Information Technology Reference
In-Depth Information
version of the CSRT, however, visual stimuli are presented on the LCD monitor and
the user is expected to step accordingly on respective positions on the MAT (see
Figure 4). In this work, the MAT showed high correlations with the laboratory-based
measure of the CSRT test. Also, the ability to reliably differentiate between fallers
and non-fallers was also confirmed. However, one of the pitfalls of this work is that
the use of the Dance Mat still imposes an obstacle on the degree of freedom and does
not allow for the collection of spatial parameters.
In [16], the authors introduced StepKinnection™, a system that uses the Microsoft
Kinect's depth and motion capture technology to measure stepping performance
through the use of a hybrid version of the Choice Stepping Reaction Time (CSRT)
task. In this system, the person stands in front of a computer screen or TV connected
to a Kinect PC. The representation of the player in the system is pair of shoes
mirroring the person's feet movements (see Figure 2). Six symmetrically distributed
square-shaped virtual panels are drawn on the screen representing the step panels
surrounding the person. The mechanics of the test are the same as the original version
of the CSRT with the exception that the person steps in space (see Figure 3).
User's actions such as a 'step' or a 'foot liftoff' are recognized by translating the
user's lower limb movements obtained by the Kinect's depth sensor into the game-
like platform. The system continuously retrieves spatial data (or skeleton frames) to
determine whether a foot is intersecting one of the virtual panels. When the
intersection of a foot and one panel is detected, the system records this action along
with a timestamp. These actions are subsequently used for the calculation of the
following time-based variables which are essential in the completion of the CSRT
test: (1) Decision Time (DT): time elapsed between the instance where the sector
turns green and the player lifts his/her foot off the central panel, (2) Movement Time
(MT): time it takes for the user to step on a coloured sector once leg movement is
initiated; (3) Response time (RT): Decision Time (DT) + Movement Time (MT). In
addition to this, complementary spatial parameters are also captured for a more
descriptive assessment of fall risk. These are: (1) Active Foot Positioning (AFP):
(x,y,z) coordinates of the user's foot when stepping on a panel; (2) Observed Step
Length (SL): distance between left and right foot while stepping on a panel; (3)
Stepping Accuracy Coefficient (SAC): Difference between expected step length and
observed step length.
The ability to collect these measures simultaneously makes this system potentially
useful in a clinical setting as it can evaluate several dimensions involved in the
assessment of fall risk in older people. However, since the Kinect is a camera-based
device restricted to process 30 skeleton frames per second, the accuracy of the
mentioned time-based measurements needs further evaluation.
