Table 1.

Summary of non-invasive techniques for measuring brain activity.

Non-invasive techniques	Main features	Pro's	Con's	References
EEG	It is the most traditional and favored method for BCI development	Systems are relatively low cost compared to others	Signal-to-noise levels are low, resulting from physiological (skull) and environmental (electrical) interferences	Coyle et al. (2007), Rossini et al. (2007), Kayagil et al. (2009), Lee et al. (2009), Sagara et al. (2009), Tai and Chau (2009), Lin et al. (2010)
	It reflects the neuronal electrical activity, triggered by external stimuli, in large cell populations	Data can be recorded simultaneously with fMRI	Has the lowest spatial resolution of the techniques
fMRI	Neuronal activity measured from changes in magnetic properties of blood oxygenated and deoxygenated hemoglobin concentrations	Spatial resolution is higher than EEG and MEG	Impractical due to limitations such as prohibitive cost, technological limitations, and ambient size, like MEG	Lebedev and Nicolelis (2006), Coyle et al. (2007), Birbaumer et al. (2008), Soraghan et al. (2008), Gerven et al. (2009), Kayagil et al. (2009), Lee et al. (2009), Sagara et al. (2009), Tai and Chau (2009)
	Activity of the deep brain structures can be sampled but temporal delays of several seconds occur	Superior at localizing specific brain activity, in real-time, for translation into a BCI	Subjects cannot talk or move around when collecting data
			Temporal resolution is poor
MEG	It reflects the magnetic field of neuronal activity,triggered by external stimuli, in large cell populations	Temporal resolution is very good	Data cannot be recorded simultaneously with fMRI	Rossini et al. (2007), Babiloni et al. (2009), Gerven et al. (2009)
	It provides three-dimensional localization of dipolar field distributions in cortical gyri and sulci	Signal characteristics are not influenced by the skull and meninges	Impractical due to limitations such as prohibitive cost, technological limitations, and ambient size, like fMRI Spatial resolution is poor
NIRS	Directly monitors cortical activity through arterial and venous hemoglobin and tissue oxygenation changes	NIRS is used in optical topography to operate BMIs by functional mapping of prefrontal cortical activity	External device control speed is limited due to metabolic response speeds, approximately 16 s are needed for list selection from a single channel	Coyle et al. (2007), Tsubone et al. (2007), Utsugi et al. (2007), Birbaumer et al. (2008), Gerven et al. (2009), Sagara et al. (2009), Tai and Chau (2009)
	Uses the near-infrared spectrum	Does not require long-term training as required by other techniques	Penetration depth of 1–3 cm can be measured but it is affected by skin color
		It is an inexpensive and mobile alternative to fMRI	Temporal resolution is similar to fMRI but spatial resolution is poorer
		Systems are relatively low cost compared to others