Figure 1.
Smartphone Brain Scanner applications running on Android devices.
Neurofeedback training and real-time 3D source reconstruction running on Android mobile devices via a wireless connection to an Emotiv or Easycap EEG systems.
Figure 2.
Overview of the layered architecture of the SBS2 framework.
Data from the connected EEG hardware are acquired and extracted by specific adapters and all subsequent processing is hardware agnostic. The empty boxes indicate the extendability of the architecture allowing additional hardware devices for data acquisition and additional processing methods.
Figure 3.
The Smartphone Brain Scanner architecture.
Data are acquired in the first layer from the EEG hardware, passed to the Data Processing Layer, and extracted. Features, as well as raw values, are then available for applications.
Figure 4.
Snapshot of the SBS2 real time brain imaging system running on a Samsung Galaxy Note 2.
EEG recorded using the Emocap [29], based on the Emotiv EEG wireless transmission setup. Visible in the picture is the entire setup required for data acquisition, processing, and visualization. The subject of the photograph has given written informed consent, as outlined in the PLOS consent form, to publish this photograph.
Figure 5.
10 Hz sinusoid is generated with a computer sound card, amplified, and fed into an oscilloscope and the EEG hardware. The device acquiring the EEG data responds to the sinusoid signal with changes of screen brightness, which is detected by a photocell connected to the oscilloscope. The time difference between the two signals is used to calculate the system delay.
Figure 6.
The system responds to the sinusoid signal peak (time 0). The red color () indicates minimal observed delay; the blue color (
) indicates jitter. Galaxy Note running Android 4.0.1, 60 Hz AMOLED screen,
,
; Nexus 7 running Android 4.1.1, 60 Hz IPS LCD screen,
,
; MacbookPro, LCD screen (60 Hz),
,
.
Figure 7.
Measured sampling frequency, including measurement resolution for three random Emotiv EEG devices, recordings for each.
All measured rates, including uncertainty, are between and
, which corresponds to
and
of nominal
. The measurements were performed with
resolution (
accuracy) on
EEG packets. All tests were performed at normal temperature on a single day. We can note consistent results within and across devices.
Figure 8.
Distances between 4-sample frames.
Red line indicates expected distance of between the groups of four
packets. The bars indicate the observed distance. We can see that the Emotiv system compensates every
samples to keep the average (black line) at the correct level.
Figure 9.
Finger-tapping results for Emotiv EEG and Biosemi standard equiment resampled to 14 channels.
Mean (solid lines) and standard deviation (dashed lines) of reconstructed current source power in the left (L) Precentral AAL regions calculated across right-cued, imagined finger-tapping conditions. Mean activity was normalized to unit at . Both activities are based on 3D reconstruction with online estimation of the
and
parameters using the Minimum Norm approch.