Aniruddha Das, Ph.D.
Associate Professor of Neuroscience and Psychiatry
Columbia University

Brain Hemodynamics: So Much More Than Just a Measure of Local Neural Firing

 

For the brain to function it needs oxygen. To support this need there exists a complex circulatory system that provides the appropriate levels of oxygenation according to changing energy needs to follow regional brain activity. For this reason the relative level of blood flow to different brain regions can serve as a correlate and measure of local activity within the brain, and it serves as the basis for blood oxygen level dependent (BOLD) functional magnetic resonance imaging (fMRI). Dr. Das' work has demonstrated that changes in blood flow are not merely a passive response to increased brain activity, but rather, may actually be involved in the anticipation of activity changes. Dr. Das' work has begun to shed light on how dynamic structural changes in the circulatory system assist neuronal function in an anticipatory fashion.

Brain imaging — such as functional magnetic resonance imaging (fMRI) — is based not on measuring neural activity but rather on brain hemodynamics, which refer to local changes in blood volume, blood flow and oxygenation. Despite the explosive current growth in the use of brain imaging, the relation between the hemodynamic imaging signal and local neural activity is still poorly understood. It is typically assumed that the imaging signal is uniformly a consequence of local neural responses. Specifically, the signal is assumed to arise when increased neural activity consumes oxygen in the blood locally and triggers a local inflow of fresh oxygenated blood. However, this assumption is based on studies in anesthetized animals. Very little is known about the links between the imaging signal and neural activity in alert animals engaged in behavioral tasks, an experimental situation that better represents the typical imaging study in humans.

Dr. Aniruddha Das' lab addresses this issue by recording neural signals through the use of implanted electrodes, while simultaneously imaging with fMRI in the brains of alert monkeys while they perform a visual discrimination task. Their studies have revealed a complex relationship between brain hemodynamics and neural activity. This complexity is evident at two levels. First, the Das lab has found that when the animals are engaged in a systematic visual task, the visual region V1 hemodynamic signal contains a strong task-related component in addition to the visually evoked response. The task-related component is a novel anticipatory signal, independent of visual input, that dilates local arteries and brings in fresh blood in time for expected visual trials. Notably, this signal is not predicted by local neural spiking, in sharp contrast to the trial-related signal that is predicted to better than 95 percent accuracy by spiking. Next, researchers in the Das lab found that even the visually evoked hemodynamic signal is likely not driven by deoxygenation in the blood as is often believed. On the contrary, visual stimulation first triggers a rapid increase in blood volume leading to an increase rather than a decrease of blood oxygenation. This stimulus-triggered blood volume increase presumably reflects an immediate dilation of local blood vessels, occurring in parallel with the increased neural activity, so as to bring additional blood to the cortex before the increased demand can lead to deoxygenation of the blood.

The Das lab is currently pursuing the consequences of this finding, both in its implications for interpreting brain images and the insights it may offer into brain processing. For brain imaging, they showed that the trial-related signal adds linearly with the stimulus-evoked signal in periodic tasks. The trial-related signal can be large, grossly distorting the net signal. However, subtracting a "blank trial" baseline leaves just the stimulus-evoked component that is robustly correlated with local spiking. The nature of the trial-related signal and its relevance to brain processing is still an open question, however. The signal is strongest when the animal is fully engaged in the task, and it changes character dramatically when the animal commits an error. However, unlike visuospatial attention, the signal is not spatially localized. It is correlated with fluctuations in heart rate and pupil dilation and thus likely reflects an autonomic process, possibly mediated by neuromodulatory input to V1. However, it is not global arousal: it is modality specific, being present in V1 only for visual and not auditory tasks. The goal of understanding these issues and exploring the significance of this novel signal forms the core of the Das research program.