The Section on Functional Imaging Methods (SFIM) continues to develop cutting edge methods at the interface of functional MRI (fMRI) technology, image processing, interpretation of the signal, and neuroscience applications. The goal of SFIM is to fill the much-needed gap of improving fMRI from both a methodological and usability standpoint.
We aim to increase the depth and breath of fMRI applications and effectively bridge the gap to clinical applications on individual patients. This report describes progress in our research. Flip Angle Selection effects on physiologic noise and BOLD sensitivity. Carried out by Javier Gonzalez-Castillo and Jerzy Bodurka. We investigate, both how the temporal signal to noise (TSNR) dependence on flip angle. Our results suggest that when physiological noise dominates, there is a wide range of angles below the Ernst angle for which TSNR suffers a negligible decrease from its maximum value. Imaging at an angle below the Ernst angle is recommended as it comes accompanied by important practical benefits such as: higher specificity of fMRI;better tissue contrast;lower susceptibility to through-plane motion;and reduced levels of heat deposition in tissue. Averaging of task-based data reveals significant activations across the whole brain. Carried out by Javier Gonzalez-Castillo collaborating with Ziad Saad and Daniel Handwerker. In this work we show that the fMRI time series has much more non-canonical information than previously thought. Using a fMRI dataset composed of 100 functional runs in each of three subjects we have shown that significant task time-locked responses of very diverse shapes (e.g., positive sustained responses, negative sustained responses, stimulus onset/offset responses) can be recorded in the majority of the brain when TSNR is sufficiently high. This work suggests that the sparseness of activations in fMRI maps result from insufficient TSNR and/or overly strict model reference functions. Moreover, we have also demonstrated that inter-regional differences in response shape can be exploited to obtain functional parcellations of the whole brain in action (e.g., while subjects are involved in a task) in a manner similar to results from resting state analysis. Resulting parcellations from task-based data were symmetrical across hemispheres, reproducible across subjects, and anatomically and functionally meaningful. This is potentially breakthrough work in that it suggests that the entire brain is activated in some manner by every task. Effect of tissue contrast on functional MRI Image Registration Carried out by Javier Gonzalez Castillo in collaboration with Kristen Duthie and Wen Ming Luh. Typical fMRI scans have parameters that create a flat image contrast in the time series images. This along with inhomogeneous signal from multi-channel coils can degrade the output of registration algorithms;and consequently degrade group analysis and any attempt to accurately localize brain function. Non-invasive ways to improve tissue contrast in fMRI images include the use of low flip angles (FAs) well below the Ernst angle and longer repetition times (TR). In this work, we use a combination of real data and simulations to show that the use of low FAs (e.g., θ≤40) significantly improves accuracy and consistency of registration for data acquired at relatively short TRs (TR≤2s). Moreover, we also show that the application of bias correction techniques significantly improves alignment both for array-coil data (known to contain high intensity inhomogeneity) as well as birdcage-coil data. What makes different resting state networks different. Carried out by Javier Gonzalez-Castillo in collaboration with Meghan Robinson. Both data-driven (e.g., ICA) and hypothesis driven (e.g., seed-based analysis) analyses produce networks that are reproducible;still little is known about the underlying temporal and spectral characteristics that makes these networks unique and so reliable. So far, we have found no repeatable temporally distinguishing characteristics of these networks other than that the regions fluctuate together. Multi-modal characterization of the effects of physical activity and motor learning on brain structure and function Carried out by Adam Thomas while collaborating as a student with Heidi Johansen-Berg at the University of Oxford. Little is known regarding the processes by which the brain remodels the different components of its architecture with learning. Recent advances offer the opportunity to measure many different aspects of brain structure and function.
The aim of this project is to characterize the different types of brain plasticity that occur under different experimental paradigms, such as aerobic exercise and motor learning, using a wide range of human neuroimaging techniques including quantitative mapping of magnetic susceptibility, myelination, and cerebral blood volume so that exactly what is changing is better understood. Applying machine learning methods to derive new measures of functional connectivity Carried out primarily by Carlton Chu, the project tries to develop a novel approach to estimate the consistency of global functional connectivity. The approach described in this paper is a combination of brain decoding and analysis of functional connectivity. We apply kernel regression methods to predict the fMRI BOLD signal of one voxel using the activation pattern of other brain regions. The new metric is based on predictive validity, which measures the robustness of the connectivity pattern. The paper, which titled Measuring the Consistency of Global Functional Connectivity using Kernel Regression Methods, was published in IEEE International Workshop on Pattern Recognition in NeuroImaging on December 2010. High-temporal resolution decoding of ocular dominant responses from the BOLD signal. Carried out by Masaya Misaki. The multi-voxel pattern analysis was utilized to decode a subtle timing difference of neural activations in the ocular dominance columns. By extracting response amplitude at variable peak time for each voxel and for each stimulus condition, the multivariate classification analysis could discriminate 100 ms onset time differences of neural activations in the ocular dominance columns. This result indicates that the BOLD signal has high-temporal and high-spatial resolution information in its multi-voxel response pattern. Cleaning up fMRI time series by multi-echo acquisition Carried out by Prantik Kundu and Souheil Inati. This approach which uses multi-echo acquisition to separate BOLD from non-BOLD signal has proven to be a highly effective method for cleanly removing, in an automated manner, artifactual ICA components and for denoising resting state fMRI time series. The functional role of the theta rhythm in hippocampal dependent memory Carried out by Raphael Kaplan with co-mentor Neil Burgess. The hippocampus is crucial for episodic or declarative memory and the theta rhythm has been implicated in mnemonic processing, but the functional contribution of theta to memory remains the subject of intense speculation. Recent evidence suggests that the human hippocampus might function as a network hub for volitional learning. Notably, hippocampal theta in behaving rodents reflects volitional movement, which has been linked to spatial exploration and memory encoding. To test the relationship between volitonal learning and volitional movement driven theta oscilations, we used a virtual spatial navigation task in humans using both MEG and fMRI respectively. In our tasks we found hippocampal volitional learning network activity and theta oscillations corresponding to both movement initiation and subsequent memory peformance.
Showing the most recent 10 out of 84 publications
