<27 Mar 2007>

The Micro-Architecture of the Cerebral Cortex: its impact on Functional Neuroimaging in Humans

Jorge Riera
Neuronal Mass Dynamics group, Dept. of Functional Brain Imaging
Institute of Development, Aging and Cancer, Tohoku University

History
The development of functional neuroimaging has been always determined by remarkable and noteworthy technological advances of an epoch. Among the most-widely used modalities are the functional magnetic resonance imaging (fMRI), the electro/magneto encephalography (EEG/MEG), the positron emission tomography (PET), the single photon emission computed tomography (SPECT) and the functional near infrared spectroscopy (fNIRs). Historically, a buzz of excitement and fashionable interest has typically followed the introduction of a particular modality, but after a massive volume of studies were performed, once its limitations were revealed, the excitement gave way to something more akin to a motivational crisis, which then catalyzed the invention of the successive modality. The discovery of EEG in 1929 by the German psychiatrist Hans Berger could be considered as the founding of functional neuroimaging. By means of EEG amplifiers, voltage differences could be recorded at a number of sites on the scalp with a temporal resolution of milliseconds. The technique was originally introduced to study the alpha rhythms in humans, but latterly it has become a powerful tool for medical diagnosis in neurology and psychiatry. After producing a profusion of fruitful results over a long period, the EEG was strongly criticized due to its susceptibility to the conductive profile of the head tissues, a fact lying beneath its inadequacy to resolve the underlying current sources inside the brain. To overcome such limitations, a more expensive functional neuroimaging modality, the MEG, came to light in the second half of 20th century as a new prospect with huge potential. David Cohen in 1968 was the first to measure the weak magnetic field induced by neuronal events. For that end, he used an external copper induction coil as the detector inside a magnetically shielded room. However, the golden era of MEG started with the later introduction of superconducting quantum interference devices (SQUID). Based on the apparent transparence of head tissues to a quasi-static magnetic field, the MEG was proclaimed as the ultimate technique for determining the current sources associated with neuronal events with a similar temporal resolution as for the EEG but a better resolution to resolve current sources in space. The popularity of MEG increased considerably when simultaneous EEG and MEG recordings using a huge number of sensors became realizable by the end of 80’s. The combination of EEG and MEG data permitted to determine not only a number of function-associated brain regions but also how these regions interacted.
The first applications of positron annihilation radiation for medical imaging came in the early part of 50’s from independent studies at the Massachusetts General Hospital and Duke University, while the concept underlying SPECT was introduced around ten years later. However, the evolution of instruments for both modalities took over several years (1969-1985) to finally achieve the existing prototypes of PET scanners and gamma cameras (SPECT). As a consequence of progresses in developing radioisotopes from the early of 70s to the middle of 1980's, MEG did not last long as the absolute modality for functional neuroimaging. The radioisotopes are unstable atoms produced in cyclotrons, which have the property to emit gamma rays and/or subatomic particles while decaying radioactively. 3D images of radioactively labeled compounds inside the brain, called radiotracers, were reconstructed with a good enough spatial resolution by observing their emission through a PET scanner or a gamma camera. For several years, PET/SPECT modalities allowed either researchers or clinicians to observe large changes in the cerebral blood flow, neurotransmitter activity, and oxygen/glucose metabolism by infusing different radioligands into the bloodstream, an issue that certain ethical committees still consider as slightly invasive. Often with a view to understanding the localization of cognitive processes in the brain, PET has provided unprecedented opportunity to psychologists and neuroscientists to bridge the gap between behaviors in humans and the neuronal events underlying mental functions. In contrast, SPECT has been more habitually used for clinical studies.
In 1990, Seiji Ogawa, a Japanese researcher, and collaborators from AT&T Bell Laboratories, discovered the key principle behind fMRI. Since then, fMRI has been regarded by many as a modality of supremacy in functional neuroimaging; in particular, due to the low invasiveness, lack of radiation exposure, and relatively wide availability. As a result of its excellent spatial resolution for localizing slow temporal fluctuations in the level of oxygenation of the blood, fMRI has been extensively used to study mechanisms underlying brain functions in healthy subjects as well as to monitor brain dysfunctions in a safe and quantitative way. Fortunately, fMRI came to light when feelings of disheartenment distressed the neuroscientists’ community, some due to the poor spatial resolution of EEG/MEG, others caused by the high production-cost and minimal availability of radioisotopes for PET/SPECT studies as well as the ethical limitations of their usage for research purposes.
These were reasons why, at that time, other groups were also working in parallel to develop a new modality based on the principles of the near infra-red spectroscopy, a technique successfully applied for chemical analysis since 1980s. A device to perform near infrared spectroscopy in humans was first invented in 1993 by Britton Chance of the University of Pennsylvania. However, it was not until the period from 1994 to 1997 that several authors finally provided evidences of its potentialities to assess brain activity through the intact skull in adult human subjects, baptizing this new modality as fNIRs. Applications of fNIRs include cognitive neuroscience, pharmacology and medical diagnostics (i.e., blood sugar and oximetry), in particular for infants.

Thanks to those original and advanced applications in functional neuroimaging, the physiological mechanisms associated with normal brain functions as well as their failures causing behavioral, neurological and psychiatric disorders are yet increasingly detailed. On the impending decades, it will aid to create more solid bases not only for a buttressing in the cognitive neurosciences but also for a revolution in the diagnosis and treatment of several of such brain disorders. Therefore, throughout the diversity in applications and customer preferences for each modality, we don’t hesitate to utter that functional neuroimaging represents the future garden where psychologists, psychiatrists, neurologists and cognitive neuroscientists will meet.

Future challenges
Despite the abovementioned advances in functional neuroimaging, there is an impressive lack of comprehension of basic phenomena triggering the signals recorded by each modality, from the molecular to the cellular substrates. Moreover, general principles of data-genesis are still largely to be discovered and modeled at different physical levels. For that end, a consensus about what large-scale neuronal activity mean is required among researchers/clinicians using unlike modalities; otherwise, it will not be clear how the data should be interpreted. An alarming issue is the imprecision with what a number of authors interpret results obtained from functional neuroimaging. The term “large-scale” is used to indicate the brain activity arising from the interactions among extensive neuronal populations with specific functions at a macroscopic level. In our opinion, authors should clearly state what has been observed with the modality in use and more conservative atmospheres are recommended when discussing their results. As pointed out by Marcus Raichle in the section about neuroimaging in “The 2006 Progress Report on Brain Research” (The Dana Alliance for Brain Initiatives): “Despite these developments, frequently stated or implied interpretations of functional imaging data suggest that, if we are not careful, functional brain imaging could be viewed as no more than a modern and extraordinarily expensive version of 19th-century phrenology, which associated certain mental faculties and character with parts of the skull.”

Members of the NMD group.
From right: Jorge Riera (Assoc. Prof.),
Enjieu Kadji Hervé. G (Post-doc),
Tina Rasmussen (Asst. Prof.),
Risa Haga (Secretary)

Understanding how phenomena originate from the mesoscopic level represents a Rosetta stone when interpreting functional neuroimaging data; and moreover, it may be of great utility to formulate cohesive forward/generative models of each modality. This constitutes the major objective of the recently created Neuronal Mass Dynamics (NMD) group at the Institute of Development, Aging and Cancer, Tohoku University. The NMD group belongs to the department of Functional Brain Imaging (Head: Prof. Ryuta Kawashima) and it has been created to provide basic foundations for functional neuroimaging as well as novel methods for data analysis to the cognitive neuroscience, language and social neuroscience groups at this department.

A member of the NMD group,
Takakuni Goto (Post-doc),
receives experimental training
in György Buzsáki’s Lab
at Center for Molecular
and Behavioral Neuroscience,
Rutgers University, Newark.

In biophysics, a mesoscopic level refers to the length scale at which one can reasonably discuss the properties of a living phenomenon without having to discuss the behavior of individual enzymes or cells. Even so, a huge amount of results obtained from comparative studies in laboratory animals using novel technologies (i.e., a- miniaturized probes chronically or acutely implantable in the brain; b- two-photon laser scanning microscopy, and c- nuclear magnetic resonance spectroscopy) provides a solid basis for future challenges. From April 2007, two of these contemporary techniques (boldface) will be available at the NMD group. Members in the NMD group will be able to combine these techniques with high-resolution manipulation systems and stereotaxic instruments to achieve in vivo small-scale imaging of the structural and functional organization of basic-networks of cells in the brain. Internal molecular mechanisms inside individual cells will be also investigated. We aim to create a more exhaustive portrait of the microscopic level, which is necessary to model the mesoscopic phenomena in the brain.

Takakuni Goto
works with a microscope
in György Buzsáki’s Lab.

The cerebral cortex constitutes the largest brain structure and the one with more implications in the genesis of functional neuroimaging data. Hence, the NMD group finds it suitable, as a beginning, to center the attention on the major physiological/anatomical issues at the microscopic level in this brain structure. In the figure below, both microscopic and mesoscopic levels in the cerebral cortex overlap with the scope of each functional neuroimaging modality in terms of spatial and temporal resolution.

The members of the NMD group will perform a variety of experiments to clarify the fundamental working principles of the most important niches in the cerebral cortex: a) the columns and minicolumns with their laminar diversity (Rockland and Ichinohe 2004), b) the tripartite functional units [e.g., the neuron-astrocyte-neuron (Volterra et al. 2002) and the neuron-astrocyte-vasculature (Cohen et al. 1996)] and c) the nerve-ending particles (i.e. synaptosomes) (Erecinska et al. 1996).

Full Size Image

 

Illustration of physical levels in the cerebral cortex and their respective spatio-temporal dimensions for either physiological/anatomical substrates or observations). The microscopic level has the lowest dimensions [space: from molecular/cellular levels to small networks of cells; time: from one millisecond to a few seconds]. The mesoscopic level comprises scales in space and time where interactions between population of cells occur [space: from minicolumns and layers to columns; time: from hundreds of milliseconds to several seconds]. The large-scale imaging modalities are schematically situated in such a 2D-plot considering their capabilities and limitations

Indeed, the NMD group has formulated three main aims (listed below with their respective sub-subjects) in order to evaluate how the micro-architecture of the cerebral cortex could impact directly on the interpretation and future evolution of the functional neuroimaging. We hope our studies will provide the foundations for upcoming modeling at the mesoscopic level. We are also interested in the reverse problem “how could the functional neuroimaging help us to elucidate such a detailed micro-architecture?” For that end, members in the NMD group will develop methods for statistical inference at the brain mesoscopic and microscopic levels from large-scale imaging data.

NMD group’s main aims

I- Principles of integration and propagation of membrane potentials in dendritic trees of different cortical neurons
- To establish general principles of electrotonic integration/propagation inside dendrites. To determine dendritic amplifiers and their role in the forward and backward propagation of action potentials.
- To study the dynamics of different types of synaptic transmission and their impact on the neuronal plasticity and neurogenesis. To characterize neurotransmitter diffusion in the synaptic clefts.
- To model signal codification and propagation within the axons of particular neurons.

II- Input/Output signal flow inside a microcolumn: the role of the laminar organization
- To establish a diagram of inputs/outputs at different cortical layers within a micro-column in order to represent canonical principles of micro-circuitry, which might have an influence on the future development of robotic engineering and neuroprosthesis.
- To establish forward/generative relationships between the neuronal population dynamics in the cerebral cortex and the EEG/MEG data, an important step to solve their respective inverse problems.
- To determine the membrane potentials in population of nonlinearly interconnected neurons from the local field potentials recorded in the neuropil at high spatial resolutions.

III- Mechanisms underlying functional hyperemia and metabolism
- To model different intrinsic pathways in the neurovascular coupling: the special neuronal networks and the neuron-astrocytes-vascular unit.
- To model oxygen extraction/consumption by the capillary network.
- To establish principles of interaction between metabolic pathways (e.g., glycolysis, TCA, oxidative phosphorylation) and several mechanisms for vascular control.
- To develop novel methods to solve the fMRI, fNIRS, PET/SPECT inverse problems.

References

Cohen Z, Bonvento G, Lacombe P, Hamel E (1996) Serotonin in the regulation of the brain microcirculation. Progress in Neurobiology 50: 335-362.

Erecinska M, Nelson D, Silver IA (1996) Metabolic and energetic properties of isolated nerve ending particles (synaptosomes). Biochimica et Biophysica Acta 1277: 13-34.

Rockland KS, Ichinohe N (2004) Some thoughts on cortical minicolumns. Exp Brain Res 158, 265-277.

Volterra A, Magistretti PJ, Haydon PG (2002). The tripartite synapse, glia in synaptic transmission. Oxford University Press, New York.