Functional MRI Method Development

National Institute Of Mental Health

Investigators

Linked publications, trials & patents

Abstract

Protocol number 93-M-0170, NCT00001360 Ultra high-resolution fMRI: With increased availability of ultra-high field (7T) human MRI scanners, fMRI spatial resolution has been pushed to the sub-millimeter domain, making it possible to resolve functional activity and connectivity across cortical depths/layers. However, despite promising potentials, the widespread application of layer fMRI in humans is tempered by limited functional resolution caused by sensitivity to larger vessels and misregistration of functional to anatomic scans caused by different pulse sequences for each. To address large vessel bias in BOLD, Dr. Yuhui Chai, a post doc in our group, developed an approach called VAPER which has shown more sensitivity to capillaries. To address the second limitation, Dr. Chai, developed a magnetization transfer weighted anatomical EPI imaging technique to allow layer analysis in native fMRI space (MT-3D-EPI), as the most accurate way to define cortical depth, obviating the need for distortion correction and registration. Application of layer fMRI: Multisensory interplay can occur in areas that are considered unisensory, such as planum temporale (PT). Dr. Chai found a division of function between visual and auditory processing in PT and distinct feedback mechanisms in different subareas. Specifically, anterior PT was activated more by auditory inputs and received feedback modulation in superficial layers, and this feedback is likely arising from top-down influences from higher-order multimodal areas. In contrast, posterior PT was preferentially activated by visual inputs and received visual feedback in both superficial and deep layers, which is likely projected directly from the early visual cortex. To our knowledge, this is the first time that sensory-specific influence in human PT was able to be delineated over the dimensions of columnar distance and laminar depth. Line scanning: In addition to imaging layer activity at 7T, we have been also working on what is known as line scanning, which adapts previous work in rodents to humans, allowing us to simultaneously achieve high spatial (200 m) and temporal resolution (100 ms) by limiting acquisition to a very small patch of cortex. The aim of this development project is to better bridge the spatiotemporal gap between human and animal cortical layer research, thus allowing for the investigation of very dynamic layer signals in cognitive tasks typically confined to human experiments. Further, the increased spatiotemporal resolution of the method lends itself to vascular modeling that can inform other depth/layer-dependent acquisition methods. Assessing and interpreting time-varying changes in fMRI signal: Recent years have seen an explosion of studies looking at time-varying aspects of functional connectivity during resting-state. These studies suggest that those temporal variations (in the orders of tens of seconds) in functional connectivity might be relevant to our understanding of healthy and diseased brain function. Yet, other studies have demonstrated that shifts in vigilance and wakefulness levels that occur as resting-state progresses can also similarly affect functional connectivity. Unfortunately, wakefulness is rarely monitored due to the need for additional concurrent recordings (e.g., eye tracking, EEG); and novel methods able to provide insights about fluctuations in wakefulness using fMRI-only data are in great demand. Recently it was reported that strong fluctuations around 0.05Hz, hypothesized to be CSF inflow, appear in the fourth ventricle (FV) when subjects fall asleep. The analysis of this slow signal could provide an easy way to evaluate wakefulness in fMRI-only data. In this project, we used the 7T resting-state sample from the Human Connectome Project, which contains eye tracking recordings that can be used to detect long periods of eye closures. Our results confirm the presence of those fluctuations in drowsy subjects, despite this data having relatively small inflow weighting. We also show that fluctuations of a similar frequency appear in large portions of grey matter with different temporal delays, and that they can substantially influence estimates of functional connectivity. Finally, we show that the temporal evolution of this signal cannot only help us reproduce previously reported overall sleep patterns in resting-state data, but also predict individual periods of eye closure with 70% accuracy, matching predictions attainable using the amplitude of the global signal (a common fMRI marker of arousal). In this manner our results demonstrate the presence of this signal in a large, publicly available fMRI sample, and its value as a marker of arousal in absence of a better metric. Dimensionality reduction of time series data: Over this year, Javier Gonzalez-Castillo and a post-bac IRTA, Isabel Fernandez, have been working on applying manifold learning techniques to dynamic functional connectivity data. We know that the cognitive states of our brain are in constant fluctuation, and we hope to develop methods that reliably capture the relationship between those states and fMRI data, so that we can better explore the neuronal correlates of spontaneous thought. In collaboration with the Machine Learning Team, we are currently exploring how model hyperparameters affect final low dimensional representations of dynamic functional connectivity data, and how such differences may affect interpretation. As an example, we are determining what the optimal dimensionality is for exploring fMRI data, and how dimensionality may vary with external task demands. As we move forward, we hope to generate a set of guidelines and heuristics that will help the neuroimaging community adopt these novel methods and apply them to the analysis and interpretation of resting state dynamics. Multi-echo fMRI We have also continued work on improving multi-echo fMRI, particularly processing of multi-echo fMRI data using tedana (TE dependent data analysis) open source software (http://tedana.readthedocs.io). This has included recent work making the software more robust, understandable, and modifiable for future improvements. We have submitted for publication a manuscript on TE-dependent analysis of multi-echo fMRI with tedana, which is under review at the Journal of Open Science Software. Comparison of fMRI with dynamic calcium imaging: Lastly, we continue to work as a co-investigators on a BRAIN grant with Elizabeth Hillman of Columbia University. We are examining wide field optical mapping data in mice using fMRI-style connectivity analysis methods. This work is helping validate fMRI approaches to connectivity analyses using invasive methods that include both calcium imaging and hemodynamic measures that are more similar to BOLD. Some of this work was presented at the BRAIN Investigators Meeting in June 2021 and it was submitted as an abstract to present at SFN 2021.