 50% correction in trunk motion was observed at the conclusion of treatment, whereas the improvement in frontal-plane pelvis position at mid-stance was not observed until the final assessment 3 weeks later. A possible explanation is inadequate pretraining hip abductor strength, which is common following amputation and is a cited cause of frontal-plane kinematic abnormalities.8,10 Aligning the trunk more vertically during single-leg stance forced the patient to position the pelvis over the prosthetic limb and use the hip abductors rather than trunk lean to control pelvis position, potentially creating a strength training stimulus for the hip abductors.10 Sufficient strength to achieve a level pelvis during mid-stance on the amputated side and a narrower step width could have then been developed over the course of the training and the subsequent 3 weeks. Patient reports of hip abductor fatigue during training provide further support for the presence of a strength training stimulus.
In addition to improvements in gait kinematics, oxygen consumption decreased up to 23%. The improvement represents a treatment effect at least 2 to 3 times greater than previously published for gait training interventions or changes in prosthetic devices and alignments.19,25–27 Although many factors could have contributed to the overall improvement, existing literature substantiates the influence of reduced trunk motion and a more normalized step width on oxygen consumption.28,29 For this patient, the greatest improvements in oxygen consumption and trunk motion occurred at the same time point, suggesting training to correct trunk abnormalities may have a greater impact on the oxygen consumption than changes in step width, which were not observed until the final test. Improved ability to capitalize on the energy-storing and return properties of the prosthetic foot also could have contributed to decreases in oxygen consumption. Previous studies have demonstrated energy-storing and return prosthetic feet reduce energy requirements at faster walking speeds compared with non–energy-storing and return devices.30 For this patient, a smaller improvement in oxygen consumption was observed at the slowest walking speed. However, a lack of appreciable change in ankle power suggests the patient did not increase the energy return from his prosthetic foot. A lack of kinetic data for multiple walking speeds limits any evaluation to determine whether improved energy return was achieved.
There are many potential factors that could have contributed to, or potentially limited, the observed improvements in biomechanical and physiological gait performance. Limitations associated with assessing the effect of a complex treatment intervention on a single patient raise several important questions to be addressed through future study.
First, the visual feedback provided a clear representation of frontal-plane motion but a limited view of motion in the sagittal or transverse plane. The lack of clear visual feedback for sagittal- and transverse-plane motions made communicating desired changes more difficult. It is possible that feedback specific to those motions might have resulted in meaningful improvements.
Second, it is impossible to determine whether the positive outcomes of the training were the direct result of cues from the therapist, self-exploration by the patient, massed treadmill walking, or other factors. Our experience and the patient's comments suggest the cueing and self-exploration are most important. Therapist feedback would appear to be most critical early in the training process to direct the patient's attention to specific areas to change and to provide potential solutions. Self-exploration by the patient can be beneficial at any time in the training, particularly when cues are not effectively achieving the desired change. The importance of self-exploration also exists in later sessions as the therapist promotes patient autonomy, awareness of internal feedback, and practice of the newly learned gait pattern by reducing the frequency of cues.
Third, prosthetists adjust limb alignment to create a safe and aesthetic gait pattern. During a conventional rehabilitation program, it is not uncommon for the prosthetic alignment to change frequently as the individual reacquires the ability to walk. In retrospect, it is possible that withholding alignment changes limited the patient's ability to adapt his gait and the overall effectiveness of the training. However, allowing alignment changes would have made it impossible to determine what real benefit the training program had for this individual.
Finally, an important observation was the carryover of changes observed in the training environment to overground walking. Although testing for biomechanical measures occurred overground, measurements of oxygen consumption were collected while walking on a treadmill. Although the patient was familiar with treadmill walking, it is possible that conducting the training and testing on a treadmill led to greater improvements in oxygen consumption than if testing had occurred overground.
Conclusions
Twelve sessions of gait training capitalizing on real-time feedback were effective in improving gait performance in a person with a unilateral transfemoral amputation. Clinically important changes in gait biomechanics and oxygen consumption were observed following the training. Retention of gains and additional improvement in several measures also were found 3 weeks after the end of training. In summary, this case report indicates the use of the VR environment–based real-time feedback holds great promise for improving the rehabilitation of individuals with an amputation by providing a treating therapist more quantitative data along with a novel and potentially more effective method for communicating desired changes.
Footnotes
-
Both authors provided concept/idea/project design, writing, data collection, project management, fund procurement, and clerical support. Dr Darter provided data analysis. Dr Wilken provided facilities/equipment and consultation (including review of manuscript before submission). Development and operation of the simulation environment was provided by Michael Vernon. Assistance with data processing was provided by Kiril and Emily Sinitski.
-
The patient in this case report was a participant in studies approved by the Institutional Review Board at Brooke Army Medical Center, Fort Sam Houston, Texas (C.2007.072t, C.2008.050dt).
-
A poster presentation of case study data was given at Virtual Rehabilitation 2009; June 20–July 2, 2009; Haifa, Israel. Platform presentations highlighting the results of multiple research studies, including case study data, were given at Combined Sections Meetings of the American Physical Therapy Association; February 17–20, 2010; San Diego, California, and February 9–12, 2011; New Orleans, Louisiana. A platform presentation highlighting the results of multiple research studies, including case study data, also was given at the International Society for Prosthetics and Orthotics World Congress; May 10–15, 2010; Leipzig, Germany.
-
Support for this project was provided by the Military Amputee Research Program through grant awards to the authors.
-
↵* Ossur Americas, 27051 Towne Centre Dr, Foothill Ranch, CA 92610.
-
↵† Motion Analysis Corporation, 3617 Westwind Blvd, Santa Rosa, CA 95403.
-
↵‡ Advanced Mechanical Technology Inc, 176 Waltham St, Watertown, MA 02472.
-
↵§ Cosmed USA Inc, 2211 N Elston Ave, Suite 305, Chicago, IL 60614.
-
↵‖ MOTEK Medical BV, Keienbergweg 77, 1101GE Amsterdam, the Netherlands.
-
↵# Vicon, 14 Minns Business Park, West Way, Oxford OX2 0JB, United Kingdom.
- Received October 29, 2010.
- Accepted May 16, 2011.
- © 2011 American Physical Therapy Association
References
Issue highlights
