A Novel Continuous Intention-Driven Rehabilitation Robot and Its Training Effectiveness - Biomechatronics in Medicine and Healthcare - page 63

In the design of the continuous intention-driven control algorithm, EMG
signals from the main antagonist muscle pairs related to the wrist flexion and
extension was detected, i.e. the flexor carpi radialis (FCR) reflecting the voluntary
effort for flexion, and the extensor carpi radialis (ECR) for extension. It was
observed clinically that persons after stroke usually had reduced muscle coordi-
nating capability in limb movements (Georgopoulos 1986; Cauraugh et al . 2000;
Dewald et al . 2001; Gribble et al . 2003), e.g., when conducting a wrist extension in
the paretic upper limb, the muscles related to wrist flexion (FCR) or elbow move-
ments (e.g., biceps brachii (BIC) and triceps brachii (TRI)) may also contract. This
is mainly caused by involuntary muscle spasticity and compensatory movements
associated with other joints after stroke (Dewald et al . 2001). Therefore, in the
robot we designed, only the EMG signal from the FCR was used to drive the motor
during the wrist flexion phase, and the ECR EMG was applied for the extension
phase, to avoid the unnecessary co-contraction from the antagonist muscle (Hu
et al . 2008).

(1) The rawFCR and ECR EMG signals were first amplifiedwith a gain of 1000 and
were band-pass filtered in 10-400 Hz band. Then, the signals were sampled at
1000 Hz through the A/D card and captured by the PC-program. The envelope
of the EMG signals was obtained by full-wave rectification and filtering with a
moving average window (100 ms). The processed EMG signals from the FCR
and ECR muscles at this step were named as, EMG i ,where i represented ECR
or FCR.
(2) The EMG signals, EMG i , were then normalized to the range 0-1, i.e.,

EMG i , rest
EMG i , IMVC −

EMG i −

i = FCR in flexion

M Flexion/Extension = (5.1)

EMG i , rest i = ECR in extension

where i represented the FCR or ECR muscle, EMG i , rest was the amplitude of
the processed EMG signal of muscle i at rest, and EMG i , IMVC was the maximal
amplitude of the processed EMG signal of muscle i during isometric maximum
voluntary contraction (IMVC, isometric maximum voluntary flexion for FCR
(IMVF) and isometric maximum voluntary extension (IMVE) for ECR) at 0 ◦ of
the wrist angle, which was detected before the training session.
(3) The assistive torque during tracking was formulated as:

G

· T IMVF · M Flexion , During the flexion tracking phase

T a = (5.2)

G · T IMVF · M Extension , During the extension tracking phase

where, T a represented the motor generated assistive torque in the flexion or
extension phase during the tracking, G is a constant gain used to adjust the
magnitude of the assistive torque; and T IMVE and T IMVF were the maximal
values of the torque during isometric extension and flexion respectively, at the
wrist angle of 0 ◦ (Hu et al . 2008). Figure 5.2 shows the representative EMG and
torque signals during isometric maximum voluntary wrist flexion and EMG
signals during wrist tracking.