The central theme of this program is modeling and visualization of spatial and temporal patterns of functional activation in the living human brain as imaged by [15O]water PET and fMRI. This program consists of two informatics components -- Project 3 (Interactive Visualization for Neuroimaging) and Core D (Neuroinformatics) and two brain/behavioral components -- Project 1 (Temporal Resolution of fMRI) and Project 2 (Reproducible Features of Functional Neuroimages); Cores B and C contain well-defined research components -- 3D brain warping (Core B) and behavioral correlates of functional neuroimaging data (Core C).

Project 1, Temporal Resolution of fMRI (S-G. Kim, PI; K. Ugurbil, Co-PI), will (i) utilize fMRI techniques to determine the physiological limitations of temporal resolution-more precisely, to define the upper limit of temporal resolution in specific brain regions during repeated movements and to improve temporal resolution for spatially separated regions by utilizing hemodynamic response data from each region and (ii) determine the reproducibility of the hemodynamic response in different brain regions by comparing the response in different cortical regions in a single subject, in the same cortical region in a single subject during different imaging sessions and in the same cortical region in different subjects. Project 1 investigators will also attempt to determine the relationship between motor behavior and the hemodynamic response; they will (i) determine the hemodynamic response in specific brain regions during single-digit movements, (ii) simulate the fMRI response by convolution of the hemodynamic response with known behavior (the inter-movement time) during single-digit movements and compare the simulated response with actual fMRI data and (iii) deconvolve the fMRI data into multiple units of the hemodynamic response and compare the result with actual behavioral data. Finally, they will study the spatiotemporal response of cortical and subcortical motor regions and cerebellum during visually-guided delayed finger movements, localize those brain regions involved in motor preparation and determine the temporal order of their activation during the motor preparation process. [Specific Aims]

Project 2, Reproducible Features of Functional Neuroimages (S.C. Strother, PI; Lars K. Hansen, Co-PI) has as its principal goal the identification of reproducible features in [15O]water PET and fMRI datasets -- within and between scanning sessions (single subject), between subjects, between participating sites and across imaging modalities (PET and fMRI at 1.5 and 4T). Reproducible features will be identified using three data analysis strategies -- linear models (e.g., MANOVA, ANCOVA, Principal Component Analysis), nonlinear models (e.g., artificial neural networks and nonlinear Fourier analyses) and model-free techniques (e.g., adaptive k-nearest neighbors). This project will also assess (i) the impact of nonlinear 3D transformations ("warps") for intersubject registration and removal of geometric distortion from 4T echo-planar fMRI images and (ii) oriental-occidental anatomical and functional differences on the reproducible features. In addition, Project 2 investigators will develop and, in collaboration with Project 3 and the Informatics Core (Core D), make available on the WWW benchmark functional activation datasets, analysis results and associated modeling software. PET and fMRI datasets will be provided by Project 1 and Core B, and data analytic results will be provided to Project 3 and Core D. [Specific Aims]

Project 3, Interactive Visualization for Neuroimaging (D.A. Rottenberg, PI) will develop an interactive visualization environment for exploring multidimensional functional data fused with a 3D structural image volume. This environment will integrate a suite of existing multidimensional display tools (surface and volume rendering, slice galleries and projections) with a rich set of navigational tools, pointers to atlas information and access to the HBP Data Repository (see Core D). The environment will be extended to manage multiple functional datasets for the comparison of differences across time, subjects, modalities and model classes. This project will also develop and evaluate techniques for region identification and description. These techniques will include (i) global and local intensity thresholds, (ii) correlation scatterplots to identify interesting feature-space regions, (iii) critical thresholds based on changes in topological features and (iv) other rule-based methods. Region descriptors, e.g., intensity and geometric descriptors, topological features and neuroanatomical labels, will be used to compute similarity metrics on focus pairs and entire datasets. Project 3 investigators will evaluate the utility of data reduction methods employing symbolic/iconic representations presented within an anatomical context. Symbolic representations will be designed to support simultaneous visualization and manipulation of up to eight 3D datasets within the interactive visualization environment. Finally, Project 3 will develop a VRML (Virtual Reality Modeling Language) environment that supports the presentation of modeling and data-analytic results ("case studies") in an interactive form that is shareable locally and over the Internet. VRML translations will be initiated from the interactive visualization environment and will be designed to minimize the need for manual editing prior to export. [Specific Aims]

Core A (D.A. Rottenberg, PI) will provide fiscal and administrative support for the P20 Program; it will organize the Annual Minneapolis Workshops, monitor progress and ensure scientific productivity.

Core B (D.A. Rottenberg, PI; B.R. Rosen, Co-PI) will (1) at UM: acquire, archive and ensure quality control of structural MRI scans of normal volunteer subjects who have undergone protocol [15O]water PET scanning at MVAMC; (2) at MVAMC: archive and ensure quality control of local PET datasets acquired using Core C protocols and perform preliminary data preprocessing, including registration of MVAMC PET, UM MRI and UM fMRI datasets; (3) at MGH: acquire structural and functional MRI datasets using Core C protocols, archive and ensure quality control of these datasets and perform preliminary data preprocessing, including registration; (4) at ARI and RH: archive and ensure quality control of local structural MRI and PET datasets acquired using protocols designed by Core C investigators and perform preliminary data preprocessing, including registration of these datasets. In addition, Core B contains a well-defined research component: Core B investigators will implement two nonlinear 3D warping algorithms for intersubject registration of MRI and PET datasets and evaluate "intersubject subspace variance" and "model prediction error" (See Project 2) as metrics for optimal warp tuning and warp selection.

Core C (J.J. Sidtis, PI) will (1) select and design stimulation paradigms to be implemented across study sites; (2) collect physiological and behavioral data during functional neuroimaging studies; (3) review, archive, analyze and distribute these data to participating sites via the Core D Librarian; (4) monitor the execution of experimental protocols across participating sites to ensure consistency in the conduct of neurobehavioral tasks and (5) develop task modifications and new protocols for multicenter multimodality studies to test hypotheses generated by Projects 1, 2 and 3. The research component of Core C will analyze demographic and performance data in relation to functional neuroimaging results and assess the reliability and validity of task performance data.

Core D (L.C. Gatewood, PI) will design and develop repositories for documentation, maintenance, update, retrieval and distribution of program software (Software Repository) and maintain databases for protocols, image datasets and analysis and visualization results (Data Repository). Core D will also provide tools to facilitate communication between program sites, projects, cores and investigators; these tools will manage internal project communications such as project notes, analysis results, interim reports, correspondence and manuscripts-in-preparation as well as the distribution of program materials (data, software, protocols, results, etc.) to program sites, collaborators and the larger neuroscientific community. Additionally, Core D will integrate software developed by projects and cores into public and private distribution sets, ensure platform-independent functionality and upgradability and coordinate testing on specific types of UNIX workstations.

---

Project 1: Temporal Resolution of fMRI

Specific Aims

1. To determine the physiological limitations of temporal resolution; more precisely, (i) to define the upper limit of temporal resolution in specific brain regions during repeated movements and (ii) to improve temporal resolution for spatially separated regions by utilizing hemodynamic response data from each region.

2. To determine the reproducibility of the hemodynamic response in different brain regions; specifically, to compare the response (i) in different cortical regions in a single subject, (ii) in the same cortical region in a single subject during different imaging sessions and (iii) in the same cortical region in different subjects. We will test the hypothesis that the hemodynamic response is reproducible only during repeated measurements in the same cortical region in the same subject.

3. To determine the relationship between motor behavior and the hemodynamic response; more precisely, (i) to determine the hemodynamic response in specific brain regions during single-digit movements, (ii) to simulate the fMRI response by convolution of the hemodynamic response with known behavior (the inter-movement time) during single-digit movements and compare the simulated response with actual fMRI data and (iii) to deconvolve the fMRI data into multiple units of the hemodynamic response and compare the result with actual behavioral data. We will test the hypothesis that the fMRI response is linearly related to behavior.

4. To determine the spatiotemporal response of cortical and subcortical motor regions and cerebellum during visually-guided delayed finger movements. Specifically, we will localize those brain regions involved in motor preparation and determine the temporal order of their activation during the motor preparation process. We will test the hypothesis that primary motor cortex, premotor cortex and supplementary motor cortex are all activated during motor preparation.

Back to Project 1 Description

---

Project 2: Reproducible Features Of Functional Neuroimages

Specific Aims

1. To identify reproducible features in [15O]water PET and in 1.5T and 4T fMRI datasets using three data analytic strategies -- linear models, artificial neural networks and other modern statistical techniques such as adaptive k-nearest neighbors;

(a) within and between scanning sessions (single subject);

(b) between subjects;

(c) between participating sites (MVAMC, ARI, RH);

(d) between neuroimaging modalities [PET and fMRI at 1.5T and 4T].

2. To assess the impact of the following factors on reproducible image features:

(a) nonlinear 3D transformations ("warps") for intersubject registration and removal of geometric distortion from 4T echo-planar fMRI images;

(b)oriental-occidental anatomical and functional differences.

3. To use the results obtained from Specific Aims 1 and 2, in collaboration with Core C investigators, to design improved experimental protocols.

4. To identify and document the analysis of benchmark functional activation datasets.

Back to Project 2 Description

---

Project 3: Interactive Visualization For Neuroimaging

Specific Aims

1. To develop an interactive visualization environment for exploring multidimensional functional data fused with a 3D structural image volume. This environment will integrate a suite of existing multidimensional display tools (surface and volume rendering, galleries of slices, projections) with a rich set of navigational tools, pointers to atlas information and access to the HBP Data Repository (See Core D). The environment will be extended to manage multiple functional datasets for comparisons across time, subjects, modalities and model classes.

2. To develop and evaluate techniques for region identification and description. These techniques will include (i) global and local intensity thresholds, (ii) correlation scatterplots to identify interesting feature-space regions and (iii) critical thresholds based on changes in warp-invariant topological features (e.g., the Euler char acteristic). Region descriptors, e.g., intensity and geometric descriptors and topological features, will be used to compute similarity metrics on focus pairs and entire datasets.

3. To evaluate the utility of data reduction methods employing symbolic/iconic representations presented within an anatomical context. Symbolic representations will be designed to support simultaneous visualization and manipulation of up to eight 3D datasets.

4. To develop a VRML (Virtual Reality Modeling Language) environment that supports the presentation of modeling and data-analytic results ("case studies") in an interactive form that is shareable locally and over the Internet. VRML translations will require minimal manual editing prior to export.

Visualization Products
Corner Cube
Corner Cube on the Web
Visualization Testbed

Flat Mapping
Flat Maps of the Human Brain
More Spherical Maps of the Brain

Symbolic Visualization
Variable symbol detail
Symbolic merge

VRML Worlds
VRML Corner Cube Model - Download
International Neuroimaging Consortium VRML Homepage

Back to Project 3 Description

---