Content uploaded by Jacques Duysens
Author content
All content in this area was uploaded by Jacques Duysens on Jan 07, 2015
Content may be subject to copyright.
Review
Cortical control of normal gait and precision stepping: An fNIRS study
Koen L.M. Koenraadt
a,
⁎, Eefje G.J. Roelofsen
a
, Jacques Duysens
a,b
, Noël L.W. Keijsers
a
a
Sint Maartenskliniek Nijmegen, Department of Research, PO box 9011, 6500 GM Nijmegen, The Netherlands
b
Katholieke Universiteit Leuven, Faculty of Kinesiology and Rehabilitation Sciences, Department of Biomedical Kinesiology, Tervuursevest 101, BE-3001 Leuven, Belgium
abstractarticle info
Article history:
Accepted 19 April 2013
Available online xxxx
Keywords:
Precision stepping
fNIRS
Gait
Motor cortex
Prefrontal cortex
Treadmill
Recently, real time imaging of the cortical control of gait became possible with functional near-infrared
spectroscopy (fNIRS). So far, little is known about the activations of various cortical areas in more complex
forms of gait, such as precision stepping. From previous work on animals and humans one would expect
precision stepping to elicit extra activity in the sensorimotor cortices (S1/M1), supplementary motor area
(SMA), as well as in prefrontal cortices (PFC). In the current study, hemodynamic changes in the PFC,
SMA, M1, and S1 were measured with fNIRS. In contrast to previous fNIRS gait studies, the technique was
optimized by the use of reference channels (to correct for superficial hemodynamic interference). Eleven
subjects randomly performed ten trials of treadmill walking at 3 km/h (normal walking) and ten trials
of 3 km/h treadmill walking on predefined spots for the left and right foot presented on the treadmill
(precision stepping). The walking trials of approximately 35 seconds were alternated with rest periods of
25–35 seconds consisting of quiet standing. The PFC revealed profound activation just prior to the onset
of both walking tasks. There was also extra activation of the PFC during the first half of the task period for
precision stepping. The SMA showed mainly increased activation prior to the start of both tasks. In contrast,
the sensorimotor cortex did not show a change in activation during either task as compared to a condition
of standing. The SMA, M1, and S1 revealed no significant differences between normal walking and precision
stepping. It was concluded that fNIRS is suited to record the planning and initiation of gait. The lack of M1/S1
activation during gait suggests that even in the current precision stepping task the control of ongoing gait
depended mostly on subcortical automatisms, while motor cortex contributions did not differ between
standing and walking.
© 2013 Elsevier Inc. All rights reserved.
Contents
Introduction ................................................................. 0
Methods .................................................................. 0
Subjects and experimental setup ..................................................... 0
NIRS imaging .............................................................. 0
Physiological and gait parameters .................................................... 0
Data analysis .............................................................. 0
Statistics ................................................................ 0
Results ................................................................... 0
Normal versus precision stepping .................................................... 0
Prefrontal cortex activation; phase effects ................................................. 0
Motor cortex activation; phase effects .................................................. 0
Physiological measures and gait characteristics .............................................. 0
Discussion .................................................................. 0
Acknowledgments .............................................................. 0
References ................................................................. 0
NeuroImage xxx (2013) xxx–xxx
⁎Corresponding author. Fax: +31 243659154.
E-mail address: k.koenraadt@maartenskliniek.nl (K.L.M. Koenraadt).
YNIMG-10382; No. of pages: 8; 4C:
1053-8119/$ –see front matter © 2013 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.neuroimage.2013.04.070
Contents lists available at SciVerse ScienceDirect
NeuroImage
journal homepage: www.elsevier.com/locate/ynimg
Please cite this article as: Koenraadt, K.L.M., et al., Cortical control of normal gait and precision stepping: An fNIRS study, NeuroImage (2013),
http://dx.doi.org/10.1016/j.neuroimage.2013.04.070
Introduction
Most of our knowledge about the control of gait originates from
animal studies whereas the cortical control in human gait is still not
entirely clear. An attempt of measuring cortical activity in real gait
in humans was made by La Fougere et al. (2010). After a walking
period of 10 minutes the conversion of intravenously injected
[
18
F]-FDG during this period was reflected by a PET scan. While infor-
mative, these data will not show the cortical activity during walking
itself. In contrast, studies with functional near-infrared spectroscopy
(fNIRS: Kurz et al., 2012; Miyai et al., 2001; Suzuki et al., 2004,
2008) and electro-encephalography (EEG: Gwin et al., 2010, 2011;
Presacco et al., 2011; Severens et al., 2012) allow to measure brain
activity during gait. In the study of Gwin et al. (2010) for example,
EEG recordings were made while subjects walked and even ran on a
treadmill. However, EEG has still its limitations, since neck muscles
and eye movements unrelated to the instructed task can affect the
quality of the recordings. Therefore, it remains a good alternative to
use fNIRS.
One of the questions that can be tackled with fNIRS concerns the
role of the sensorimotor cortex during different complexities of gait
in human. It has long been known from animal studies that the pri-
mary motor cortex (M1) is not essential for automated unperturbed
gait (Liddell and Phillips, 1944) but it is increasingly important for
precision stepping (such as walking on a ladder, Armstrong, 1988).
In normal walking sequential activations (i.e. mainly at the end of the
swing phase) of subgroups of neurons in M1 were regularly demon-
strated (Armstrong, 1986; Beloozerova and Sirota, 1993). In humans
the loss of motor cortex affects locomotion more severely, thereby
indicating an increasingly important contribution of the cortex in the
control of gait (Duysens et al., 2013). However, the recording studies
in human show a mixed picture. Some authors found that M1 is not ac-
tivated in imagined gait (Bakker et al., 2008; la Fougere et al., 2010),
whereas in PET studies with real walking M1 did reveal activation
(la Fougere et al., 2010; Tashiro et al., 2001). In the studies with fNIRS,
an involvement of the motor cortex is controversial. Several studies
revealed sensorimotor cortex activity during normal gait (Kurz et al.,
2012; Miyai et al., 2001) and backwards walking (Kurz et al., 2012).
However, Suzuki et al. (2004) revealed no effect of walking speed on
the sensorimotor cortex activation while clear SMA and prefrontal
cortex activations changes were demonstrated. In agreement with
this, the EEG records from Presacco et al. (2011) showed relatively little
activity over the motor cortex. It may be argued that such M1 activa-
tions require the use of more complex gait patterns (along the line of
the animal studies mentioned above). Hence, fNIRS studies on gait
including more complex forms of walking are particularly valuable to
throw light on the involvement of M1 in gait.
In the present study the focus is on precision stepping in compari-
son with normal gait. It is often assumed that cortical activity during a
movement implies deliberate conscious control, whereas subcortical
and spinal networks are responsible for automatic movements that re-
quire little conscious attention. In single-unit recording studies in cats
it was shown that undemanding steady state walking involves rela-
tively little cortical activity (Drew et al., 2004). Single-unit recording
studies further showed that pyramidal tract neurons in the motor
cortex are mostly active when vision is used to adapt gait in conditions
of gait challenges (Amos et al., 1990; uneven terrain, obstacle, etc.:
Beloozerova and Sirota, 1993; Drew, 1988, 1993; Drew et al., 2002;
Drew et al., 1996; Marple-Horvat et al., 1993; Widajewicz et al.,
1994). Furthermore, under these conditions, the activity in the motor
cortex “contributes primarily to the execution of the gait modifications
rather than to their planning”(Drew et al., 2008).
In humans the recording of activity over the motor cortex has only
be achieved sparsely. In the fNIRS study of Kurz et al. (2012) the
idea of introducing a form of precision stepping was achieved by
having the subjects walk forwards and backwards on a treadmill.
Gait variability was found to be greater during backward walking
compared to forward walking (Hoogkamer et al., 2012; Kurz et al.,
2012). The greater gait variability was reflected in higher cortical ac-
tivity (Kurz et al., 2012). In addition to the primary sensorimotor cor-
tices, the supplementary motor area (SMA) and prefrontal cortices
(PFC) are likely to play a role in more complex forms of gait. The
PFC is seen to be recruited in fNIRS studies when there is a high atten-
tion demand on the gait task (such as with increasing speed: Suzuki
et al., 2004). In the fNIRS study of Holtzer et al. (2011), increased
activations in the prefrontal cortex (PFC) were seen when walking
was combined with talking. In general, the PFC has been reported
as being typically active during attention demanding tasks (Wood
and Grafman, 2003). The SMA has been shown to play an important
role during normal gait in several fNIRS studies, such as those from
Suzuki et al. (2004, 2008). The area is known to be involved in the se-
lection, planning, and coordination of voluntary movements. This was
supported by the findings of a profound activation of the SMA in
the period prior to and around the start of locomotor tasks (Mihara
et al., 2008; Sahyoun et al., 2004). More generally, the SMA is also
known to be important for interlimb coordination of rhythmic arm
and leg movements (Debaere et al., 2001)(Debaere et al., 2001). In
fNIRS studies on difficult types of walking (such as backward walking)
Kurz et al. (2012) found increased SMA activity (along with other
premotor activity) for the more difficult task of backward walking.
Hence, the PFC and SMA seem important during gait and particularly
during complex gait.
The goal of the present study was to advance our knowledge of
precision stepping in comparison with normal gait, in line with the
historic importance of this task to evaluate the role of the different
motor areas in gait (Drew et al., 2004). fNIRS data were collected
from motor related cortical areas (sensorimotor and supplementary
motor areas) and prefrontal cortices during rest periods, steady state
walking on a treadmill at 3 km/h, and a precision stepping task on
the treadmill. Based on the literature described above it is hypothe-
sized that the latter task would require substantial activation in PFC
since it is an attention-demanding task. Furthermore, the activation
of the primary sensorimotor areas (M1 and S1) is likely to increase
with the more complex precision stepping task. Finally, the activation
of SMA is expected prior to and around the start of the locomotion
tasks and might also play a more prominent role while performing
the precision stepping task. An additional benefit of the present
study is that the fNIRS technique is currently improved considerably
by the use of reference channels to correct for superficial hemody-
namic interference. This interference is mainly caused by systemic
interference arising from cardiac activity, respiration, and other homeo-
static processes (Diamond et al., 2009; Obrig et al., 2000; Toronov et al.,
2000), which are very likely to interfere during gait. Several studies
have recently suggested to use reference channels (Gagnon et al.,
2012; Saager et al., 2011; Zhang et al., 2005), but has never been used
in walking studies.
Methods
Subjects and experimental setup
Eleven healthy subjects (3 males, 8 females) with a mean age of
23 years (SD: 4) participated in the present study. The study was
conducted according to the Declaration of Helsinki. Written informed
consent was obtained prior to the experiment. The subjects performed
two different locomotor tasks on a programmable treadmill (Forcelink,
Culemborg, The Netherlands). The two conditions consisted of 1) nor-
mal walking at 3 km/h and 2) precision stepping at 3 km/h that forced
the subjects to step on predefined spots on the treadmill. Each task pe-
riod was 35 seconds of duration (including instruction and starting
and stopping of the treadmill) and waspreceded with a baseline period
varying between 25 and 35 seconds. During the baseline periods
2K.L.M. Koenraadt et al. / NeuroImage xxx (2013) xxx–xxx
Please cite this article as: Koenraadt, K.L.M., et al., Cortical control of normal gait and precision stepping: An fNIRS study, NeuroImage (2013),
http://dx.doi.org/10.1016/j.neuroimage.2013.04.070
subjects were not allowed to hold on to the safety bars at the treadmill.
In total 10 trials of each condition were performed randomly and a
short break of approximately 3 minutes was given after 10 trials.
A video projector (Dell 2400MP, The Netherlands) attached on
the ceiling above the treadmill was used to present the rectangles
for the intended right (red rectangle) and left (green rectangle)
steps during the complex task. This procedure was extracted from
Bank et al (2011) who also projected rectangles as stepping stones
on the treadmill for gait rehabilitation purposes. Variation in step
width and step length was induced by presenting the rectangles of
each foot at 5 predefined positions in the frontal plane and 5 different
positions in the sagittal plane, respectively. The positions in the sagit-
tal plane were based on the step length of each individual that was
measured in a 1-minute treadmill walking session and for each step
the position of the rectangle in the sagittal plane was randomly ad-
justed to −30%, −15%, −0%, +15% or +30% of the individual step
length. Variation in position of the rectangle in the frontal plane was
based on an estimation of preferred step width in humans of 29 cm
(Donelan et al., 2001, 2004).For each step, the position of the rectangle
in the frontal plane was randomly adjusted with −25 cm, −12.5 cm,
0 cm, +12.5 cm, or +25 cm. Once a rectangle was projected at
the predefined spot it moved with the same speed and direction as
the treadmill. The video projector was also used for the presenta-
tion of the task instructions “Precision stepping”and “Walking”
(for 2 seconds) and a fixation cross on the treadmill during the normal
walking task and the baseline periods in between.
NIRS imaging
NIRS measurements were conducted with the use of two continuous-
wave NIRS instruments (Oxymon, Artinis Medical Systems, Zetten,
The Netherlands). The NIRS equipment used two wavelengths, 858
and 764 nm, and the data were sampled at 25 Hz. A soft and in size
adjustable headband was placed tightly around the participants head.
Subsequently, a six-channel motor cortex unit and a three-channel
prefrontal cortex unit were attached to the headband to prevent for
displacements.
The six-channel motor cortex unit included four long-separation
channels (channel 1–4) with an interoptode distance of 30 mm and
two short separation channels (channel 5–6) with an interoptode
distance of 10 mm (Fig. 1). The setup was created using two receivers
and five transmitting optodes, which were fixed in 10 mm thick foam
with holes for the optodes and Velcro was used to fasten the foam
to the headband. The four long-separation channels covered the S1,
M1, and supplementary motor areas (SMA and (pre-)SMA) of the
left hemisphere with respect to the Cz position of the International
10–20 system (Okamoto et al., 2004).
The prefrontal cortex unit consisted of two long-separation chan-
nels with an interoptode distance of 40 mm (lower- and upper
channel) and one short separation channel with an interoptode dis-
tance of 10 mm (channel 3 right panel of Fig. 1). This setup was cre-
ated by one receiver and three transmitting optodes, which were
placed on the prefrontal cortex with one channel overlapping the
Fp1-postion of the international 10–20 system on the forehead. The
holes for the optodes of that channel were incorporated into the
headband. The remaining optodes were fixed using holes in foam
and Velcro was used to fix the foam to the headband.
Physiological and gait parameters
Prior to the start of the experiment subjects were equipped with
two tri-axial accelerometers (Analog Devices, ADXL335) on the
shoes of both feet closely located above the head of metatarsal II
in order to calculate the step length from the 1-minute treadmill
walking part and to detect gait variability afterwards. Subsequently,
the NIRS setup was attached to the subjects head. A finger cuff was
fixed on the middle finger of the left hand to continuously monitor
the blood pressure (Finapres Medical Systems BV, Amsterdam, The
Netherlands). All data were sampled at 250 Hz.
Data analysis
The Oxymon software preprocessed the NIRS signals by converting
the changes in optical density in changes in HbO and HbR using the
modified Beer–Lambert law and the age dependent path length factor
(Duncan et al., 1996). After the measurements, analysis of HbO and
HbR signals was performed using a customized code implemented in
MatLab (R2007b). Firstly, a second order low pass Butterworth filter
with a cut off frequency of 1.25 Hz was conducted to reduce high fre-
quency noise. In addition, a second order high pass Butterworth filter
with a cut off frequency of 0.01 Hz was used to reduce low frequency
drift caused by the NIRS system. Subsequently, the short separation
channels (channel 3 in prefrontal cortex and channel 5 and 6 in motor
cortex, see Fig. 1) were used to remove hemodynamic changes insuper-
ficial tissue layers. The short-distance signal was scaled by a factor
and subtracted from the nearest long-distance signal. The scaling factor
was obtained during the 1 min rest period by matching the short-
separation signal with the long-separation channel data (Gagnon
et al., 2012; Saager et al., 2011). After the correction for superficial
interference, a second order low pass Butterworth filter with a cut off
frequency of 1 Hz was conducted. Finally, in order to compare the
data between all subjects, the maximal concentration change in HbO
and HbR overall trials and channelswas determined for eachindividual.
Subsequently, the individual data were normalized by dividing the
individual mean hemodynamic response amplitudes of all channels
by the corresponding (HbO or HbR) maximum concentration change
throughout the whole experiment. This procedure was used to decrease
the amplitude differences across subjects.
For the next steps of the data analysis the task period was divided
into three phases of 12.5 seconds; 1) a pre-task phase running from
12.5 seconds prior to the task instruction onto the task instruction,
2) an early-task phase running from 6 to 18.5 seconds after the start
of the task instruction, and 3) a late-task phase from 18.5 to 31 seconds
after the start of the task instruction. There was a 6 seconds period be-
tween the pre- and early-task and this consisted of 2 seconds of task
instruction succeeded by 4 seconds of treadmill acceleration to reach
5
61234 1
2
3
Cz Fp1
Fig. 1. Optodes configuration. Motor cortex setup (left panels) and prefrontal cortex
setup (right panels) of the fNIRS optodes. The upper panels show the optodes with
respect to the Cz and Fp1 locations of the International 10–20 system. In the lower
panels the channel numbers 1 to 4 represent the S1, M1, SMA and (pre-)SMA channel
for the motor cortex setup and 1 to 2 the lower and upper channel on the prefrontal
cortex. Channel 5 and 6 for the motor cortex setup and channel 3 for the prefrontal
setup represent the reference channels.
3K.L.M. Koenraadt et al. / NeuroImage xxx (2013) xxx–xxx
Please cite this article as: Koenraadt, K.L.M., et al., Cortical control of normal gait and precision stepping: An fNIRS study, NeuroImage (2013),
http://dx.doi.org/10.1016/j.neuroimage.2013.04.070
a constant velocity of 3 km/h (see Fig. 2). From the processed fNIRS
signal, the HbO and HbR concentrations were averaged over the
pre-task, the early-task, and late-task period for each channel and
each trial.
To analyze the accelerometer data, a peak detection procedure
was written in MatLab (R2007b) to determine all foot contacts.
Subsequently, the mean step time for each trial was calculated by
averaging the step times during the whole task period (early-task
and late-task). Finally, for each trial the gait variability was calculated
by taking the standard deviation of the step times.
The increase in mean blood pressure was calculated by subtraction
of the pre-task mean blood pressure from the mean blood pressure
during the whole task period for each trial. In addition, the heart
rate (beats/min) was derived from the blood pressure data and ana-
lyzed the same way as the blood pressure.
Statistics
One-way ANOVAs with the different task phases (pre-, early-, and
late-task) as repeated measures were used to determine whether a
response was seen in the hemodynamic response (HbO or HbR) in
each channel. Bonferroni corrections for multiple comparisons were
applied during the post hoc analyses. To determine differences in
conditions paired T-tests were performed between normal walking
and precision stepping for the early-task and late-task periods.
Differences between the physiological and gait measures for normal
walking and precision stepping were tested with paired T-tests over
the whole task period of 25 seconds (early- and late-task). In addi-
tion, paired T-tests were performed to indicate differences between
the pre-task and whole task period in blood pressure and heart rate.
Results
Fig. 3 shows the average HbO and HbR concentration changes of
the motor cortices (S1, M1, SMA, and (pre-) SMA in the left four
panels) for the whole study population. The right two panels present
the average HbO and HbR concentration changes on the upper and
lower channel of the PFC. The two SMA channels demonstrate compa-
rable changes over time, with an increase in HbO and a decrease in
HbR during the pre-task phase. Subsequently, the HbO concentration
decreases and the HbR concentration increases during the early-task
phase, finally reaching a plateau during the late-task phase. The M1
and S1 channels, on the other hand, show a peak in HbO concentra-
tions right after the treadmill reached its constant speed followed
by a small undershoot directly after the peak and a plateau during
the late-task phase. In addition, the HbR concentrations in M1 and
S1 increase during the early- and late-task phase. The hemodynamic
concentration changes in M1, S1, and (pre-)SMA over time are com-
parable between normal walking and precision stepping. Finally, the
PFC channels reveal an increase of HbO and a decrease in HbR during
the pre-task, a decrease in HbO and an increase in HbR during early-
task and no obvious changes during the late-task phase. For both PFC
channels it can be noticed that HbO concentrations for precision
stepping rise above HbO of normal walking right after the task in-
struction until a few seconds after the treadmill reaches constant
speed. Moreover, for the upper prefrontal channel the precision
stepping task reveals HbR concentrations lower compared to the
normal walking task during the early- and late-task.
Normal versus precision stepping
Statistical analyses of the differences between normal and preci-
sion stepping revealed a significant larger HbR decrease during
the early-task for precision stepping compared to normal walking
(pb0.05) in the upper PFC channel (Fig. 4). For the late-task no
significant difference was found (p= 0.26). In addition, no signifi-
cant difference was found for the HbR of the lower channel (early-task:
p= 0.49; late-task: p= 0.17) and for the HbO at both channels
(p-values ranging from 0.12 to 0.79). In contrast to the PFC, the motor
cortex channels revealed no significant difference between the two
conditions for the early- and late-task phases (p-values ranging from
0.12 to 0.96).
Prefrontal cortex activation; phase effects
Mean differences in HbO and HbR between the pre-task, and
either the early- or late-task are shown in Fig. 4 together with
the standard deviations along all subjects. For normal walking, the
one-way RM ANOVAs on the HbO data revealed no significant differ-
ences between the pre-, early-, and late-task phase on both channels
(lower channel: F(2,10) = 2.9, p= 0.08; upper channel: F(2,10) =
2.8, p= 0.08). Precision stepping, on the other hand, did reveal a
Velocity (km/h)
0
3
0-12.5 18.56 31
Time (s)
INSTRUCTION
Pre-
Task
Early-
Task
Late-
Task
Fig. 2. Timing of the experiment. The gray blocks indicate the three different phases
of the task used in the data-analysis. The dotted black line indicates the velocity of
the treadmill. After the 2 seconds task instruction, the treadmill reached a constant
speed of 3 km/h after 4 seconds.
Fig. 3. Mean group hemodynamic responses. HbO (solid) and HbR (dotted) relative
concentration changes over time for normal walking (NW, black lines) and precision
stepping (PS, gray lines) are presented for the four channels of the motor cortex
setup (left four panels) and for the two channels of prefrontal cortex setup (right
two panels). Vertical black lines represent in chronological order the start of the in-
struction (first solid line), treadmill reaching constant speed (first dotted line), slow
down of the treadmill (last solid line), and during the last vertical line the treadmill
comes to a standstill. Average SDs across the complete time course of HbO ranged
from 0.17 to 0.24 for the motor related channels and from 0.27 to 0.35 for the prefron-
tal channels. For HbR time courses the average SDs ranged from 0.19 to 0.28 for motor
and 0.28 to 0.33 for the prefrontal channels. Note that, instead of opposite changes in
HbO and HbR, parallel changes seem to occur in the S1 and M1 channels. This has
been addressed in many previous studies and might be the result of several underlying
mechanisms (Hoshi, 2007; Sato et al., 2005; Yamada et al., 2012; Yamamoto and Kato, 20 02).
4K.L.M. Koenraadt et al. / NeuroImage xxx (2013) xxx–xxx
Please cite this article as: Koenraadt, K.L.M., et al., Cortical control of normal gait and precision stepping: An fNIRS study, NeuroImage (2013),
http://dx.doi.org/10.1016/j.neuroimage.2013.04.070
significant phase effect in HbO on the upper channel (F(2,10) = 4.6,
pb0.05). However, post hoc analyses revealed no significant dif-
ference between the different phases (see Fig. 4). For HbR, the
upper channel revealed significant phase effects for normal walking
(F(2,10) = 10.3, pb0.001) and precision stepping (F(2,10) = 7.8,
pb0.005). Post hoc analyses revealed larger HbR concentrations for
early- and late-task compared to pre-task during normal walking,
while during precision stepping only the late-task HbR was signifi-
cantly larger compared to the pre-task phase. Individual analysis
revealed one subject that demonstrated no phase effects on both nor-
mal and precision stepping.
Motor cortex activation; phase effects
For S1, M1, SMA and (pre-)SMA, the mean differences and stan-
dard deviation of the HbO and HbR concentrations between the
pre-task and early-task and pre-task and late-task are shown in
Fig. 5. Normal walking revealed a significant phase effect of the HbO
data for the SMA channel (F(2,10) = 11.7, pb0.001). Post hoc anal-
yses revealed significant larger pre-task HbO responses compared
to early- and late-task. The same omnibus effect (F(2,10) = 8.2,
pb0.005) and post hoc results were seen in the precision stepping
HbO analyses for the SMA channel. For HbR during normal walking sig-
nificant omnibus effects were found for the channels S1 (F(2,10) = 4.4,
pb0.05), M1/SMA (F(2,10) = 8.2, pb0.005), and SMA (F(2,10) =4.7,
pb0.05). For precision stepping only the (pre-)SMA channel showed
significance (F(2,10) = 6.1, pb0.01). However, post hoc analyses
revealed no significant differences, although several trends were
noticed (as shown in Fig. 5). Most remarkably, for the (pre-)SMA in
both conditions the pre-task revealed a trend of larger HbR compared
to the early- and late-task. Individual analysis revealed only one subject
that revealed an absence of phase effects on the motor cortical channels
during normal walking. This subject revealed significant differences
between the phases for the prefrontal cortical channels. During the
precision stepping task all subjects revealed significant phase effects in
motor cortical areas.
Physiological measures and gait characteristics
The HR increased from 76 (±9) beats/min (during the rest period
before normal walking) to 83 (± 7) beats/min during the normal
walking task (pb0.01) and from 77 (± 9) beats/min (during the
rest period before precision stepping)to85(±8)beats/minduringthe
precision stepping task (pb0.01). No significant difference was found
between the increase in HR for the two conditions (p= 0.12). The
mean BP increased from 89 (± 13) mmHg during the rest period be-
fore normal walking to 90 (±14) mmHg during the normal walking
task (pb0.01) and from 88 (±12) mmHg during the rest period before
precision stepping to 90 (±14) mmHg during the precision stepping
task (pb0.05). No significant (p= 0.21) difference was found be-
tween the increase for normal walking and precision stepping.
The accelerometer data revealed mean step times of 0.66 (±0.02)
seconds for normal walking and 0.66 (± 0.04) seconds for preci-
sion stepping. The step time variability of 0.09 seconds (±0.02)
for precision stepping was significant larger (pb0.001) compared
to a step time variability of 0.04 seconds (±0.01) for the normal
walking task.
Discussion
The present fNIRS study examined hemodynamic responses in
multiple cortical areas before and during treadmill walking at 3 km/h
(“normal walking”) and a precision steppingtask at 3 km/h (“precision
stepping”). Reference optodes were used to correct for superficial
Fig. 4. Average hemodynamic responses for the prefrontal cortex. Mean HbO (upper panel) and HbR (lower panel) responses during the early- and late-task in comparison to
the pre-task are shown for the two prefrontal cortex channels. NW = normal walking, PS = precision stepping. For the comparisons within one condition, the asterisk (*) indicates
a significant difference with a pb0.0167 and the plus sign (+) indicates a trend with a p-value b0.05. For the comparisons between the conditions, the “x”indicates a significant
difference with pb0.05. One of the two channels was positioned across the Fp1 position of the International 10–20 system (i.e. the lower channel) and the other channel was
positioned approximately 1 cm more caudally (the upper channel).
5K.L.M. Koenraadt et al. / NeuroImage xxx (2013) xxx–xxx
Please cite this article as: Koenraadt, K.L.M., et al., Cortical control of normal gait and precision stepping: An fNIRS study, NeuroImage (2013),
http://dx.doi.org/10.1016/j.neuroimage.2013.04.070
hemodynamic interferences. The current study revealed increased acti-
vation (increased HbO and/or decreased HbR) in the SMA channels
prior to the start of normal walking and precision stepping. The senso-
rimotor cortex channels (S1 and M1), on the other hand, revealed no
differences betweenthe rest periods (i.e. standing) and normal walking
or precision stepping. The prefrontal cortex revealed enlarged oxygen-
ation priorto the task compared to the last half of thetask phase for nor-
mal walking and precision stepping. In addition, for precision stepping
the prefrontal cortex showed a prolonged activation during the first
half of the task.
With respect to the main purpose of the present study (changes
between normal walking and precision stepping), we demonstrated
that we introduced successfully more step time variability during
the precision stepping task. Considering the hemodynamic changes,
the precision stepping task revealed more PFC activation during the
first half of the task compared to normal walking. Furthermore, an
increase in activity, as indicated by an increase in HbO and a decrease
in HbR, occurred mainly before the start of both normal walking
andprecision stepping. The prefrontal cortex is known to be activated
during attention demanding tasks (Wood and Grafman, 2003; Yogev-
Seligmann et al., 2008). This has also been shown with fNIRS. In pos-
tural tasks for example, Mihara et al. (2008) showed fNIRS activity in
the PFC in conjunction with postural perturbations provided the sub-
jects were warned beforehand. This preparatory activation is most
likely related to the “allocation of attention”as typically found in the
dorsolateral PFC (Luks et al., 2007; Mihara et al., 2008). This type of
activation is also found during task execution with increasing complex-
ity. Recently, Holtzer et al. (2011) found an increased PFC activity during
walking while talking in comparison to normal walking. Our findings
of a prolonged activation of the PFC for the precision stepping task are
in line with these findings and indicate that more attention wasneeded
to perform precision stepping in comparison to normal walking.
In contrast to the PFC, the sensorimotor cortices revealed hardly
any significant hemodynamic changes during normal walking and
precision stepping, although a peak in HbO concentrations was no-
ticed at the beginning of the task. For S1, only the HbR concentration
changes for normal walking indicated somewhat more activity prior to
the task compared to the second half of the normal walking period. It
should be emphasized that these walking data were compared to a pre-
ceding period of standing. This is important since standing by itself may
elicit considerable activity. Indeed, in a previous fNIRS study, Mihara
et al. (2008) demonstratedan essential role of the sensorimotor cortices
in balance control. Since subjects were standing during the rest periods
in the present study, it is likely that S1 and M1 were activated during
the rest period and this may be part of the reason why no further incre-
ment in activity was seen during the walking task (ceiling effect). Previ-
ous fMRI, PET, and SPECT studies (Bakker et al., 2008; Dobkin et al.,
2004; Fukuyama et al., 1997; la Fougere et al., 2010) used rest periods
consisting of supine position and therefore no activation of S1 and M1
was expected during rest in these studies. Nevertheless, some fNIRS
studies were able to detect sensorimotor cortex activity related to the
legs with experimental gait paradigms comparable to the present
study (Kurz et al., 2012; Miyai et al., 2001). However, the slow walking
speeds of 1 km/h (Miyai et al., 2001)and1.6km/h(Kurz et al., 2012)
are remarkable in these studies. A walking speed more comparable
with the preferred walking speed is likely to decrease the M1 involve-
ment during gait since then locomotion depends more on subcortical
structures (such as CPGs, see den Otter et al., 2004; Duysens and Van
de Crommert, 1998; Nielsen, 2003). Furthermore, our negative findings
for M1 are in line with those found in an fNIRS study of Suzuki et al.
(2004) during 3 km/h and 5 km/h treadmill walking and with those
found by the EEG study of Presacco et al. (2011) at a maximum speed
of 2.4 km/h. The lack of additional M1 recruitment during precision
stepping is also in agreement with previous cat work (Armstrong and
Fig. 5. Average hemodynamic responses for the sensorimotor channels. Average HbO (upper panel) and HbR (lower panel) responses during the early- and late-task in comparison
to the pre-task are shown for the four sensorimotor channels. NW = normal walking, PS = precision stepping. * indicates a significant difference with a pb0.0167. + indicates a
trend with a pb0.05.
6K.L.M. Koenraadt et al. / NeuroImage xxx (2013) xxx–xxx
Please cite this article as: Koenraadt, K.L.M., et al., Cortical control of normal gait and precision stepping: An fNIRS study, NeuroImage (2013),
http://dx.doi.org/10.1016/j.neuroimage.2013.04.070
Drew, 1984) that demonstrated no changes in neuronal firing rates
when the animals increased walking speed or walked uphill.
In contrast to the sensorimotor cortices, the SMA revealed distinct
activation primarily before and around the start of the task for
both conditions. Since the SMA and (pre-)SMA channels revealed
comparable hemodynamic time courses they are discussed as one
area. The SMA has previously been reported to be involved in locomo-
tion (Fukuyama et al., 1997; Miyai et al., 2001). In the present study
there was also substantial activity of SMA during the rest periods
before gait. This might be explained by the balance control necessary
during the rest periods. Mihara et al. (2008) identified the SMA as in-
volved in balance control. However, since our precision stepping task
requested more balance control and since no differences were seen
on the SMA between normal walking and precision stepping this
explanation seems somewhat inappropriate for the present findings.
Another explanation originates from the role in motor preparation
of this specific cortical area. Sahyoun et al. (2004) revealed preparato-
ry activation of the SMA before actual foot flexion and extension in the
fMRI. In contrast, after the movement onset the SMA is no longer very
active. This was seen in the present experiments but also in earlier
findings (la Fougere et al., 2010; Suzuki et al., 2008). For example,
the [
18
F]-FDG PET study of la Fougere et al. (2010) failed to show
SMA activity after a 10 min walking period. Nevertheless, an fNIRS
study of Kurz et al. (2012) demonstrated an increase in SMA activity
during backwards walking compared to normal walking. The absence
of additional SMA activity during precision stepping in the present
study might be caused by the differences in speed of the treadmill
(as mentioned above for M1 discrepancies). The 3 km/h walking
speed in the present study for both normal walking and precision
stepping is much closer to the preferred walking speed in human com-
pared to the 1.6 km/h used in the study of Kurz et al. (2012). Presum-
ably, the more preferred walking speed in the present study resulted
in less dependence on the SMA for planning of the movement. Another
explanation for the discrepancy between our findings and those of
Kurz et al. (2012) might originate from the fact that they did not use
reference channels to correct for superficial interferences. Therefore,
factors such as blood pressure changes might have influenced the
hemodynamic responses and thereby the outcome in their study
(Gagnon et al., 2012; Saager et al., 2011). Since we did use reference
channels in the present study, we concluded the SMA to be mainly
active before the start of the task period in walking at 3 km/h, most
likely due to a preparatory/initiating function.
One of the major limitations of the present study is the small num-
ber of optodes used (as compared to some other gait related studies
such as Miyai et al (2001) and Suzuki et al. (2004, 2008)). Therefore,
not all the cortical areas involved in gait could be recorded. For exam-
ple, the parietal cortex is not measured although this area can be
expected to be very important in precision stepping (Drew et al.,
2004). Instead of increasing the number of cortical areas studied,
we chose to sacrifice some optodes to create reference channels in
order to correct for superficial hemodynamic interferences (Gagnon
et al., 2012; Saager et al., 2011). Since Gagnon et al. (2012) empha-
sized that systemic interference seems inhomogeneous across the
scalp, three short separation reference channels were created for six
long distance channels. For the small number of channels used, this
seems appropriate but it is recognized that for larger number of chan-
nels one could use other approaches. For example, the principal com-
ponent analysis as described in Zhang et al. (2005) and also used in
the fNIRS gait study of Kurz et al. (2012) seems a solid alternative
approach to correct for systemic interference when using a large
number of channels covering a large cortical area. A second (related)
limitation is that it is difficult to be certain about the areas recorded
from in view of the small number of optodes. For example, given
the position of channel 3 of the motor cortex setup in Fig. 2 it may
be argued that this channel was related to M1 and SMA activity rather
than purely SMA. In addition, the SMA channels might also cover
parts of the dorsal premotor cortex. Future work might therefore
focus on better methods of identification, for example one may profit
from combining different approaches, such as additional fMRI scans as
used by Kleinschmidt et al. (1996). Finally, other continuous wave
fNIRS issues like differences in optical path length between subjects,
interindividual differences in the type and time course of the hemody-
namic response, and the absence of absolute measures of the chromo-
phores might have influenced the results (Strangman et al., 2003).
In addition, the slow hemodynamic response following brain activity
makes fNIRS unsuitable to study cortical activation changes within the
different phases of the gait cycle or initiation of gait. For this purpose
EEG seems a more proper approach, as demonstrated by Gwin et al.
(2010, 2011). Despite these limitations, it is clear that fNIRS studies
definitely have a place in gait research, in particular now that the
method can beimproved with the addition of reference channels to cor-
rect for superficial hemodynamic interferences. This paves the way for
applications of fNIRS in future research to provide insight in the neural
mechanisms of movement disorders and future applications of fNIRS
in gait rehabilitation, for example for use in a brain–computer interface.
Acknowledgments
The authors gratefully acknowledge the support of theBrainGain
Smart Mix Programme of The Netherlands Ministry of Economic Affairs
and The Netherlands Ministry of Education, Culture and Science. We
would also like to thank Jan van Erp from TNO (Zeist, The Netherlands)
for the use of an additional fNIRS system.
Conflict of interest
The authors declare no conflict of interest.
References
Amos, A., Armstrong, D.M., Marple-Horvat, D.E., 1990. Changes in the discharge
patterns of motor cortical neurones associated with volitional changes in stepping
in the cat. Neurosci. Lett. 109, 107–112.
Armstrong, D.M., 1986. Supraspinal contributions to the initiation and control of loco-
motion in the cat. Prog. Neurobiol. 26, 273–361.
Armstrong, D.M., 1988. The supraspinal control of mammalian locomotion. J. Physiol.
405, 1–37.
Armstrong, D.M., Drew, T., 1984. Discharges of pyramidal tract and other motor cortical
neurones during locomotion in the cat. J. Physiol. 346, 471–495.
Bakker, M., DeLange, F.P., Helmich, R.C.,Scheeringa, R., Bloem, B.R., Toni, I., 2008. Cerebral
correlatesof motor imagery of normal and precision gait. Neuroimage 41, 998–1010.
Bank, P.J., Roerdink, M., Peper, C.E., 2011. Comparing the efficacy of metronome beeps
and stepping stones to adjust gait: steps to follow! Exp. Brain Res. 209, 159–169.
Beloozerova, I.N., Sirota, M.G., 1993. The role of the motor cortex in the control of accu-
racy of locomotor movements in the cat. J. Physiol. 461, 1–25.
Debaere, F., Swinnen, S.P., Beatse, E., Sunaert, S., Van, H.P., Duysens, J., 2001. Brain areas
involved in interlimb coordination: a distributed network. Neuroimage 14, 947–958.
den Otter, A.R., Geurts,A.C., Mulder, T., Duysens, J., 2004. Speed related changes inmuscle
activity from normal to very slow walking speeds. Gait Posture 19, 270–278.
Diamond, S.G., Perdue, K.L., Boas, D.A., 2009. A cerebrovascular response model for
functional neuroimaging including dynamic cerebral autoregulation. Math. Biosci.
220, 102–117.
Dobkin, B.H., Firestine, A., West, M., Saremi, K., Woods, R., 2004. Ankle dorsiflexion as an
fMRI paradigm to assay motor control for walking during rehabilitation. Neuroimage
23, 370–381.
Donelan, J.M., Kram, R., Kuo, A.D., 2001. Mechanical and metabolic determinants of
the preferred step width in human walking. Proc. Biol. Sci. 268, 1985–1992.
Donelan, J.M., Shipman, D.W., Kram, R., Kuo, A.D., 2004. Mechanical and metabolic re-
quirementsfor active lateral stabilization in human walking. J. Biomech. 37, 827–835.
Drew, T., 1988. Motor cortical cell discharge during voluntary gait modification. Brain
Res. 457, 181–187.
Drew, T., 1993. Motor cortical activity during voluntary gait modifications in the cat. I.
Cells related to the forelimbs. J. Neurophysiol. 70, 179–199.
Drew, T., Jiang, W., Kably, B., Lavoie, S., 1996. Role of the motor cortex in the control of
visually triggered gait modifications. Can. J. Physiol. Pharmacol. 74, 426–442.
Drew, T., Jiang, W., Widajewicz, W., 2002. Contributions of the motor cortex to the con-
trol of the hindlimbs during locomotion in the cat. Brain Res. Brain Res. Rev. 40,
178–191.
Drew, T., Prentice, S., Schepens, B., 2004. Cortical and brainstem control of locomotion.
Prog. Brain Res. 143, 251–261.
Drew, T., Andujar, J.E., Lajoie, K., Yakovenko, S., 2008. Cortical mechanisms involved in
visuomotor coordination during precision walking. Brain Res. Rev. 57, 199–211.
7K.L.M. Koenraadt et al. / NeuroImage xxx (2013) xxx–xxx
Please cite this article as: Koenraadt, K.L.M., et al., Cortical control of normal gait and precision stepping: An fNIRS study, NeuroImage (2013),
http://dx.doi.org/10.1016/j.neuroimage.2013.04.070
Duncan, A., Meek, J.H., Clemence, M., Elwell, C.E., Fallon, P., Tyszczuk, L., Cope, M.,
Delpy, D.T., 1996. Measurement of cranial optical path length as a function of age
using phase resolved near infrared spectroscopy. Pediatr. Res. 39, 889–894.
Duysens, J., Van de Crommert, H.W., 1998. Neural control of locomotion; The central
pattern generator from cats to humans. Gait Posture 7, 131–141.
Duysens, J., Severens, M., Nienhuis, B., 2013. How can active cycling produce less brain
activity than passive cycling? Clin. Neurophysiol. 124 (2), 217–218 (Feb).
Fukuyama, H., Ouchi, Y., Matsuzaki, S., Nagahama, Y., Yamauchi, H., Ogawa, M., Kimura,
J., Shibasaki, H., 1997. Brain functional activity during gait in normal subjects:
a SPECT study. Neurosci. Lett. 228, 183–186.
Gagnon, L., Cooper, R.J., Yucel, M.A., Perdue, K.L., Greve, D.N., Boas, D.A., 2012. Short
separation channel location impacts the performance of short channel regression
in NIRS. Neuroimage 59, 2518–2528.
Gwin, J.T., Gramann, K., Makeig, S., Ferris, D.P., 2010. Removal of movement artifact
from high-density EEG recorded during walking and running. J. Neurophysiol.
103, 3526–3534.
Gwin, J.T., Gramann, K., Makeig, S., Ferris, D.P., 2011. Electrocortical activity is coupled
to gait cycle phase during treadmill walking. Neuroimage 54, 1289–1296.
Holtzer, R., Mahoney, J.R., Izzetoglu, M., Izzetoglu, K., Onaral, B., Verghese, J., 2011.
fNIRS study of walking and walking while talking in young and old individuals.
J. Gerontol. A Biol. Sci. Med. Sci. 66, 879–887.
Hoogkamer, W., Massaad, F., Jansen, K., Bruijn, S.M., Duysens, J., 2012. Selective bilateral
activation of leg muscles after cutaneous nerve stimulation during backward
walking. J. Neurophysiol. 108, 1933–1941.
Hoshi, Y., 2007. Functional near-infrared spectroscopy: current status and future pros-
pects. J. Biomed. Opt. 12, 062106.
Kleinschmidt, A., Obrig, H., Requardt, M., Merboldt, K.D., Dirnagl, U., Villringer, A.,
Frahm, J., 1996. Simultaneous recording of cerebral blood oxygenation changes
during human brain activation by magnetic resonance imaging and near-infrared
spectroscopy. J. Cereb. Blood Flow Metab. 16, 817–826.
Kurz, M.J., Wilson, T.W., Arpin, D.J., 2012. Stride-time variability and sensorimotor
cortical activation during walking. Neuroimage 59, 1602–1607.
la Fougere, C., Zwergal, A., Rominger, A., Forster, S., Fesl, G., Dieterich, M., Brandt, T.,
Strupp, M., Bartenstein, P., Jahn, K., 2010. Real versus imagined locomotion:
a [18F]-FDG PET-fMRI comparison. Neuroimage 50, 1589–1598.
Liddell, E.G., Phillips, C.G., 1944. Pyramidal section in the cat. Brain 67, 1–9.
Luks, T.L., Simpson, G.V., Dale, C.L., Hough, M.G., 2007. Preparatory allocation of atten-
tion and adjustments in conflict processing. Neuroimage 35, 949–958.
Marple-Horvat, D.E., Amos, A.J., Armstrong, D.M., Criado, J.M., 1993. Changes in the
discharge patterns of cat motor cortex neurones during unexpected perturbations
of on-going locomotion. J. Physiol. 462, 87–113.
Mihara, M., Miyai, I., Hatakenaka, M., Kubota, K., Sakoda, S., 2008. Role of the prefrontal
cortex in human balance control. Neuroimage 43, 329–336.
Miyai, I., Tanabe, H.C., Sase, I., Eda, H., Oda, I., Konishi, I., Tsunazawa, Y., Suzuki, T.,
Yanagida, T., Kubota, K., 2001. Cortical mapping of gait in humans: a near-infrared
spectroscopic topography study. Neuroimage 14, 1186–1192.
Nielsen, J.B., 2003. How we walk: central control of muscle activity during human
walking. Neuroscientist 9, 195–204.
Obrig, H., Neufang, M., Wenzel, R., Kohl, M., Steinbrink, J., Einhaupl, K., Villringer, A.,
2000. Spontaneous low frequency oscillations of cerebral hemodynamics and me-
tabolism in human adults. Neuroimage 12, 623–639.
Okamoto, M., Dan, H., Sakamoto, K., Takeo, K., Shimizu, K., Kohno, S., Oda, I., Isobe, S.,
Suzuki, T., Kohyama, K., Dan, I., 2004. Three-dimensional probabilistic anatomical
cranio-cerebral correlation via the international 10–20 system oriented for trans-
cranial functional brain mapping. Neuroimage 21, 99–111.
Presacco, A., Goodman, R., Forrester, L., Contreras-Vidal, J.L., 2011. Neural decoding of
treadmill walking from noninvasive electroencephalographic signals. J. Neurophysiol.
106, 1875–1887.
Saager, R.B., Telleri, N.L., Berger, A.J., 2011. Two-detector Corrected Near Infrared
Spectroscopy (C-NIRS) detects hemodynamic activation responses more robustly
than single-detector NIRS. Neuroimage 55, 1679–1685.
Sahyoun, C., Floyer-Lea, A., Johansen-Berg, H., Matthews, P.M., 2004. Towards an un-
derstanding of gait control: brain activation during the anticipation, preparation
and execution of foot movements. Neuroimage 21, 568–575.
Sato, H., Fuchino, Y., Kiguchi, M., Katura, T., Maki, A., Yoro, T., Koizumi, H., 2005.
Intersubject variability of near-infrared spectroscopy signals during sensorimotor
cortex activation. J. Biomed. Opt. 10, 44001.
Severens, M., Nienhuis, B., Desain, P., Duysens, J., 2012. Feasibility of measuring event
related desynchronization with electroencephalography during walking. Conf.
Proc. IEEE Eng. Med. Biol. Soc. 2764–2767. http://dx.doi.org/10.1109/EMBC.2012.
6346537.
Strangman, G., Franceschini, M.A., Boas, D.A., 2003. Factors affecting the accuracy of
near-infrared spectroscopy concentration calculations for focal changes in oxygen-
ation parameters. Neuroimage 18, 865–879.
Suzuki, M., Miyai, I., Ono, T., Oda, I., Konishi, I., Kochiyama, T., Kubota, K., 2004. Prefron-
tal and premotor cortices are involved in adapting walking and running speed on
the treadmill: an optical imaging study. Neuroimage 23, 1020–1026.
Suzuki, M., Miyai, I., Ono, T., Kubota, K., 2008. Activities in the frontal cortex and gait
performance are modulated by preparation. An fNIRS study. Neuroimage 39,
600–607.
Tashiro, M., Itoh, M., Fujimoto, T., Fujiwara, T., Ota, H., Kubota, K., Higuchi, M., Okamura,
N., Ishii, K., Bereczki, D., Sasaki, H., 2001. 18F-FDG PET mapping of regional brain
activity in runners. J. Sports Med. Phys. Fitness 41, 11–17.
Toronov, V., Franceschini, M.A., Filiaci, M., Fantini, S., Wolf, M., Michalos, A., Gratton, E.,
2000. Near-infrared study of fluctuations in cerebral hemodynamics during rest and
motor stimulation: temporal analysis and spatial mapping. Med. Phys. 27, 801–815.
Widajewicz, W., Kably, B., Drew, T., 1994. Motor cortical activity during voluntary
gait modifications in the cat. II. Cells related to the hindlimbs. J. Neurophysiol.
72, 2070–2089.
Wood, J.N., Grafman, J., 2003. Human prefrontal cortex: processing and representational
perspectives. Nat. Rev. Neurosci. 4, 139–147.
Yamada, T., Umeyama, S., Matsuda, K., 2012. Separation of fNIRS signals into functional
and systemic components based on differences in hemodynamic modalities. PLoS
One 7, e50271.
Yamamoto, T., Kato, T., 2002. Paradoxical correlation between signal in functional mag-
netic resonance imaging and deoxygenated haemoglobin content in capillaries:
a new theoretical explanation. Phys. Med. Biol. 47, 1121–1141.
Yogev-Seligmann, G., Hausdorff, J.M., Giladi, N., 2008. The role of executive function
and attention in gait. Mov. Disord. 23, 329–342.
Zhang, Y., Brooks, D.H., Franceschini, M.A., Boas, D.A., 2005. Eigenvector-based spatial fil-
tering for reduction of physiological interference in diffuse optical imaging. J. Biomed.
Opt. 10, 11014.
8K.L.M. Koenraadt et al. / NeuroImage xxx (2013) xxx–xxx
Please cite this article as: Koenraadt, K.L.M., et al., Cortical control of normal gait and precision stepping: An fNIRS study, NeuroImage (2013),
http://dx.doi.org/10.1016/j.neuroimage.2013.04.070
