Received: 19 June 2018 

|
  Revised: 7 September 2018 
|  Accepted: 12 September 2018

DOI: 10.1111/sms.13303

ORIGINAL ARTICLE

High‐density electromyography activity in various hamstring

exercises

András Hegyi1   |  Dániel Csala2  |  Annamária Péter1   |  Taija Finni1   | 

Neil J Cronin1

1
Neuromuscular Research Center, Faculty
of Sport and Health Sciences, University of
Abstract
Jyvaskyla, Finland Proximal‐distal differences in muscle activity are rarely considered when defining
2
Department of Biomechanics, University the activity level of hamstring muscles. The aim of this study was to determine the
of Physical Education, Budapest, Hungary inter‐muscular and proximal‐distal electromyography (EMG) activity patterns of
Correspondence hamstring muscles during common hamstring exercises. Nineteen amateur athletes
András Hegyi, Neuromuscular Research without a history of hamstring injury performed 9 exercises, while EMG activity was
Center, Faculty of Sport and Health
recorded along the biceps femoris long head (BFlh) and semitendinosus (ST) mus-
Sciences, University of Jyvaskyla, Finland.
Email: andras.a.hegyi@jyu.fi cles using 15‐channel high‐density electromyography (HD‐EMG) electrodes. EMG
activity levels normalized to those of a maximal voluntary isometric contraction
(%MVIC) were determined for the eccentric and concentric phase of each exercise
and compared between different muscles and regions (proximal, middle, distal)
within each muscle. Straight‐knee bridge, upright hip extension, and leg curls exhib-
ited the highest hamstrings activity in both the eccentric (40%‐54%MVIC) and con-
centric phases (69%‐85%MVIC). Hip extension was the only BF‐dominant exercise
(Cohen’s d = 0.28 (eccentric) and 0.33 (concentric)). Within ST, lower distal than
middle/proximal activity was found in the bent‐knee bridge and leg curl exercises (d
range = 0.53‐1.20), which was not evident in other exercises. BFlh also displayed
large regional differences across exercises (d range = 0.00‐1.28). This study demon-
strates that inter‐muscular and proximal‐distal activity patterns are exercise‐depend-
ent, and in some exercises are affected by the contraction mode. Knowledge of
activity levels and relative activity of hamstring muscles in different exercises may
assist exercise selection in hamstring injury management.

KEYWORDS
heterogeneous activity, injury reduction, rehabilitation

1  |   IN TRO D U C T ION play,5 suggesting suboptimal loading in the rehabilitation

process.
Hamstring strain is the most frequent injury in sports in- Some interventions implementing eccentric exercises
volving high‐speed running.1,2 For example in football, seem to mitigate hamstring injury occurrence.6-10 In addition
this type of injury results in a substantial player time to low strength and short muscle length,11,12 neural inhibi-
loss,1 decreased team performance,3 and significant fi- tion13 and imbalances between the activity level of ham-
nancial burdens on teams.4 Re‐injury rate can be as high string muscles14 are also associated with hamstring injury.
as 24% and is typical in the early stages of return to Proper exercise selection potentially allows the clinician to

|
34     wileyonlinelibrary.com/journal/sms
© 2018 John Wiley & Sons A/S. Scand J Med Sci Sports. 2019;29:34–43.
Published by John Wiley & Sons Ltd
HEGYI et al.   
|
   35

better succeed in (re‐)injury prevention, but this is challeng-

2.2  |  Study protocol
ing for many reasons. For example, non‐uniform adaptations
to exercise interventions11,15,16 may be associated with non‐ The study was performed in the mid‐season when the fre-
uniform hamstring activity patterns across exercises.11,17-19 quency of intense strength training was minimized. Participants
Moreover, study results are inconsistent concerning which refrained from additional strengthening exercises during the
hamstring muscles are activated in different exercises, as study to minimize training effects. Prior to data collection, 12‐
well as the extent of activation,20 and it is questionable repetition maximum load (12RM) was defined for 9 hamstring
whether these differences are real or at least partly reflect exercises across 4‐5 sessions (4‐7 days in‐between). The exam-
the (in)accuracy with which different methods can define ined exercises were good morning (GM), unilateral Romanian
muscle activity. deadlift (RDL), cable pendulum (CP), bent‐knee bridge (BB),
Electromyography (EMG) is the most commonly used 45° hip extension (45HE), prone leg curl (PLC), slide leg curl
method to examine hamstring muscle activity.20 In con- (SLC), upright hip extension conic‐pulley (UHC), and straight‐
ventional EMG studies, electrodes are placed over the knee bridge (SB) (Figure 1 and Video S1). In each session
mid‐belly of hamstring muscles, ignoring possible prox- except the last one, 2‐3 randomly selected exercises were prac-
imal‐distal differences in muscle activity. Studies have ticed, and then, 12RM was tested,23 while exercise technique
shown non‐uniform proximal‐distal metabolic activity was assessed and (if needed) corrected by an experienced prac-
patterns within hamstring muscles.18,19,21 Similarly, during titioner to ensure standard technical performance. Unilateral
two common hamstring exercises, we recently observed exercises were performed with the dominant (kicking) leg (4
large differences in muscle activity within the semiten- left, 15 right). In the last familiarization session, maximal vol-
dinosus (ST) and biceps femoris long head (BFlh) using untary isometric contractions (MVICs) were practiced.
high‐density EMG (HD‐EMG).22 Due to such regional In the main testing session, after preparation and warm‐
differences, spatially robust methods may improve under- up, participants performed knee flexion and hip extension
standing of hamstrings activity patterns. This would po- MVICs for the purpose of EMG normalization, followed
tentially allow the clinician to selectively activate specific by 6 repetitions of each exercise in a random order. The
muscles or muscle regions. warm‐up consisted of cycling, dynamic stretching (5 min-
In this study, we aimed to define the excitation level utes each), and then 10 submaximal hip extension and
of ST and BFlh muscles in the eccentric and concentric knee flexion contractions performed in a custom‐made
phases of 9 typical hamstring exercises. We also tested dynamometer (UniDrive, University of Jyväskylä),24 with
whether the relative activity of these muscles is similar the intensity increasing from ~30 to ~90% MVIC. In the
in the eccentric and concentric phases, as well as whether dynamometer where MVICs were performed, participants
proximal‐distal activity patterns are similar across exer- lay prone with the trunk and hip fixed to the dynamome-
cises. According to the study aims, exercises were chosen ter bench in neutral position. In the dominant (measured)
that include clear eccentric and concentric phases (ie, at the leg, the knee joint was positioned in 20° of flexion while
muscle‐tendon unit level), and which are generally used in the other leg was extended. For knee flexion MVICs, the
hamstring injury management. lever arm of the dynamometer was fixed ~5 cm above the
lateral malleolus. For hip extension MVICs, the lever arm
was strapped just above the knee joint fold, and partici-
2  |  M ATE R IA L S A N D ME T HODS pants were asked to maintain 20° of knee flexion, which
was confirmed before each contraction using a goniometer.
2.1  | Participants For both hip extension and knee flexion MVICs, two rep-
Nineteen young male amateur athletes (mean ± standard etitions were performed, followed by a third if peak torque
deviation, age 26.1 ± 3.2 years, body mass 80.2 ± 14.1 kg, differed by>5% between the first two contractions. For each
height 178.3 ± 9.3 cm) from high injury‐risk sports (9 contraction, maximum effort was maintained for 2 seconds
soccer, 6 Gaelic football, and 4 rugby players) and expe- and 2 minutes rest was applied between contractions. A
rienced at performing hamstring exercises participated in simultaneous performance of knee flexion and hip exten-
this study. Exclusion criteria were history of hamstring sion was also performed, wherein the participants reached
strain, previous anterior cruciate ligament or lower back maximum effort in both tasks simultaneously, which was
injury, and cardiovascular or musculo‐skeletal disorders. maintained for 2 seconds. For this task, the dynamome-
Participants received detailed information about the study ter lever arm was fixed ~5 cm above the lateral malleolus
before they gave written informed consent. Testing pro- and the thigh was tightly fixed to the bench. Thereafter, 6
cedures were approved by the ethics committee of the repetitions of the 9 selected exercises were performed in
University of Jyväskylä and performed according to the random order, at 12 RM load. For the exercises, hip and
Declaration of Helsinki. knee goniometers were aligned with the trochanter major
 |
36       HEGYI et al.

and lateral epicondyle of the femur, respectively. Both the channel 8‐9 from the distal end of the array was aligned with
eccentric and concentric phases were performed in 2 sec- the midpoint along the ischial tuberosity‐popliteal fossa dis-
onds, controlled with a metronome. Four‐minute rest was tance, while in ST the EMG array was placed 1 cm below the
applied between exercises. Hip and knee joint angles were tendinous inscription which was located relatively proximally.
recorded as well as BFlh and ST EMG activity. Participants Arrays were fixed over the skin using adhesive foam and tape.
reported no substantial fatigue throughout the testing. EMG arrays were connected to an amplifier, and signals were
digitized (EMG‐USB 12‐bit A/D converter, OT Bioelettronica)
for recording in BioLab software (v3.1, OT Bioelettronica). To
2.3  |  Data collection
maintain skin‐electrode contact, electrode cavities were filled
To determine correct HD‐EMG array positioning, B‐mode 2D with 20 µL conductive gel. A reference electrode was placed
ultrasonography (Aloka α10, Tokyo, Japan) was used to define over the contralateral wrist. Signal quality was confirmed during
and mark the borders of the BFlh and ST muscles as well as submaximal contractions. EMG data were sampled at 2048 Hz
the location of their distal musculo‐tendinous junctions. After and amplified by a factor of 1000. During the measurements, 15
skin preparation, a 15‐channel EMG array (10‐mm inter‐elec- differential channels were recorded from each muscle.
trode distance, OT Bioelettronica, Torino, Italy) was secured During MVICs, hip extension and knee flexion forces
over each muscle (Figure 2) so that the electrodes were as far were measured with the dynamometer strain gauge at a
away from the muscle borders as possible, to minimize cross sampling frequency of 1000 Hz, digitized (EMG‐USB
talk. Electrode positioning was standardized so that in BFlh 12‐bit A/D converter, OT Bioelettronica) and recorded in

F I G U R E 1   Nine typical rehabilitation exercises examined in this study. GM, good morning; RDL, unilateral Romanian deadlift; CP, cable
pendulum; BB, bent‐knee bridge; 45HE, 45° hip extension; PLC, prone leg curl; SLC, slide leg curl; UHC, upright hip extension conic‐pulley; SB,
straight‐knee bridge
HEGYI et al.   
|
   37

(RMS) EMG activity was calculated from a 1‐second stable

force plateau for each EMG channel. From the exercises,
RMS activity was calculated in the entire eccentric and con-
centric phase (ie, ~2 seconds for each) for each EMG chan-
nel based on hip and knee joint angular displacement. RMS
values across the eccentric and concentric phases of the six
repetitions were averaged, respectively, and expressed as a
percentage of the highest RMS activity of the corresponding
EMG channel during any of the MVIC tasks (%MVIC).
Activity for each muscle was determined for the eccentric
and concentric phases separately as the average RMS activity
of all 15 channels along the corresponding muscle, which is
hereafter referred to as overall activity. To determine the ac-
tivity level of different muscle regions, average activity was
calculated for channels 1‐5 (distal region), 6‐10 (middle re-
gion), and 11‐15 (proximal region).
To provide estimates of hip extension and knee flexion
strength, maximal torque during the isometric contractions
was calculated as the maximum instantaneous force multi-
plied by the respective lever arm. The highest torque of all rep-
etitions was used for the hip extension and knee flexion tasks.

2.5  |  Statistical analysis

Normal distributions of studentized residuals were confirmed
using Shapiro‐Wilk test and Q‐Q plots. For each exercise and
contraction mode, the difference between BFlh and ST overall
activity was tested with paired samples t test in SPSS (IBM,
Armonk, NY, USA). Significance level was set at P < 0.05.
F I G U R E 2   High‐density electromyography (HD‐EMG) arrays
(A) were attached and secured (B) over the semitendinosus (ST)
Contraction mode*region interaction for each exercise and
and the long head of the biceps femoris (BFlh) to comprehensively region*exercise interactions for each contraction mode were
describe muscle activity levels during each exercise tested for each muscle with repeated‐measures ANOVA.
If Mauchly’s test of sphericity was violated (P < 0.05),
Greenhouse‐Geisser adjustment was applied. Differences
BioLab software in synchrony with the EMG signals. Lever were located after Bonferroni correction. Cohen’s d ± 90%
arms were measured to calculate torque. For hip extension, confidence intervals (90% CI) were calculated to determine
the lever arm was measured as the distance between the the magnitude of differences using a custom spreadsheet.25
trochanter major and the middle of the strain gauge. For Differences were considered as trivial (<0.2), small (≥0.2),
knee flexion, the lever arm was measured as the distance moderate (≥0.5), or large (≥0.8). Differences where 90% CIs
between the lateral epicondyle of the femur and the middle overlapped both 0.2 and −0.2 were considered unclear.26
of the strain gauge. During muscle contractions, force‐time
curve feedback was provided.
Hip and knee joint angles were recorded using cus- 3  |  RESULTS
tom‐made electro‐goniometers (University of Jyväskylä,
Finland). Angle data were digitized by the A/D converter Maximal hip extension and knee flexion torque dur-
of the EMG system and recorded in BioLab software si- ing the isometric contractions were 236.5 ± 84.1 Nm and
multaneously with the EMG data. 153.3 ± 59.2 Nm (mean ± standard deviation), respectively.

2.4  |  Data analysis 3.1  |  Overall activity

A 10‐500 Hz fourth‐order zero‐phase band‐pass Butterworth BFlh overall activity level ranged across exercises from an
filter was used to filter EMG data in MATLAB (MathWorks average of 17%‐54% in the eccentric and 32%‐83% in the
Inc, Natick, MA, USA). For MVICs, root‐mean‐square concentric phase, relative to MVIC (Figure 3). In ST, activity
 |
38       HEGYI et al.

T A B L E 1   Differences (Cohen's
Eccentric Concentric
d ± 90% confidence limits) between BFlh
Straight‐knee bridge (SB) 0.19 ± 0.37T −0.09 ± 0.36U and ST muscles in the eccentric and
Upright hip extension conic‐pulley (UHC) 0.11 ± 0.33U −0.16 ± 0.29T concentric phase of hamstring exercises
T
Slide leg curl (SLC) 0.12 ± 0.25 −0.26 ± 0.28S
Prone leg curl (PLC) 0.17 ± 0.20T −0.35 ± 0.27S
45° hip extension (45HE) 0.28 ± 0.28 S
0.33 ± 0.24S
Bent‐knee bridge (BB) −0.17 ± 0.27T −0.24 ± 0.25S
Cable pendulum (CP) −0.02 ± 0.43U 0.01 ± 0.38U
T
Unilateral Romanian deadlift (RDL) −0.19 ± 0.24 −0.11 ± 0.22T
Good morning (GM) −0.21 ± 0.19S −0.09 ± 0.25T
Positive values: biceps femoris long head > semitendinosus (BFlh > ST)
Negative values: biceps femoris long head < semitendinosus (BFlh < ST)
T, trivial difference; S, small difference between muscles; U, unclear.
P < 0.05.

levels of 19%‐51% in the eccentric and 33%‐85% in the con-

centric phase were observed (Figure 3).
The only exercise with higher activity in BFlh com-
pared to ST was 45HE: in both the concentric and eccen-
tric phases, small differences between muscles were found
(d = 0.28 ± 0.28 and 0.33 ± 0.24, respectively), which
reached statistical significance in the concentric but not the
eccentric phase (P = 0.026 and 0.100, respectively). ST ac-
tivity was higher than BFlh activity in the eccentric phase of
GM (d = 0.21 ± 0.19) and concentric phase of PLC, SLC and
BB exercises (d = 0.35 ± 0.27, 0.26 ± 0.28, and 0.24 ± 0.25,
respectively), from which only PLC reached statistical signif-
icance (P = 0.036, 0.118, and 0.107, respectively). Between‐
muscle differences are presented in Table 1.

3.2  |  Regional activity patterns

Mean and standard deviation of regional activity levels are
shown in Figure 4. Different exercises showed distinct re-
gional patterns both in ST (P < 0.001 in both eccentric and
concentric) and in BFlh (eccentric: P = 0.001, concentric:
P < 0.001). The contraction mode affected the regional ac-
tivity pattern of ST in BB, HE, PLC, and SLC (P = 0.001,
P = 0.040, P < 0.001, and P < 0.001, respectively), and
the regional activity pattern of BFlh in UHC, PLC, SB,
and SLC (P = 0.012, P < 0.001, P = 0.016, and P = 0.009,
respectively).
Lower activity in the distal compared to the middle
F I G U R E 3   Electromyography (EMG) activity levels in the
or proximal regions was found in BB, PLC, and SLC (d
eccentric (A) and concentric (B) phase of each exercise. Mean and
range = 0.53‐1.20, P < 0.05), in both the eccentric and
standard deviation are presented. Data represent the average of 15
concentric phases. In all other exercises, no or only small
EMG channels along each muscle. Dotted lines represent equal activity
level between the two muscles when normalized to maximal voluntary
differences between distal vs other regions were found (d
isometric activity (MVIC). GM, good morning; RDL, unilateral range = 0.00‐0.40, P > 0.05). Similarly in BFlh, a large range
Romanian deadlift; CP, cable pendulum; BB, bent‐knee bridge; 45HE, in the magnitude of regional differences was observed across
45° hip extension; PLC, prone leg curl; SLC, slide leg curl; UHC, exercises (difference between regions, d range = 0.02‐1.28),
upright hip extension conic‐pulley; SB, straight‐knee bridge with PLC displaying the largest differences between muscle
HEGYI et al.   
|
   39

F I G U R E 4   Mean and standard deviation of the normalized activity level (%MVIC, maximal voluntary isometric contraction) in the
proximal, middle, and distal regions of each muscle during the eccentric and concentric phase of each exercise. GM, good morning; RDL, unilateral
Romanian deadlift; CP, cable pendulum; BB, bent‐knee bridge; 45HE, 45° hip extension; PLC, prone leg curl; SLC, slide leg curl; UHC, upright
hip extension conic‐pulley; SB, straight‐knee bridge

regions (d range = 0.41‐1.28). Differences are detailed in In addition to recent studies using muscle functional
Table 2. magnetic resonance imaging (mfMRI)18,19,21 and our pre-
vious results using HD‐EMG,22 the exercise‐dependent
changes in proximal‐distal activity patterns observed in
4  |   D IS C U SSION this study reinforce the notion that spatially robust methods
are needed to accurately describe the activity level of ST
In the current study, muscle activity patterns were determined and BFlh muscles. This is further supported by the sub-
in 9 typical hamstring exercises using HD‐EMG while taking stantially different proximal‐distal EMG activity patterns
proximal‐distal differences into account. Small differences between muscles in most of the exercises. This was most
between the activity levels of BFlh and ST muscles were pronounced in BB, wherein regional differences were mod-
observed in the concentric phase of 45HE, SLC, PLC, and erate‐to‐large in ST but trivial in BFlh. This phenomenon
BB, from which the only BFlh‐dominant exercise—45HE— likely leads to a non‐systematic error when the activity lev-
showed a difference in the eccentric phase. Proximal‐distal els of these muscles are compared based on a small region
distribution of EMG signals varied substantially across ex- of the muscle.
ercises and showed different patterns between ST and BFlh Similar to previous studies,17,27 we found high normalized
muscles. activity levels in SB, SLC, and PLC. Additionally, during
T A B L E 2   Regional differences in the electromyography activity level of hamstring muscles in the eccentric and concentric phase of hamstring exercises
|

Eccentric Concentric
40      

Semitendinosus Biceps femoris long head Semitendinosus Biceps femoris long head

Region middle proximal middle proximal middle proximal middle proximal

Straight‐knee bridge (SB)
Distal −0.40 ± 0.42S 0.04 ± 0.67U 0.31 ± 0.24S −0.26 ± 0.32S −0.29 ± 0.39S 0.21 ± 0.79U 0.54 ± 0.31M 0.10 ± 0.44U
S M M
Middle – 0.44 ± 0.43 – −0.58 ± 0.27 – 0.50 ± 0.59 – −0.45 ± 0.33S
Upright hip extension conic‐pulley (UHC)
Distal −0.06 ± 0.41U −0.17 ± 0.34T 0.22 ± 0.19S −0.19 ± 0.30T 0.01 ± 0.40U 0.02 ± 0.38U 0.40 ± 0.23S 0.25 ± 0.41S
Middle – −0.12 ± 0.30T – −0.42 ± 0.26S – 0.01 ± 0.33U – −0.15 ± 0.34T
Slide leg curl (SLC)
Distal 0.53 ± 0.33M 0.63 ± 0.32M 0.09 ± 0.23T −0.16 ± 0.28T 0.95 ± 0.43L 1.02 ± 0.32L −0.31 ± 0.26S −0.49 ± 0.28S
Middle – 0.11 ± 0.30T – −0.25 ± 0.25S – 0.07 ± 0.41U – −0.18 ± 0.20T
Prone leg curl (PLC)
Distal 0.62 ± 0.29M 0.79 ± 0.31M −0.46 ± 0.28S −0.87 ± 0.30L 1.03 ± 0.28L 1.15 ± 0.30L −0.84 ± 0.32L −1.28 ± 0.30L
Middle – 0.17 ± 0.30T – −0.41 ± 0.20S – 0.11 ± 0.39U – −0.45 ± 0.24S
45° hip extension (45HE)
Distal −0.06 ± 0.22T 0.02 ± 0.23U 0.13 ± 0.18T −0.46 ± 0.28S −0.08 ± 0.24T 0.17 ± 0.26T 0.21 ± 0.17S −0.14 ± 0.24T
T M S
Middle – 0.08 ± 0.18 – −0.59 ± 0.23 – 0.26 ± 0.19 – −0.35 ± 0.23S
Bent‐knee bridge (BB)
Distal 1.03 ± 0.34L 1.20 ± 0.44L 0.02 ± 0.21T −0.02 ± 0.22U 0.98 ± 0.30L 1.13 ± 0.37L 0.13 ± 0.23T 0.13 ± 0.27T
U U U
Middle – 0.16 ± 0.44 – −0.05 ± 0.26 – 0.14 ± 0.37 – 0.00 ± 0.30U
Cable pendulum (CP)
Distal −0.08 ± 0.47U −0.27 ± 0.36S 0.00 ± 0.14T −0.39 ± 0.18S −0.06 ± 0.45U 0.00 ± 0.42U 0.07 ± 0.25T −0.28 ± 0.35S
Middle – −0.19 ± 0.53U – −0.39 ± 0.15S – 0.06 ± 0.44U – −0.35 ± 0.34S
Unilateral Romanian deadlift (RDL)
Distal 0.20 ± 0.36S 0.14 ± 0.30T 0.06 ± 0.13T −0.29 ± 0.17S 0.21 ± 0.32S 0.26 ± 0.32S 0.10 ± 0.14T −0.26 ± 0.19S
Middle – −0.06 ± 0.24T – −0.35 ± 0.13S – 0.05 ± 0.26U – −0.36 ± 0.18S
Good morning (GM)
Distal 0.20 ± 0.42U 0.12 ± 0.33U 0.00 ± 0.13T −0.42 ± 0.18S 0.21 ± 0.42U 0.25 ± 0.48U 0.04 ± 0.14T −0.47 ± 0.17S
Middle – −0.08 ± 0.25T – −0.42 ± 0.14S – 0.05 ± 0.40U – −0.51 ± 0.20M
Cohen’s d ± 90% confidence limits. Positive and negative differences correspond to higher activity level in the relatively more proximal and distal regions, respectively.
T, trivial difference; S, small difference; M, moderate difference; L, large difference between regions; U, unclear.
P < 0.05.
HEGYI et al.
HEGYI et al.   
|
   41

UHC, which has not been the focus of many experiments, the higher in SB, UHC, SLC, and PLC in our study, this does
activity level exceeded 80%MVIC in the concentric phase. not necessarily imply that the eccentric phase of these exer-
High activity levels in these exercises may facilitate training‐ cises can more effectively elongate BFlh fascicles. Askling
induced adaptations in the hamstrings, although adaptations et al7,8 demonstrated that exercises performed at longer mus-
in response to these exercises are unclear. In accordance with cle operating lengths are more effective for injury prevention
previous literature,28 particularly low overall hamstrings ac- than those requiring hamstrings to operate at a shorter length.
tivity was observed in GM, which is apparently associated Muscle length is clearly longer in 45HE compared to all four
with low hamstring muscle forces in this exercise.29 Exercises of the aforementioned high‐activity exercises. Nonetheless,
inducing limited hamstrings activity are likely suboptimal to Nordic hamstring exercise also seems to reduce hamstring in-
facilitate meaningful muscle adaptations. juries,6,9,10 even though the operating length is likely similar
The relevance of the relative roles of individual hamstring to that in SLC and PLC. Future studies should further clarify
muscles in hamstring injury is yet to be clarified. Training which of these exercises are the most beneficial to mitigate
interventions should target the mitigation of injury‐risk fac- injury‐risk factors.
tors. An imbalance between BFlh and ST muscle activity During rehabilitation, it may be of value to know regional
level seems to be associated with hamstring injuries.14 Thus, activity patterns relative to the injury site to enable selective
balanced strengthening of these muscles should be a training activation of the injured muscle region. In 80% of running‐
goal. Although conventional EMG studies are not in agree- type hamstring injuries, the BFlh is affected primarily and
ment, previous mfMRI studies suggest that BFlh is relatively typically at the proximal site.32 Within the BFlh, the proximal
more active in hip‐dominant exercises, while ST is relatively region seems to be the most challenging to activate since this
more active in knee‐dominant exercises.20 Based on the cur- region did not show higher activity compared to the distal or
rent study, it seems rather challenging to preferentially acti- middle regions in any of the exercises in the current study. On
vate BFlh. Previously, mfMRI showed relatively high activity the contrary, lunge19 and CP21 have been shown to activate
in BFlh compared to ST in 45HE,17 which is confirmed by the proximal BFlh in mfMRI studies. In the current study, CP
our results. Other hip‐dominant exercises did not induce showed the lowest activity in the proximal region. In any case,
higher activity in BFlh than in ST in this study. in both lunge33 and CP, the overall hamstrings activity level is
Contraction mode–dependent between‐muscle activ- rather low, likely limiting meaningful adaptations in response
ity patterns were observed in some exercises in the current to these exercises. Manipulating the shin angle during a lunge
study. In the concentric phase, three exercises—SLC, PLC, may expose the hamstrings to substantially higher forces,29
and UHC—showed higher activity in ST compared to BFlh. likely increasing hamstrings activity. However, it is unclear
However, this difference was not evident in the eccentric whether this manipulation alters the proximal‐distal activity
phase of these exercises. This is inconsistent with previous re- pattern. Future studies should examine whether targeting the
sults concerning eccentric PLC (120% concentric 1RM)18,30 injured muscle region during the rehabilitation process ac-
and the mechanically similar high‐load eccentric‐only Nordic celerates the restoration of muscle function after a hamstring
hamstring exercise,17,22,31 which seem to selectively activate injury.
ST. This discrepancy may be explained by the substantially It should be mentioned that some discrepancies exist
lower load applied in the current study. Similar to these ex- when comparing some of our results with some previous
ercises, no between‐muscle differences were found in the mfMRI findings. Contrary to our finding that there are only
eccentric phase of SB, BB, or one‐leg RDL. Based on the trivial differences between ST and BFlh muscle activity lev-
current study, these exercises should be used when balanced els in RDL, this exercise has been suggested to be a BFlh‐
eccentric activation of ST and BFlh muscles is of interest. dominant exercise based on mfMRI data.34 However, in that
However, it is also likely important to include exercises with study, the exercise was performed bilaterally and included
a relatively high overall hamstrings activity level to better fa- only 6 participants. In any case, in our study, hamstrings ac-
cilitate muscle adaptations. The above observations suggest tivity levels were 21% and 43% in the eccentric and concen-
that ST‐BFlh muscle selectivity cannot always be predicted tric phases of RDL, the second lowest out of the examined
based solely on the hip‐ or knee‐dominant nature of the exer- exercises, likely minimizing the clinical relevance of this
cise and may be affected by different neural control strategies difference. On the contrary, hamstrings activity was particu-
in the eccentric and concentric phases. larly high in SB. In the current study, we did not detect clear
In BFlh, eccentric stimuli may be of particular importance differences between muscles in SB, contrary to Bourne et
to elicit fascicle lengthening, which seems to reduce the risk al35 who found higher metabolic activity in ST compared to
for hamstring injury.12 45HE exhibited the largest activity in BFlh, although the between‐muscle difference seems to be
BFlh relative to ST and has already been shown to effectively smaller compared to most of the other exercises previously
increase BFlh fascicle length.11 Although activity level was examined with mfMRI.20 These discrepancies may arise
 |
42       HEGYI et al.

from methodological issues: both mfMRI and EMG have ORCID

limitations when comparing the relative contribution of dif-
András Hegyi  http://orcid.org/0000-0002-3663-0288
ferent hamstring muscles. Metabolic activity estimated by
mfMRI is sensitive to glycolysis,36 vascular dynamics,37 and Annamária Péter  http://orcid.org/0000-0002-5197-6869
fiber type proportions,38 which may differ between muscles Taija Finni  http://orcid.org/0000-0002-7697-2813
and individuals. With respect to EMG, it is not clear whether Neil J Cronin  http://orcid.org/0000-0002-5332-1188
reference contractions used for normalization activate all ex-
amined hamstring muscles to a similar extent. Accordingly,
to examine the relative contribution of different hamstring R E F E R E NC E S
muscles using these methods, it is likely most appropriate to 1. Ekstrand J, Walden M, Hagglund M. Hamstring injuries have in-
compare within the same individuals and measurement ses- creased by 4% annually in men's professional football, since 2001:
sion across exercises. a 13‐year longitudinal analysis of the UEFA Elite Club injury
As a possible limitation of this study, surface EMG is prone to study. Br J Sports Med. 2016;50(12):731‐737.
cross talk. To minimize this effect, we used HD‐EMG electrodes 2. Opar DA, Drezner J, Shield A, et al. Acute hamstring strain injury
with a relatively shallow pick‐up area and 10‐mm inter‐electrode in track‐and‐field athletes: A 3‐year observational study at the
Penn Relay Carnival. Scand J Med Sci Sports. 2014;24(4):e254
distance,39 ensured correct electrode location using ultrasonog-
–e259.
raphy, and measured male athletes with a relatively thin subcuta-
3. Hagglund M, Walden M, Magnusson H, Kristenson K,
neous layer overlying the target muscles. Furthermore, recording Bengtsson H, Ekstrand J. Injuries affect team performance
from 15 cm along each muscle likely minimized the effect of negatively in professional football: an 11‐year follow‐up of
muscle movement relative to the skin, which is considered an in- the UEFA Champions League injury study. Br J Sports Med.
herent limitation of surface EMG. Additionally, muscle regions 2013;47(12):738‐742.
were covered to a slightly different extent across individuals due 4. Ekstrand J. Keeping your top players on the pitch: the key to
to differences in muscle length relative to the length of the EMG football medicine at a professional level. Br J Sports Med.
2013;47(12):723‐724.
arrays. As an additional limitation, we measured amateur ath-
5. Malliaropoulos N, Isinkaye T, Tsitas K, Maffulli N. Reinjury after
letes without a history of hamstring injury, so our results may not acute posterior thigh muscle injuries in elite track and field ath-
be directly applicable to other populations, for example, injured letes. Am J Sports Med. 2011;39(2):304‐310.
and/or professional athletes. 6. Arnason A, Andersen TE, Holme I, Engebretsen L, Bahr R.
Prevention of hamstring strains in elite soccer: an intervention
study. Scand J Med Sci Sports. 2008;18(1):40‐48.
4.1  | Perspectives 7. Askling CM, Tengvar M, Tarassova O, Thorstensson A. Acute
hamstring injuries in Swedish elite sprinters and jumpers: a pro-
HD‐EMG revealed exercise‐specific inter‐ and intramuscular
spective randomised controlled clinical trial comparing two reha-
hamstring activity patterns in 9 typical hamstring exercises.
bilitation protocols. Br J Sports Med. 2014;48(7):532‐539.
This study also revealed that the relative activity of differ- 8. Askling CM, Tengvar M, Thorstensson A. Acute hamstring inju-
ent hamstring muscles may differ between the eccentric and ries in Swedish elite football: a prospective randomised controlled
concentric phases of an exercise. These findings highlight clinical trial comparing two rehabilitation protocols. Br J Sports
the potential impact of exercise selection procedure on ham- Med. 2013;47(15):953‐959.
strings strengthening. The clinical implications of heteroge- 9. Petersen J, Thorborg K, Nielsen MB, Budtz‐Jorgensen E, Holmich
neous hamstrings EMG activity should be further examined, P. Preventive effect of eccentric training on acute hamstring inju-
ries in men's soccer: a cluster‐randomized controlled trial. Am J
as well as the mechanisms and functional relevance of het-
Sports Med. 2011;39(11):2296‐2303.
erogeneous activity.
10. van der Horst N, Smits DW, Petersen J, Goedhart EA, Backx FJ.
The preventive effect of the nordic hamstring exercise on ham-
string injuries in amateur soccer players: a randomized controlled
ACKNOWLEDGEMENT
trial. Am J Sports Med. 2015;43(6):1316‐1323.
We acknowledge Johan Lahti for assisting with teaching and 11. Bourne MN, Duhig SJ, Timmins RG, et al. Impact of the Nordic
monitoring proper exercise technique throughout the study. hamstring and hip extension exercises on hamstring architecture
We also thank Antoine Nordez (EA4344, Nantes) for assist- and morphology: implications for injury prevention. Br J Sports
Med. 2017;51(5):469‐477.
ing with research equipment.
12. Timmins RG, Bourne MN, Shield AJ, Williams MD, Lorenzen
C, Opar DA. Short biceps femoris fascicles and eccentric knee
CONFLICT OF INTEREST flexor weakness increase the risk of hamstring injury in elite
football (soccer): a prospective cohort study. Br J Sports Med.
The authors have no professional relationships with any com- 2016;50(24):1524‐1535.
pany or manufacturer who would benefit from the current 13. Opar DA, Williams MD, Timmins RG, Dear NM, Shield AJ. Knee
study results. flexor strength and biceps femoris electromyographical activity is
HEGYI et al.   
   43
|
lower in previously strained hamstrings. J Electromyogr Kinesiol. 29. Schellenberg F, Taylor WR, Lorenzetti S. Towards evidence
2013;23(3):696‐703. based strength training: a comparison of muscle forces during
14. Schuermans J, Van Tiggelen D, Danneels L, Witvrouw E. deadlifts, goodmornings and split squats. BMC Sports Sci Med
Biceps femoris and semitendinosus–teammates or competitors? Rehabil. 2017;9:13.
New insights into hamstring injury mechanisms in male foot- 30. Ono T, Okuwaki T, Fukubayashi T. Differences in activation pat-
ball players: a muscle functional MRI study. Br J Sports Med. terns of knee flexor muscles during concentric and eccentric ex-
2014;48(22):1599‐1606. ercises. Res Sports Med. 2010;18(3):188‐198.
15. Potier TG, Alexander CM, Seynnes OR. Effects of eccen- 31. Fernandez‐Gonzalo R, Tesch PA, Linnehan RM, et al. Individual
tric strength training on biceps femoris muscle architec- muscle use in hamstring exercises by soccer players assessed
ture and knee joint range of movement. Eur J Appl Physiol. using functional MRI. Int J Sports Med. 2016;37(7):559‐564.
2009;105(6):939‐944. 32. De Smet AA, Best TM. MR imaging of the distribution and loca-
16. Timmins RG, Ruddy JD, Presland J, et al. Architectural Changes tion of acute hamstring injuries in athletes. AJR Am J Roentgenol.
of the Biceps Femoris Long Head after Concentric or Eccentric 2000;174(2):393‐399.
Training. Med Sci Sports Exerc. 2016;48(3):499‐508. 33. Jonhagen S, Halvorsen K, Benoit DL. Muscle activation and
17. Bourne MN, Williams MD, Opar DA, Al Najjar A, Kerr GK, length changes during two lunge exercises: implications for reha-
Shield AJ. Impact of exercise selection on hamstring muscle acti- bilitation. Scand J Med Sci Sports. 2009;19(4):561‐568.
vation. Br J Sports Med. 2017;51(13):1021‐1028. 34. Ono T, Higashihara A, Fukubayashi T. Hamstring functions during
18. Kubota J, Ono T, Araki M, Torii S, Okuwaki T, Fukubayashi T. hip‐extension exercise assessed with electromyography and mag-
Non‐uniform changes in magnetic resonance measurements of netic resonance imaging. Res Sports Med. 2011;19(1):42‐52.
the semitendinosus muscle following intensive eccentric exercise. 35. Bourne M, Williams M, Pizzari T, Shield A. A functional MRI
Eur J Appl Physiol. 2007;101(6):713‐720. exploration of hamstring activation during the supine bridge exer-
19. Mendiguchia J, Garrues MA, Cronin JB, et al. Nonuniform cise. Int J Sports Med. 2018;39(2):104‐109.
changes in MRI measurements of the thigh muscles after 36. Fleckenstein JL, Haller RG, Lewis SF, et al. Absence of exer-
two hamstring strengthening exercises. J Strength Cond Res. cise‐induced MRI enhancement of skeletal muscle in McArdle's
2013;27(3):574‐581. disease. J Appl Physiol. 1991;71(3):961‐969.
20. Bourne MN, Timmins RG, Opar DA, et al. An evidence‐based 37. Patten C, Meyer RA, Fleckenstein JL. T2 mapping of muscle.
framework for strengthening exercises to prevent hamstring in- Semin Musculoskelet Radiol. 2003;7(4):297‐305.
jury. Sports Med. 2018;48(2):251‐267. 38. Jenner G, Foley JM, Cooper TG, Potchen EJ, Meyer RA.
21. Mendez‐Villanueva A, Suarez‐Arrones L, Rodas G, et al. MRI‐ Changes in magnetic resonance images of muscle depend
based regional muscle use during hamstring strengthening exer- on exercise intensity and duration, not work. J Appl Physiol.
cises in elite soccer players. PLoS One. 2016;11(9):e0161356. 1994;76(5):2119‐2124.
22. Hegyi A, Peter A, Finni T, Cronin NJ. Region‐dependent ham- 39. De Luca CJ, Kuznetsov M, Gilmore LD, Roy SH. Inter‐elec-
strings activity in Nordic hamstring exercise and stiff‐leg deadlift trode spacing of surface EMG sensors: reduction of cross-
defined with high‐density electromyography. Scand J Med Sci talk contamination during voluntary contractions. J Biomech.
Sports. 2018;28(3):992‐1000. 2012;45(3):555‐561.
23. Baechle TR, Earle RW. Essentials of Strength Training and
Conditioning, 3rd edn. Champaign (IL): Human Kinetics;
2008:397‐399.
24. Komi PV, Linnamo V, Silventoinen P, Sillanpaa M. Force and SUPPORTING INFORMATION
EMG power spectrum during eccentric and concentric actions.
Additional supporting information may be found online in
Med Sci Sports Exer. 2000;32(10):1757‐1762.
25. Hopkins W. Spreadsheets for analysis of controlled trials with ad-
the Supporting Information section at the end of the article. 
justment for a predictor. Sportscience. 2006;10:46‐50.
26. Batterham AM, Hopkins WG. Making meaningful inferences
about magnitudes. Int Sports Physiol Perform. 2006;1(1):50‐57. How to cite this article: Hegyi A, Csala D, Péter A,
27. Zebis MK, Skotte J, Andersen CH, et al. Kettlebell swing tar- Finni T, Cronin NJ. High‐density electromyography
gets semitendinosus and supine leg curl targets biceps femoris: activity in various hamstring exercises. Scand J Med Sci
an EMG study with rehabilitation implications. Br J Sports Med. Sports. 2019;29:34–43. https://doi.org/10.1111/
2013;47(18):1192‐1198. sms.13303
28. Vigotsky AD, Harper EN, Ryan DR, Contreras B. Effects of
load on good morning kinematics and EMG activity. PeerJ.
2015;3:e708.