- Email: [email protected]
A virtual avatar to facilitate gait rehabilitation post-stroke
Abstracts / Gait & Posture 39S (2014) S1–S141
S51
Fig. 1. Sagittal hip, knee and ankle angles for overground walking (solid lines) and GRAIL walking (dashed lines), averaged over all subjects, with SD (overground in light grey, top; GRAIL in dark grey, bottom).
tated in stance (left leg only). All mean differences were in the order of 1-3 degrees. GGI scores were significantly higher on the GRAIL (right 53.2, left 31.1 versus 15.7 for both left and right overground). Discussion and conclusions: In conclusion, apart from step width, GRAIL and overground walking were very similar, with only minor kinematic differences. A higher GGI on the GRAIL was expected since overground was used as reference, but the difference was relatively large, due to the summation of all small differences. When choosing between overground and GRAIL walking for clinical gait analysis, the potential differences should be weighed against the gains of using a treadmill, e.g. the availability of many steps, and advanced modes like perturbations and realtime feedback. Our findings indicate that GRAIL walking may be used for clinical gait analysis instead of overground walking, but further study is needed. In ongoing analyses, we are investigating the individual effects of VR and self-paced mode on gait parameters (including kinetics), in typically developing children and children with cerebral palsy; and we are comparing spatiotemporal parameters of both labs to walking outside.
Reference [1] Schutte, et al. An index for quantifying deviations from normal gait. Gait Posture 2000;11:25–31.
http://dx.doi.org/10.1016/j.gaitpost.2014.04.070 066 A virtual avatar to facilitate gait rehabilitation post-stroke Heather Thikey 1,∗ , Frederike van Wijck 2 , Madeleine Grealy 3 , Phil Rowe 1 1 Biomedical Engineering Department, University of Strathclyde, Glasgow, UK 2 Institute for Applied Health Research, Glasgow Caledonian University, Glasgow, UK 3 School of Psychological Sciences and Health, University of Strathclyde, Glasgow, UK
Introduction and aim: Augmented visual feedback has the potential to compensate for impairments in intrinsic feedback mechanisms after a stroke and highlight salient information to facilitate motor learning. Although there is much literature to support its use with the healthy population, there is limited evidence for its use with stroke patients [1]. Athletes use feedback via motion capture techniques to enhance their understanding of how to optimise movement patterns. A feasibility study was carried out to explore whether a novel visual feedback aid could facilitate improvements in mobility outcomes after a stroke. Patients and their therapists were shown a virtual avatar that could mimic their
Fig. 1. Virtual avatar with fan target for knee flexion exercise in sitting position.
movements in real-time, with the option to display range of motion targets at joint centres (see Fig. 1). Patients/materials and methods: Twelve stroke participants (5 males, 7 females) with an average age of 59.8 years (range 26–82 years) and average time of 49.4 days (range 20–95 days) post-stroke were enrolled in a feasibility RCT. Participants were randomised into one of three groups for six weeks: an experimental group that received twelve one-hour gait-orientated physiotherapy sessions using the virtual avatar; a control I group that matched the experimental group for type and intensity of physiotherapy without the avatar; and a control II group that received standard care only. An 8camera (v100:R2) Optitrack optical tracking system (NaturalPoint, Inc.) was used to provide movement data required to drive the patient avatar. A detailed protocol is available by Thikey et al. [2]. Outcomes were measured at baseline and post intervention. Primary outcomes were walking velocity, step length and spatial and temporal symmetry. Data acquisition was carried out using a 2D video-based gait assessment system [3]. Results: The changes made in spatiotemporal gait parameters by the different groups between baseline and post-intervention assessments are summarised in Table 1. All groups demonstrated improvements in most parameters, with changes in spatial and temporal symmetry being relatively small. Importantly, all experimental participants were able to engage with the virtual avatar and expressed that they found it useful to see how they moved. Discussion and conclusions: This feasibility study suggests the use of a virtual avatar could be a useful adjunct to gait rehabilitation after a stroke. Although study numbers are small, results suggest this feedback does not inhibit recovery and has the potential to enhance mobility outcomes. Consequent data analysis will explore patient-reported and qualitative outcomes with regards to engagement, motivation and understanding of rehabilitation.
S52
Abstracts / Gait & Posture 39S (2014) S1–S141
Table 1 Gains [percentage improvement gains] in spatiotemporal gait parameters over 6 week trial period. Group
Walking velocity (m/s)
Step length (m)
Spatial symmetry
Temporal symmetry
Experimental (n = 4)
0.275 ± 0.292 95% CI: −0.189 to 0.740 [103.4 ± 132.1%]
0.153 ± 0.119 95% CI: −0.037 to 0.342 [50.3 ± 48.0%]
0.028 ± 0.031 95% CI: −0.022 to 0.077 [6.06 ± 7.66%]
0.010 ± 0.022 95% CI: −0.024 to 0.044 [1.938 ± 3.893%]
Control I (n = 4)
0.111 ± 0.112 95% CI: −0.067 to0.289 [27.8 ± 32.5%]
0.045 ± 0.031 95% CI: −0.004 to 0.094 [12.6 ± 9.19%]
−0.013 ± 0.036 95% CI: −0.070 to 0.045 [−2.41 ± 7.30%]
0.015 ± 0.037 95% CI: −0.044 to 0.074 [2.190 ± 6.169%]
Control II (n = 4)
0.245 ± 0.213 95% CI: −0.093 to 0.584 [83.6 ± 81.0%]
0.090 ± 0.101 95% CI: −0.071 to 0.251] [35.9 ± 43.5%]
0.008 ± 0.029 95% CI: −0.038 to 0.053 [0.008 ± 6.38%]
−0.008 ± 0.013 95% CI: −0.021 to 0.026 [0.482 ± 2.955%]
Reference [1] Van Vliet, Wulf. Disabil Rehabil 2006;28(13–14):831–40. [2] Thikey H, et al. Trials 2012;13:163. [3] Ugbolue, et al. Gait Posture 2013 [in press].
http://dx.doi.org/10.1016/j.gaitpost.2014.04.071 067 Ultrasound use improves the accuracy of surface electromyography placement over rectus femoris in children with cerebral palsy N. Blucher ∗ , A. Bass, G. Holmes, B. Mwaura Kimani, D. Trinca Alder Hey Children’s Hospital, Liverpool, England, United Kingdom Introduction and aim: Rectus femoris (RF) transfer to Sartorius is one treatment option for children with Cerebral Palsy (CP) who have a stiff knee gait [1], therefore assessment of RF function is important in selecting patients for this procedure. The assessment of RF depends on the quality of the signal recorded from sEMG electrodes which in turn depends on the position of the electrode on the skin overlying the muscle. The SENIAM recommendations for sEMG electrode placement are widely accepted, but have only been validated in healthy adult subjects [2]. The aim of this study was to validate the SENIAM recommendations for surface electromyography placement (sEMG) over rectus femoris (RF) muscle in healthy children and in children with cerebral palsy(CP) during gait analysis and compare placement using these guidelines to using ultrasonography. Patients/materials and methods: The study included 10 healthy children volunteers and 10 CP children volunteers, aged 8–12 years. All the children attended the same centre for gait analysis. Ethical approval was gained and parental consent was sought before the enrolment into the study. All the CP children had spastic diplegia, were GMFCS levels I–II, had not previously undergone surgery and were able to cooperate and follow instructions. Routine laboratory protocol for fully instrumented 3D gait analysis was carried out, including a standardized physical examination. RF electrodes were placed following SENIAM recommendations: midpoint between the anterior inferior iliac spine and the superior surface of the patella. This skin at this point was marked. RF was then identified by ultrasound. The medial and lateral edges of RF was identified and marked on the skin in the same horizontal plane as the previous sEMG electrode position. The distance between the lateral edge of RF and the position of the sEMG electrode as per SENIAM guidelines and the width of RF was measured, to the nearest millimetre and recorded. Results: We considered ‘ideal electrode’ position to be at halfway between the edges of RF (i.e 50%). The mean percentage difference in distance from the ‘ideal electrode’ position as measured
by ultrasound to electrode placement following SEMIAN guidelines was 2.7% in the healthy children group compared with 19.5% in the CP group. By performing unpaired t tests we showed that there was no significant difference between the mean electrode position using SEMIAN guidelines and ‘ideal electrode’ position in the healthy children (p = 0.0531), however the mean electrode position using SEMIAN guidelines in the CP patients was significantly different from the ‘ideal electrode’ position (p = 0.0001). Discussion and conclusions: SENIAM recommendations for sEMG electrode placement over RF muscle were validated in 10 healthy children. We showed that ultrasonography improves the accuracy of sEMG electrode placement in children with CP, who can exhibit anatomical variation due to their condition. Accurate electrode placement will ensure that a more accurate signal is recorded which may have a direct clinical bearing on the decision to proceed with surgical intervention. Reference [1] Gage JR, Perry J, Hicks RR, Koop S, Werntz JR. Rectus femoris transfer to improve knee function of children with cerebral palsy. Dev Med Child Neurol 1987;29(2):159–66. [2] Hermens HJ, Freriks B, Disselhorst-Klug C, Rau G. Development of recommendations for SEMG sensors and senson placement procedures. J Electromyogr Kinesiol 2000;10:361–74.
http://dx.doi.org/10.1016/j.gaitpost.2014.04.072 068
Session 4C: Orthopaedics 2
Outcome predictors for multilevel surgery in children with cerebral palsy/spastic diplegia I. Hinden ∗ , J. Stebbins, N. Thompson, T. Theologis Oxford Gait Lab, Nuffield Orthopaedic Centre, OUH NHS Trust, UK Introduction and aim: Single event multilevel surgery (SEMLS) is commonly performed to correct orthopaedic deformities in cerebral palsy. Although the outcome has been shown to be positive overall [1], a significant element of unpredictability is still present for individual cases. In order to improve individual outcomes, it would be helpful to determine which subject and surgical characteristics are associated with improvements following surgery. The aim of this study was therefore to perform a retrospective analysis on all patients who had undergone multi-level surgery at our institution, to determine the relationship between selected characteristics and functional outcome using gait analysis. Patients/materials and methods: A retrospective cohort study was conducted on data collected in the Gait Lab between 1994 and 2012. Subjects included 160 spastic diplegic cerebral palsy patients
S51
Fig. 1. Sagittal hip, knee and ankle angles for overground walking (solid lines) and GRAIL walking (dashed lines), averaged over all subjects, with SD (overground in light grey, top; GRAIL in dark grey, bottom).
tated in stance (left leg only). All mean differences were in the order of 1-3 degrees. GGI scores were significantly higher on the GRAIL (right 53.2, left 31.1 versus 15.7 for both left and right overground). Discussion and conclusions: In conclusion, apart from step width, GRAIL and overground walking were very similar, with only minor kinematic differences. A higher GGI on the GRAIL was expected since overground was used as reference, but the difference was relatively large, due to the summation of all small differences. When choosing between overground and GRAIL walking for clinical gait analysis, the potential differences should be weighed against the gains of using a treadmill, e.g. the availability of many steps, and advanced modes like perturbations and realtime feedback. Our findings indicate that GRAIL walking may be used for clinical gait analysis instead of overground walking, but further study is needed. In ongoing analyses, we are investigating the individual effects of VR and self-paced mode on gait parameters (including kinetics), in typically developing children and children with cerebral palsy; and we are comparing spatiotemporal parameters of both labs to walking outside.
Reference [1] Schutte, et al. An index for quantifying deviations from normal gait. Gait Posture 2000;11:25–31.
http://dx.doi.org/10.1016/j.gaitpost.2014.04.070 066 A virtual avatar to facilitate gait rehabilitation post-stroke Heather Thikey 1,∗ , Frederike van Wijck 2 , Madeleine Grealy 3 , Phil Rowe 1 1 Biomedical Engineering Department, University of Strathclyde, Glasgow, UK 2 Institute for Applied Health Research, Glasgow Caledonian University, Glasgow, UK 3 School of Psychological Sciences and Health, University of Strathclyde, Glasgow, UK
Introduction and aim: Augmented visual feedback has the potential to compensate for impairments in intrinsic feedback mechanisms after a stroke and highlight salient information to facilitate motor learning. Although there is much literature to support its use with the healthy population, there is limited evidence for its use with stroke patients [1]. Athletes use feedback via motion capture techniques to enhance their understanding of how to optimise movement patterns. A feasibility study was carried out to explore whether a novel visual feedback aid could facilitate improvements in mobility outcomes after a stroke. Patients and their therapists were shown a virtual avatar that could mimic their
Fig. 1. Virtual avatar with fan target for knee flexion exercise in sitting position.
movements in real-time, with the option to display range of motion targets at joint centres (see Fig. 1). Patients/materials and methods: Twelve stroke participants (5 males, 7 females) with an average age of 59.8 years (range 26–82 years) and average time of 49.4 days (range 20–95 days) post-stroke were enrolled in a feasibility RCT. Participants were randomised into one of three groups for six weeks: an experimental group that received twelve one-hour gait-orientated physiotherapy sessions using the virtual avatar; a control I group that matched the experimental group for type and intensity of physiotherapy without the avatar; and a control II group that received standard care only. An 8camera (v100:R2) Optitrack optical tracking system (NaturalPoint, Inc.) was used to provide movement data required to drive the patient avatar. A detailed protocol is available by Thikey et al. [2]. Outcomes were measured at baseline and post intervention. Primary outcomes were walking velocity, step length and spatial and temporal symmetry. Data acquisition was carried out using a 2D video-based gait assessment system [3]. Results: The changes made in spatiotemporal gait parameters by the different groups between baseline and post-intervention assessments are summarised in Table 1. All groups demonstrated improvements in most parameters, with changes in spatial and temporal symmetry being relatively small. Importantly, all experimental participants were able to engage with the virtual avatar and expressed that they found it useful to see how they moved. Discussion and conclusions: This feasibility study suggests the use of a virtual avatar could be a useful adjunct to gait rehabilitation after a stroke. Although study numbers are small, results suggest this feedback does not inhibit recovery and has the potential to enhance mobility outcomes. Consequent data analysis will explore patient-reported and qualitative outcomes with regards to engagement, motivation and understanding of rehabilitation.
S52
Abstracts / Gait & Posture 39S (2014) S1–S141
Table 1 Gains [percentage improvement gains] in spatiotemporal gait parameters over 6 week trial period. Group
Walking velocity (m/s)
Step length (m)
Spatial symmetry
Temporal symmetry
Experimental (n = 4)
0.275 ± 0.292 95% CI: −0.189 to 0.740 [103.4 ± 132.1%]
0.153 ± 0.119 95% CI: −0.037 to 0.342 [50.3 ± 48.0%]
0.028 ± 0.031 95% CI: −0.022 to 0.077 [6.06 ± 7.66%]
0.010 ± 0.022 95% CI: −0.024 to 0.044 [1.938 ± 3.893%]
Control I (n = 4)
0.111 ± 0.112 95% CI: −0.067 to0.289 [27.8 ± 32.5%]
0.045 ± 0.031 95% CI: −0.004 to 0.094 [12.6 ± 9.19%]
−0.013 ± 0.036 95% CI: −0.070 to 0.045 [−2.41 ± 7.30%]
0.015 ± 0.037 95% CI: −0.044 to 0.074 [2.190 ± 6.169%]
Control II (n = 4)
0.245 ± 0.213 95% CI: −0.093 to 0.584 [83.6 ± 81.0%]
0.090 ± 0.101 95% CI: −0.071 to 0.251] [35.9 ± 43.5%]
0.008 ± 0.029 95% CI: −0.038 to 0.053 [0.008 ± 6.38%]
−0.008 ± 0.013 95% CI: −0.021 to 0.026 [0.482 ± 2.955%]
Reference [1] Van Vliet, Wulf. Disabil Rehabil 2006;28(13–14):831–40. [2] Thikey H, et al. Trials 2012;13:163. [3] Ugbolue, et al. Gait Posture 2013 [in press].
http://dx.doi.org/10.1016/j.gaitpost.2014.04.071 067 Ultrasound use improves the accuracy of surface electromyography placement over rectus femoris in children with cerebral palsy N. Blucher ∗ , A. Bass, G. Holmes, B. Mwaura Kimani, D. Trinca Alder Hey Children’s Hospital, Liverpool, England, United Kingdom Introduction and aim: Rectus femoris (RF) transfer to Sartorius is one treatment option for children with Cerebral Palsy (CP) who have a stiff knee gait [1], therefore assessment of RF function is important in selecting patients for this procedure. The assessment of RF depends on the quality of the signal recorded from sEMG electrodes which in turn depends on the position of the electrode on the skin overlying the muscle. The SENIAM recommendations for sEMG electrode placement are widely accepted, but have only been validated in healthy adult subjects [2]. The aim of this study was to validate the SENIAM recommendations for surface electromyography placement (sEMG) over rectus femoris (RF) muscle in healthy children and in children with cerebral palsy(CP) during gait analysis and compare placement using these guidelines to using ultrasonography. Patients/materials and methods: The study included 10 healthy children volunteers and 10 CP children volunteers, aged 8–12 years. All the children attended the same centre for gait analysis. Ethical approval was gained and parental consent was sought before the enrolment into the study. All the CP children had spastic diplegia, were GMFCS levels I–II, had not previously undergone surgery and were able to cooperate and follow instructions. Routine laboratory protocol for fully instrumented 3D gait analysis was carried out, including a standardized physical examination. RF electrodes were placed following SENIAM recommendations: midpoint between the anterior inferior iliac spine and the superior surface of the patella. This skin at this point was marked. RF was then identified by ultrasound. The medial and lateral edges of RF was identified and marked on the skin in the same horizontal plane as the previous sEMG electrode position. The distance between the lateral edge of RF and the position of the sEMG electrode as per SENIAM guidelines and the width of RF was measured, to the nearest millimetre and recorded. Results: We considered ‘ideal electrode’ position to be at halfway between the edges of RF (i.e 50%). The mean percentage difference in distance from the ‘ideal electrode’ position as measured
by ultrasound to electrode placement following SEMIAN guidelines was 2.7% in the healthy children group compared with 19.5% in the CP group. By performing unpaired t tests we showed that there was no significant difference between the mean electrode position using SEMIAN guidelines and ‘ideal electrode’ position in the healthy children (p = 0.0531), however the mean electrode position using SEMIAN guidelines in the CP patients was significantly different from the ‘ideal electrode’ position (p = 0.0001). Discussion and conclusions: SENIAM recommendations for sEMG electrode placement over RF muscle were validated in 10 healthy children. We showed that ultrasonography improves the accuracy of sEMG electrode placement in children with CP, who can exhibit anatomical variation due to their condition. Accurate electrode placement will ensure that a more accurate signal is recorded which may have a direct clinical bearing on the decision to proceed with surgical intervention. Reference [1] Gage JR, Perry J, Hicks RR, Koop S, Werntz JR. Rectus femoris transfer to improve knee function of children with cerebral palsy. Dev Med Child Neurol 1987;29(2):159–66. [2] Hermens HJ, Freriks B, Disselhorst-Klug C, Rau G. Development of recommendations for SEMG sensors and senson placement procedures. J Electromyogr Kinesiol 2000;10:361–74.
http://dx.doi.org/10.1016/j.gaitpost.2014.04.072 068
Session 4C: Orthopaedics 2
Outcome predictors for multilevel surgery in children with cerebral palsy/spastic diplegia I. Hinden ∗ , J. Stebbins, N. Thompson, T. Theologis Oxford Gait Lab, Nuffield Orthopaedic Centre, OUH NHS Trust, UK Introduction and aim: Single event multilevel surgery (SEMLS) is commonly performed to correct orthopaedic deformities in cerebral palsy. Although the outcome has been shown to be positive overall [1], a significant element of unpredictability is still present for individual cases. In order to improve individual outcomes, it would be helpful to determine which subject and surgical characteristics are associated with improvements following surgery. The aim of this study was therefore to perform a retrospective analysis on all patients who had undergone multi-level surgery at our institution, to determine the relationship between selected characteristics and functional outcome using gait analysis. Patients/materials and methods: A retrospective cohort study was conducted on data collected in the Gait Lab between 1994 and 2012. Subjects included 160 spastic diplegic cerebral palsy patients