- Email: [email protected]
Quantifying lateral pelvic displacement during walking
Clinical Biomechanics13(1998)371-373
ELSEVIER
Brief report Quantifying lateral pelvic displacement during walking K.J. Dodd,“* T.V. Wrigley,h P.A. Goldie,” M.E. Morris,“c C.D. Grant” “School of Physiotherapy, La Trohe University, Bundoora, 3083, Vie., Australia “Centre for Rehabilitation, Exercise and Sport Science, Victoria University of Technology, Melbourne, Vie. 3000. Australia ‘Kingston Centre, Cheltenham, 3192, Australia
Received11September1996;accepted18December1997
Abstract Objective. The purpose of this investigation was to test a new procedure for quantifying lateral pelvic displacement during
walking. Design. A quasi-experimental design was used to quantify the gait of 18 unimpaired people and one person with hemiplegia. Background. Although previous techniques provided useful information on amplitude of lateral pelvic displacement, they did not consider step-to-step variations in walking direction or enable quantification of symmetry. Methods. Three-dimensional motion analysis was used to collect the coordinates of light-reflective markers placed on the scarum and heels of each subject. Subjects performed one 10 m overground walk at their preferred speed. Amplitude and symmetry of lateral pelvic displacement were quantified relative to the step-to-step variation in the path of motion (base of support). Results. The mean amplitude of lateral pelvic displacement for the unimpaired group was 40.8 mm, and symmetry was 3.1 mm. The amplitude of lateral pelvic displacement for the hemiplegic person was 88.4 mm. Symmetry was 30.9 mm, with deviation toward the non-paretic side. Conclusion. The new procedure provided information on the amplitude and symmetry of lateral pelvic displacement in unimpaired adults and was sensitive to deviations of a person with a walking abnormality.
Relevance Treatment of atypical lateral pelvic displacement is frequently an aim of stroke rehabilitation. Therefore, it is important to have objective, accurate methods of quantification. 0 1998 Elsevier Science Ltd. All rights reserved. Kqwords:
Gait; Locomotion;Pelvis;Hemiplegia
1. Introduction
Lateral displacement of the pelvis is an integral feature of human locomotion [l]. To maintain balance, the pelvis and u-pper body normahy move from sideto-side. The side-to-side motion of the pelvis is termed lateral pelvic displacement (LPD). In pathologies such as hemiplegic stroke, the amplitude and symmetry of LPD are frequently atypical on clinical examination [2,3]. Treatment of LPD disorders has been hampered by the absence of practical objective methods to quantify changes in performance. * Correspondenceand reprint requeststo: K.J. Dodd.
Existing methods of measuring LPD were problematic because they: (1) did not consider the lateral stepto-step variations in the walking direction, thereby precluding quantification of LPD in relation to the base of support [4-61; (2) did not permit quantification of symmetry of LPD [4-61; and (3) used fixed laboratory coordinate systemsto define the three-dimensional (3-D) axes of pelvic motion. Therefore, the effect of deviations in walking direction from the laboratory fixed axes of motion on measurement of frontal plane pelvic motion were not considered [4-61. The purpose of this investigation was to describe and test a new procedure for quantifying amplitude and symmetry of LPD which addressed the above issues.
0268-0033/98/$19.00 t- 0.000 1998ElsevierScienceLtd. All rights reserved PII: SO268-0033(98)00101-0
372
K.J. Dodd et aLlClinical Biomechanics 13 (1998) 371-373
2. Methods Seven men and 11 women without history of neurological or orthopaedic conditions known to affect walking were recruited. Mean age was 26.5 yr (range 1.55-1.95 m, SD = 0.11 m) and mean height was 1.71 m (range 1.55-1.95, SD = 0.11). A 66-year-old male with a left hemiplegia was also measured. He clinically demonstrated increased amplitude and symmetry of LPD. A PEAK Performance Technologies motion measurement system (Englewood, Colorado, USA) was used to reconstructed the 3-D motion of the body. The PEAK tracked the trajectory of 30 mm, passive, lightreflective markers attached to the skin overlying the sacrum at the level of S2 and to each heel. Three genlocked Panasonic PAL F15 video-cameras, with shutter speeds set at l/250 s, captured images at 50 fields per second. Camera one was placed 5.5 m directly behind the centre of the recording area. Cameras two and three were placed 6.5 m and 7.1 m from the centre of the recording area, forming a 71” angle to each other. Pressure-sensitive microswitches were attached to the soles of the feet to determine timing of initial foot contact. Subjects were instructed to ‘walk to the end of the walkway at a comfortable pace’. Data was collected from the calibrated middle 3.5 m length of the 10 m walkway. The digitised 3-D positions of the markers were smoothed using a Butterworth fourth-order, zero-lag, recursive, low-pass filter. Set at an optimal cut-off frequency using the procedure described by Jackson [7]. The mean root-mean-square calibration error was 3.0 mm (range 2.0-5.8 mm, SD = 1.0 mm) in the x-dimension (forward), 1.6 mm (range 0.8-2.2 mm, SD = 0.4 mm) in the y-dimension (vertically upwards), and 2.3 mm (range 1.7-3.9 mm, SD = 0.6 mm) in the z-dimension (lateral) (based on the direct linear transformation method). To assessthe difference between the traditional and new methods, the amplitude of LPD was calculated using both methods. The traditional method of quantifying sacral marker displacement in relation to the laboratory-fixed 3-D coordinate system was termed gross LPD. Amplitude was the mean of gross LPD peak-to-peak z-displacements (Fig. 1). Using the new method, the 3-D axes of motion were related to an individual’s general walking direction. In software, the x-axis of the 3-D coordinate system was rotated coincident with the straight-line connecting the mid-position between the heel markers at the first initial foot contact and the last recorded initial foot contact. This line was termed the direction of walking (Fig. 2). The step-to-step variation in the path of motion was determined by calculating the mid-point between the
(b) 401
-
Net lateral pelvic displacement -
Gmss lateral pelvic displacement
(S2 marker)
Fig. 1. (a,b) Amplitude of LPD for two unimpaired subjects derived using the new method (net LPD) and the traditional method (gross LPD).
two heels at each initial foot contact, and linearly interpolating the points between these mid-point coordinates. To calculate the magnitude of the step-to-step variation in the path of motion, the z-distance of the mid-point between the heels from the direction of walking was determined at each initial foot contct. For each step, this z-position was subtracted from the z-coordinate of the mid-point beween the heels at the preceding initial foot contact. A mean absolute stepto-step variation in the path of motion was derived for each individual. Net LPD described LPD around the step-to-step variation in the path of motion (Fig. 2). This was calculated by subtracting the step-to-step variation
Net lateral pelvic displacement
-30 ---
Mean axis of net lafeml pelvic displacement
----)
Step-to-step
variation
in the path of motion
Fig. 2. Net LPD around the step-to-step variation in the path of motion, mean axis of net LPD, and symmetry of LPD in an unimpaired person.
K.J. Dodd et al./Clinicul Biomrchnics
(z-value) from the gross LPD (z-value). Amplitude was the mean of the peak-to-peak net LPD values. The symmetry of LPI) was the z-difference (millimetres) between the step-to-step variation in the path of motion and the mean value (axis) of net LPD (Fig. 2). 3. Results
For the unimpaired group, the means for amplitude using the new method (40.8 mm, SD = 12.9 mm) and the traditional method (40.9 mm, SD = 13.3 mm) were very similar, and no systematic difference between methods was detected (tu7)= -0.17; p = 0.86). However, a random difference between the methods was found on inspection of individual scores. The mean difference between the methods (0.1 mm) and the standard deviation of the difference (3.5 mm) were used to calculate the 95% confidence interval, and the limits of agreement extended from -6.8 to 7.0 mm. The group mean step-to-step variation in the path of motion was 13.5 mm (SD = 5.0 mm). The group mean symmetry of LPD was 3.1 mm (SD = 7.8 mm). For the person with hemiplegia, the amplitude was 88.4 mm using the new method and 84.4 mm using the traditional method. Symmetry was 30.9 mm with deviation towards the non-paretic side. The step-to-step variation in the path of motion score was 18.4 mm. 4. Discussion
The new method enabled rotation of the 3-D axes of motion to correspond to each individual’s walking direction, rather than quantifying LPD using static laboratory-fixed Icoordinate systems [4-61. With fixed systems, pelvic out-of-plane motion can cause inaccuracy of quantification of frontal plane (z-axis) pelvic displacement. The new method enabled more accurate quantification of LPD. The development of an internal reference point (the step-to-step variation in the path of motion) was an advance which enabled quantification of symmetry and amplitude of LPI) in relation to the base of support. It is important to control for step-to-step variation in the
13 (1998) 371-373
373
path of motion because variations introduce measurement noise, which may prevent the accurate quantification of LPD. The group mean step-to-step variation (13.5 mm) was relatively large in proportion to the group mean and range of amplitude and symmetry scores (amplitude: mean = 40.8 mm, range = 27.274.1 mm; symmetry: mean = 3.1 mm, range = -7.519.3 mm). Thus, the proportion of measurement noise to genuine score may be large. The results using the two methods differed by up to 7.0 mm, and the limits of agreement extended from -6.8 to 7.0 mm. These results indicate the effect that variations in the path of motion may have on the accurate quantification of LPD. The new method also permitted quantification of symmetry. The results showed that people normally have symmetrical patterns of LPD. In contrast, the person with hemiplegia showed asymmetry with a relatively large deviation towards the non-paretic side (30.9 mm). The group mean amplitude of LPD was within the 40-50 mm range reported in previous studies [4,5], confirming that normal amplitude of LPD is relatively small. The person with hemiplegia demonstrated a larger (88.4 mm) amplitude of LPD. Whether this is typical of the hemiplegic population awaits further investigation. References
[II
lnman VT, Ralston HJ, Todd F. Human Walking. Baltimore: Williams and Wilkins, 1981. PI Carr JH, Shepherd RB. A Motor Relearning Programme for Stroke. 2nd ed. London: Heinemann Physiotherapy, 1987. (31 Bobath B. Adult Hemiplegia: Evaluation and Treatment. 3rd ed. London: Heinemann Medical Books, 1990. 141 Stokes VP, Andersson C, Forssberg H. Rotational and translational movement features of the pelvis and thorax during adult human locomotion. Journal of Biomechanics 1988;22:( 1):43-50. [51 Waters RL, Morris J, Perry J. Translational motion of the head and trunk during normal *alking. Journal of Biomechanics 1973;6:167-172. PI Zarrugh MY, Radcliffe CW. Computer generation of human gait kinematics. Journal of Biomechanics 1979;12:99-111. [71 Jackson KM. Fitting of mathematical functions in biomechanical data. IEEE Transactions Biomed Eng 1979;26(2): 122-124.
ELSEVIER
Brief report Quantifying lateral pelvic displacement during walking K.J. Dodd,“* T.V. Wrigley,h P.A. Goldie,” M.E. Morris,“c C.D. Grant” “School of Physiotherapy, La Trohe University, Bundoora, 3083, Vie., Australia “Centre for Rehabilitation, Exercise and Sport Science, Victoria University of Technology, Melbourne, Vie. 3000. Australia ‘Kingston Centre, Cheltenham, 3192, Australia
Received11September1996;accepted18December1997
Abstract Objective. The purpose of this investigation was to test a new procedure for quantifying lateral pelvic displacement during
walking. Design. A quasi-experimental design was used to quantify the gait of 18 unimpaired people and one person with hemiplegia. Background. Although previous techniques provided useful information on amplitude of lateral pelvic displacement, they did not consider step-to-step variations in walking direction or enable quantification of symmetry. Methods. Three-dimensional motion analysis was used to collect the coordinates of light-reflective markers placed on the scarum and heels of each subject. Subjects performed one 10 m overground walk at their preferred speed. Amplitude and symmetry of lateral pelvic displacement were quantified relative to the step-to-step variation in the path of motion (base of support). Results. The mean amplitude of lateral pelvic displacement for the unimpaired group was 40.8 mm, and symmetry was 3.1 mm. The amplitude of lateral pelvic displacement for the hemiplegic person was 88.4 mm. Symmetry was 30.9 mm, with deviation toward the non-paretic side. Conclusion. The new procedure provided information on the amplitude and symmetry of lateral pelvic displacement in unimpaired adults and was sensitive to deviations of a person with a walking abnormality.
Relevance Treatment of atypical lateral pelvic displacement is frequently an aim of stroke rehabilitation. Therefore, it is important to have objective, accurate methods of quantification. 0 1998 Elsevier Science Ltd. All rights reserved. Kqwords:
Gait; Locomotion;Pelvis;Hemiplegia
1. Introduction
Lateral displacement of the pelvis is an integral feature of human locomotion [l]. To maintain balance, the pelvis and u-pper body normahy move from sideto-side. The side-to-side motion of the pelvis is termed lateral pelvic displacement (LPD). In pathologies such as hemiplegic stroke, the amplitude and symmetry of LPD are frequently atypical on clinical examination [2,3]. Treatment of LPD disorders has been hampered by the absence of practical objective methods to quantify changes in performance. * Correspondenceand reprint requeststo: K.J. Dodd.
Existing methods of measuring LPD were problematic because they: (1) did not consider the lateral stepto-step variations in the walking direction, thereby precluding quantification of LPD in relation to the base of support [4-61; (2) did not permit quantification of symmetry of LPD [4-61; and (3) used fixed laboratory coordinate systemsto define the three-dimensional (3-D) axes of pelvic motion. Therefore, the effect of deviations in walking direction from the laboratory fixed axes of motion on measurement of frontal plane pelvic motion were not considered [4-61. The purpose of this investigation was to describe and test a new procedure for quantifying amplitude and symmetry of LPD which addressed the above issues.
0268-0033/98/$19.00 t- 0.000 1998ElsevierScienceLtd. All rights reserved PII: SO268-0033(98)00101-0
372
K.J. Dodd et aLlClinical Biomechanics 13 (1998) 371-373
2. Methods Seven men and 11 women without history of neurological or orthopaedic conditions known to affect walking were recruited. Mean age was 26.5 yr (range 1.55-1.95 m, SD = 0.11 m) and mean height was 1.71 m (range 1.55-1.95, SD = 0.11). A 66-year-old male with a left hemiplegia was also measured. He clinically demonstrated increased amplitude and symmetry of LPD. A PEAK Performance Technologies motion measurement system (Englewood, Colorado, USA) was used to reconstructed the 3-D motion of the body. The PEAK tracked the trajectory of 30 mm, passive, lightreflective markers attached to the skin overlying the sacrum at the level of S2 and to each heel. Three genlocked Panasonic PAL F15 video-cameras, with shutter speeds set at l/250 s, captured images at 50 fields per second. Camera one was placed 5.5 m directly behind the centre of the recording area. Cameras two and three were placed 6.5 m and 7.1 m from the centre of the recording area, forming a 71” angle to each other. Pressure-sensitive microswitches were attached to the soles of the feet to determine timing of initial foot contact. Subjects were instructed to ‘walk to the end of the walkway at a comfortable pace’. Data was collected from the calibrated middle 3.5 m length of the 10 m walkway. The digitised 3-D positions of the markers were smoothed using a Butterworth fourth-order, zero-lag, recursive, low-pass filter. Set at an optimal cut-off frequency using the procedure described by Jackson [7]. The mean root-mean-square calibration error was 3.0 mm (range 2.0-5.8 mm, SD = 1.0 mm) in the x-dimension (forward), 1.6 mm (range 0.8-2.2 mm, SD = 0.4 mm) in the y-dimension (vertically upwards), and 2.3 mm (range 1.7-3.9 mm, SD = 0.6 mm) in the z-dimension (lateral) (based on the direct linear transformation method). To assessthe difference between the traditional and new methods, the amplitude of LPD was calculated using both methods. The traditional method of quantifying sacral marker displacement in relation to the laboratory-fixed 3-D coordinate system was termed gross LPD. Amplitude was the mean of gross LPD peak-to-peak z-displacements (Fig. 1). Using the new method, the 3-D axes of motion were related to an individual’s general walking direction. In software, the x-axis of the 3-D coordinate system was rotated coincident with the straight-line connecting the mid-position between the heel markers at the first initial foot contact and the last recorded initial foot contact. This line was termed the direction of walking (Fig. 2). The step-to-step variation in the path of motion was determined by calculating the mid-point between the
(b) 401
-
Net lateral pelvic displacement -
Gmss lateral pelvic displacement
(S2 marker)
Fig. 1. (a,b) Amplitude of LPD for two unimpaired subjects derived using the new method (net LPD) and the traditional method (gross LPD).
two heels at each initial foot contact, and linearly interpolating the points between these mid-point coordinates. To calculate the magnitude of the step-to-step variation in the path of motion, the z-distance of the mid-point between the heels from the direction of walking was determined at each initial foot contct. For each step, this z-position was subtracted from the z-coordinate of the mid-point beween the heels at the preceding initial foot contact. A mean absolute stepto-step variation in the path of motion was derived for each individual. Net LPD described LPD around the step-to-step variation in the path of motion (Fig. 2). This was calculated by subtracting the step-to-step variation
Net lateral pelvic displacement
-30 ---
Mean axis of net lafeml pelvic displacement
----)
Step-to-step
variation
in the path of motion
Fig. 2. Net LPD around the step-to-step variation in the path of motion, mean axis of net LPD, and symmetry of LPD in an unimpaired person.
K.J. Dodd et al./Clinicul Biomrchnics
(z-value) from the gross LPD (z-value). Amplitude was the mean of the peak-to-peak net LPD values. The symmetry of LPI) was the z-difference (millimetres) between the step-to-step variation in the path of motion and the mean value (axis) of net LPD (Fig. 2). 3. Results
For the unimpaired group, the means for amplitude using the new method (40.8 mm, SD = 12.9 mm) and the traditional method (40.9 mm, SD = 13.3 mm) were very similar, and no systematic difference between methods was detected (tu7)= -0.17; p = 0.86). However, a random difference between the methods was found on inspection of individual scores. The mean difference between the methods (0.1 mm) and the standard deviation of the difference (3.5 mm) were used to calculate the 95% confidence interval, and the limits of agreement extended from -6.8 to 7.0 mm. The group mean step-to-step variation in the path of motion was 13.5 mm (SD = 5.0 mm). The group mean symmetry of LPD was 3.1 mm (SD = 7.8 mm). For the person with hemiplegia, the amplitude was 88.4 mm using the new method and 84.4 mm using the traditional method. Symmetry was 30.9 mm with deviation towards the non-paretic side. The step-to-step variation in the path of motion score was 18.4 mm. 4. Discussion
The new method enabled rotation of the 3-D axes of motion to correspond to each individual’s walking direction, rather than quantifying LPD using static laboratory-fixed Icoordinate systems [4-61. With fixed systems, pelvic out-of-plane motion can cause inaccuracy of quantification of frontal plane (z-axis) pelvic displacement. The new method enabled more accurate quantification of LPD. The development of an internal reference point (the step-to-step variation in the path of motion) was an advance which enabled quantification of symmetry and amplitude of LPI) in relation to the base of support. It is important to control for step-to-step variation in the
13 (1998) 371-373
373
path of motion because variations introduce measurement noise, which may prevent the accurate quantification of LPD. The group mean step-to-step variation (13.5 mm) was relatively large in proportion to the group mean and range of amplitude and symmetry scores (amplitude: mean = 40.8 mm, range = 27.274.1 mm; symmetry: mean = 3.1 mm, range = -7.519.3 mm). Thus, the proportion of measurement noise to genuine score may be large. The results using the two methods differed by up to 7.0 mm, and the limits of agreement extended from -6.8 to 7.0 mm. These results indicate the effect that variations in the path of motion may have on the accurate quantification of LPD. The new method also permitted quantification of symmetry. The results showed that people normally have symmetrical patterns of LPD. In contrast, the person with hemiplegia showed asymmetry with a relatively large deviation towards the non-paretic side (30.9 mm). The group mean amplitude of LPD was within the 40-50 mm range reported in previous studies [4,5], confirming that normal amplitude of LPD is relatively small. The person with hemiplegia demonstrated a larger (88.4 mm) amplitude of LPD. Whether this is typical of the hemiplegic population awaits further investigation. References
[II
lnman VT, Ralston HJ, Todd F. Human Walking. Baltimore: Williams and Wilkins, 1981. PI Carr JH, Shepherd RB. A Motor Relearning Programme for Stroke. 2nd ed. London: Heinemann Physiotherapy, 1987. (31 Bobath B. Adult Hemiplegia: Evaluation and Treatment. 3rd ed. London: Heinemann Medical Books, 1990. 141 Stokes VP, Andersson C, Forssberg H. Rotational and translational movement features of the pelvis and thorax during adult human locomotion. Journal of Biomechanics 1988;22:( 1):43-50. [51 Waters RL, Morris J, Perry J. Translational motion of the head and trunk during normal *alking. Journal of Biomechanics 1973;6:167-172. PI Zarrugh MY, Radcliffe CW. Computer generation of human gait kinematics. Journal of Biomechanics 1979;12:99-111. [71 Jackson KM. Fitting of mathematical functions in biomechanical data. IEEE Transactions Biomed Eng 1979;26(2): 122-124.