RICERCA ITALIANA     

TEBAM: study, development and physiological-clinical validation of a multi-modal methodology for the 3D "True Electrical Brain Activity Mapping" in normal and pathological subjects

Università degli Studi di Trieste

Abstract

The aim of this Research Project is to study and develop an advanced tool, TEBAM (True Electrical Brain Activity Mapping), based on the multimodal integration of Magnetic Resonance (MR) neuroimaging methods with neurophysiological data, in particular with electroencephalographic data, to allow true 3D electrical brain activity mapping in a very refined spatio-temporal scale. This tool will be able to be used be used either in normal conditions (on healthy subjects) or in critical conditions as in the presence of morphologic brain pathologies (expansive morphological lesions altering functionally and anatomically the encephalon).
This research will be carried out by three Research Units, characterized by complementary scientific backgrounds in bioengineering and radiological fields, the fusion of which is essential for the achievement of the research objectives. The three research units will collaborate for the adjustment of three-dimensional anatomo-functional models based on new RM multimodal methods, in particular DT-MRI and RM methods alternative to those currently in use for Functional Imaging, to be integrated with neurophysiological three-dimensional data to obtain the above cited tool. These models will be able to include the individual tissue conductivity characteristics with their electric anisotropy and with the unpredictable additional variations caused by the presence of a morphologic brain pathology. This project has in fact among its objectives the development of an original method which will make it possible to obtain individual tissue conductivity data evaluated starting from Magnetic Resonance Imaging measurements (DT-MRI), basing on the quantitative deduction of tissues' electrical conductivity tensor starting from water self-diffusion tensor. The innovation with respect to the current state of the art consists, on one side, in the introduction of the pathology in the systems for the analysis and reconstruction of the sources of electrical brain activity and, on the other side, in the introduction of the above mentioned original method allowing the use of individual conductivity data evaluated from imaging measurements instead of mean values available from the literature, as currently performed in the scientific community and implemented in commercially available products for the analysis of the sources of brain activity.
The tool the realization of which is the objective of this Project should integrate the so obtained models with electroencephalographic data to achieve 3D brain activity source mapping using an approach able also to incorporate anatomical constraints on the location of activation sites. By means of the collaboration of all the Research Units, the most favorable conditions for a "constructive" integration of the potentials of the single Magnetic Resonance imaging methods will be determined by adapting the contribution of each method to the diverse experimental conditions and specific features of the regional anatomy of the brain. By this way a precise, powerful and flexible tool will be obtained for true 3D brain activity source mapping, based upon a cross-platform structure that will also be able to run on the high performance systems of the inter-university consortium CINECA. The use of multiprocessor supercomputers is requested in case of heavy computational load, as happens for highly complex anatomo-physiological situations with requests of high spatio-temporal resolution. By this way this tool will also be able to be used for providing innovative services in health environment for true 3D brain activity mapping by means of the inter-university consortium CINECA.
The final goal is to conclude the research with the writing and the deposit of an Italian national and international patent under the care of the Bioengineering Research Unit. <<<

Principal Investigator

Paolo INCHINGOLO Università degli Studi di TRIESTE

Research Objectives

The objective of this Research Project is to develop an advanced tool, TEBAM (True Electrical Brain Activity Mapping), for an investigation approach with multimodal non invasive techniques, integrating different kinds of neurophysiological and neuroimaging data, allowing true 3D electrical brain activity mapping in a refined temporal scale.
This tool will be able to be used be used either in normal conditions (on healthy subjects) or in critical conditions as in the presence of morphologic brain pathologies (expansive morphological lesions altering functionally and anatomically the encephalon). This tool will be based on a three-dimensional anatomo-functional model of the head based on the multimodal integration of Magnetic Resonance methods. Such model will be able to include the individual tissue conductivity characteristics with their electric anisotropy and with the unpredictable additional variations caused by the presence of a morphologic brain pathology. This requirement is based on the fact that the knowledge of the electrical conductivity properties of the tissues is a key point to find the relationship between electromagnetic fields generated by a tissue, e.g., the measure of electroencephalographic potentials, and the neural sources the activation of which generates the measured electroencephalographic signals, as source mapping accuracy heavily depends on the accuracy of the conductivity values assigned to the tissues in the volume conductor head model. The attempt of characterizing the active brain zones starting form the associated electroencephalographic potential measurements could find great improvements from the availability of measuring tissue electrical conductivity non-invasively, instead of basing upon mean conductivity values taken from the literature, as it is done at present in the scientific community and it is implemented in commercially available products for the analysis of the sources of brain activity. For this reason, among the objectives of this Project, the development of an original method should be mentioned which will make it possible to obtain individual tissue conductivity data evaluated starting from Magnetic Resonance Imaging measurements (DT-MRI), by means of the quantitative deduction of tissues' electrical conductivity tensor starting from water self-diffusion tensor. It should also be noticed that in the specific pathological context, there are alterations in the electrical properties of non pathological tissues due to the presence of a lesion in the encephalon. Furthermore, an expansive lesion, occupying space, deforms and compresses the soft brain structures, altering also the conductivity values of the surrounding normal tissues and causing conductivity inhomogeneities within a certain brain structure. The innovation with respect to the current state of the art consists, on one side, in the introduction of the pathology in the systems for the analysis and reconstruction of the sources of electrical brain activity and, on the other side, in the introduction of the above mentioned original method which allows using individual conductivity data evaluated from imaging measurements instead of mean values available from the literature. The tool the realization of which is the objective of this Project should integrate the model so obtained with three-dimensional neurophysiological data and in particular with electroencephalographic data to achieve 3D brain activity source mapping using an approach able also to incorporate anatomical constraints on the location of activation sites. Finally, by means of the collaboration of all the Research Units, the most favorable conditions for a "constructive" integration of the potentials of the single Magnetic Resonance imaging methods will be determined by adapting the contribution of each method to the diverse experimental conditions and specific features of the regional anatomy of the brain. By this way a precise, powerful and flexible tool will be obtained for true 3D brain activity source mapping. This tool will solve bioelectrical problems using finite differences algorithms (FDM), allowing a flexibility in model characteristics definition higher than the one that could be obtained with any other approach. Even in front of a high computational load, there are highly qualifying and innovative points: the easiness of implementation of anisotropic structures' modeling and of gradients of conductivity variation within such structures, the easy increment or reduction of model complexity (number of compartments) and of spatial resolution of the volume conductor head model used by the numerical solvers. Such approach will also allow making bioimages' segmentation procedures virtually not necessary for the construction of the volume conductor head model to be used in brain electrical activity mapping procedures, with great advantages in particular when using highly complex head models. The Bioengineering Unit of Trieste will also integrate in this tool a suite of instruments for integrated data and structures visualization (many of these have been already realized), suitable to give an adequate information "feed-back" in the analysis of the results of electrical brain activity mapping. Also a specific software for the visualization of the information relative to the diffusion tensor and tissue conductivity will be achieved. Such instruments will be based on powerful and flexible freeware and open-source tools for graphics pipelines development, like VTK (Visualization Toolkit) and OpenGL.
The software tool that will be created for 3D electrical brain activity mapping will be based upon a cross-platform structure and will be able to run on mono or multi-processor PC with Windows or Unix O.S., and on Unix-like high performance systems of the inter-university consortium CINECA. The use of multiprocessor supercomputers is requested in case of heavy computational load, as happens for highly complex anatomo-physiological situations with requests of high spatio-temporal resolution. By this way this tool will also be able to be used for providing innovative services in health environment for true 3D brain activity mapping by means of the inter-university consortium CINECA.
The benefits of this study are expected either for basic knowledge or in the field of clinical applications, in particular for neurosurgical applications or for the possibility of piloting prosthetic limbs by amputated or spinal cord-injured subjects.
The final goal is to conclude the research with the writing and the deposit of an Italian national and international patent under the care of the Bioengineering Research Unit. <<<

First Results

On the basis of the results obtained during the first phase of the Project by the U-RADIOL-PI Unit about the new experimental solutions for the technologies of diffusion tensor imaging, the U-RADIO-FISIO-TS Unit will develop in collaboration with the U-BIOING-TS Unit an original method which will make it possible to obtain individual tissue conductivity data starting from DT-MRI imaging measurements. This innovative technique will allow the quantitative deduction of tissues' electrical conductivity tensor starting from water self-diffusion tensor, obtained from DT Magnetic Resonance Imaging measurements. The U-BIOING-TS Unit, after having defined by means of a simulation analysis the necessary characteristics ad the possible validity limits, will build basing on this result a numerical 3D phantom of the head (named "volume conductor head model") mimicking the real shape of the head and of its structures with the actual conductivity values of its tissues and eventually their anisotropy, even in the presence of morphologic pathologies in the brain. The fusion of the results achieved by the three Research Units during the first phase of this Project will allow to obtain an extremely precise and refined volume conductor head model for the 3D mapping of the sources of electrical brain activity performed starting from electroencephalographic measurements.At the end of the second phase of the Project, the realization of an advanced tool for true 3D electrical brain activity mapping is expected, named "TEBAM"- True Electrical Brain Activity Mapping, which may be used either in normal conditions or in critical condition as in the presence of pathologies (expansive morphological lesions altering functionally and anatomically the encephalon). This tool is based on a three-dimensional anatomo-functional model of the head based on the multimodal integration of Magnetic Resonance methods. Such volume conductor head model will be able to include the individual tissue conductivity characteristics with their electric anisotropy and with the additional variations caused by the presence of the morphologic pathology. The innovation with respect to the current state of the art consists, on one side, in the introduction of the pathology in the systems for the analysis and reconstruction of the sources of electrical brain activity and, on the other side, in the introduction of an original method which allows using individual conductivity data evaluated from imaging measurements instead of mean values available from the literature.
The integration with three-dimensional neurophysiological data and in particular with electroencephalographic data allows brain activity source mapping using an approach able to incorporate anatomical constraints on the location of activation sites. Furthermore, the U-BIOING-TS Unit will integrate in this tool a suite of instruments for integrated data and structures visualization, suitable to give an adequate information "feed-back" in the analysis of the results of electrical brain activity mapping. Such instruments will be based on powerful and flexible freeware and open-source tools for graphics pipelines development, like VTK (Visualization Toolkit) and OpenGL, and will be integrated in the bioelectrical phenomena simulation system developed by the research unit. Also a specific software for the visualization of the information relative to the diffusion tensor and tissue conductivity will be achieved.
The software tool that will be created for brain activity mapping will be based upon a cross-platform structure and will be able to run on mono or multi-processor PC with Windows or Unix O.S., and on Unix-like high performance systems of the inter-university consortium CINECA. The use of supercomputers is requested in case of heavy computational load, as happens for highly complex anatomo-physiological situations with requests of high spatio-temporal resolution. this way this tool will also be able to be used for providing innovative services in health environment for true brain activity mapping by means of the inter-university consortium CINECA and the MIUR.
The final goal is to conclude the research with the writing and the deposit of an Italian national and international patent under the care of the U-BIOING-TS Unit. <<<

Timescale

24 months

National and international background

Over the last decade tremendous progress has been made in the development of techniques and methods for producing macroscopic images of human brain activity. Optimal technology for the study of cerebral functionality in humans should be characterized by a very good spatial and temporal resolution, to be able to detect and localize even quick variations in cerebral activity (1). Nevertheless the State of the Art of technology imposes the following limits: a) high temporal resolution is a property typical only of some functional exams, like electroencephalography (EEG) and magnetoencephalography (MEG); b) high spatial resolution on the other side is a property of structural exams: it is very good for typical neuroimaging techniques like Computed Tomography (CT) and Magnetic Resonance Imaging (MRI); c) functional neuroimaging techniques exploring haematic perfusion and cerebral metabolism, like Positron Emission Tomography (PET), Single Photon Emission Computed Tomography (SPECT) and functional Magnetic Resonance (fMRI) only guarantee a low spatial resolution and a limited temporal resolution (1). Only by combining different structural (anatomical) and functional methods for brain exploration it is possible to obtain at the same time a high spatial and temporal resolution.
It must be also noted that functional neuroimaging techniques allow indirect visualization of the activity of groups of neurons through the correlated variations in haematic flux, metabolism and oxygenation in the brain; these techniques allow the localization of active zones in the brain, so they are functional due to a correlation between activation and function. Although their spatial resolution is better in comparison with EEG (at least in standard modalities), these methods are not adequate to follow the activity evolution due to a heavy temporal under-sampling.
The study of brain functionality based on functional exams is not constrained by the large time constant associated to haematic flux variations, being these exams instead linked to the measurement of electro and/or magnetic field alterations associated to the generation of electrical signals from groups of neurons, directly related to the manifestation of the functionality of such groups (1). These exams then give important information about brain functioning, as in the case of neural activities sequences, allowing then to study the origin of some neurological disorders like epilepsy and brain tumors, or locations of focal activities (2-6).
Interesting complementarities do then exist between electrophysiological techniques used in functional exams and neuroimaging techniques: the former allow a direct measurement of neural activity and are characterized by an optimal temporal resolution, while the latter are characterized by a very good spatial resolution.
The integration of structural (neuroimages) and functional (electrophysiological) exams, especially if the latter are obtained with a large number of electrodes, allows to localize and reconstruct with high spatial precision the single cerebral functional processes with an unchallenged temporal resolution.
In this research project an approach based on anatomical-functional integration was chosen, aiming to co-elaborate anatomical information (bioimages) and functional data (multi-channel EEG measurements) registered on the scalp to reconstruct and visualize with the best spatial and temporal resolution the "in vivo" brain activity within the specific patient's anatomical structure, and to deepen the knowledge of the mechanics underlying cerebral functions both in physiological and in pathological conditions.
This procedure of reconstruction of sources of brain activity is named "solution of the inverse bioelectric problem", so called in opposition to the "solution of the forward bioelectric problem" related to the simulation of EEG signals generated by neural sources. Source location and characteristics are of great interest both for basic research and for clinical applications (2-8). In this latter case accurate information about source localization may help in pre-surgical planning for the removal of brain lesions, avoiding to damage the functionality of important brain areas (those containing the sources) near the lesion (9).
The procedure of reconstruction of the neural sources generating the EEG potentials recorded on the scalp requires postulates for a source model and a volume conductor head model, allowing to approximate in the best way the true electrical and geometrical properties of the head under examination.
Source reconstruction accuracy can be quantified by parameters as localization error and intensity estimate error; source reconstruction accuracy is affected by a number of factors: currently many studies have been conducted upon effects provoked by various causes, like head modeling errors, source modeling errors, errors in measurement electrodes positioning or errors due to noise in the recorded EEG (10-26).
Errors in head volume conductor modeling depend from differences between the real head and its model. The model is composed by compartments corresponding to the various anatomical structures, differing for the electrical conductivity value of the tissue they represent. Approximation of the model with respect to the real head depends from the number of compartments considered in the model, and from morphology and conductivity values assigned to such compartments. These factors then influence source reconstruction accuracy. Studies conducted so far in the scientific community concerned the problem of geometric shape and of conductivity of realistic head models, limiting however the study to head models in normal conditions, for which volumetric and electrical characterization of the head is highly predictable (27-31). In a pathological context it must be noted that brain lesions have an electrical conductivity widely different from the conductivity of normal, non pathological, brain tissues and so they are a possible compartment in the head model (32-33). Research studies conducted so far by the unit of Trieste allowed to point out the characteristics of a volume conductor head model with reference to electrical and geometrical characterization of possible morphological brain lesions. Results obtained considering both "ideal" noiseless data (multichannel EEG signals) and noisy data clearly indicate that, in the presence of pathological malformations, it is necessary to use models of high complexity, opposite to the practice, in use in the scientific community and implemented in commercial products, of reducing the volume conductor head model to a scheme with only three compartments (scalp-skull-brain) (34-43).
Given the evident higher difficulty caused by a pathological situation, only few research groups dealt with specific problems linked to functional mapping in the presence of pathologies, concentrating instead on more general problems like algorithms' efficiency.
In particular, it must be underlined the almost complete absence of literature about modeling of anisotropic structures, as anatomical structures like muscles (fibers), brain white matter (44) and bone (45-46), and about modeling of inhomogeneous structures, as in the case of infiltrating lesions or of pressure exercised by calcified tumors upon nearby non pathological soft tissues. The attempt of characterizing active brain areas starting from measurements of the associated electroencephalographic potential measurements would find great advantage from the ability of non-invasively measuring the electrical conductivity properties of the tissues instead of basing on mean conductivity values taken from the literature, as it is usually performed by the scientific community and it is implemented in commercial products for the analysis of the sources of electrical brain activity.
As far as this subject is concerned, it should be noticed that the survey of the methods for the study of the brain using Magnetic Resonance (MR) has taken advantage in the last decade of powerful analysis tools for the cerebral structure and function. Among the methods so far optimised for the structural analysis, it appears particularly notable the study of the diffusion of water molecules through the internal cerebral tissues. The diffusion of water in the brain is affected by a number of factors dealing with both structure and function and its study may reveal analytical information about the fine structure of the cerebral tissue and the underlying neuronal dynamics.
The innovative structural information achievable by means of the study of the diffusion, concerns the capability of evaluating analytically the directionality of the diffusion with the "diffusion tensor" (47). Since the water diffusion tends to be uniformly directed or isotropic in the cerebral interstitial tissues, while following the principal axis of neuronal prolongations, the main factor determining the directionality of the diffusion is the presence of axons within the bundles of neuronal fibers. It is, thus, easy to accept that the detailed study of the diffusion tensor by means of "fiber tracking" algorithm allows the reconstruction of the fibers' pathway within the white matter in the brain (44). Similarly to the other MR techniques, the study of the diffusion tensor and the fiber tracking may also be carried on the entire volume of the brain, gathering three-dimensional informations that can be co-registered with volumetric data of the cerebral anatomy using high resolution MR acquisition. In this research project we aim to develop, among the other topics, an original method which would allow us to obtain individual conductivity data evaluated starting from imaging Magnetic Resonance measurements by quantitatively deducing tissues' electric conductivity tensor starting from water self-diffusion tensor. This way it is easy to produce three-dimensional cerebral models that simultaneously contain analytical information about the cerebral cortical surface and the bundles of associative and projective fibers in the white matter. This opens for the functional MR a new "connectivistic" perspective, where the "diffusion tensor" is capable of joining the functional study to the study of the connections among each single functionally active region. For what concerns the functional information, the study of the diffusion exploits the hydro-electrical homeostasis of cerebral tissues and is commonly adopted in the neuro-radiological diagnosis, particularly of the ischemic vascular deseases in their acute phase (48). Nevertheless, recent findings (49) suggest that the sensitivity of the method can be sufficient to detect slight variations of the neuronal and interstitial hydro-electrical micro-environment that occur after neuronal functional activation. The interest evoked by the functional application of the diffusion is justified by the possibility of make visible to the MR new classes of functional phenomena, enriching significantly the spatial and temporal resolution of the method, and improving the accuracy because of the strictly mono-parametric nature of the diffusion phenomena.
The study of the functional changes in the diffusion is, this, configured as a possible candidate to expand the most widespread strategy for the acquisition of functional information, based on the BOLD (Blood Oxygenation Level Dependent) (50) contrast, in which the signal changes occurring in T2* weighted images are related to the local changes in the deoxy-hemoglobin content.
The BOLD signal depends simultaneously by several regional paramaters, like the blood flow, the blood volume and the oxygen consumption. For these reasons the BOLD contrast, although tremendously useful for the identification of the cerebral regions that are sites of functional activity, has important intrinsic limitations for the comprehension and the complete characterization of the measured signal. On the contrary, it is possible to obtain "mono-parametric" information, and thus more analytically predictable, by adopting advanced MR techiniques that enable the selective study of functional variations of flow and regional brain blood volume and are named "perfusion MRI". These methods also offer clear advantages in terms of accuracy of the localization, since they do not suffer of the errors induced by the venous artifacts. The use of functional methods based on perfusion and diffusion as well as on the evaluation of metabolic functions (rather than on BOLD phenomena) requires to reformulate the strategies for data processing and statistical analysis of the functional temporal series.
Functional information based on diffusion, as all MR functional methods, can produce 3D models of brain activity that are easy to integrate with anatomical RM models and with neuro-physiological data as electroencephalography (EEG).
The integration with EEG is strongly recommended since it improves our capability to solve the "inverse problem" by spatial constraints to the localization of the sources of neuronal activity.
The association with neuro-physiology in manifold and is based on the creation of multimodal models of the brain: it is possible to create anatomical 3D models of the cortical surface, than to integrate them with "tractography" (able to depict white matter connections and fibres) and with functional information contained in 3D models of brain activity. <<<