Over the past year, the Unit on Functional MRI Methods has been making progress along several research avenues ? all with the purpose of better understanding the information content contained in the time series fMRI signal, and to subsequently devise methods to better extract or make use of this novel information. Specifically, several of the research avenues have included: 1) characterization of the spatial extent of the hemodynamic response and spatial heterogeneity hemodynamic response magnitude and linearity using well understood visual stimuli, 2) investigation of the source (neuronal, hemodynamic, or both) of the observed nonlinearity of the BOLD (Blood Oxygenation Level Dependent) response to stimuli of varying duration, 3) determining the optimal paradigm timing for detection and estimation of the hemodynamic response in the context of jittered event related fMRI, 4) characterization of the hemodynamic latency distribution across different cortical regions, 5) creation of brain activation paradigms and processing techniques that allow robust extraction of hemodynamic latency and width modulation with task modulation 6) characterization of the spatial and temporal characteristics of baseline fluctuations in the fMRI time series, 7) from an understanding of these time series fluctuations characteristics, derivation a simple relationship that predicts optimal resolution for fMRI as a function of image signal to noise, 8) creation of methods to separate neuronally mediated fMRI time series fluctuations from non-neuron fluctuations, 9) creation of MRI pulse sequences which detect neuronal current-induced signal changes directly, potentially bypassing BOLD contrast completely. For research avenue number 1, the following manuscript have been written and is in second revision: Z. S. Saad, K. M. Ropella, E. A. DeYoe, P. A. Bandettini, The spatial extent of the BOLD response. NeuroImage (second revision). In this manuscript, the primary point that is made is that the spatial extent of BOLD response is significantly greater than what is typically measured in a standard fMRI experiment. After collection of 5 times more data than is typically collected, we found the area of activation to continue to grow, with many voxels still moving towards significance. This not only has implications on methods which compare sizes of activation across populations but also on interpretation of what the BOLD signal is truly indicative of (neuronal firing, field potentials, or subthreshold potential changes). Also for avenue number 1, the following manuscript, R. M. Birn, Z. Saad, P. A. Bandettini, Spatial heterogeneity of the nonlinear dynamics in the fMRI BOLD response. NeuroImage, 14: 817-826, (2001), was published, demonstrating that the BOLD response dynamics (magnitude changes with stimuli duration ? from 100 ms to 3 sec) are not only nonlinear (a greater response observed with short stimuli duration than is predicted by a linear system) but the nonlinearity varies across cortical region in a manner that is not correlated with underlying hemodynamics. This is an argument that the nonlinearity is neuronal rather than hemodynamic in origin. For research avenue number 2, the following manuscript was published: P. A. Bandettini and L. G. Ungerleider, From neuron to BOLD: new connections. Nature Neuroscience, 4: 864-866, (2001). Assuming a linear relationship between neuronal activity and hemodynamic changes, we deconvolved the neuronal input from the hemodynamic response and obtained a transiently high level of neuronal input that closely resembled that published recently by Logothetis et al. For research avenue number 3, the following manuscript was published: R. M. Birn, R. W. Cox, P. A. Bandettini, Detection versus estimation in Event-Related fMRI: choosing the optimal stimulus timing. NeuroImage 15: 262-264, (2002). This paper describes the optimal stimulus timing when performing event-related fMRI. The original contribution is that the optimal timing depends on whether one is interested in optimizing the paradigm for hemodynamic estimation or for hemodynamic change detection. For research avenue number 4, ongoing work is being carried out with post doc, Dr. Patrick Bellgowan and special volunteer, Dr. Marta Maeron, to establish the degree of and source of the variation in latency and width across brain regions and across individual voxels within brain regions. Methods for calibration are also being developed towards the goal to push the temporal resolution of fMRI to the order of 100 ms rather than that of seconds. For research avenue number 5, Patrick Bellgowan is finishing preparation for a manuscript to be submitted to PNAS. It is: P.S.F. Bellgowan, Z. S. Saad, P. A. Bandettini, Voxel-wise estimation of hemodynamic onset delay and width modulation due to word rotation and lexical decision-making. In this article specific regions involved with word rotation and lexical decision making are differentiated by the correlation of the latency of the decision making for each task type with the latency and width modulation of the hemodynamic response. Areas that show increases in width with reaction time are bottlenecks in processing whereas areas that show increases latency with reaction time are areas of downstream processing. The accuracy of this technique is on the order of 100 ms. It represents a new method by which fMRI paradigms can be designed and analyzed to extract previously unobtainable information. For research area number 6, graduate student Natalia Petridou has characterized the spatial distribution and temporal characteristics of physiological fluctuations. A novel method to obtain the relative contribution of physiological fluctuation power on a voxel wise basis was developed as well. Based on these studies, research area 7 was developed, demonstrating that as image signal to noise is increased, gains in temporal signal to noise at a constant resolution are limited. The only advantage of increasing image signal to noise (i.e. local rf coils and/or higher field strength) is to increase resolution such that thermal and physiological noise contribute approximately equally. She has also been working on ways to separate, based on echo time dependence (susceptibility related changes show echo time dependence whereas other changes due to respiration, movement, or heart beat do not show an echo time dependence), the fluctuations due to spontaneous neuronal activity from fluctuations due to other physiologic processes ? research area 8. I believe that this technique has significant potential in regard to accurate mapping of epileptic foci using fMRI. Three manuscripts are currently being written on these topics. For research area number 9, one manuscript has been published: J. Bodurka, P. A. Bandettini. Toward direct mapping of neuronal activity: MRI detection of ultra weak transient magnetic field changes, Magn. Reson. Med 47: 1052-1058, (2002). This manuscript is a seminal piece of work describing and demonstrating the feasibility of using MRI to detect neuronal currents, at their source. The effects of these currents are typically detected on the scalp using MEG and EEG. If these techniques can be translated to human use, it would be a breakthrough in functional imaging. Work is ongoing to develop this technique further for human use.
Showing the most recent 10 out of 36 publications
