Computational Intelligence in Gait and Movement Pattern Analysis - Computational Intelligence in Biomedical Engineering - page 303

Commercial systems are available for both tethered and wireless surface
and fine-wire electromyography recording. Although raw EMG signals
(Figure 7.7a) can show approximate on/off timing and activity levels, these
signals are usually further processed to gain more information prior to their
application. This includes minimizing both low- and high-frequency noise
using digital filters (low and high pass) and rectification to convert negative
signals to positive values. Low-pass filters allow for high-frequency noise to be
eliminated, such as that from electrical equipment, whereas high-pass filters
minimize the low-frequency noise due to sources such as movement between
the electrode and the skin or movement of the cable connecting the electrode
to the amplifier. Refer to Chapter 5 for details of EMG signal processing.
A low-pass filtered EMG signal is known as a linear envelope ; this closely
approximates the muscle forces and is frequently used in gait applications.
Figure 7.8 illustrates the activity of major lower-limb muscle groups over the
gait cycle. Within the human body, there are both two-joint and single-joint
muscles. Single-joint muscles affect motion at one joint only, thus, tibialis ante-
rior activity at the start of the gait cycle keeps the foot dorsiflexed and ensures
proper heel contact (see Figure 7.8). A two-joint muscle can have two functions
at the two joints. For example, the gastrocnemius may work either as an ankle
plantarflexors during the push-off phase or as a knee flexor (see Figure 7.8).
7.2.5 Combined Output
Motion data are frequently collected in conjunction with force and EMG data,
and postprocessing is often undertaken to derive useful parameters. Motion
and force data, for example, can be combined using anthropometric measures
to calculate internal joint moments and the energy and power around ankle,
knee, and hip joints using inverse dynamics (Winter, 1990). Figure 7.9 and
7.5d illustrate typical 3-D output from a commercial motion analysis system
(Optotrak motion analysis system and Visual3D software) displaying kine-
matic, kinetic, and muscle activity information during gait.
There are also techniques to superimpose force, force vectors derived from
the GRF data, and EMG information onto the video image in real time. This
type of analysis can assist the study of force vectors relative to lower-limb
joint positions. There have been attempts to combine information from two or
more biomechanical measurement systems into an integrated output to allow
visualization of the motion and associated forces (Begg et al., 1991; Roberts
et al., 1995; Rowe, 1996). Figure 7.10 illustrates output from the system devel-
oped by Begg et al. (1991), which uses a combination of video and force
platform information (force vector). One advantage is that this output can
reveal effects of the foot-ground reaction forces on the individual lower-limb
joints. This can be achieved by observing the type (flexing/extending) of
joint moment. Moments can be estimated by studying the moment arms and
magnitude of the resultant GRF. Such real-time analysis can be helpful for
the assessment of orthotic and prosthetic interventions. From the two sagit-
tal plane outputs in Figure 7.10, although Figure 7.10a illustrates a flexing