Biomedical Engineering Reference
In-Depth Information
Figure 4.6
Comparison of fingertip force of stroke patient P3 (female, 83 years old and
6 years post-stroke) between the beginning (left), the middle (middle) and the end (right)
of the training while the subject had to reach and maintain a target force (between dotted
lines) with the index finger (AF in the figure). Forces of the AF and the NAF as well as the
combined force are displayed.
to compose a specific word associated with a picture, e.g. the name of a city.
While fingers were maintained in a fixed position by the HandCARE, letters were
selected by isometrically applying a certain amount of force with a specific finger,
i.e. Active Finger (AF), and by minimizing the force applied by the other fingers,
i.e. Non- Active Fingers (NAF). The training was focused on the tripod thumb-
index-middle as these fingers are most commonly used in ADL.
One stroke subject with limited finger impairment performed this exercise
during the entire 8 weeks of therapy. With training, the number of successful
trials (i.e. correctly selected letter) increased for these three fingers
(+
16% for the
+
+
thumb,
29% for the middle finger). The time to perform
a successful trial was also significantly reduced
28% for the index, and
, which indicates that the
subject felt more comfortable with the exercise and was able to better control forces
generated by these fingers. This suggests that an intense isometric force training
promotes recovery of finger function.
Figure 4.6
illustrates how the force applied by the NAF tends to decrease with
training when the index finger produces force.
A similar exercise to train the regulation of power grasp force with the
HapticKnob illustrated improvement in the ability to generate, quickly adjust and
maintain precise grasping force.
(−
45%
)
4.4.3 Evolution in Muscle Activity Patterns
To determine whether post-stroke subjects were more limited in their ability
to modify patterns of muscle activation than age-matched healthy subjects, we
computed the principal components of the root-mean-squared (RMS) electromyo-
graphy (EMG) of 8 muscles of the hand and forearm during the performance
of the different exercises. We assumed that greater limitation would be evident
as a higher percentage of the variance being accounted for by fewer principal
components (PCs). We compared the number of PCs needed to account for at least
