Impact of exercise selection on hamstring muscle

activation

Author
Bourne, Matthew N, Williams, Morgan D, Opar, David A, Al Najjar, Aiman, Kerr, Graham K,
Shield, Anthony J

Published
2017

Journal Title
British Journal of Sports Medicine

Version
Version of Record (VoR)

DOI
https://doi.org/10.1136/bjsports-2015-095739

Copyright Statement
© The Author(s) 2017. This is the author-manuscript version of this paper. It is posted here
with permission of the copyright owner(s) for your personal use only. No further distribution
permitted. For information about this journal please refer to the publisher’s website or contact
the author(s).

Downloaded from
http://hdl.handle.net/10072/372340

Griffith Research Online

2 Impact of exercise selection on hamstring muscle activation

3 Authors:
4
5 Matthew N. Bourne1,2,3, Morgan D. Williams4, David A. Opar5, Aiman Al Najjar6,
6 Graham K. Kerr1,2, Anthony J. Shield1,2.
7
8 1School of Exercise and Nutrition Science, Faculty of Health, Queensland University of

9 Technology, Brisbane, Australia.

12 3 Queensland Academy of Sport, Centre of Excellence for Applied Sport Science Research,

13 Brisbane, Australia.

14 4School of Health, Sport and Professional Practice, Faculty of Life Sciences and Education,

15 University of South Wales, Wales, United Kingdom

16 5School of Exercise Sciences, Australian Catholic University, Melbourne, Australia.

17 6Centre for Advanced Imaging, University of Queensland, Brisbane, Australia.

18
19 Corresponding Author:

20 Dr Anthony Shield

21 School of Exercise and Nutrition Sciences and the Institute of Health and Biomedical
22 Innovation,
23 Queensland University of Technology, Victoria Park Road, Kelvin Grove, 4059,
24 Brisbane, Queensland, Australia.
25 Email: [email protected]
26 Ph: +61 7 3138 5829
27 Fax: +61 7 3138 3980

28 Key Words

29 Injury prevention, rehabilitation, physical therapy

1
30 ABSTRACT

31 Objective: To determine the extent to which different strength training exercises selectively

32 activate the commonly injured biceps femoris long head (BFLH) muscle. Methods: This two-

33 part observational study recruited 24 recreationally active males. Part 1 explored the

34 amplitudes and the ratios of lateral to medial hamstring (BF/MH) normalised

35 electromyography (nEMG) during the concentric and eccentric phases of 10 common

36 strength training exercises. Part 2 used functional magnetic resonance imaging (fMRI) to

37 determine the spatial patterns of hamstring activation during two exercises which i) most

38 selectively, and ii) least selectively activated the BF in part 1. Results: Eccentrically, the

39 largest BF/MH nEMG ratio was observed in the 45° hip extension exercise and the lowest

40 was observed in the Nordic hamstring (NHE) and bent-knee bridge exercises. Concentrically,

41 the highest BF/MH nEMG ratio was observed during the lunge and 45° hip extension and the

42 lowest was observed for the leg curl and bent-knee bridge. fMRI revealed a greater BFLH to

43 semitendinosus activation ratio in the 45° hip extension than the NHE (p<0.001). The T2

44 increase after hip extension for BFLH, semitendinosus and semimembranosus muscles were

45 greater than that for BFSH (p<0.001). During the NHE, the T2 increase was greater for the

46 semitendinosus than for the other hamstrings (p≤0.002). Conclusion: This investigation

47 highlights the non-uniformity of hamstring activation patterns in different tasks and suggests

48 that hip extension exercise more selectively activates the BFLH while the NHE preferentially

49 recruits the semitendinosus. These findings have implications for strength training

50 interventions aimed at preventing hamstring injury.

51 What are the new findings?

52  The hamstrings are activated non-uniformly during hip- and knee-based exercises
 Hip extension exercise more evenly activates the three long heads of the hamstrings
and the Nordic hamstring exercise preferentially recruits the semitendinosus (ST)

2
53 INTRODUCTION

54 Paragraph 1 Hamstring strain injuries (HSIs) are commonly experienced by athletes

55 involved in running-based sports. They are the most prevalent injury in track and field,[1]

56 Australian Rules football,[2] and soccer[3] and up to 30% recur within 12 months.[4]

57 Upwards of 80% of HSIs involve the biceps femoris long head (BFLH) muscle[5-7] and most

58 injuries are thought to occur during the late swing phase of high-speed running.[8] During

59 this phase of the gait cycle, the BFLH reaches its peak length and develops maximal force

60 while undergoing a forceful eccentric contraction to decelerate the shank for foot strike,[9]

61 and it is thought that these conditions may at least partly explain its propensity for injury. It

62 has also been reported that prior BFLH injury is associated with a degree of neuromuscular

63 inhibition[10 11] and prolonged atrophy[12], which suggests that current rehabilitation

64 practices do not adequately restore function to this muscle.

65 Paragraph 2 It has been proposed that hamstring weakness is a risk factor for future strain

66 injury[6 13 14] and interventions aimed at increasing strength, particularly eccentric knee

67 flexor strength, have been effective in reducing HSI rates in several sports.[15-18] However,

68 despite an increased focus on hamstring strength in prophylactic programs,[19] exercise

69 selection is often implemented on the basis of clinical recommendations and assumptions

70 rather than empirical evidence.[20 21] There is currently a very small body of work on the

71 activation patterns of the hamstrings during commonly employed exercises. Studies using

72 functional magnetic resonance imaging (fMRI) have shown that activation differs within and

73 between hamstring muscles during different tasks.[11 22-24] For example, the

74 semitendinosus (ST) appears to be selectively activated during the Nordic hamstring exercise

75 (NHE)[11] and the eccentric prone leg curl,[24] while the semimembranosus (SM) is

76 preferentially recruited during the stiff leg deadlift.[23] Surface electromyography (sEMG)

77 has been used extensively in the analysis of hamstring exercises.[23-26] However, these

3
 78 studies are sometimes contradictory and are often inconsistent with the results from fMRI.[11

79 22-25 27] The lack of complete agreement between fMRI and sEMG might reflect the

80 different physiological basis of each technique.[28] Surface EMG amplitude is sensitive to

81 the electrical activity generated by active motor units and is detected by electrodes overlying

82 the skin.[29] This provides valuable information on the neural strategies involved during

83 muscle activation with high temporal resolution, but is prone to cross talk[29] and cannot

84 discriminate between closely approximated segments of muscles[30] such as the medial

85 hamstrings (semimembranosus and semitendinosus). By contrast, fMRI is a relatively

86 new[31] technique which reflects the metabolic activity associated with exercise.[28] Muscle

87 activation is associated with a transient increase in the transverse (T2) relaxation time of

88 tissue water, which can be detected from signal intensity changes in fMR images. These T2

89 shifts, which, like sEMG, increase in proportion to exercise intensity,[31 32] can be mapped

90 in cross-sectional images of muscles and therefore provide significantly greater spatial clarity

91 than sEMG.[28 30]

92 Paragraph 3 An improved understanding of the patterns of hamstring muscle activation

93 during common strength training exercises may enable practitioners to make better informed

94 decisions regarding exercise selection in injury prevention and rehabilitation programs. These

95 data may also inform the design of training studies aimed at investigating the chronic

96 adaptations induced by different exercises. The purpose of this two-part study was to

97 determine the extent to which different exercises selectively activate the commonly injured

98 BFLH. Part 1 used sEMG to determine the amplitude and ratio of lateral to medial hamstring

99 activation during 10 commonly employed exercises. Based on these findings, part 2

100 employed fMRI to map muscle activation during two exercises that appeared to a) most

101 selectively; and b) least selectively activate the BF according to sEMG. We hypothesised that

102 the patterns of hamstring muscle activation would be non-uniform between exercises and, on

4
103 the basis of previous work,[23] that more selective activation of the BFLH would be observed

104 during hip-extension exercise.

105 METHODS

106 Participants

107 Paragraph 4 Twenty-four recreationally active male athletes (age, 24.4 ± 3.3 years, height,

108 181.8 ± 6.1 cm, weight, 85.2 ± 13.4 kg) participated in this study. Eighteen athletes (age, 23.9

109 ± 3.1, height, 180.6 ± 5.9, weight, 86.0 ± 14.8) participated in part 1 and ten athletes (age,

110 24.6 ± 4.0, height, 183.5 ± 7.0, weight, 83.5 ± 8.7) participated in part 2. A priori sample size

111 estimates were based on 1) the capacity to detect a 10% difference in the ratio of BF to MH

112 (BF/MH) sEMG amplitude between exercises;[25] and 2) an effect size of 1.0 in the

113 percentage change in T2 relaxation time between muscles,[11] at a power of 0.80 and with

114 p<0.05. Participants were free from soft tissue and orthopaedic injuries to the trunk, hips and

115 lower limbs at the time of testing and had no known history of cardiovascular, metabolic or

116 neurological disorders. Participants had no history of HSI in the previous 18 months and had

117 never suffered an anterior cruciate ligament injury. Prior to testing, all participants completed

118 a cardiovascular screening questionnaire to make sure it was safe for them to perform intense

119 exercise and those who were involved in part 2 also completed a standard MRI screening

120 questionnaire to ensure it was safe for them to enter the magnetic field. All participants

121 provided written, informed consent for this study, which was approved by the Queensland

122 University of Technology Human Research Ethics Committee and the University of

123 Queensland Medical Research Ethics Committee.

124 Study Design

125 Paragraph 5 This cross-sectional study involved two parts. In the first we explored the

126 sEMG amplitudes and ratios of BF to medial hamstring (MH) sEMG activity during ten

5
127 commonly employed strength training exercises. Based on these findings, part 2 involved an

128 fMRI investigation of two exercises which appeared to a) most selectively, and b) least

129 selectively activate the BF muscle during eccentric contractions.

130 PART 1

131 Paragraph 6 On a separate day prior to experimental testing, participants were familiarised

132 with the exercises used in this investigation. All were shown a demonstration of each exercise

133 (Figure 1) and performed several practice repetitions while receiving verbal feedback from

134 the investigators. Once the participant could complete the exercise with appropriate

135 technique, the loads were progressively increased until an approximate 12RM load was

136 determined (unless the exercise was already supramaximal, ie. NHE and glute-ham-raise). On

137 the day of testing, participants reported to the laboratory and were prepared for sEMG

138 measurement. The testing session began with two maximal voluntary isometric contractions

139 (MVICs) for the hamstrings. Subsequently, participants completed a single set of six

140 repetitions of each exercise, each with the predetermined 12RM load, in randomised order.

141 All data were sampled from a randomly selected limb (dominant or non-dominant), which

142 was the exercised limb during all unilateral movements and all testing sessions were

143 supervised by the same investigator (MNB) to ensure consistency of procedures.

144

145 Exercise Protocol

146 Paragraph 7 The 10 exercises were chosen based on a review of the scientific literature[23

147 25 27]. They included the bilateral and unilateral stiff-leg deadlift, hip hinge, lunge, unilateral

148 bent and straight knee bridges, leg curl, 45° hip extension, glute-ham-raise and the NHE.

149 Unless the exercise was explosive (hip hinge) or supramaximal and eccentric-only (NHE and

6
150 glute-ham raise) participants completed both the concentric and eccentric phases of each

151 exercise using a 12-RM load at a constant pace (~2s up and ~2s down).

152

153 INSERT FIGURE 1

154

155 Electromyography

156 Paragraph 8 Bipolar pre-gelled Ag/AgCl sEMG electrodes (10mm diameter, 15mm

157 interelectrode distance) (Ambu, BlueSensor N) were used to record electromyographical

158 activity from the BF and MH . The skin of the participants was shaved, lightly abraded and

159 cleaned with alcohol before electrodes were placed on the posterior thigh, midway between

160 the ischial tuberosity and tibial epicondyles. Electrodes were oriented parallel to the line

161 between these two landmarks, as per SENIAM guidelines,[33] and secured with tape to

162 minimise motion artefact. The reference electrode was placed on the ipsilateral head of the

163 fibula. Muscle bellies of the BF and MH were identified via palpation and correct placement

164 was confirmed by observing active external and internal rotation of the knee in 90° of

165 flexion.[10] During all exercise trials, hip and knee joint angles were measured

166 simultaneously with sEMG data using two digital goniometers. The hip sensor’s axis of

167 rotation was aligned with the greater trochanter of the femur and the knee sensor was

168 positioned superficial to the lateral femoral epicondyle.

169 Maximal voluntary contraction

170 Paragraph 9 Surface EMG activity was recorded during MVICs of the hamstrings using a

171 custom-made device which was fitted with two uniaxial load cells.[34] Participants lay prone

172 with their hips in 0° of flexion and knees fully extended (180°), with their ankles secured in

7
173 immoveable yokes and were asked to perform forceful knee flexion while investigators

174 provided strong verbal encouragement. After 1-2 warm-up contractions, participants

175 completed two 3-4s MVICs, with 30-sec of rest separating each attempt. The contraction that

176 elicited the highest average amplitude for the BF and MH was used to represent the maximal

177 EMG amplitude.

178 Data analysis

179 Paragraph 10 All sEMG and joint angle data were sampled at 1 kHz through a 16-bit

180 PowerLab 26T AD unit (ADInstruments, New South Wales, Australia) (amplification =

181 1000; common mode rejection ratio = 110dB) and analysed using LabChart 8.0 (AD

182 Instruments, New South Wales, Australia). Raw sEMG data were filtered using a Bessel filter

183 (frequency bandwidth = 10-500Hz) and then full wave rectified. Joint angle data were used to

184 determine the concentric and eccentric phases of each repetition for each exercise. For each

185 phase, the filtered sEMG signal was normalised to values obtained during MVIC and these

186 normalised sEMG (nEMG) values were averaged across the six repetitions.

187

188 Statistical analysis

189 Paragraph 11 Data were analysed using JMP version 10.02 (SAS Institute Inc, 2012).

190 Descriptive statistics were calculated for mean nEMG amplitudes of BF and MH for the

191 concentric and eccentric phases of each exercise and an activation ratio was determined by

192 dividing the average BF nEMG amplitude by the average MH nEMG amplitude (BF/MH);

193 ratios >1.0 indicated that the BF was more active than the MH muscles. For both the

194 concentric and eccentric phases, repeated measures linear mixed models fitted with the

195 restricted maximum likelihood method were used to determine differences between exercises.

8
196 For this analysis, exercise was the fixed factor and participant identity the random factor.

197 When a significant main effect was observed for exercise, post hoc t-tests with Bonferroni

198 corrections were used to identify the source and reported as mean differences with 95% CIs.

199 For these analyses, the Bonferroni adjusted p value was set at <0.002.

200

201 PART 2

202 Paragraph 12 A cross-sectional design was used to map the spatial patterns of hamstring

203 muscle activation during the 45° hip extension and NHE. These exercises were chosen

204 because they a) most selectively (45° hip extension) and b) least selectively (NHE) activated

205 the BF muscle during eccentric contractions according to sEMG. Participants completed two

206 separate exercise sessions, separated by at least six days (14 ± 5 days), with each session

207 involving one of the aforementioned exercises. Functional MRI scans of both thighs were

208 acquired before and immediately after each exercise bout. All testing sessions were

209 supervised by the same investigator (MNB).

210 Exercise Protocol

211 Paragraph 13 A depiction of the 45° hip extension and NHE can be found in Figure 1. All

212 exercise was completed using the same equipment as that used in part 1. Participants

213 completed five sets of 10 repetitions of each exercise with one-minute rest intervals between

214 sets. The higher volume of exercise (compared to part 1) was necessary because transient T2

215 changes reflect fluid shifts associated with glycolysis and have a higher detection threshold

216 than sEMG.[28] All subjects completed 50 repetitions successfully. During the rest periods,

217 participants remained in a seated position (for the hip extension exercise) or lay prone (NHE)

218 to minimise activation of the hamstrings. The 45° hip extension exercise was performed

9
219 unilaterally (with the limb chosen randomly) with a starting load corresponding to each

220 participant’s approximate 12-RM (median = 10kg; range = 10 to 20kg). However if the

221 participant could no longer complete the exercise with the allocated load, the weight was

222 gradually reduced by increments of 5kg until it could be completed at the desired speed (2sec

223 up and 2 sec down), which was controlled by an electronic metronome. The NHE was

224 performed bilaterally with body weight only. Participants received verbal support from the

225 investigators throughout all exercise sessions to promote maximal effort. All participants

226 were returned to the scanner immediately following the cessation of exercise and post-

227 exercise scans began within 148.6 ± 24sec (mean ± SD).

228 Functional muscle magnetic resonance imaging (fMRI)

229 Paragraph 14 All fMRI scans were performed using a 3-Tesla (Siemens TrioTim, Germany)

230 imaging system with a spinal coil. The participant was positioned supine in the magnet bore

231 with their knees fully extended and hips in neutral and straps were secured around both limbs

232 to prevent any undesired movement. Consecutive T2-weighted axial images were acquired of

233 both limbs beginning at the level of the iliac crest and finishing distal to the tibial plateau

234 using a 180 x 256 image matrix. Images were acquired before and immediately after exercise

235 using a Car-Purcel-Meiboom-Gill (CPMG) spin-echo pulse sequence and the following

236 parameters: transverse relaxation time (TR) = 2540ms; echo time (TE) = 8, 16, 24, 32, 40, 48

237 and 56ms; number of excitations = 1; slice thickness = 10mm; interslice gap = 10mm; field of

238 view = 400 x 281.3mm). The total acquisition time for each scan was 6min 24s. A localiser

239 adjustment (20s) was applied prior to the first sequence of each scan to standardise the field

240 of view and to align collected images between the pre- and post- exercise scans.[11] To

241 minimise any inhomogeneity in MR images caused by dielectric resonances at 3T, a (B1)

242 filter was applied to all scans; this is a post-processing image filter that improves the image

243 signal intensity profile without affecting the image contrast. In addition, to ensure that the

10
244 signal intensity profile of T2-weighted images was not disrupted by anomalous fluid shifts,

245 participants were instructed to avoid any exhaustive resistance training of the lower limbs in

246 the week preceding testing, and were seated for a minimum of 15 minutes[23] before pre-

247 exercise imaging.

248 Paragraph 15 For each exercise session, the T2 relaxation times of each hamstring muscle

249 were measured in T2-weighted images acquired before and after exercise to evaluate the

250 degree of muscle activation during exercise. All fMRI scans were transferred to a Windows

251 computer in the digital imaging and communications in medicine (DICOM) file format. The

252 T2 relaxation times of each hamstring muscle (BFLH, BFSH, ST and SM) were measured in

253 five axial slices, corresponding to 30, 40, 50, 60 and 70% of thigh length; these values were

254 determined relative to the distance between the inferior margin of the ischial tuberosity (0%)

255 and the superior border of the tibial plateau (100%).[11 23] Image analysis software (Sante

256 Dicom Viewer and Editor, Cornell University) was used to measure the signal intensity of

257 each hamstring muscle in the exercised limb in both the pre- and post-exercise scans. The

258 signal intensity was measured manually in each slice using a circular region of interest

259 (ROI)[27] which was placed in a homogenous region of contractile tissue in each muscle

260 belly (avoiding fat, aponeurosis, tendon, bone and blood vessels). The size of each ROI

261 varied (0.2 to 5.6 cm2) based on the cross-sectional area and the amount of homogeneous

262 tissue available in each slice. The signal intensity reflected the mean value of all pixels within

263 the ROI and was measured across seven echo times (8, 16, 24, 32, 40, 48, 56ms). To calculate

264 the T2 relaxation time for each ROI, the signal intensity value at each echo time was fitted to

265 a mono-exponential decay model using a least squares algorithm:

266 [(SI= M  exp(echo time / T2)][23]

11
267 where SI is the signal intensity at a specific echo time, and M represents the pre-exercise

268 fMRI signal intensity. To assess the extent to which each ROI was activated during exercise,

269 the mean percentage change in T2 was calculated as:

270 [(mean post-exercise T2 / mean pre-exercise T2) x 100].

271 To provide a meaningful measure of whole-muscle activation, the percentage change in T2

272 relaxation time for each hamstring muscle was evaluated using ROIs from all five thigh

273 levels. Previous studies have demonstrated excellent reliability of T2 relaxation time

274 measures with intra-class correlation coefficients ranging from 0.87 to 0.94.[28 35]

275

276 Statistical analysis

277 Paragraph 16 Absolute T2 values before and after each exercise session were reported

278 descriptively as mean ± SD. Repeated measures linear mixed models fitted with the restricted

279 maximum likelihood (REML) method were used to determine the spatial activation patterns

280 of the hamstring muscles during the 45° hip extension and NHE. The percentage change in

281 T2 relaxation time was compared between each hamstring muscle (BFLH, BFSH, ST and SM)

282 for both exercises. For this analysis, muscle was the fixed factor and both participant identity

283 and participant identity x muscle the random factors. When a significant main effect was

284 detected for muscle, post hoc t tests with Bonferroni corrections were used to determine the

285 source; the adjusted alpha was set at p<0.008. Given that the two examined exercises differed

286 in intensity and contraction mode(s), it was not appropriate to directly compare the magnitude

287 of the T2 shifts between exercises.[36] Instead, repeated measures linear mixed models fitted

288 with the REML method were used to determine differences in the ratio of BF to ST (BFLH/ST

289 and BFSH/ST) and SM to ST (SM/ST) percentage change in T2 relaxation time between

290 exercises. For these analyses exercise was the fixed factor and participant identity the random

291 factor. When a main effect was found for exercise, post hoc t tests were again used to

12
292 determine the source and reported as mean difference (and 95% CI). Alpha was set at p<0.05

293 for these analyses.

294

295

296

297

298

299

300

13
301 RESULTS

302 Levels of hamstring muscle activation

303 Paragraph 17 Average BF muscle activity ranged from 21.4% (lunge) to 99.3% (unilateral

304 straight knee bridge) MVIC during the concentric phase and 10.7% (hip hinge) to 71.9%

305 (NHE) during the eccentric phase. Average MH muscle activity ranged from 18.1% (lunge)

306 to 120.7% (leg curl) during the concentric phase and 11.6% (hip hinge) to 101.8% (NHE)

307 during the eccentric phase.

308

309 Concentric biceps femoris to medial hamstring (BF:MH) activation ratio

310 Paragraph 18 The concentric BF/MH activation ratio for each exercise can be found in

311 Figure 2a. A significant main effect was detected between exercises (p <0.001) with post hoc

312 t tests showing that the BF/MH ratio was greater during the lunge than the leg curl (mean

313 difference = 0.8, 95% CI = 0.5 to 1.1, p<0.001) and bent-knee bridge (mean difference = 0.7,

314 95% CI = 0.4 to 1.1, p<0.001). Similarly, the BF/MH ratio was greater in the 45° hip

315 extension exercise than the leg curl (mean difference = 0.6, 95% CI = 0.3 to 1.0, p<0.001)

316 and bent-knee bridge (mean difference = 0.6, 95% CI = 0.2 to 0.9, p=0.001).

317

318 Eccentric biceps femoris to medial hamstring (BF:MH) activation ratio

319 Paragraph 19 The eccentric BF/MH activation ratio for each exercise can be found in Figure

320 2b. A significant main effect was observed for exercise (p<0.001) with post hoc analyses

321 revealing that the BF/MH ratio was significantly greater in the 45° hip extension than the

322 NHE (mean difference = 0.7, 95% CI = 0.4 to 1.0, p<0.001), bent-knee bridge (mean

14
323 difference = 0.7, 95% CI = 0.4 to 1.0, p<0.001), leg curl (mean difference = 0.6, 95% CI =

324 0.3 to 0.9, p<0.001) and the glute-ham raise (mean difference = 0.6, 95% CI = 0.3 to 0.9,

325 p<0.001). No other between-exercise differences were observed once adjusted for multiple

326 comparisons (p>0.002).

327

328 INSERT FIGURE 2

329

330 Percentage change in T2 relaxation time following the 45° hip extension exercise

331 Paragraph 20 A significant main effect was observed for muscle (p<0.001) with post hoc t

332 tests revealing that the exercise-induced T2 changes in the BFSH were significantly lower than

333 those observed for the BFLH (mean difference = 60.7%, 95% CI = 41.3 to 80.1%, p<0.001),

334 ST (mean difference = 78.0%, 95% CI = 58.4 to 97.6%, p<0.001) and SM muscles (mean

335 difference = 49.8%, 95% CI = 30.1 to 69.5%, p<0.001) (Figure 3). The T2 change for ST was

336 significantly greater than SM (mean difference = 28.2%, 95% CI = 9.2 to 47.1%, p=0.005)

337 however, no difference was observed between the BFLH and SM (p=0.245) or between the

338 BFLH and ST muscles (p=0.067). Absolute T2 values before and after the hip extension

339 exercise are reported in Table 1.

340

341 INSERT FIGURE 3

342

343 Percentage change in T2 relaxation time following the Nordic hamstring exercise

344 Paragraph 21 A main effect was detected for muscle (p<0.001). Post hoc analyses showed

345 that the T2 changes induced by exercise within the ST were significantly larger than those
15
346 observed for the BFLH (mean difference = 29.8%, 95% CI = 20.5 to 39.2%, p<0.001), BFSH

347 (mean difference = 16.2%, 95% CI = 6.4 to 26.0%, p=0.002) and SM (mean difference =

348 29.9%, 95% CI = 20.4 to 39.4%, p<0.001) muscles (Figure 4). In addition, the T2 increase

349 observed for BFSH was significantly greater than for the BFLH (mean difference = 13.7%,

350 95% CI = 3.9 to 23.4%, p=0.008) and SM (mean difference = 13.8, 95% CI = 3.8 to 23.7,

351 p=0.008) muscles. No difference was observed between the BFLH and SM muscles (p=0.982).

352 The absolute T2 values before and after the NHE are reported in Table 1.

353

354 INSERT FIGURE 4

355

356 Comparison of hamstring activation ratios between exercises

357 Paragraph 22 When comparing the BFLH/ST ratio, a significant main effect was observed

358 for exercise (p<0.001) with post hoc analyses revealing a significantly greater ratio during

359 45° hip extension exercise than during the NHE (mean difference = 0.7, 95% CI = 0.6 to 0.9,

360 p<0.001) (Figure 5).

361

362 INSERT FIGURE 5

363

364 Paragraph 23 A significant main effect was also detected for exercise when comparing the

365 BFSH/ST ratio (p<0.001). Post hoc t tests demonstrated that this ratio was significantly greater

366 during the NHE than during the 45° hip extension exercise (mean difference = 0.42, 95% CI

367 = 0.24 to 0.62, p<0.001).When comparing the SM/ST ratio a significant main effect was

16
368 detected for exercise (p<0.001) with post hoc t tests showing relatively higher ratios during

369 the 45° hip extension than during the NHE (mean difference = 0.51, 95% CI = 0.39 to 0.64,

370 p<0.001).

371

372 Table 1. T2 relaxation time values measured before (T2 Pre) and immediately after (T2 Post)

373 the 45° hip extension and Nordic hamstring exercise (NHE) sessions.

45° hip extension NHE

T2 Pre (ms) T2 Post (ms) T2 Pre (ms) T2 Post (ms)

BFLH 35.61 (±8.13) 62.56 (±17.34) 35.84 (±8.83) 39.76 (±11.13)

BFSH 31.03 (±3.85) 36.17 (±4.65) 32.36 (±6.05) 40.57 (±9.64)

ST 38.68 (±11.89) 74.18 (±20.06) 37.31 (±10.74) 53.04 (±19.97)

SM 44.26 (±13.23) 71.09 (17.48) 41.39 (±10.87) 45.4 (±11.67)

374
375 Data are presented as mean values (±SD). BFLH, biceps femoris long head; BFSH, biceps
376 femoris short head; ST, semitendinosus; SM, semimembranosus.

377

378

17
379 DISCUSSION

380 Paragraph 24 The primary aim of this study was to determine movements that most

381 selectively activate the commonly injured BFLH. The results support the hypothesis that

382 hamstring activation patterns differ markedly between exercises and provide evidence to

383 suggest that hip extension exercise more selectively targets the BFLH than the NHE.

384 Paragraph 25 The NHE has been shown, in a number of studies,[15 16 18] to be very

385 effective at reducing HSIs in soccer players as long as compliance is adequate.[37] However,

386 we[11] and others[22] have previously reported that the NHE preferentially activates the ST

387 and this might be interpreted as evidence that the exercise is sub-optimal to protect against

388 running-related strain injury. It is entirely possible that the NHE confers injury-preventive

389 benefits via an improved load-bearing capacity of its agonist,[38] however, in this study, we

390 have provided EMG evidence which shows, despite the relatively selective activation of the

391 ST, that the lateral hamstrings were still strongly activated during the NHE. Indeed, BF

392 nEMG was higher during the NHE than during the eccentric phase of any other exercise and

393 the evidence for this exercise’s protective effects[15 16 18] suggests that eccentric actions

394 alone in a training program are sufficient to make the hamstrings more resistant to strain

395 injury. High levels of BF nEMG during the NHE are consistent with previous

396 investigations[25] and are the result of the supramaximal intensity of the exercise, which

397 potentially explains why high levels of BF nEMG were also observed in the eccentric glute-

398 ham-raise. High levels of BF nEMG in concentric actions were observed in several other

399 exercises including the straight-knee bridge, leg curl and the 45° hip extension which

400 corroborates previous observations.[25] However, the importance of hamstring activation

401 patterns during concentric actions remains unclear from the perspective of injury prevention.

402 Paragraph 26 While high levels of nEMG are an important stimulus for improving strength

403 and voluntary activation,[39] exercise selectivity may still have important implications for

18
404 rehabilitation. For example, inhibition of previously injured BF muscles during eccentric

405 actions has been reported many months after rehabilitation,[10 11 40] and it has been

406 proposed[41] that these deficits might partly explain observations of persistent eccentric knee

407 flexor weakness,[10] BFLH atrophy[12] and a chronic shortening of BFLH fascicles.[42] These

408 data[10-12 40 42] are consistent with the possibility that conventional rehabilitation strategies

409 may not adequately target the commonly injured BFLH. Previous studies have shown that the

410 ratio of lateral to medial hamstring (BF/MH) sEMG varies with foot rotation[43] and differs

411 between exercises.[25] In the current study, the eccentric phase of the 45° hip extension

412 exercise exhibited the greatest BF/MH nEMG ratio (1.5 ± 0.1) while the NHE (0.8 ± 0.1) and

413 bent-knee bridge exercises (0.8 ± 0.1) displayed the lowest ratios. These observations were

414 confirmed in the subsequent fMRI analysis whereby the ratio of BFLH to ST in the 45° hip

415 extension exercise (0.96 ± 0.09) was markedly higher than that observed for the NHE (0.23 ±

416 0.08). It is also noteworthy that the eccentric phase of other hip-oriented exercises (straight-

417 knee bridge, unilateral and bilateral stiff-leg deadlift and hip hinge) displayed BF/MH nEMG

418 ratios >1.0. In contrast, the eccentric phase of exercises that involved significant movement at

419 the knee (leg curl, glute-ham-raise, bent-knee bridge and NHE) had higher levels of medial

420 hamstring nEMG (BF/MH ratio <1.0). These data suggest that hamstring activation strategies

421 are partly dependent on the joints involved in each movement. During concentric

422 contractions, the most selective BF activation was observed in the lunge exercise which

423 corroborates a previous fMRI investigation.[27] However, it is important to consider that the

424 lunge also exhibited the lowest BF nEMG amplitude (21.4 ± 7.4%) of any exercise which

425 likely renders it an inadequate stimulus for improving strength or stimulating hypertrophy in

426 this muscle.[39] Interestingly, the exercise that least selectively activated the BF during

427 concentric contractions was the leg curl, which mimics the joint positions and hamstring

428 muscle-tendon lengths experienced in the NHE.

19
429 Paragraph 27 The mechanism for higher levels of BFLH activity during hip extension-

430 oriented movements remains unclear, however, it is possible that hamstring muscle moment

431 arms play a role. For example, the BFLH exhibits a larger moment arm at the hip than at the

432 knee[44] and therefore possesses a greater mechanical advantage at this joint. As a result, the

433 BFLH undergoes significantly more shortening during hip extension than knee flexion.[44] By

434 contrast, the ST displays a larger sagittal plane moment arm at the knee than both BFLH and

435 SM,[44] which may explain its preferential recruitment during movements at this joint, such

436 as the NHE and leg curl exercises. It is also noteworthy that the ST is a fusiform muscle with

437 long fibre lengths and many sarcomeres in series, which potentially makes it well-suited to

438 forceful eccentric contractions[45] such as those experienced in the NHE. Further work is

439 needed to clarify the mechanisms underpinning these unique strategies of hamstring

440 activation during hip and knee movements.

441 Paragraph 28 The current findings are different to some others that have investigated

442 hamstring activation patterns. For example, Zebis and colleagues[25] recently reported that

443 both the NHE and the prone isokinetic leg curl were performed with very similar levels of ST

444 and BFLH nEMG. However, in the current investigation, the NHE and leg curl exercises

445 resulted in more selective activation of the medial hamstrings and, in the case of the NHE, the

446 fMRI results also suggest selective use of the ST muscle. Differences between these studies

447 may conceivably be related to participant sex (females[25] versus males in the current study),

448 electrode placement, and the fact that this earlier work did not differentiate between the

449 concentric and eccentric phases of each exercise. However, it is also important to consider

450 that sEMG does not have the spatial resolution of fMRI and cannot reliably distinguish

451 between neighbouring muscles,[30] such as the long and short heads of BF or the ST and SM,

452 which appear to display distinct activation magnitudes.[11 23 24] These data highlight the

20
453 limitations of relying exclusively on sEMG to infer strategies of hamstring muscle activation

454 during exercise and suggest the need for more spatially robust methods in future work.

455 Paragraph 29 In interpreting the results of this study, it is important to consider that sEMG

456 and fMRI techniques measure different aspects of muscle activity. The absence of T2

457 relaxation time changes in people with McCardle’s disease[46] suggests that fMRI is

458 sensitive to glycolysis and it is thought that the osmotic fluid shifts which persist after

459 exercise and give rise to T2 changes are a consequence of the accumulation of glycolytic

460 metabolites.[36] Fortunately, the proportion of Type II glycolytic fibres does not appear to

461 vary across the hamstring muscles[47] so this is unlikely to be a confounding factor in this

462 study. However, exercise induced changes in T2 will be influenced by contraction mode

463 because concentric work is characterised by higher nEMG amplitudes[29] and is markedly

464 less efficient than eccentric work against the same loads.[48] As a consequence, the

465 differences in T2 relaxation time changes after the 45° hip extension exercise which involved

466 concentric and eccentric actions and the almost entirely eccentric NHE do not reflect only the

467 levels of voluntary muscle activation. Instead, fMRI can offer insights into the relative

468 metabolic activity and reliance upon different hamstring muscles in each exercise. According

469 to fMRI, the NHE involves preferential ST use with modest use of the other hamstrings,

470 while the 45° hip extension exercise appears to heavily recruit both the BFLH and ST muscles.

471 These observations are largely consistent with the sEMG component of this study, which also

472 suggested higher activation of the medial than lateral hamstrings in the NHE and more even

473 activation of the medial and lateral hamstrings in the 45° hip extension.

474 Paragraph 30 Characterising the activation patterns of the hamstrings during different tasks

475 is an important first step in identifying exercises worthy of further investigation, however,

476 electrical or metabolic activity of muscles should not be the only factors considered in

477 exercise selection. Indeed, despite the BFLH being more active in hip extension, there is

21
478 currently no evidence to suggest that training with this exercise actually leads to a reduction

479 in the risk of HSI. Further work is required to understand how the hamstrings adapt to this

480 and other exercises and adaptation is influenced by a range of factors, such as contraction

481 mode[49] and range of motion,[45] which were not a part of the current investigation. For

482 example, there is little reason to believe that concentric or concentrically-biased exercise is

483 effective in HSI prevention or rehabilitation programs.[17] Indeed, there is evidence that

484 concentric training shortens BFLH fascicles[49] and shifts knee flexor torque-joint angle

485 relationships towards shorter muscle lengths[50] and neither of these adaptations are

486 considered beneficial for HSI prevention.[7 51] Because eccentric and concentric training

487 programs appear to have opposing effects on fascicle lengths,[49] it is possible that exercises

488 combining contraction modes may have minimal or at least blunted effects on muscle

489 architecture. Future studies are needed to assess the impact of certain exercises on known or

490 proposed risk factors for HSI such as eccentric strength[6] and fascicle lengths,[7] and only

491 then will there be sufficient evidence to justify use of those exercises in intervention studies

492 aimed at reducing the risk of injury. Based on the current findings, for example, it seems

493 logical to compare the effects of training programs including the NHE and the 450 hip

494 extension exercises on the abovementioned variables.

495 Limitations

496 Paragraph 31 Given the high cost of fMRI, it was not possible to include all participants in

497 both parts of the experiment. Therefore, comparing the results of part 1 and 2 should be done

498 with caution. Furthermore, all of our participants were recreationally active so it remains to

499 be seen if these findings can be applied to more highly trained athletes. We have previously

500 shown that recreationally active young men with a history of unilateral hamstring strain

501 exhibited less T2 change in previously injured muscles than in their uninjured homologous

502 muscles from the contralateral limbs after performing the NHE.[22] More work will be

22
503 needed to establish whether the patterns of selective muscle activation observed in the current

504 study are also evident in athletes with a history of strain injury.[10 11 40] Lastly, it should be

505 acknowledged that the T2 response to an exercise stimulus is highly dynamic and can be

506 influenced by a range of factors such as the metabolic capacity and vascular dynamics of the

507 active tissue.[28 36] We attempted to minimise this by recruiting only male participants with

508 a similar age and training status.

509 Conclusion

510 Paragraph 32 The current study suggests that the patterns of hamstring muscle activation are

511 heterogeneous across a range of different strength exercises. We have provided sEMG

512 evidence to suggest that, during eccentric contractions, hip extension exercise more

513 selectively activates the lateral hamstrings while knee flexion-oriented exercises

514 preferentially recruit the medial hamstrings. However, despite being the least selective

515 activator of the BF, the NHE still elicited higher levels of BF nEMG during eccentric actions

516 than any other exercise which may help to explain how it confers HSI-preventive benefits.[15

517 16 18] The results of the fMRI investigation largely confirm our initial sEMG observations,

518 showing that, relative to the ST, the BFLH was ~4 times more active in hip extension than the

519 NHE. However, they also show that the BFLH, BFSH, ST and SM display distinct patterns of

520 muscle use during different tasks. Collectively, the results of this study highlight the

521 limitations of relying on a single method to infer strategies of muscle activation and suggest

522 that the hip extension exercise may be useful for improving strength and voluntary activation

523 of the commonly injured BFLH. Future work is needed to determine the effect of this and

524 other exercises on hamstring architecture and morphology before we can justify their

525 inclusion in interventions aimed at reducing the risk of HSI.

526

23
527
How might it impact upon clinical practice in the future?
528
 Hamstring injury prevention and rehabilitation exercises can potentially be targeted to
529
the site of injury.
530
 Hip extension exercise may be more useful than the NHE for selectively activating
531 the commonly injured BFLH.
532

533

534

535

536

537

538

539

540

541

542

543

544

545

546

547

548

549

550

551

24
552 ACKNOWLEDGEMENTS

553 We thank the Queensland Academy of Sport’s Centre of Excellence for Applied Sport

554 Science Research for funding this investigation. The authors also acknowledge the facilities,

555 and the scientific and technical assistance of the National Imaging Facility at the Centre for

556 Advanced Imaging, University of Queensland.

557 CONTRIBUTORS

558 MB was the principle investigator and was involved with study design, recruitment, analysis and

559 manuscript write up. MW, DO, GK and TS were involved with the study design, analysis and

560 manuscript preparation. AA was involved in fMRI data acquisition. All authors had full access to all

561 of the data (including statistical reports and tables) in the study and can take responsibility for the

562 integrity of the data and the accuracy of the data analysis.

563 TRANSPARENCY DECLARATION

564 The lead author* (MB) affirms that this manuscript is an honest, accurate, and transparent account of

565 the study being reported; that no important aspects of the study have been omitted; and that any

566 discrepancies from the study as planned (and, if relevant, registered) have been explained. * = The

567 manuscript’s guarantor.

568 COPYRIGHT DECLARATION

569 The Corresponding Author has the right to grant on behalf of all authors and does grant on behalf of

570 all authors, a worldwide licence to the Publishers and its licensees in perpetuity, in all forms, formats

571 and media (whether known now or created in the future), to i) publish, reproduce, distribute, display

572 and store the Contribution, ii) translate the Contribution into other languages, create adaptations,

573 reprints, include within collections and create summaries, extracts and/or, abstracts of the

574 Contribution, iii) create any other derivative work(s) based on the Contribution, iv) to exploit all

575 subsidiary rights in the Contribution, v) the inclusion of electronic links from the Contribution to third

25
576 party material where-ever it may be located; and, vi) licence any third party to do any or all of the

577 above.

578 DATA SHARING

579 Consent was not obtained for data sharing but the presented data are anonymised and risk of

580 identification is low.

581 FUNDING

582 This study was funded by the Queensland Academy of Sport’s Centre of Excellence for Applied

583 Sports Science Research.

584 COMPETING INTERESTS

585 None declared. All authors have completed the Unified Competing Interest form

586 at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare

587 that (1) the Queensland Academy of Sport’s Centre of Excellence for Applied Sports Science

588 Research funded this study; (2) MB, MW, DO, GK, AA and TS have no relationships with companies

589 that might have an interest in the submitted work in the previous 3 years; (3) their spouses, partners,

590 or children have no financial relationships that may be relevant to the submitted work; and (4) MB,

591 MW, DO, GK, AA and TS have no non-financial interests that may be relevant to the submitted work.

592

593 ETHICAL CLEARANCE

594 All participants provided written, informed consent for this study, which was approved by the

595 Queensland University of Technology Human Research Ethics Committee and the University of

596 Queensland Medical Research Ethics Committee.

597

598

26
599

600

601

602

603

27
604 REFERENCES

605 1. Opar DA, Drezner J, Shield A, et al. Acute hamstring strain injury in track-and-field

606 athletes: A 3-year observational study at the Penn Relay Carnival. Scand J Med Sci

607 Sports 2013; 24 (4): 254–259.

608 2. Orchard JW, Seward H, Orchard JJ. Results of 2 decades of injury surveillance and public

609 release of data in the Australian Football League. Am J Sports Med 2013;41(4):734-

610 41.

611 3. Ekstrand J, Hagglund M, Walden M. Injury incidence and injury patterns in professional

612 football: the UEFA injury study. Br J Sports Med 2011;45(7):553-8.

613 4. Orchard J, Best TM. The management of muscle strain injuries: an early return versus the

614 risk of recurrence. Clin J Sport Med 2002;12(1):3-5.

615 5. Koulouris G, Connell DA, Brukner P, et al. Magnetic Resonance Imaging Parameters for

616 Assessing Risk of Recurrent Hamstring Injuries in Elite Athletes. Am J Sports Med

617 2007;35(9):1500-06.

618 6. Opar DA, Williams MD, Timmins RG, et al. Eccentric Hamstring Strength and Hamstring

619 Injury Risk in Australian Footballers. Med Sci Sports Exerc 2014;47(4):857-65.

620 7. Timmins R, Bourne M, Shield A, et al. Biceps femoris long head architecture, eccentric

621 knee flexor strength and hamstring injury risk in professional football (soccer): a

622 prospective cohort study. Medicine and science in sports and exercise. Br J Sports

623 Med In press.

624 8. Schache AG, Wrigley TV, Baker R, et al. Biomechanical response to hamstring muscle

625 strain injury. Gait Posture 2009;29(2):332-8.

626 9. Chumanov ES, Heiderscheit BC, Thelen DG. Hamstring musculotendon dynamics during

627 stance and swing phases of high-speed running. Med Sci Sports Exerc 2011;43(3):525-

628 32.

28
629 10. Opar DA, Williams MD, Timmins RG, et al. Knee flexor strength and bicep femoris

630 electromyographical activity is lower in previously strained hamstrings. J

631 Electromyogr Kinesiol 2013; 23(3):696-703.

632 11. Bourne M, Opar,DA, Williams,MD, Al Najjar, A, Shield, AJ Muscle activation patterns

633 in the Nordic hamstring exercise: Impact of prior strain injury. Scand J Med Sci

634 Sports 2015 doi: 10.1111/sms.12494.

635 12. Silder A, Heiderscheit B, Thelen D, et al. MR observations of long-term musculotendon

636 remodeling following a hamstring strain injury. Skeletal Radiol 2008;37:1101-09.

637 13. Croisier JL, Ganteaume S, Binet J, et al. Strength imbalances and prevention of hamstring

638 injury in professional soccer players. Am J Sports Med 2008;36(8):1469.

639 14. Thorborg K. What are the most important risk factors for hamstring muscle injury? Clin J

640 Sport Med 2014;24(2):160-1.

641 15. Arnason A, Andersen TE, Holme I, et al. Prevention of hamstring strains in elite soccer:

642 an intervention study. Scand J Med Sci Sports 2008;18(1):40-8.

643 16. Petersen J, Thorborg K, Nielsen MB, et al. Preventive Effect of Eccentric Training on

644 Acute Hamstring Injuries in Men’s Soccer: A Cluster-Randomized Controlled Trial.

645 Am J Sports Med 2011;39(11):2296-303.

646 17. Askling CM, Tengvar M, Tarassova O, et al. Acute hamstring injuries in Swedish elite

647 sprinters and jumpers: a prospective randomised controlled clinical trial comparing

648 two rehabilitation protocols. Br J Sports Med 2014;48(7):532-9.

649 18. van der Horst N, Smits DW, Petersen J, et al. The preventive effect of the nordic

650 hamstring exercise on hamstring injuries in amateur soccer players: a randomized

651 controlled trial. Am J Sports Med 2015;43(6):1316-23.

29
652 19. Heiderscheit BC, Sherry MA, Silder A, et al. Hamstring strain injuries:

653 Recommendations for diagnosis, rehabilitation and injury prevention. J Orthop Sports

654 Phys Ther 2010;40(2):67.

655 20. Malliaropoulos N, Mendiguchia J, Pehlivanidis H, et al. Hamstring exercises for track and

656 field athletes: injury and exercise biomechanics, and possible implications for

657 exercise selection and primary prevention. Br J Sports Med 2012;46(12):846-51.

658 21. Guex K, Millet GP. Conceptual framework for strengthening exercises to prevent

659 hamstring strains. Sports Med 2013;43(12):1207-15.

660 22. Mendiguchia J, Arcos AL, Garrues MA, et al. The use of MRI to evaluate posterior thigh

661 muscle activity and damage during Nordic hamstring exercise. J Strength Cond Res

662 2013:27 (12):3426–3435.

663 23. Ono T, Higashihara A, Fukubayashi T. Hamstring functions during hip-extension

664 exercise assessed with electromyography and magnetic resonance imaging. Res Sports

665 Med 2011;19(1):42-52.

666 24. Ono T, Okuwaki T, Fukubayashi T. Differences in activation patterns of knee flexor

667 muscles during concentric and eccentric exercises. Res Sports Med 2010;18(3):188-

668 98.

669 25. Zebis MK, Skotte J, Andersen CH, et al. Kettlebell swing targets semitendinosus and

670 supine leg curl targets biceps femoris: an EMG study with rehabilitation implications.

671 Br J Sports Med 2013;47(18):1192-8.

672 26. Ditroilo M, De Vito G, Delahunt E. Kinematic and electromyographic analysis of the

673 Nordic Hamstring Exercise. J Electromyogr Kinesiol 2013;23(5):1111-8.

674 27. Mendiguchia J, Garrues MA, Cronin JB, et al. Non-uniform changes in MRI

675 measurements of the thigh muscles following two hamstring strengthening exercises.

676 J Strength Cond Res 2013; 27(3):574-81

30
677 28. Cagnie B, Elliott JM, O'Leary S, et al. Muscle functional MRI as an imaging tool to

678 evaluate muscle activity. J Orthop Sports Phys Ther 2011;41(11):896-903.

679 29. Farina D, Merletti R, Enoka RM. The extraction of neural strategies from the surface

680 EMG. J Appl Physiol (1985) 2004;96(4):1486-95.

681 30. Adams GR, Duvoisin MR, Dudley GA. Magnetic resonance imaging and

682 electromyography as indexes of muscle function. J Appl Physiol (1985)

683 1992;73(4):1578-83.

684 31. Fleckenstein JL, Canby RC, Parkey RW, et al. Acute effects of exercise on MR imaging

685 of skeletal muscle in normal volunteers. AJR Am J Roentgeno 1988;151(2):231-7.

686 32. Fisher MJ, Meyer RA, Adams GR, et al. Direct relationship between proton T2 and

687 exercise intensity in skeletal muscle MR images. Invest Radiol 1990;25(5):480-5.

688 33. Hermens HJ, Freriks B, Disselhorst-Klug C, et al. Development of recommendations for

689 SEMG sensors and sensor placement procedures. J Electromyogr Kinesiol

690 2000;10:361-74.

691 34. Opar DA, Piatkowski T, Williams MD, et al. A novel device using the Nordic hamstring

692 exercise to assess eccentric knee flexor strength: a reliability and retrospective injury

693 study. J Orthop Sports Phys Ther 2013;43(9):636-40.

694 35. Cagnie B, Dickx N, Peeters I, et al. The use of functional MRI to evaluate cervical flexor

695 activity during different cervical flexion exercises. J Appl Physiol (1985)

696 2008;104(1):230-5.

697 36. Patten C, Meyer RA, Fleckenstein JL. T2 mapping of muscle. Semin Musculoskelet

698 Radiol 2003;7(4):297-305.

699 37. Goode AP, Reiman MP, Harris L, et al. Eccentric training for prevention of hamstring

700 injuries may depend on intervention compliance: a systematic review and meta-

701 analysis. J Orthop Sports Phys Ther 2015;49(6):349-56.

31
702 38. Schuermans J, Van Tiggelen D, Danneels L, et al. Biceps femoris and semitendinosus-

703 teammates or competitors? New insights into hamstring injury mechanisms in male

704 football players: a muscle functional MRI study. Br J Sports Med 2014;48(22):1599-

705 606.

706 39. Kraemer WJ, Adams K, Cafarelli E, et al. American College of Sports Medicine position

707 stand. Progression models in resistance training for healthy adults. Med Sci Sports Ex

708 2002;34(2):364-80.

709 40. Opar DA, Williams MD, Timmins RG, et al. Rate of torque and electromyographic

710 development during anticipated eccentric contraction is lower in previously strained

711 hamstrings. Am J Sports Med 2013;41(1):116-25.

712 41. Fyfe JJ, Opar DA, Williams MD, et al. The role of neuromuscular inhibition in hamstring

713 strain injury recurrence. J Electromyogr Kinesiol. 2013;23(3):523-30.

714 42. Timmins RG, Shield AJ, Williams MD, et al. Biceps femoris long-head architecture: a

715 reliability and retrospective injury study. Med Sci Sports Ex 2015;47(5):905-13.

716 43. Lynn SK, Costigan PA. Changes in the medial-lateral hamstring activation ratio with foot

717 rotation during lower limb exercise. J Electromyogr Kinesiol 2009;19(3):e197-205.

718 44. Thelen DG, Chumanov ES, Hoerth DM, et al. Hamstring muscle kinematics during

719 treadmill sprinting. Med Sci Sports Ex 2005;37(1):108-14.

720 45. Lieber RL, Friden J. Functional and clinical significance of skeletal muscle architecture.

721 Muscle Nerve 2000;23(11):1647-66.

722 46. Fleckenstein JL, Haller RG, Lewis SF, et al. Absence of exercise-induced MRI

723 enhancement of skeletal muscle in McArdle's disease. J Appl Physiol (1985)

724 1991;71(3):961-9.

725 47. Garrett WE, Jr., Califf JC, Bassett FH. Histochemical correlates of hamstring injuries. Am

726 J Sports Med 1984;12(2):98-103.

32
727 48. Shellock FG, Fukunaga T, Mink JH, et al. Acute effects of exercise on MR imaging of

728 skeletal muscle: concentric vs eccentric actions. AJR Am J Roentgenol

729 1991;156(4):765-8.

730 49. Timmins RG, Ruddy JD, Presland J, et al. Architectural changes of the biceps femoris

731 after concentric or eccentric training. Med Sci Sports Ex 2015 [Epub ahead of print].

732 50. Kilgallon M, Donnelly AE, Shafat A. Progressive resistance training temporarily alters

733 hamstring torque-angle relationship. Scand J Med Sci Sports 2007;17(1):18-24.

734 51. Brockett CL, Morgan DL, Proske U. Predicting hamstring strain injury in elite athletes.

735 Med Sci Sports Ex 2004;36(3):379-87.

736

737

738

739

740

741

742

743

744

33
745 Figure 1. The 10 examined exercises. (a) bilateral stiff-leg deadlift, (b) hip hinge, (c)

746 unilateral stiff-leg deadlift, (d) lunge, (e) unilateral bent knee bridge, (f) unilateral straight

747 knee bridge, (g) leg curl, (h) 45° hip extension, (i) glute-ham-raise, (j) Nordic hamstring

748 exercise (NHE).

749 Figure 2. Biceps femoris (BF) to medial hamstring (MH) normalised EMG (nEMG)

750 relationship for the (a) concentric and (b) eccentric phases of each exercise. (SDL) Bilateral

751 stiff-leg deadlift, (HH) hip hinge, (USDL) unilateral stiff-leg deadlift, (L) lunge, (bKb)

752 unilateral bent knee bridge, (SKB) unilateral straight knee bridge, (LC) leg curl, (HE) 45° hip

753 extension, (GHR) glute-ham-raise, (NHE) Nordic hamstring exercise. Exercises to the left of

754 and above the 45° line exhibited higher levels of BF than MH nEMG and exercises to the

755 right and below the line displayed higher levels of MH than BF nEMG.

756 Figure 3. Percentage change in fMRI T2 relaxation times of each hamstring muscle

757 following the 45° hip extension exercise. Values are expressed as mean percentage change

758 compared to values at rest. ** indicates significantly different from ST, BFLH and SM

759 (p<0.001). * indicates significantly different from ST (p=0.005). Error bars depict standard

760 error. BFLH, biceps femoris long head; BFSH, biceps femoris short head; ST, semitendinosus;

761 SM, semimembranosus.

762 Figure 4. Percentage change in fMRI T2 relaxation times of each hamstring muscle

763 following the Nordic hamstring exercise. Values are expressed as mean percentage change

764 compared to values at rest. ** indicates significantly different from BFLH, BFSH and SM

765 (p≤0.002). * indicates significantly different from BFLH and SM (p=0.008) Error bars depict

766 standard error. BFLH, biceps femoris long head; BFSH, biceps femoris short head; ST,

767 semitendinosus; SM, semimembranosus.

34
768 Figure 5. Ratio of biceps femoris long head (BFLH) to semitendinosus (ST) (BFLH/ST)

769 percentage change in fMRI T2 relaxation times following the 45° hip extension and the

770 Nordic hamstring exercise (NHE). * indicates a significant difference between exercises

771 (p<0.001). Error bars depict standard error.

772

773

35