16.11.2012 Views

Brain–Computer Interfaces - Index of

Brain–Computer Interfaces - Index of

Brain–Computer Interfaces - Index of

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

254 K.J. Miller and J.G. Ojemann<br />

In a one dimensional experiment, the learning process is rapid (3–24 min,<br />

see [8]). Though not always highly focal, we have observed cortical activity that<br />

increases with improved performance (Fig. 7b). Anecdotally, this is associated<br />

with a psychological “de-linking” <strong>of</strong> the screening behavior (e.g., hand movement<br />

imagery) and the cursor control, where the cursor gains some kind <strong>of</strong> abstract, toollike,<br />

representation with the subject. They simply will the cursor to move to the<br />

left or the right, and there is an increase in power in the appropriate feature. Recent<br />

results (an example is illustrated in Fig. 7c) have demonstrated efficacy using ECoG<br />

for two-dimensional control [10]. This requires two independent signals to drive the<br />

two degrees (typically up-down and left-right cursor position, as shown in Fig. 7c).<br />

The motor system allows for some intuitive, separable features, using, for example,<br />

hand and mouth, which are distinct when high-frequency ECoG features are used<br />

[6, 36, 37]. Despite success, simultaneous 2D control has been less robust than for<br />

one dimension. One reason for this is that it requires a good deal more mental coordination<br />

by the subject. For example, in this simple form <strong>of</strong> a 2D target task BCI,<br />

there is fundamental uncertainty in the null condition. When increased power in one<br />

channel drives the cursor up and a lack <strong>of</strong> power down, with increased power in<br />

another channel driving the cursor to the left, then weak power in both channels<br />

could be ambiguous when targeting a down or right sided target. It takes practice<br />

to coordinate the two null directions. Even if the signals are not overlapping<br />

during the screening for the two different tasks, when combined to achieve twodimensional<br />

cursor control, they may have some initial interdependence. Through<br />

feedback, however, these can be segregated by some individuals [37, 38].<br />

6 Case Study<br />

We conclude this chapter with a step-by-step example <strong>of</strong> successful implementation<br />

<strong>of</strong> electrocorticographic BCI.<br />

Background: The patient is an 18 year old right handed woman with intractable<br />

seizures related to a left frontal lesion. Electrodes were placed to map seizures and<br />

localize speech function prior to resective surgery, and the array covered speech<br />

areas and primary, peri-central, sensorimotor cortex. This excellent coverage <strong>of</strong><br />

motor cortex made her an ideal candidate for multiple degree <strong>of</strong> freedom cursor<br />

control. Signals were split outside <strong>of</strong> the patient, with a common ground and<br />

reference to the clinical system. The experiments were conducted in three stages<br />

spanning two sessions in the same day. The subject was positioned comfortably<br />

in her hospital bed throughout the experimental process, and the experimental setting<br />

is illustrated in Fig. 5. The electrode localization and cortical plotting were<br />

performed using the LOC package [16]. The three steps that the experiment was<br />

constructed upon were cue-based screening for feature identification, independent<br />

one-dimensional cursor-to-target feedback, and combined two-dimensional<br />

cursor-to-target feedback.<br />

Figure 7a, Screening for feature identification: In response to 3 s cues, she repetitively<br />

moved either her hand or her tongue in repetition, resting for 3 s in between

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!