Advances in Neural Signal Processing
16
The trial period was split into two time windows: forward period (1000 ms
after the first sound (1–1000 ms)) and comeback period (1000 ms after the second
sound for each trial (1001–2000 ms)). These two time windows represent two
distinct movements in each trial. Forward period includes the movement from the
starting position to the ending position, while comeback period includes the movement back from the ending position to the starting position (see Figure 1).
Statistical analysis was performed on specific time windows in both forward
period and comeback period. Time windows were first defined by the accelerometer
data. Our accelerometer can detect not only the acceleration but also the position in
space starting from a baseline reference position. We observed that the participants
reached the starting point and ending point slightly after the actual sound presentation. We extracted data from the accelerometer from each trial and observed that
the delay between sound presentation and actual start of the movement from the
starting point was 190 ms while between sound presentation and start of actual
movement from the ending point was 160 ms (see Figure 2).
Finally, statistical analysis was performed using STATISTICA software
(StatSoft, Inc., Tulsa, OK, USA).
2.3.2.2 Source localization
Using low-resolution brain electromagnetic tomography (LORETA) [43], it is
possible to solve the inverse problem in EEG and localize generators of electrophysiological components of EEG signal in a specific frequency band.
After time-frequency analysis, we performed source localization analysis using
LORETA in order to observe differences between vertical and diagonal movements
for theta (4–7 Hz), alpha (8–12 Hz), and beta (13–30 Hz) activity. Source localization analysis was conducted in specific time windows using a data-driven approach,
according to what we observed in the time-frequency analysis.
More specifically, we focused on the time windows previously observed in timefrequency analysis for each specific frequency band. Therefore, we compared the
generator of theta in diagonal and vertical movements during the planning of movement in forward period (between 100 and 300 ms), of alpha during the two peaks
of activity in comeback period (200–400 ms; 650–850 ms) and of beta in forward
period (320–520 ms) and comeback period (220–420 ms). Specifically, analyzed frequencies were theta (7 Hz), alpha (11 Hz), and beta (two frequencies, 19 and 23 Hz).
We performed one-tailed t-test comparisons based on the time-frequency
observed activity pattern (i.e., diagonal-related activity greater than vertical-related
activity or vice versa). Therefore, for theta, we expected diagonal > vertical; for
alpha, we expected diagonal > vertical in P1 and vertical > diagonal in P2; and for
beta, we expected diagonal > vertical in both time windows.
Statistical analysis was conducted using subject-wise normalization, and results
are expressed as a t-test on the logarithmically transformed data. Nonparametric randomized permutation and probability threshold corrections were performed [44].
3. Results
3.1 Time-frequency results
3.1.1 Theta (4–7 Hz)
We selected a time window ranging from −100 to +100 ms around the peaks
recorded by accelerometer and then extracted and analyzed the activity. We
16
The trial period was split into two time windows: forward period (1000 ms
after the first sound (1–1000 ms)) and comeback period (1000 ms after the second
sound for each trial (1001–2000 ms)). These two time windows represent two
distinct movements in each trial. Forward period includes the movement from the
starting position to the ending position, while comeback period includes the movement back from the ending position to the starting position (see Figure 1).
Statistical analysis was performed on specific time windows in both forward
period and comeback period. Time windows were first defined by the accelerometer
data. Our accelerometer can detect not only the acceleration but also the position in
space starting from a baseline reference position. We observed that the participants
reached the starting point and ending point slightly after the actual sound presentation. We extracted data from the accelerometer from each trial and observed that
the delay between sound presentation and actual start of the movement from the
starting point was 190 ms while between sound presentation and start of actual
movement from the ending point was 160 ms (see Figure 2).
Finally, statistical analysis was performed using STATISTICA software
(StatSoft, Inc., Tulsa, OK, USA).
2.3.2.2 Source localization
Using low-resolution brain electromagnetic tomography (LORETA) [43], it is
possible to solve the inverse problem in EEG and localize generators of electrophysiological components of EEG signal in a specific frequency band.
After time-frequency analysis, we performed source localization analysis using
LORETA in order to observe differences between vertical and diagonal movements
for theta (4–7 Hz), alpha (8–12 Hz), and beta (13–30 Hz) activity. Source localization analysis was conducted in specific time windows using a data-driven approach,
according to what we observed in the time-frequency analysis.
More specifically, we focused on the time windows previously observed in timefrequency analysis for each specific frequency band. Therefore, we compared the
generator of theta in diagonal and vertical movements during the planning of movement in forward period (between 100 and 300 ms), of alpha during the two peaks
of activity in comeback period (200–400 ms; 650–850 ms) and of beta in forward
period (320–520 ms) and comeback period (220–420 ms). Specifically, analyzed frequencies were theta (7 Hz), alpha (11 Hz), and beta (two frequencies, 19 and 23 Hz).
We performed one-tailed t-test comparisons based on the time-frequency
observed activity pattern (i.e., diagonal-related activity greater than vertical-related
activity or vice versa). Therefore, for theta, we expected diagonal > vertical; for
alpha, we expected diagonal > vertical in P1 and vertical > diagonal in P2; and for
beta, we expected diagonal > vertical in both time windows.
Statistical analysis was conducted using subject-wise normalization, and results
are expressed as a t-test on the logarithmically transformed data. Nonparametric randomized permutation and probability threshold corrections were performed [44].
3. Results
3.1 Time-frequency results
3.1.1 Theta (4–7 Hz)
We selected a time window ranging from −100 to +100 ms around the peaks
recorded by accelerometer and then extracted and analyzed the activity. We