Principles of Multimodal Functional Imaging and Data Integration

Abstract

In a system as complex as the human brain, one cannot conceive of meaningful events involving a change in a single observable (physiological parameter). Therefore, achieving the ultimate aim of a complete understanding of brain events and brain activity in general will require the integration of a variety of observations related to these events. Multimodal imaging, or more generally measurements whereby data from various types of instruments are brought together, has arisen partly from this realisation, partly because some events are best observed in one modality and the investigator is interested in another (e.g. a more recently developed modality), and to be honest sometimes as a response to the technical challenge of combining modalities for simultaneous observations. Fundamentally, multimodal imaging should allow the investigator to address the question: what happens to brain observable Z when observable X changes (or event Y occurs)? In the second half of the twentieth century, and particularly since the 1990s, a rapid development of noninvasive functional and structural brain imaging methods has occurred. While some of these developments have resulted from gradual improvements in some methods, other developments have led to completely new approaches for measuring brain activity, affording new types of information about the brain. In the former case, the older methods were eventually replaced [e.g. scintigraphic methods by positron emission tomography (PET), and SPECT or low-field MRI (magnetic resonance imaging) by higher field MRI]. In the latter case, however, newer developments have not replaced older ones; rather, they have been added to an ever-larger orchestra of functional and structural neuroimaging methods consisting of techniques that offer complementary information about the brain. Table 1 gives an overview of currently available methods for noninvasive brain imaging and the principle that each exploits.