Abstract
This thesis describes studies on the electric potential of a neuronal population of about half a million neurons. It is essential to understand neurophysiology at the millimeter scale since ECoG and fMRI studies have shown that some functional units are specifically defined at this scale. While ECoG measures neuronal activity,
... read more
fMRI only reflects its vascular and metabolic correlate and it is not well understood how this relates to the underlying neurophysiology. In the setting of motor and language function we used ECoG to define the dominant neurophysiological processes at the scale of millimeters: high frequency changes that are related to local neuronal activity and low frequency changes that reflect more global processes. We studied their interaction and investigated how these are related to the fMRI BOLD change to gain a better understanding of the role of these processes. The studies described in this thesis were done in a clinical setting with a brain computer interface (BCI) application in mind. These studies are relevant for at least two questions from the BCI field. First, which part of the ECoG signal is optimal for BCI and whether the optimal location for a BCI device can be predicted with fMRI. This thesis has shown several characteristics of the fMRI and ECoG signal that can help to answer these questions. First, chapter 4 showed that in the subjects who were scanned on a 1.5 T scanner instead of on a 3 T scanner, the signal in primary sensorimotor cortex did not correlate as well with high frequencies. It is thus necessary to have a good signal to noise ratio to be able to use fMRI as a localizer for high frequencies. Second, chapter 4 shows that large fMRI signal changes in primary sensorimotor areas corresponded best with high frequency increases, similar as what we have shown in chapter 3 for premotor cortex during motor imagery and for dorsolateral prefrontal cortex during working memory 58. However, in nonprimary areas, there may be BOLD changes that do not correspond to high frequency power change, but only to low frequency power changes. It is thus essential to choose areas with a primary involvement in a task, with a large BOLD signal change, before implantation of an electrode is considered. Furthermore, it is often assumed for brain computer interfaces that every action is linked to the same neuronal activity. Chapter 5 shows a drastic decrease in high frequency activity during fast repeated movement. It may be much more difficult to classify these faster movements compared to slower movement. In short, using fMRI to localize target areas for BCI seems possible when people are scanned on a 3T machine or at even higher field strength, provided that a) an electrode is placed on an area primarily involved in the task at hand, b) high frequency power changes are used to control a BCI and c) repetition effects are well understood before an electrode is placed.
show less
