Annals of Biomedical Engineering ( 2014) DOI: 10.1007/s10439-014-1052-2
A Headform for Testing Helmet and Mouthguard Sensors that Measure Head Impact Severity in Football Players GUNTER P. SIEGMUND,1,2 KEVIN M. GUSKIEWICZ,3 STEPHEN W. MARSHALL,4 ALYSSA L. DEMARCO,1 and STEPHANIE J. BONIN5 1 MEA Forensic Engineers & Scientists, 11-11151 Horseshoe Way, Richmond, BC V7A 4S5, Canada; 2School of Kinesiology, University of British Columbia, Vancouver, BC, Canada; 3Matthew A. Gfeller Sport-Related Traumatic Brain Injury Research Center, Department of Exercise and Sport Science, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA; 4 Department of Epidemiology, Gillings School of Global Public Health, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA; and 5MEA Forensic Engineers & Scientists, Laguna Hills, CA, USA
(Received 17 March 2014; accepted 31 May 2014) Associate Editor Stefan M Duma oversaw the review of this article.
Abstract—A headform is needed to validate and compare helmet- and mouthguard-based sensors that measure the severity and direction of football head impacts. Our goal was to quantify the dynamic response of a mandibular loadsensing headform (MLSH) and to compare its performance and repeatability to an unmodified Hybrid III headform. Linear impactors in two independent laboratories were used to strike each headform at six locations at 5.5 m/s and at two locations at 3.6 and 7.4 m/s. Impact severity was quantified using peak linear acceleration (PLA) and peak angular acceleration (PAA), and direction was quantified using the azimuth and elevation of the PLA. Repeatability was quantified using coefficients of variation (COV) and standard deviations (SD). Across all impacts, PLA was 1.6 ± 1.8 g higher in the MLSH than in the Hybrid III (p = 0.002), but there were no differences in PAA (p = 0.25), azimuth (p = 0.43) and elevation (p = 0.11). Both headforms exhibited excellent or acceptable repeatability for PLA (HIII:COV = 2.1 ± 0.8%, MLSH:COV = 2.0 ± 1.2%, p = 0.98), but site-specific repeatability ranging from excellent to poor for PAA (HIII:COV = 7.2 ± 4.0%, MLSH:COV = 8.3 ± 5.8%, p = 0.58). Direction SD were generally 10%).3,10 The COVs for PLA and PAA were calculated for each series of five repeated tests for all eight impact conditions in each lab. COV is a poor measure of repeatability for angles (e.g., azimuth and elevation) because the value of the mean used in the denominator depends on the reference frame definition. As a result, SD was used to quantify angular variability. To confirm that our 5-min inter-test interval did not alter the head/neck response, we computed the differences in PLA and PAA between consecutive tests under identical conditions (4 differences for every 5 consecutive tests; n = 64 differences per variable per headform, pooled across labs) and then computed the 95th percentile confidence interval (95% CI) of the mean of these differences to ensure it contained zero. Systematic changes in the response of the head and/or neck would be expected to generate non-zero differences across all consecutive impacts. Differences in the performance of the MLSH and Hybrid III headforms within each lab were compared using separate two-tailed two-sample t tests for each impact condition (n = 5 trials per headform per lab). Differences in repeatability of the two headforms were assessed using the COVs of PLA and PAA and the SD of azimuth and elevation for each impact condition. Each COV and SD was treated as an independent measurement of repeatability. Differences between the COVs and SD for the MLSH and Hybrid III across all test conditions (n = 16 pooled across labs) were tested using separate two-tailed paired t tests for PLA, azimuth and elevation. Since the radii of the two 3–2–2–2 arrays were expected to affect the COV of PAA, separate paired t tests were conducted within each lab (n = 8 conditions per lab) for this variable. Significance was set to p < 0.05 for
all tests. An adjustment for multiple comparisons was not used because we wanted to highlight possible differences between the headforms.
RESULTS Of the 160 impacts performed in both labs, two trials were discarded (Lab 1, Hybrid III, side impact, high speed; and Lab 2, MLSH, crown impact, medium speed) because their impact speeds fell outside the prescribed range. For the remaining trials, the COVs for impact speed within each block remained below 2% (average ± SD of 0.64 ± 0.39%). The raw data for the MLSH was considerably noisier than the raw data from the Hybrid III headform (Fig. 3a). This noise occurred after the primary head impact kinematics and was effectively eliminated by the digital filter (Fig. 3b). Peak values taken from the filtered data were used for the remainder of the analysis. Power spectra of the resultant linear accelerations showed similar signal power for the MLSH and Hybrid III headform over the frequency range of interest (0–300 Hz) for all but the crown impact location (Fig. 4). Greater power in the spectra above 300 Hz for the MLSH’s raw data was consistent with the high frequency noise seen in its kinematic data (Fig. 3a). No systematic changes in the head/neck response characteristics were observed between consecutive tests of the same test condition. For the MLSH headform, the median time between tests was 5.6 min (range 5.0– 17.1 min) with an average change between consecutive tests of 0.01 ± 2.68 g (95% CI 20.80 to 0.83 g) and 17 ± 360 rad/s2 (95% CI 292 to 126 rad/s2). For the Hybrid III headform, the median time between tests was 5.5 min (range 5.1–21.6 min) with an average change between consecutive tests of 0.18 ± 2.14 g (95% CI 20.47 to 0.82 g) and 15 ± 332 rad/s2 (95% CI 285 to 116 rad/s2). A similarly non-significant change was observed when only trials with inter-test intervals between 5 and 6 min were analyzed (n = 38 of 62 for MLSH, n = 42 of 62 for Hybrid III). PLAs of the MLSH were 1.6 ± 1.8 g (range 20.4 to 5.9 g) larger than the Hybrid III headform when averaged across all impact conditions (p = 0.002; Table 1; Fig. 5). Similar comparisons at the individual impact locations within each lab showed that only 4 of the 16 test conditions were different (Table 1). Peak angular accelerations between the two headforms were not different when averaged across all impact conditions (144 ± 481 rad/s2, range 2702 to 1375 rad/s2, p = 0.25), but four site-specific differences were present (Table 1; Fig. 6). Across all impact conditions, neither azimuth (0.3 ± 1.5, p = 0.43) nor elevation
Headform for Testing Head Impact Sensors Angular Acceleration (krad/s2)
Linear Acceleration (g)
MLSH
Hybrid III
(a) Raw Data 50
3
0
0
−50
−3
50
10
0
0
−50
−10
MLSH
Hybrid III
(b) Filtered Data
X
Y
Z
50
3
0
0
−50
−3
50
3
0
0
−50
−3
Resultant
MLSH
(c) Raw Data - with a mouthguard, helmet and snug chinstrap 50
3
0
0
−50
-3 0
20
40 60 Time (ms)
80
100
0
20
40 60 Time (ms)
80
100
FIGURE 3. Comparison of the raw and filtered data for a forehead impact. Within the raw (a) and filtered (b) panels, linear acceleration (left column) and angular acceleration (right column) are shown for the Hybrid III (top row) and MLSH (bottom row). The three components and the resultant accelerations are shown. The lower panel (c) shows the effect of a mouthguard, helmet and snug chinstrap on the high frequency noise in the raw data at a similar impact speed.
(0.6 ± 1.4, p = 0.11) was different between the MLSH and Hybrid III headforms despite most of the individual impact conditions showing small (£3.7) but significant differences (Table 2). Across both labs, the COVs for PLA indicated excellent repeatability (£3%) for 13 and 14 of the 16 impact conditions for the MLSH and Hybrid III headforms respectively (Table 1). The remaining 5 impact conditions achieved acceptable repeatability (3 < COV £ 7). For peak angular acceleration (PAA), the repeatability for both headforms ranged from excellent to poor (Table 1). The front boss, rear boss and crown locations in both labs were responsible for all but one of the marginal and poor repeatability values. When compared across all conditions, there
were no differences in the COVs between the two headforms for linear acceleration (Hybrid III: 2.1 ± 0.8%, MLSH 2.0 ± 1.2%, p = 0.98). For angular acceleration, there were also no differences in the COVs between the two headforms when using either the standard array in the Hybrid III (Lab 1: Hybrid III: 4.7 ± 2.0%, MLSH: 6.9 ± 3.1%, p = 0.07) or the compact array in the Hybrid III (Lab 2: Hybrid III: 7.2 ± 4.0%, MLSH: 8.3 ± 5.8%, p = 0.58). The SD for azimuth and elevation were highest for the crown impact location (3.0 ± 0.9), but generally remained below 1 for the other sites (0.5 ± 0.2) (Table 2). There was no difference between the azimuth and elevation SD between the two headforms (Az: p = 0.54, El: p = 0.80).
Power
Power
SIEGMUND et al.
10
2
10
0
10
−2
10
−4
10
2
10
0
10
−2
10
−4
FH - Forehead
FB - Front Boss
SD - Side
RB - Rear Boss
RR - Rear
CR - Crown
10
1
2
3
10 10 Frequency (Hz)
10
1
2
3
10 10 Frequency (Hz)
10
1
MLSH MLSH-filtered Hybrid III Hybrid III-filtered
2
3
10 10 Frequency (Hz)
FIGURE 4. Power spectra for the MLSH (red) and Hybrid III (blue) for 100 ms of raw data (solid line) and low-pass filtered data (dashed line) for the six impact locations at 5.5 m/s. The data represent the mean power of the five trials at each condition.
TABLE 1. Mean 6 SD, COV and difference (D) between the MLSH and Hybrid III headforms for PLA and PAA at all impact sites and speeds. PAA (krad/s2)
PLA (g) Hybrid III (a) Lab 1 Forehead 57.2 Front Boss 82.6 Side 73.7 Rear boss 67.5 Rear 73.2 Crown 41.9 (b) Lab 2 Forehead 57.2 Front boss 74.4 Side 78.2 Rear boss 72.1 Rear 71.9 Crown 38.4 (c) Side impact High speed 112.6 Medium 73.7 Low 42.9 (d) Rear boss impact High speed 109.1 Medium 72.1 Low 49.3
COV
MLSH
COV
D
Hybrid III
COV
MLSH
COV
D
± ± ± ± ± ±
1.1 1.2 1.8 0.6 0.9 1.0
1.9 1.5 2.5 0.9 1.2 2.3
58.3 83.4 76.0 68.5 72.8 42.0
± ± ± ± ± ±
1.1 1.6 1.6 0.6 1.1 0.3
1.9 1.9 2.1 0.9 1.5 0.8
1.1 0.8 2.3 1.0* 20.4 0.1
3.19 4.44 4.93 3.19 3.83 2.34
± ± ± ± ± ±
0.09 0.34 0.22 0.09 0.16 0.17
2.7 7.7 4.5 2.7 4.1 7.4
3.23 3.77 4.43 3.03 3.80 2.50
± ± ± ± ± ±
0.09 0.30 0.52 0.25 0.28 0.22
2.9 8.1 11.7 8.4 7.4 8.9
0.04 20.68* 20.50 20.16 20.03 0.16
± ± ± ± ± ±
2.1 0.7 1.1 1.4 2.1 1.1
3.7 1.0 1.4 1.9 2.9 2.9
57.9 75.5 80.8 75.4 77.8 39.1
± ± ± ± ± ±
1.3 0.9 0.5 3.9 1.4 1.1
2.2 1.2 0.6 5.1 1.8 2.8
0.7 1.1 2.5* 3.3 5.9* 0.7
3.23 5.26 5.64 3.29 3.82 2.86
± ± ± ± ± ±
0.14 0.33 0.26 0.36 0.21 0.42
4.3 6.3 4.6 10.9 5.5 14.8
3.29 4.55 5.38 2.96 3.75 4.23
± ± ± ± ± ±
0.14 0.36 0.28 0.45 0.21 0.24
4.2 7.9 5.2 15.1 5.7 5.6
0.06 20.70* 20.26 20.33 20.07 1.37*
± 3.9 6 1.8 ± 0.6
3.5 2.5 1.5
113.9 ± 1.8 76.0 6 1.6 43.4 ± 0.5
1.6 2.1 1.3
1.2 2.3 0.5
6.27 ± 0.34 4.93 6 0.22 3.33 ± 0.12
5.4 4.5 3.5
5.82 ± 0.30 4.43 6 0.52 3.28 ± 0.10
5.2 11.7 3.0
20.45 20.50 20.05
± 2.2 6 1.4 ± 0.9
2.0 1.9 1.8
113.7 ± 3.6 75.4 6 3.9 49.1 ± 1.9
3.1 5.1 3.9
4.6* 3.3 20.2
4.40 ± 0.37 3.29 6 0.36 2.12 ± 0.05
8.4 10.9 2.5
4.15 ± 0.81 2.96 6 0.45 1.67 ± 0.05
19.6 15.1 3.2
20.26 20.33 20.44*
The two italic values represent the largest COV for each lab and the bold value rows are repeated to allow easy comparison with the other speed tests. D is the difference between the MLSH and the Hybrid III headforms, calculated as (MLSH–Hybrid III). *Significant differences, p £ 0.05, using a two-sample t test.
Headform for Testing Head Impact Sensors
(a) Laboratory 1 Forehead
Front Boss
Side
Rear Boss MLSH Hybrid III
Linear Acceleration (g)
100 50 0 Rear
Crown
Side - low
Side - high
Forehead
Front Boss
Side
Rear Boss
Rear
Crown
Rear Boss - low
Rear Boss - high
100 50 0
(b) Laboratory 2
Linear Acceleration (g)
100 50 0 100 50 0
0
5
10
15
0
5
10
15
0
5
10
15
0
5
10
15
Time (ms)
FIGURE 5. Comparison of the resultant linear accelerations for the MLSH (red) and Hybrid III (blue) for all eight impact conditions in both laboratories. Each graph contains five MLSH traces and five Hybrid III traces.
DISCUSSION Our goal was to compare the peak kinematic response and repeatability of the MLSH and an unmodified Hybrid III headform when mounted on a Hybrid III neck and exposed to padded head impacts on a linear impactor. Based on our tests at six impact locations in two laboratories, we found site-specific differences in the linear and angular acceleration responses between the two headforms, but no differences in the repeatability of two headforms. Overall, both headforms exhibited excellent/acceptable repeatability (COV < 7%) for peak resultant linear acceleration and low variability (
A Headform for Testing Helmet and Mouthguard Sensors that Measure Head Impact Severity in Football Players GUNTER P. SIEGMUND,1,2 KEVIN M. GUSKIEWICZ,3 STEPHEN W. MARSHALL,4 ALYSSA L. DEMARCO,1 and STEPHANIE J. BONIN5 1 MEA Forensic Engineers & Scientists, 11-11151 Horseshoe Way, Richmond, BC V7A 4S5, Canada; 2School of Kinesiology, University of British Columbia, Vancouver, BC, Canada; 3Matthew A. Gfeller Sport-Related Traumatic Brain Injury Research Center, Department of Exercise and Sport Science, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA; 4 Department of Epidemiology, Gillings School of Global Public Health, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA; and 5MEA Forensic Engineers & Scientists, Laguna Hills, CA, USA
(Received 17 March 2014; accepted 31 May 2014) Associate Editor Stefan M Duma oversaw the review of this article.
Abstract—A headform is needed to validate and compare helmet- and mouthguard-based sensors that measure the severity and direction of football head impacts. Our goal was to quantify the dynamic response of a mandibular loadsensing headform (MLSH) and to compare its performance and repeatability to an unmodified Hybrid III headform. Linear impactors in two independent laboratories were used to strike each headform at six locations at 5.5 m/s and at two locations at 3.6 and 7.4 m/s. Impact severity was quantified using peak linear acceleration (PLA) and peak angular acceleration (PAA), and direction was quantified using the azimuth and elevation of the PLA. Repeatability was quantified using coefficients of variation (COV) and standard deviations (SD). Across all impacts, PLA was 1.6 ± 1.8 g higher in the MLSH than in the Hybrid III (p = 0.002), but there were no differences in PAA (p = 0.25), azimuth (p = 0.43) and elevation (p = 0.11). Both headforms exhibited excellent or acceptable repeatability for PLA (HIII:COV = 2.1 ± 0.8%, MLSH:COV = 2.0 ± 1.2%, p = 0.98), but site-specific repeatability ranging from excellent to poor for PAA (HIII:COV = 7.2 ± 4.0%, MLSH:COV = 8.3 ± 5.8%, p = 0.58). Direction SD were generally 10%).3,10 The COVs for PLA and PAA were calculated for each series of five repeated tests for all eight impact conditions in each lab. COV is a poor measure of repeatability for angles (e.g., azimuth and elevation) because the value of the mean used in the denominator depends on the reference frame definition. As a result, SD was used to quantify angular variability. To confirm that our 5-min inter-test interval did not alter the head/neck response, we computed the differences in PLA and PAA between consecutive tests under identical conditions (4 differences for every 5 consecutive tests; n = 64 differences per variable per headform, pooled across labs) and then computed the 95th percentile confidence interval (95% CI) of the mean of these differences to ensure it contained zero. Systematic changes in the response of the head and/or neck would be expected to generate non-zero differences across all consecutive impacts. Differences in the performance of the MLSH and Hybrid III headforms within each lab were compared using separate two-tailed two-sample t tests for each impact condition (n = 5 trials per headform per lab). Differences in repeatability of the two headforms were assessed using the COVs of PLA and PAA and the SD of azimuth and elevation for each impact condition. Each COV and SD was treated as an independent measurement of repeatability. Differences between the COVs and SD for the MLSH and Hybrid III across all test conditions (n = 16 pooled across labs) were tested using separate two-tailed paired t tests for PLA, azimuth and elevation. Since the radii of the two 3–2–2–2 arrays were expected to affect the COV of PAA, separate paired t tests were conducted within each lab (n = 8 conditions per lab) for this variable. Significance was set to p < 0.05 for
all tests. An adjustment for multiple comparisons was not used because we wanted to highlight possible differences between the headforms.
RESULTS Of the 160 impacts performed in both labs, two trials were discarded (Lab 1, Hybrid III, side impact, high speed; and Lab 2, MLSH, crown impact, medium speed) because their impact speeds fell outside the prescribed range. For the remaining trials, the COVs for impact speed within each block remained below 2% (average ± SD of 0.64 ± 0.39%). The raw data for the MLSH was considerably noisier than the raw data from the Hybrid III headform (Fig. 3a). This noise occurred after the primary head impact kinematics and was effectively eliminated by the digital filter (Fig. 3b). Peak values taken from the filtered data were used for the remainder of the analysis. Power spectra of the resultant linear accelerations showed similar signal power for the MLSH and Hybrid III headform over the frequency range of interest (0–300 Hz) for all but the crown impact location (Fig. 4). Greater power in the spectra above 300 Hz for the MLSH’s raw data was consistent with the high frequency noise seen in its kinematic data (Fig. 3a). No systematic changes in the head/neck response characteristics were observed between consecutive tests of the same test condition. For the MLSH headform, the median time between tests was 5.6 min (range 5.0– 17.1 min) with an average change between consecutive tests of 0.01 ± 2.68 g (95% CI 20.80 to 0.83 g) and 17 ± 360 rad/s2 (95% CI 292 to 126 rad/s2). For the Hybrid III headform, the median time between tests was 5.5 min (range 5.1–21.6 min) with an average change between consecutive tests of 0.18 ± 2.14 g (95% CI 20.47 to 0.82 g) and 15 ± 332 rad/s2 (95% CI 285 to 116 rad/s2). A similarly non-significant change was observed when only trials with inter-test intervals between 5 and 6 min were analyzed (n = 38 of 62 for MLSH, n = 42 of 62 for Hybrid III). PLAs of the MLSH were 1.6 ± 1.8 g (range 20.4 to 5.9 g) larger than the Hybrid III headform when averaged across all impact conditions (p = 0.002; Table 1; Fig. 5). Similar comparisons at the individual impact locations within each lab showed that only 4 of the 16 test conditions were different (Table 1). Peak angular accelerations between the two headforms were not different when averaged across all impact conditions (144 ± 481 rad/s2, range 2702 to 1375 rad/s2, p = 0.25), but four site-specific differences were present (Table 1; Fig. 6). Across all impact conditions, neither azimuth (0.3 ± 1.5, p = 0.43) nor elevation
Headform for Testing Head Impact Sensors Angular Acceleration (krad/s2)
Linear Acceleration (g)
MLSH
Hybrid III
(a) Raw Data 50
3
0
0
−50
−3
50
10
0
0
−50
−10
MLSH
Hybrid III
(b) Filtered Data
X
Y
Z
50
3
0
0
−50
−3
50
3
0
0
−50
−3
Resultant
MLSH
(c) Raw Data - with a mouthguard, helmet and snug chinstrap 50
3
0
0
−50
-3 0
20
40 60 Time (ms)
80
100
0
20
40 60 Time (ms)
80
100
FIGURE 3. Comparison of the raw and filtered data for a forehead impact. Within the raw (a) and filtered (b) panels, linear acceleration (left column) and angular acceleration (right column) are shown for the Hybrid III (top row) and MLSH (bottom row). The three components and the resultant accelerations are shown. The lower panel (c) shows the effect of a mouthguard, helmet and snug chinstrap on the high frequency noise in the raw data at a similar impact speed.
(0.6 ± 1.4, p = 0.11) was different between the MLSH and Hybrid III headforms despite most of the individual impact conditions showing small (£3.7) but significant differences (Table 2). Across both labs, the COVs for PLA indicated excellent repeatability (£3%) for 13 and 14 of the 16 impact conditions for the MLSH and Hybrid III headforms respectively (Table 1). The remaining 5 impact conditions achieved acceptable repeatability (3 < COV £ 7). For peak angular acceleration (PAA), the repeatability for both headforms ranged from excellent to poor (Table 1). The front boss, rear boss and crown locations in both labs were responsible for all but one of the marginal and poor repeatability values. When compared across all conditions, there
were no differences in the COVs between the two headforms for linear acceleration (Hybrid III: 2.1 ± 0.8%, MLSH 2.0 ± 1.2%, p = 0.98). For angular acceleration, there were also no differences in the COVs between the two headforms when using either the standard array in the Hybrid III (Lab 1: Hybrid III: 4.7 ± 2.0%, MLSH: 6.9 ± 3.1%, p = 0.07) or the compact array in the Hybrid III (Lab 2: Hybrid III: 7.2 ± 4.0%, MLSH: 8.3 ± 5.8%, p = 0.58). The SD for azimuth and elevation were highest for the crown impact location (3.0 ± 0.9), but generally remained below 1 for the other sites (0.5 ± 0.2) (Table 2). There was no difference between the azimuth and elevation SD between the two headforms (Az: p = 0.54, El: p = 0.80).
Power
Power
SIEGMUND et al.
10
2
10
0
10
−2
10
−4
10
2
10
0
10
−2
10
−4
FH - Forehead
FB - Front Boss
SD - Side
RB - Rear Boss
RR - Rear
CR - Crown
10
1
2
3
10 10 Frequency (Hz)
10
1
2
3
10 10 Frequency (Hz)
10
1
MLSH MLSH-filtered Hybrid III Hybrid III-filtered
2
3
10 10 Frequency (Hz)
FIGURE 4. Power spectra for the MLSH (red) and Hybrid III (blue) for 100 ms of raw data (solid line) and low-pass filtered data (dashed line) for the six impact locations at 5.5 m/s. The data represent the mean power of the five trials at each condition.
TABLE 1. Mean 6 SD, COV and difference (D) between the MLSH and Hybrid III headforms for PLA and PAA at all impact sites and speeds. PAA (krad/s2)
PLA (g) Hybrid III (a) Lab 1 Forehead 57.2 Front Boss 82.6 Side 73.7 Rear boss 67.5 Rear 73.2 Crown 41.9 (b) Lab 2 Forehead 57.2 Front boss 74.4 Side 78.2 Rear boss 72.1 Rear 71.9 Crown 38.4 (c) Side impact High speed 112.6 Medium 73.7 Low 42.9 (d) Rear boss impact High speed 109.1 Medium 72.1 Low 49.3
COV
MLSH
COV
D
Hybrid III
COV
MLSH
COV
D
± ± ± ± ± ±
1.1 1.2 1.8 0.6 0.9 1.0
1.9 1.5 2.5 0.9 1.2 2.3
58.3 83.4 76.0 68.5 72.8 42.0
± ± ± ± ± ±
1.1 1.6 1.6 0.6 1.1 0.3
1.9 1.9 2.1 0.9 1.5 0.8
1.1 0.8 2.3 1.0* 20.4 0.1
3.19 4.44 4.93 3.19 3.83 2.34
± ± ± ± ± ±
0.09 0.34 0.22 0.09 0.16 0.17
2.7 7.7 4.5 2.7 4.1 7.4
3.23 3.77 4.43 3.03 3.80 2.50
± ± ± ± ± ±
0.09 0.30 0.52 0.25 0.28 0.22
2.9 8.1 11.7 8.4 7.4 8.9
0.04 20.68* 20.50 20.16 20.03 0.16
± ± ± ± ± ±
2.1 0.7 1.1 1.4 2.1 1.1
3.7 1.0 1.4 1.9 2.9 2.9
57.9 75.5 80.8 75.4 77.8 39.1
± ± ± ± ± ±
1.3 0.9 0.5 3.9 1.4 1.1
2.2 1.2 0.6 5.1 1.8 2.8
0.7 1.1 2.5* 3.3 5.9* 0.7
3.23 5.26 5.64 3.29 3.82 2.86
± ± ± ± ± ±
0.14 0.33 0.26 0.36 0.21 0.42
4.3 6.3 4.6 10.9 5.5 14.8
3.29 4.55 5.38 2.96 3.75 4.23
± ± ± ± ± ±
0.14 0.36 0.28 0.45 0.21 0.24
4.2 7.9 5.2 15.1 5.7 5.6
0.06 20.70* 20.26 20.33 20.07 1.37*
± 3.9 6 1.8 ± 0.6
3.5 2.5 1.5
113.9 ± 1.8 76.0 6 1.6 43.4 ± 0.5
1.6 2.1 1.3
1.2 2.3 0.5
6.27 ± 0.34 4.93 6 0.22 3.33 ± 0.12
5.4 4.5 3.5
5.82 ± 0.30 4.43 6 0.52 3.28 ± 0.10
5.2 11.7 3.0
20.45 20.50 20.05
± 2.2 6 1.4 ± 0.9
2.0 1.9 1.8
113.7 ± 3.6 75.4 6 3.9 49.1 ± 1.9
3.1 5.1 3.9
4.6* 3.3 20.2
4.40 ± 0.37 3.29 6 0.36 2.12 ± 0.05
8.4 10.9 2.5
4.15 ± 0.81 2.96 6 0.45 1.67 ± 0.05
19.6 15.1 3.2
20.26 20.33 20.44*
The two italic values represent the largest COV for each lab and the bold value rows are repeated to allow easy comparison with the other speed tests. D is the difference between the MLSH and the Hybrid III headforms, calculated as (MLSH–Hybrid III). *Significant differences, p £ 0.05, using a two-sample t test.
Headform for Testing Head Impact Sensors
(a) Laboratory 1 Forehead
Front Boss
Side
Rear Boss MLSH Hybrid III
Linear Acceleration (g)
100 50 0 Rear
Crown
Side - low
Side - high
Forehead
Front Boss
Side
Rear Boss
Rear
Crown
Rear Boss - low
Rear Boss - high
100 50 0
(b) Laboratory 2
Linear Acceleration (g)
100 50 0 100 50 0
0
5
10
15
0
5
10
15
0
5
10
15
0
5
10
15
Time (ms)
FIGURE 5. Comparison of the resultant linear accelerations for the MLSH (red) and Hybrid III (blue) for all eight impact conditions in both laboratories. Each graph contains five MLSH traces and five Hybrid III traces.
DISCUSSION Our goal was to compare the peak kinematic response and repeatability of the MLSH and an unmodified Hybrid III headform when mounted on a Hybrid III neck and exposed to padded head impacts on a linear impactor. Based on our tests at six impact locations in two laboratories, we found site-specific differences in the linear and angular acceleration responses between the two headforms, but no differences in the repeatability of two headforms. Overall, both headforms exhibited excellent/acceptable repeatability (COV < 7%) for peak resultant linear acceleration and low variability (
