A Digital Stereo Microscope Platform for Microsurgery Training - Technologies and Applications of Artificial Intelligence - page 300

The experiment consists of placing the task targets (example task targets are shown
in Fig.1) under the software defined microscope. Then a predefined surgical move-
ment is made using the digital forceps.
The training surgeon uses the digital forceps shown in Fig.3(b) to trace, locate,
place accurately and with less number of movements. The movement data is cap-
tured by the microscope system and is analyzed as described in section 3.2 and 3.3.
The surgeon may visualize the movement profile with the quality markings and tries
to make better movements. For each surgeon, we conducted 40 motion challenges. In
the subsequent section, the data corresponding to one surgeon is used to explain the
measures.

3.2 Motion Signal Characteristics

The frequency spectrum of the motion signal is shown in Fig. 4, the motion signal has
two parts, the lower frequency ( < 25 Hz) is a combination of voluntary motion signal
and physiological tremor. Signal components above 25 Hz are instrument and elec-
tromagnetic noise. The noise may be filtered out through frequency based lowpass
filter.

Fig. 4. Frequency spectrum of hand motion

Another characteristics of the motion is the size of the motion which is analyzed by
segmenting the motion into a number of movements. A movement is defined as a
sequence of spatial points wherein between any two points Pi and Pi+1, the following
conditions hold.

|Pi, Pi+1| > ʴ (3)

Each movement represent one dexterous action. The size of the movement is
represented by (P 1 , P 2 , …. P n ) its path length defined as

L 1,n = |P 2 - P 1 | + ….. |P i+1 - P i | + …. |P n - P n-1 | (4)

Using the size of the movement a histogram is computed and from the histogram the phy-
siological process can be understood. As shown in Fig. 5, the histogram is multi-modal