Fourier Approach for Functional Imaging

Excerpt

Determining the function of the brain is one of the important goals of neuroscience and its associated clinical disciplines of neurology, psychiatry, neurosurgery, and psychology. Functional brain mapping or neuroimaging is a set of imaging modalities that permits localization and visualization of brain activity. Neuroimaging has the potential of allowing observation of the function of the human nervous system, and to aid in clinical diagnosis and monitor ongoing neural function.

There are at least a dozen of neuroimaging approaches that allow monitoring of neural functions. These approaches differ very widely in their spatial and temporal resolution, level of invasiveness, and costs. They can be divided into two groups: (1) ones that measure neural activity (emission of electromagnetic fields) directly and (2) ones that rely on indirect measurements of metabolic signals. The techniques that fall into the first group typically are not used for neuroimaging because they are either very invasive, for example, electrophysiology, or provide very low spatial resolution, for example, electroencephalography (EEG) or magnetoencephalography (MEG). The other group consists of a variety of techniques that either use radioactive labels such as positron emission tomography (PET) or use signals of endogenous nature and include functional magnetic resonance imaging (fMRI) and optical imaging of intrinsic signals (ISOI). The latter group is particularly attractive for brain mapping because it provides moderate to high spatial resolution and does not require introduction of labels. What is characteristic of this group is that it measures hemodynamic signals, that is, the signals related to changes in blood oxygenation, blood volume, and blood flow rate [1,2]. As a consequence the measured signals are contaminated by the cardiovascular artifacts such as heart beat, respiration, and vasomotion [3]. The conventional imaging protocols are implemented in the time domain where randomized stimulus presentation is followed by data collection and averaging. As the result the evoked activity cannot be straightforwardly disassociated from the noise. The standard approach to noise reduction requires massive averaging, which in turn leads to long experimental runs.

Recently, we developed a rather general methodology that effectively removes the physiological noise [4]. A simple observation that the components of the noise are cyclic suggested implementation of the imaging protocol in the frequency domain instead of the time domain. The noise components appear localized in rather narrow frequency bands in the frequency domain, thus periodic stimulation at a frequency far from these bands will evoke a periodic response which is nearly free of the cardiovascular contaminations. Finally, a straightforward Fourier analysis extracts the parameters of the evoked response.

In this chapter we present the basics of the Fourier approach, which can be applied to any imaging modality that suffers from cyclic contaminations, and demonstrate its use for rapid acquisition of cortical maps by optical imaging of intrinsic signals.