- Email: [email protected]
Hip joint center location from palpable bony landmarks—A cadaver study
Pergamon
J. Biomechanics, Vol. 28, No. 8, pp. 995~ 998. 1995 Copyright C 1995 Elsevier Science Ltd Printed in Great Bntam. All r~ghtd reamed CO21 9290/95 $9.50 + Oil
0021-9290(94)00149-9
TECHNICAL
NOTE
HIP JOINT CENTER LOCATION FROM PALPABLE LANDMARKS-A CADAVER STUDY
BONY
Geoffrey K. Seidel,* David M. Marchinda,t Marcel Dijkers*$ and Robert W. Soutas-Littlepi *Department of Physical Medicine and Rehabilitation, Wayne State University, Detroit, Michigan; tEast Lansing, Michigan; SRehabilitation Institute of Michigan, Detroit, Michigan; §Biomechanics Evaluation Laboratory, Michigan State University, East Lansing, Michigan, U.S.A. Abstract-An anatomical anthropometric study of adult human cadaveric pelves was performed to investigate the relationship between hip joint center (HJC) and selected aspects of pelvic geometry. Sixty-five pelves (35 female and 30 male) were examined. Measurements of pelvic geometry and HJC (center of bony acetabular rim) were taken from bony landmarks of de-fleshed pelves. Correlation analysis revealed that HJC cannot be accurately located as a function of pelvic width alone, but requires estimation as a function of pelvic height and depth as well. HJC was optimally located relative to the respective ASIS: 14% of pelvic width medial (mean error 0.58 cm), 34% of pelvic depth posterior (mean error 0.30 cm), and 79% of pelvic height inferior (mean error 0.35 cm). No significant differences were found between males and females in HJC estimation. Keywords: Hip center; Pelvic landmarks. INTRODUCTION
The location of hip joint center (HJC) is important in defining the kinematics and kinetics of the hip during gait. Small errors in locating HJC result in large errors in joint contact force (Crowninshield et al., 1977) and muscle force generating capacity (Delp and Maloney, 1993) computations. Various techniques are utilized to estimate the location of HJC in three dimensions. Previous efforts to locate HJC including the use of moire fringes (Ellis et al., 1979), intercepts of helical axes (Blankevoort et al., 1990), and the use of sphere centers (Cappozzo, 1984) are not applicable with limitations of joint range of motion in many clinical settings. Crowninshield et al. (1978) utilized radiographic analysis; however, the risks of radiation exposure as well as technical difficulties related to conversion of planar images into coordinate data makes this technique undesirable in routine gait analysis. Palpable bony landmarks are commonly utilized to estimate HJC for routine gait analysis as proposed by Andriacchi et al. (1980) and Tylkowski et al. (1982). Bell et al. (1989,199O) found that HJC estimation as fixed proportions of pelvic width minimized errors in HJC estimation. Our objective was to establish if the error of HJC estimation along each axis is minimized with fixed percentages of a single pelvic parameter (pelvic width) or if other pelvic measurements are needed to estimate more precisely HJC. Additionally, do gender differences in pelvic shape effect HJC estimation? METHODS
Human cadavers at Wayne State University School of Medicine, Detroit, Michigan from April 1991 through JanReceived in final form 30 October 1994. Address correspondence to: Geoffrey K. Seidel, M.D., 590 Shenandoah Dr. Clawson, MI 48017-1346, U.S.A.
uary 1993 were examined after anatomical dissection. Cadaveric pelvic specimens were excluded from the study if there were gross pelvic abnormalities or acetabular surgery. Sixty-five cadavers (35 female and 30 male), 122 sides, were examined. The ages ranged from 53 to 99 yr (mean 75.1, SD. 12.6), representing skeletally mature adults. Pelves were removed from the cadavers and de-fleshed to allow accurate measurement of bony landmarks. Each pelvis was placed on a flat surface with both ASISs and the pubis in contact with a data recording sheet. The points of contact of both the ASISs and the pubis were traced on the data record sheet to ensure the pelvis did not move during the pelvic measurement procedure. The frontal plane was defined as the plane passing through both ASISs and the pubic symphysis. The coordinate system was defined with its origin at the respective ASIS side being measured: y-axis mediolateral (positive medial-defined by ASISs), z-axis superoinferior (positive inferior), and x-axis anteroposterior (positive posterior). All measurements were made, with a millimeter ruler, from bony surfaces. Measurements were made of acetabular depth (ADEP) and diameter (ADIA) to ensure that the acetabulum could be considered a hemisphere and HJC was defined as the center of the acetabular rim (labrum removed). The acetabular depth was measured to be on average 49% of the acetabular diameter. This supported the assumption that the acetabulum is a hemisphere and that the location of HJC can be defined as the center of the acetabular rim. HJC-x was defined as the perpendicular distance from the frontal plane to the center of the bony acetabular rim. The position of HJC-x was noted on the frontal plane, the pelvis was lifted and the specific contact points of each ASIS and the pubis were noted within the traced areas of the data recording sheet. Other pelvic measurements included oelvic width (ASIS-ASIS), pelvic height (perpendicular from pubic center to the inter-ASIS line). ASIS to oubic center (ING). HJC to Adductor magnus origin (HJC-a’) ASIS and pubic center to the intersection of HJC-x with the frontal plane (AS-x and PS-x, respectively). Pelvic depth was measured. with a cali995
Technical Note
996
per accurate to a millimeter, in an oblique fashion from the ASIS to the posterior superior iliac spine (PSIS). The pelvic depth (PD) measurement was added to the measurement protocol after a preliminary analysis (N = 30) of the relationship between PW and HJC-x. With the ASIS location known, the HJC-x projection of the HJC onto the frontal plane, and the pubic center (PC) defined as the center of the pubic symphysis. the HJC-y (medial measurement) and the HJC-z (height measurement) were calculated. The first author made all measurements reported. lnterrater reliability was assessedwith the assistance of two additional measurers who were instructed in the measurement methodology and then each allowed to select a group of pelvic specimens at random to reproduce all measurements (measurer 1: 20 sides and measurer 2: 17 sides). Mean intermeasurer differences were less than one millimeter for each measurement. Data plots failed to reveal any significant systematic or random error. Intraclass correlation coefficients for the pelvic measurements were high indicating a reliable measurement procedure (Table 1). Randomly selected hips from all complete pelves and incomplete pelves were statistically analyzed. No violations of normality were observed, and outlier data were included in the anlayses. Female pelves on average were significantly but proportionally smaller than male pelves in all measurements except pelvic width (PW), HJC-y and PS-x. HJC location as a function of pelvic geometry was similar for both genders. Mean pelvic geometric parameters (HJC-y/PW, HJC-x/PW, HJC-z/PW, HJC-x/pelvic depth (PD), and HJC-z/pelvic height (PH)) for males were within one percent of values of females, and none of the differences was statistically significant (Student’s t-test, minimum N =35). Therefore, pelvic measurement data for the two genders were pooled for further analysis. Measured distances from the ASIS to ipsilateral HJC along x, y, and z axes were divided by PW (ASIS separation), PD. and PH and were correlated. The absolute values of differences between estimated and true HJC coordinate measurements were calculated and analyzed with paired Student t-tests to assess the relative accuracy of various methods of estimating HJC location.
Table 1. Descriptive statistics of pelvic measurements (in centimeters upper panel and percent lower panel) of randomly selected hips of pooled male and female data with mean. standard deviation (S.D.), number of cases, and intraclass correlation coefficients (ICC). ICC, a measure of interrater reliability, were significant for all variables listed (p < 0.001). Algorithm ICC (1, 1) was used in all cases(With 3 measurers and 37 sides), except for the values marked with “*“. which were determined using algorithm ICC (2, 1) (with 2 measurers and 17 sides). Refer to the methods for the meaning of abbreviations -Mean SD. Number ICC PW PH PD ADIA ADEP HJC-a ING
HJC-x HJC-y HJC-z PC-X AS-x HJC-s/PW HJC-y/PW HJC-z/PW HJC-z/PH HJC-x/PD
23.8 9.0 16.4 4.9
1.7 0.8 0.9 0.4
65 65 35 52
0.99 0.99 0.99* 0.95
2.4 8.2
0.3 0.8
0.90 0.93 0.98 0.96 0.93 0.91 0.90 0.88 0.97 0.90 0.92 0.76 0.92*
15.0
0.9
5.1 3.4 8.8 8.0
0.5 0.9 0.8 0.6 0.7
52 60 65 65 65 65 65 65
24% 14% 30% 79% 34%
3% 3% 4% 5% 2%
65 65 65 65 35
7.1
(N =65, NS, Fig. 2(A)). These correlations suggest that HJCx and HJC-z do not vary predictably with pelvic width and thus cannot be accurately located as a function of PW. No significant relationship was found between PW and PH (r=0.12, N=65) or PH and PD (r=0.18, N=35). PW correlated with PD r=0.52 (N= 35, p < 0.001). HJC-x correlated with PD at r=0.54 (N =35, p=O.OOOl, Fig. l(B)) demonstrating a linear relationship between HJC-x and PD. HJC-z RESULTS strongly correlated-with PH at I =0.81 (N = 65, p < 0.0001, Fig. 2(B)) demonstrating a linear relationship between HJC-z Correlation analysis demonstrated that HJC should not and PH. HJC-x and HJC-y were not significantly correlated be determined as a function of pelvic width alone. HJC-y with PH, and HJC-y and HJC-z were not significantly correstrongly correlated with PW at r = 0.85 (N = 65, p < O.OOOl), lated with PD. HJC-x did not correlate with PW at I= - 0.17 (N =65, NS, As a functions of PW alone, our study revealed that HJC Fig. l(A)), and HJC-z did not correlate with PW at I = 0.01 was located 14% (SD. 3%) of PW medially, 24% (S.D. 3%)
HJC-x WITH PELVIC WIDTH BY SEX
HJC-x WITH PELVIC DEPTH BY SEX ac
. MALE
A FEMALE
l
N=l22 10
19
20
21
22
22 PELWC
24
21
WIDTH
26 (em)
27
28
2,
MALE
A FEMALE
1 N-U
3 30
14
14.1
is
16.6
ia PELVIC
10.6 DEPTH
17 17.6 (cm)
IO
Fig. 1. Plots of HJC-x (all sides) with pelvic width (A) and pelvic depth (B) by sex in centimeters. HJC-x is not correlated with pelvic width but highly correlated with pelvic depth.
16.~
Technical Note of PW posteriorly, and 30% (S.D. 4%) of PW inferiorly relative to the ASIS. As a function of three pelvic parameters (width for HJC-y, height for HJC-z and depth for HJC-x), we found that HJC was located 14% (S.D. 3%) of PW medially, 34% (S.D. 2%) of PD posteriorly, and 79% (S.D. 5%) of PH inferiorly. The relative error in HJC estimation along each axis was minimized with HJC estimation from pelvic width for HJCy, pelvic height for HJC-z and pelvic depth for HJC-x (Table 2). Using the estimated location of HJC-x as 24% of PW resulted in a mean error of 0.49 cm (S.D. 0.34; max 1.49). The mean error using the location of HJC-x as 34% of PD yielded a mean error of 0.30 cm (S.D. 0.23; max 0.89). The calculation of HJC-x as a function of PD resulted in a smaller error (0.2 cm) than calculation of HJC-x as a function of PW (p < O.C001). Estimating HJC-z as a function of PW (30%) resulted in a mean error of 0.75 cm (S.D. 0.56, max 2.26), while calculation as a function of PH (79%) revealed a mean error of 0.35 cm (S.D. 0.28; max 1.29). The pelvic height method of calculating HJC-z reduced the mean error by a mean of 0.40 cm (p < 0.0001). Calculating HJC-y as a function of pelvic width (14%) yielded a mean error of 0.58 cm (S.D. 0.42; max 2.18). DISCUSSION
In gait analysis studies the accurate location of HJC is critical for precise computation of hip muscle moments, joint intersegmental resultant forces, and joint motion. A reliable technique for locating HJC is crucial to minimize the errors associated with HJC displacements from true position. Bell et al.‘s (1990) evaluation of Cappozzo’s rotational method concluded that the use of palpable bony landmarks is the most accurate and efficient way of determining HJC on a subject-specific basis. Bell et al. (1989,199O) estimated HJC as a constant proportion of PW for each axis. Our correlation analysis revealed that both HJC-x and HJC-z do not vary in a predictable linear relationship with PW (Fig. l(A) and 2(A)), while HJCq does strongly vary predictably with PW. These correlations indicate that errors are expected when locating HJC as a function of PW on any axis except the mediolateral axis. We found HJC to be located relative to ASK at 30% (S.D. 4%) of PW distally, 14% (S.D. 3%) of PW medially, and 24% (SD. 2%) of PW posteriorly. The differences in the posterior percentage as compared to the Bell et al. (1989, 1990) results could be due to a difference in sample size or the
Table 2. Errors in estimating HJC-x, HJC-z, and HJC-y resulting from various alogrithms. In all instances, the error was calculated as the absolute value of the specific estimate value minus the true value. Errors were minimized with estimation of HJC-x based on pelvic depth and HJC-z based on pelvic height Absolute value HJC algorithm Bell et HJC-x Bell et HJC-x
N
(1990) 19% PW al. (1989) 22% PW
1.19 0.51 0.32
2.19
(2)
65
0.58 0.44 0.03
1.51
Seidel et al. (1994) (3) HJC-x 24% PW
65
0.49 0.34 0.03
1.49
Seidel et al. (1994) (4) HJC-x 34% PD Bell et al. (1989, 1990) & Seidel et al. (1994) (5) HJC-z 30% PW
35
0.30 0.23 0.00
0.89
65
0.75 0.56 0.01
2.26
65
0.35 0.28 0.00
1.29
65
0.58 0.42 0.00
2.18
al.
Seidel et al. (1994) (6) HJC-z 80%PH Bell et al. (1989, 1900) & Seidel et al. (1994) HJC-y 14% PW
Comparisons of estimates using paired t-test. (2) compared to (1): p < 0.0001 (3) compared to (2): p < 0.08 (4) compared to (2): p < 0.0001 (4) compared to (3): p < 0.0001 (5) compared to (6): p < 0.0001
differences inherent in taking direct pelvic measurements vs radiographic measurements. The average percentages were, however, rather similar to Bell et al. (1989, 1990). Although this confirms an auerage location of HJC as a function of PW, it does not relate accurately on a subject-specific basis since there is no correlation of HJC-x and HJC-z with PW. Also, the mean error in location of HJC measurements using
HJC-z WITH PELVIC HEIGHT BY SEX
6 9.5
’ MALE
A FEMALE
l
9
MALE
. . ’ A.. .‘A.1 ;m. .
6
A :: 7
i
7.5
‘! Y I
6.5
1
6 FEMALE
6.5
g
Max
65
10
2 I’
Mean SD. Min
(1)
HJC-z WITH PELVIC WIDTH BY SEX
A
997
6 iA
5.5
5
a I?r
4 18
N=12:
Nr122 19
20
21
22
25 23 24 PELVIC WIDTH
26 (cm)
27
26
26
30
4.5
7
7.5
a
6.5 PELVIC
9 9.5 10 HEIGHT (cm)
10.5
Fig. 2. Plots of HJC-z (all sides) with pelvic width (A) and pelvic height (B) by sex ,in centimeters. HJC-z is not correlated with pelvic width but highly correlated with pelvic height.
il.l 11
998
Technical Note
PW alone was significantly higher than the error in locating HJC relative to pelvic geometric measurements. The calculation of HJC-x as a function of PD and HJC-z as a function of PH resulted in significantly smaller errors than the calculation of both HJC-x and HJC-z as a function of PW. The correlation between HJC-x and PD was r=0.54. Although this correlation was not as high as the HJC-y (r=O.85) and HJC-z (r=0.81) correlations, it is markedly higher than the correlation between HJC-x and PW (r= - 0.17). This ‘lower’ correlation of HJC-x and PD is most likely due to the fact that the measurement of PD (ASIS to PSIS) is not a true perpendicular PD measurement from the frontal plane. The apparent relationship between PW and PD with r =0.52, may reflect the oblique measurement of PD. Further study will need to determine which PD parameter (ASIS to PSIS vs the perpendicular distance from the frontal plane to the PSIS) will best estimate HJC-x. Since, in our study, the female pelvic measurements were proportionally smaller than those of males (consistent with Brinkmann et al., 1981) locating HJC as a function of pelvic geometric measurements can be performed similarly for both genders. Measurements were taken only on skeletally mature adults; therefore, the applicability of our results to HJC location in children is unknown. In gait analysis procedures, current pelvic targeting protocols utilize spherical reflective markers on the ASISs and the sacrum. In oivo assessment of HJC-y is based upon the distance between the centers of each ASIS marker. HJC-x may be obtained by establishing the distance between the centers of equal sized spherical markers placed on the ASIS and ipsilateral PSIS for the hip joint in question or with the use of obstetrical calipers to measure the distance from the PSIS to the ASIS to establish pelvic depth. The ratio of HJC-x with pelvic depth is assumed to remain constant as soft tissues add equally over the ASIS and PSIS. HJC-z may be obtained from pelvic height by measuring the distance from the ASIS to the public symphysis (the midpoint between the pubic tubercles) and the use of the mean of two position vectors from each ASIS as the midpoint. Soft tissue depth over the pubis is not a factor in pelvic height assessment. HJC estimation relative to the ASISs as a function PW alone results in significant deviations from the true HJC location. HJC estimation from pelvic width, height and depth parameters optimizes HJC- location relative-to true HJC. This study and Bell et al. (1989,199O) propose the same method to locate HJC along the mediolateral axis as 14% of pelvic width (SD. 3%) relative to the ASIS. However, in
contrast to Bell et al. (1989. 1990) this study proposes the location of HJC along the anteroposterior axis as 34% of pelvic depth (S.D. 2%) relative to the ASIS and the location of HJC along the superoinferior axis as 79% of pelvic height (S.D. 5%) relative to the ASIS. Acknowledgements-Rowe extend out appreciation to the Department of Anatomy and the Mortuary Staff, Wayne State University, School of Medicine for their assistance with this project. REFERENCES
Andriacchi, T. P., Andersson, G. B. J., Fermier, R. W., Stern. D. and Galante, J. 0. (1980) A study of lower-limb mechanics during stair climbing. J. Bone Jt Surg 62-A, 749-757. Bell, A. L., Brand, R. A. and Pedersen, D. R. (1989) Prediction of hip joint center location from external landmarks. Hum. Mvmt. Sci. 8, 3-16. Bell, A. L., Pedersen, D. R. and Brand, R. A. (1990) A comparison of the accuracy of several hip center location prediction methods. J. Biomechanics 23, 617-621. Blankevoort, L., Huiskes, R. and de Lange, A. (1990) Helical axes of passive knee joint motions. J. Biomechanics 23, 1219-1229. Brinckmann, P., Hoefert, H. and Jongen, H. Th. (1981) Sex differences in the skeletal geometry of the human pelvis and hip joint. J. Biomechanics 14, 427-430. Cappozzo, A. (1984) Gait analysis methodology. Hum. Mvmt.
Sci. 3, 27-50.
Crowninshield, R. D., Johnston, R. C., Andrews, J. G. and Brand, R. A. (1977) The effect of joint center on hip kinetics. Proc. 23rd Annual ORS Convention 47-49. Las Vegas, Nevada. Crowninshield, R. D., Johnston, R. C., Andrews, J. G. and Brand, R. A. (1978) A biomechanical investigation of the human hip. J. Biomechanics 11, 75-85. Delp, S. L. and Maloney, W. (1993) Effects of hip center location on the moment-generating capacity of the muscles. J. Biomechanics 26, 485-499. Ellis, M. I., Seedhom, B. B., Amis, A. A., Dowson, D. and Wright, V. (1979) Forces in the knee joint shilst rising from normal and motorized chairs. Engng. Med. 8, 3340. Tylkowski, C. M., Simon, S. R. and Mansour, J. M. (1982) Internal rotation gait in spastic cerebral palsy in the hip. Proc.
10th
Open
Scientijc
Meeting
of the Hip
Society
(Edited by Nelson, J. P.) pp. 89-125. Mosby. St. Louis.
J. Biomechanics, Vol. 28, No. 8, pp. 995~ 998. 1995 Copyright C 1995 Elsevier Science Ltd Printed in Great Bntam. All r~ghtd reamed CO21 9290/95 $9.50 + Oil
0021-9290(94)00149-9
TECHNICAL
NOTE
HIP JOINT CENTER LOCATION FROM PALPABLE LANDMARKS-A CADAVER STUDY
BONY
Geoffrey K. Seidel,* David M. Marchinda,t Marcel Dijkers*$ and Robert W. Soutas-Littlepi *Department of Physical Medicine and Rehabilitation, Wayne State University, Detroit, Michigan; tEast Lansing, Michigan; SRehabilitation Institute of Michigan, Detroit, Michigan; §Biomechanics Evaluation Laboratory, Michigan State University, East Lansing, Michigan, U.S.A. Abstract-An anatomical anthropometric study of adult human cadaveric pelves was performed to investigate the relationship between hip joint center (HJC) and selected aspects of pelvic geometry. Sixty-five pelves (35 female and 30 male) were examined. Measurements of pelvic geometry and HJC (center of bony acetabular rim) were taken from bony landmarks of de-fleshed pelves. Correlation analysis revealed that HJC cannot be accurately located as a function of pelvic width alone, but requires estimation as a function of pelvic height and depth as well. HJC was optimally located relative to the respective ASIS: 14% of pelvic width medial (mean error 0.58 cm), 34% of pelvic depth posterior (mean error 0.30 cm), and 79% of pelvic height inferior (mean error 0.35 cm). No significant differences were found between males and females in HJC estimation. Keywords: Hip center; Pelvic landmarks. INTRODUCTION
The location of hip joint center (HJC) is important in defining the kinematics and kinetics of the hip during gait. Small errors in locating HJC result in large errors in joint contact force (Crowninshield et al., 1977) and muscle force generating capacity (Delp and Maloney, 1993) computations. Various techniques are utilized to estimate the location of HJC in three dimensions. Previous efforts to locate HJC including the use of moire fringes (Ellis et al., 1979), intercepts of helical axes (Blankevoort et al., 1990), and the use of sphere centers (Cappozzo, 1984) are not applicable with limitations of joint range of motion in many clinical settings. Crowninshield et al. (1978) utilized radiographic analysis; however, the risks of radiation exposure as well as technical difficulties related to conversion of planar images into coordinate data makes this technique undesirable in routine gait analysis. Palpable bony landmarks are commonly utilized to estimate HJC for routine gait analysis as proposed by Andriacchi et al. (1980) and Tylkowski et al. (1982). Bell et al. (1989,199O) found that HJC estimation as fixed proportions of pelvic width minimized errors in HJC estimation. Our objective was to establish if the error of HJC estimation along each axis is minimized with fixed percentages of a single pelvic parameter (pelvic width) or if other pelvic measurements are needed to estimate more precisely HJC. Additionally, do gender differences in pelvic shape effect HJC estimation? METHODS
Human cadavers at Wayne State University School of Medicine, Detroit, Michigan from April 1991 through JanReceived in final form 30 October 1994. Address correspondence to: Geoffrey K. Seidel, M.D., 590 Shenandoah Dr. Clawson, MI 48017-1346, U.S.A.
uary 1993 were examined after anatomical dissection. Cadaveric pelvic specimens were excluded from the study if there were gross pelvic abnormalities or acetabular surgery. Sixty-five cadavers (35 female and 30 male), 122 sides, were examined. The ages ranged from 53 to 99 yr (mean 75.1, SD. 12.6), representing skeletally mature adults. Pelves were removed from the cadavers and de-fleshed to allow accurate measurement of bony landmarks. Each pelvis was placed on a flat surface with both ASISs and the pubis in contact with a data recording sheet. The points of contact of both the ASISs and the pubis were traced on the data record sheet to ensure the pelvis did not move during the pelvic measurement procedure. The frontal plane was defined as the plane passing through both ASISs and the pubic symphysis. The coordinate system was defined with its origin at the respective ASIS side being measured: y-axis mediolateral (positive medial-defined by ASISs), z-axis superoinferior (positive inferior), and x-axis anteroposterior (positive posterior). All measurements were made, with a millimeter ruler, from bony surfaces. Measurements were made of acetabular depth (ADEP) and diameter (ADIA) to ensure that the acetabulum could be considered a hemisphere and HJC was defined as the center of the acetabular rim (labrum removed). The acetabular depth was measured to be on average 49% of the acetabular diameter. This supported the assumption that the acetabulum is a hemisphere and that the location of HJC can be defined as the center of the acetabular rim. HJC-x was defined as the perpendicular distance from the frontal plane to the center of the bony acetabular rim. The position of HJC-x was noted on the frontal plane, the pelvis was lifted and the specific contact points of each ASIS and the pubis were noted within the traced areas of the data recording sheet. Other pelvic measurements included oelvic width (ASIS-ASIS), pelvic height (perpendicular from pubic center to the inter-ASIS line). ASIS to oubic center (ING). HJC to Adductor magnus origin (HJC-a’) ASIS and pubic center to the intersection of HJC-x with the frontal plane (AS-x and PS-x, respectively). Pelvic depth was measured. with a cali995
Technical Note
996
per accurate to a millimeter, in an oblique fashion from the ASIS to the posterior superior iliac spine (PSIS). The pelvic depth (PD) measurement was added to the measurement protocol after a preliminary analysis (N = 30) of the relationship between PW and HJC-x. With the ASIS location known, the HJC-x projection of the HJC onto the frontal plane, and the pubic center (PC) defined as the center of the pubic symphysis. the HJC-y (medial measurement) and the HJC-z (height measurement) were calculated. The first author made all measurements reported. lnterrater reliability was assessedwith the assistance of two additional measurers who were instructed in the measurement methodology and then each allowed to select a group of pelvic specimens at random to reproduce all measurements (measurer 1: 20 sides and measurer 2: 17 sides). Mean intermeasurer differences were less than one millimeter for each measurement. Data plots failed to reveal any significant systematic or random error. Intraclass correlation coefficients for the pelvic measurements were high indicating a reliable measurement procedure (Table 1). Randomly selected hips from all complete pelves and incomplete pelves were statistically analyzed. No violations of normality were observed, and outlier data were included in the anlayses. Female pelves on average were significantly but proportionally smaller than male pelves in all measurements except pelvic width (PW), HJC-y and PS-x. HJC location as a function of pelvic geometry was similar for both genders. Mean pelvic geometric parameters (HJC-y/PW, HJC-x/PW, HJC-z/PW, HJC-x/pelvic depth (PD), and HJC-z/pelvic height (PH)) for males were within one percent of values of females, and none of the differences was statistically significant (Student’s t-test, minimum N =35). Therefore, pelvic measurement data for the two genders were pooled for further analysis. Measured distances from the ASIS to ipsilateral HJC along x, y, and z axes were divided by PW (ASIS separation), PD. and PH and were correlated. The absolute values of differences between estimated and true HJC coordinate measurements were calculated and analyzed with paired Student t-tests to assess the relative accuracy of various methods of estimating HJC location.
Table 1. Descriptive statistics of pelvic measurements (in centimeters upper panel and percent lower panel) of randomly selected hips of pooled male and female data with mean. standard deviation (S.D.), number of cases, and intraclass correlation coefficients (ICC). ICC, a measure of interrater reliability, were significant for all variables listed (p < 0.001). Algorithm ICC (1, 1) was used in all cases(With 3 measurers and 37 sides), except for the values marked with “*“. which were determined using algorithm ICC (2, 1) (with 2 measurers and 17 sides). Refer to the methods for the meaning of abbreviations -Mean SD. Number ICC PW PH PD ADIA ADEP HJC-a ING
HJC-x HJC-y HJC-z PC-X AS-x HJC-s/PW HJC-y/PW HJC-z/PW HJC-z/PH HJC-x/PD
23.8 9.0 16.4 4.9
1.7 0.8 0.9 0.4
65 65 35 52
0.99 0.99 0.99* 0.95
2.4 8.2
0.3 0.8
0.90 0.93 0.98 0.96 0.93 0.91 0.90 0.88 0.97 0.90 0.92 0.76 0.92*
15.0
0.9
5.1 3.4 8.8 8.0
0.5 0.9 0.8 0.6 0.7
52 60 65 65 65 65 65 65
24% 14% 30% 79% 34%
3% 3% 4% 5% 2%
65 65 65 65 35
7.1
(N =65, NS, Fig. 2(A)). These correlations suggest that HJCx and HJC-z do not vary predictably with pelvic width and thus cannot be accurately located as a function of PW. No significant relationship was found between PW and PH (r=0.12, N=65) or PH and PD (r=0.18, N=35). PW correlated with PD r=0.52 (N= 35, p < 0.001). HJC-x correlated with PD at r=0.54 (N =35, p=O.OOOl, Fig. l(B)) demonstrating a linear relationship between HJC-x and PD. HJC-z RESULTS strongly correlated-with PH at I =0.81 (N = 65, p < 0.0001, Fig. 2(B)) demonstrating a linear relationship between HJC-z Correlation analysis demonstrated that HJC should not and PH. HJC-x and HJC-y were not significantly correlated be determined as a function of pelvic width alone. HJC-y with PH, and HJC-y and HJC-z were not significantly correstrongly correlated with PW at r = 0.85 (N = 65, p < O.OOOl), lated with PD. HJC-x did not correlate with PW at I= - 0.17 (N =65, NS, As a functions of PW alone, our study revealed that HJC Fig. l(A)), and HJC-z did not correlate with PW at I = 0.01 was located 14% (SD. 3%) of PW medially, 24% (S.D. 3%)
HJC-x WITH PELVIC WIDTH BY SEX
HJC-x WITH PELVIC DEPTH BY SEX ac
. MALE
A FEMALE
l
N=l22 10
19
20
21
22
22 PELWC
24
21
WIDTH
26 (em)
27
28
2,
MALE
A FEMALE
1 N-U
3 30
14
14.1
is
16.6
ia PELVIC
10.6 DEPTH
17 17.6 (cm)
IO
Fig. 1. Plots of HJC-x (all sides) with pelvic width (A) and pelvic depth (B) by sex in centimeters. HJC-x is not correlated with pelvic width but highly correlated with pelvic depth.
16.~
Technical Note of PW posteriorly, and 30% (S.D. 4%) of PW inferiorly relative to the ASIS. As a function of three pelvic parameters (width for HJC-y, height for HJC-z and depth for HJC-x), we found that HJC was located 14% (S.D. 3%) of PW medially, 34% (S.D. 2%) of PD posteriorly, and 79% (S.D. 5%) of PH inferiorly. The relative error in HJC estimation along each axis was minimized with HJC estimation from pelvic width for HJCy, pelvic height for HJC-z and pelvic depth for HJC-x (Table 2). Using the estimated location of HJC-x as 24% of PW resulted in a mean error of 0.49 cm (S.D. 0.34; max 1.49). The mean error using the location of HJC-x as 34% of PD yielded a mean error of 0.30 cm (S.D. 0.23; max 0.89). The calculation of HJC-x as a function of PD resulted in a smaller error (0.2 cm) than calculation of HJC-x as a function of PW (p < O.C001). Estimating HJC-z as a function of PW (30%) resulted in a mean error of 0.75 cm (S.D. 0.56, max 2.26), while calculation as a function of PH (79%) revealed a mean error of 0.35 cm (S.D. 0.28; max 1.29). The pelvic height method of calculating HJC-z reduced the mean error by a mean of 0.40 cm (p < 0.0001). Calculating HJC-y as a function of pelvic width (14%) yielded a mean error of 0.58 cm (S.D. 0.42; max 2.18). DISCUSSION
In gait analysis studies the accurate location of HJC is critical for precise computation of hip muscle moments, joint intersegmental resultant forces, and joint motion. A reliable technique for locating HJC is crucial to minimize the errors associated with HJC displacements from true position. Bell et al.‘s (1990) evaluation of Cappozzo’s rotational method concluded that the use of palpable bony landmarks is the most accurate and efficient way of determining HJC on a subject-specific basis. Bell et al. (1989,199O) estimated HJC as a constant proportion of PW for each axis. Our correlation analysis revealed that both HJC-x and HJC-z do not vary in a predictable linear relationship with PW (Fig. l(A) and 2(A)), while HJCq does strongly vary predictably with PW. These correlations indicate that errors are expected when locating HJC as a function of PW on any axis except the mediolateral axis. We found HJC to be located relative to ASK at 30% (S.D. 4%) of PW distally, 14% (S.D. 3%) of PW medially, and 24% (SD. 2%) of PW posteriorly. The differences in the posterior percentage as compared to the Bell et al. (1989, 1990) results could be due to a difference in sample size or the
Table 2. Errors in estimating HJC-x, HJC-z, and HJC-y resulting from various alogrithms. In all instances, the error was calculated as the absolute value of the specific estimate value minus the true value. Errors were minimized with estimation of HJC-x based on pelvic depth and HJC-z based on pelvic height Absolute value HJC algorithm Bell et HJC-x Bell et HJC-x
N
(1990) 19% PW al. (1989) 22% PW
1.19 0.51 0.32
2.19
(2)
65
0.58 0.44 0.03
1.51
Seidel et al. (1994) (3) HJC-x 24% PW
65
0.49 0.34 0.03
1.49
Seidel et al. (1994) (4) HJC-x 34% PD Bell et al. (1989, 1990) & Seidel et al. (1994) (5) HJC-z 30% PW
35
0.30 0.23 0.00
0.89
65
0.75 0.56 0.01
2.26
65
0.35 0.28 0.00
1.29
65
0.58 0.42 0.00
2.18
al.
Seidel et al. (1994) (6) HJC-z 80%PH Bell et al. (1989, 1900) & Seidel et al. (1994) HJC-y 14% PW
Comparisons of estimates using paired t-test. (2) compared to (1): p < 0.0001 (3) compared to (2): p < 0.08 (4) compared to (2): p < 0.0001 (4) compared to (3): p < 0.0001 (5) compared to (6): p < 0.0001
differences inherent in taking direct pelvic measurements vs radiographic measurements. The average percentages were, however, rather similar to Bell et al. (1989, 1990). Although this confirms an auerage location of HJC as a function of PW, it does not relate accurately on a subject-specific basis since there is no correlation of HJC-x and HJC-z with PW. Also, the mean error in location of HJC measurements using
HJC-z WITH PELVIC HEIGHT BY SEX
6 9.5
’ MALE
A FEMALE
l
9
MALE
. . ’ A.. .‘A.1 ;m. .
6
A :: 7
i
7.5
‘! Y I
6.5
1
6 FEMALE
6.5
g
Max
65
10
2 I’
Mean SD. Min
(1)
HJC-z WITH PELVIC WIDTH BY SEX
A
997
6 iA
5.5
5
a I?r
4 18
N=12:
Nr122 19
20
21
22
25 23 24 PELVIC WIDTH
26 (cm)
27
26
26
30
4.5
7
7.5
a
6.5 PELVIC
9 9.5 10 HEIGHT (cm)
10.5
Fig. 2. Plots of HJC-z (all sides) with pelvic width (A) and pelvic height (B) by sex ,in centimeters. HJC-z is not correlated with pelvic width but highly correlated with pelvic height.
il.l 11
998
Technical Note
PW alone was significantly higher than the error in locating HJC relative to pelvic geometric measurements. The calculation of HJC-x as a function of PD and HJC-z as a function of PH resulted in significantly smaller errors than the calculation of both HJC-x and HJC-z as a function of PW. The correlation between HJC-x and PD was r=0.54. Although this correlation was not as high as the HJC-y (r=O.85) and HJC-z (r=0.81) correlations, it is markedly higher than the correlation between HJC-x and PW (r= - 0.17). This ‘lower’ correlation of HJC-x and PD is most likely due to the fact that the measurement of PD (ASIS to PSIS) is not a true perpendicular PD measurement from the frontal plane. The apparent relationship between PW and PD with r =0.52, may reflect the oblique measurement of PD. Further study will need to determine which PD parameter (ASIS to PSIS vs the perpendicular distance from the frontal plane to the PSIS) will best estimate HJC-x. Since, in our study, the female pelvic measurements were proportionally smaller than those of males (consistent with Brinkmann et al., 1981) locating HJC as a function of pelvic geometric measurements can be performed similarly for both genders. Measurements were taken only on skeletally mature adults; therefore, the applicability of our results to HJC location in children is unknown. In gait analysis procedures, current pelvic targeting protocols utilize spherical reflective markers on the ASISs and the sacrum. In oivo assessment of HJC-y is based upon the distance between the centers of each ASIS marker. HJC-x may be obtained by establishing the distance between the centers of equal sized spherical markers placed on the ASIS and ipsilateral PSIS for the hip joint in question or with the use of obstetrical calipers to measure the distance from the PSIS to the ASIS to establish pelvic depth. The ratio of HJC-x with pelvic depth is assumed to remain constant as soft tissues add equally over the ASIS and PSIS. HJC-z may be obtained from pelvic height by measuring the distance from the ASIS to the public symphysis (the midpoint between the pubic tubercles) and the use of the mean of two position vectors from each ASIS as the midpoint. Soft tissue depth over the pubis is not a factor in pelvic height assessment. HJC estimation relative to the ASISs as a function PW alone results in significant deviations from the true HJC location. HJC estimation from pelvic width, height and depth parameters optimizes HJC- location relative-to true HJC. This study and Bell et al. (1989,199O) propose the same method to locate HJC along the mediolateral axis as 14% of pelvic width (SD. 3%) relative to the ASIS. However, in
contrast to Bell et al. (1989. 1990) this study proposes the location of HJC along the anteroposterior axis as 34% of pelvic depth (S.D. 2%) relative to the ASIS and the location of HJC along the superoinferior axis as 79% of pelvic height (S.D. 5%) relative to the ASIS. Acknowledgements-Rowe extend out appreciation to the Department of Anatomy and the Mortuary Staff, Wayne State University, School of Medicine for their assistance with this project. REFERENCES
Andriacchi, T. P., Andersson, G. B. J., Fermier, R. W., Stern. D. and Galante, J. 0. (1980) A study of lower-limb mechanics during stair climbing. J. Bone Jt Surg 62-A, 749-757. Bell, A. L., Brand, R. A. and Pedersen, D. R. (1989) Prediction of hip joint center location from external landmarks. Hum. Mvmt. Sci. 8, 3-16. Bell, A. L., Pedersen, D. R. and Brand, R. A. (1990) A comparison of the accuracy of several hip center location prediction methods. J. Biomechanics 23, 617-621. Blankevoort, L., Huiskes, R. and de Lange, A. (1990) Helical axes of passive knee joint motions. J. Biomechanics 23, 1219-1229. Brinckmann, P., Hoefert, H. and Jongen, H. Th. (1981) Sex differences in the skeletal geometry of the human pelvis and hip joint. J. Biomechanics 14, 427-430. Cappozzo, A. (1984) Gait analysis methodology. Hum. Mvmt.
Sci. 3, 27-50.
Crowninshield, R. D., Johnston, R. C., Andrews, J. G. and Brand, R. A. (1977) The effect of joint center on hip kinetics. Proc. 23rd Annual ORS Convention 47-49. Las Vegas, Nevada. Crowninshield, R. D., Johnston, R. C., Andrews, J. G. and Brand, R. A. (1978) A biomechanical investigation of the human hip. J. Biomechanics 11, 75-85. Delp, S. L. and Maloney, W. (1993) Effects of hip center location on the moment-generating capacity of the muscles. J. Biomechanics 26, 485-499. Ellis, M. I., Seedhom, B. B., Amis, A. A., Dowson, D. and Wright, V. (1979) Forces in the knee joint shilst rising from normal and motorized chairs. Engng. Med. 8, 3340. Tylkowski, C. M., Simon, S. R. and Mansour, J. M. (1982) Internal rotation gait in spastic cerebral palsy in the hip. Proc.
10th
Open
Scientijc
Meeting
of the Hip
Society
(Edited by Nelson, J. P.) pp. 89-125. Mosby. St. Louis.