fMRI Modeling & Signal Characteristics

The basis of fMRI experiments is a tight coupling of neural activity with the BOLD response in both space and time. However, cross-modal negative BOLD signal may also be caused by blood-steal effects. Here, we measure the negative BOLD signal using ultra-high field fMRI and the neuroelectrical activity as measured by EEG in the same subjects. Neural deactivation is found in the EEG-based electrical neuroimaging maps in regions partially overlapping with those where the negative BOLD signal is found, implying a possible neural basis for the negative BOLD response.

It is a common practice to apply a short repetition time (TR) for acquiring more fMRI volumes within a given total imaging time, thus gaining contrast-to-noise ratio (CNR). However, both the task-evoked BOLD (T2/T2*) effect and flow-related component (R1) increases could contribute to the total percentage change of fMRI signal. This study aims to quantitatively evaluate the BOLD and flow contributions in the fMRI signal detected by gradient echo EPI in the human visual cortex during visual stimulation at 4T. The results show a substantially large flow-related contribution in the measured fMRI signal when TR is short. The observed flow-related enhancement in fMRI signal is likely attributed by perfusion change, and it benefits fMRI mapping in two aspects: improved CNR and specificity. The finding also suggests that the flow-related component needs to be considered in BOLD quantification (e.g., calibration of CMRO2).

To study the aging effect on neural mechanism of visual repetition, we performed an fMRI study. Participants saw inside and outside scenes repeatedly for five blocks, and the first two blocks (eight repetitions for each scene) were compared with the last two blocks for Young and Elderly groups. The results showed that the Elderly group has longer time-constant for the visual repetition effect to take place within fusiform areas and the occipital cortex. The results supported the conclusion that the aging effect on visual repetition is represented by the compromised ‘slew rate’ i.e., temporal responsivity in the fusiform gyri.

The purpose of this study was to evaluate the hemodynamic response function (HRF) change after inhalation of different carbon dioxide concentrations. Using event-related method investigate the transient hemodynamic response function by short-time visual stimulus with different CO2 concentrations. Our results show that the peak of HRF curve is decreased and delayed with increased inhaled CO2 concentrations, and the width of HRF curve is wider with higher inhaled CO2 concentration.

Previous studies found that attention produces robust BOLD modulation with only modest increase in firing activity in primary visual cortex. This could be due to a pure preemptive CBF increase with no relation to local neuronal firing. We tested the presence of such a mechanism using combined CBF/BOLD measurements to estimate relative CMRO2. Results show that attentional enhancement of V1 activity involves an increase of both metabolic activity and blood flow, rather than a preemptive increase in blood flow alone. The ratio of CBF to CMRO2 change appears to be higher when the stimulus is unattended than when attended.

Functional magnetic resonance imaging (fMRI) is widely used to map brain function. Nevertheless, it does measure neural activity only indirectly via hemodynamic changes. Here we performed fMRI in combination with 1H-MRS in order to study the relationships between the vascular and metabolic response of the brain to a visual stimulation paradigm specifically designed to partly disentangle spiking and synaptic activity within the primary visual cortex. Our results, though preliminary, confirm that the energetics of the stimulated brain contains more information than that revealed by fMRI alone, thereby indicating an uncoupling between hemodynamics and metabolism upon brain activation.

Quantification of brain hemodynamic parameters during functional activation is essential for understanding biophysical mechanisms behind blood oxygenation level depend (BOLD) phenomenon. Models of BOLD signal are often derived based on a relationship established by PET studies between cerebral blood volume (CBV) and cerebral blood flow (CBF), ignoring that only portion of CBV - deoxyhemoglobin-containing blood volume (DBV) affects BOLD signal. In this study, we directly measure DBV during visual activation in human visual cortex area using qBOLD-fMRI technique. We demonstrate for the first time that DBV decreases during functional activation – an effect opposite to well known increase in CBV.

Recent DfMRI studies preformed at 3T have observed an increase in the fractional BOLD signal change with increasing b value during functional stimulation. The origin of this effect remains controversial with both cell swelling and vascular sources being offered as explanations. If this effect is truly due to cell swelling, increasing the static magnetic field will not alter the effect. We present DfMRI results obtained at 4T which shows a constant BOLD signal change with increasing b value. This result is consistent with a vascular source for a varying DfMRI response with diffusion weighting seen at 3T.

Resting oscillatory patterns in cortical activity can originate by both network- and metabolism-related mechanisms. In particular, recent evidences suggest that the cell metabolic state exert an indirect control over the intrinsic network responsivity of the brain, much likely via astrocytic intracellular calcium (Ca2+)-mediated gliotransmission. Here we examined theoretically the contribution of astrocytes in the generation of the fMRI signal changes in the absence of focal neuronal stimulations. We found that oscillations in brain electrical, metabolic and vascular activity, as revealed by BOLD fMRI, can be qualitatively and quantitatively explained by calcium-mediated coupling between neuroglial activation and metabolism.

While existing functional MRI techniques can reliably detect stimulus-induced activation in gray matter, activity in white matter regions has not been readily measured. A recent study has reported subtle increases

in fractional anisotropy (FA) in specific white matter pathways in response to motor or visual stimulation in human subjects. In the current study, we aimed to validate these findings with a direct cortical stimulation paradigm in rats. Our functional DTI approach revealed significant but variable FA changes in restricted corpus callosum regions, which demonstrates that functional DTI enables detection of white matter activity in response to cortical stimulation.
1123. Whole-Brain Mapping of Venous Vessel Size in Humans Using the Hypercapnia-Induced BOLD Effect

Thies Halvor Jochimsen^1,2, Dimo Ivanov², Derek V M Ott², Wolfgang Heinke³, Robert Turner², Harald E. Möller², Jürgen R. Reichenbach<sup>1

1</sup>Medical Physics Group, Department of Diagnostic and Interventional Radiology, Jena University Hospital, Jena, Germany; ²Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany; ³Department of Anesthesiology and Intensive Care Medicine, University of Leipzig, Medical Faculty, Leipzig, Germany

It is demonstrated that non-invasive mapping of the venous microstructure (vessel radius) is possible by using the hypercapnia-induced BOLD effect. Furthermore, it is shown that maps of venous blood volume and vessel density can be obtained from the same experimental setup. These parameters are important for characterizing tumor angiogenesis and type.

A limitation of T2*w BOLD fMRI is the confounding contribution of signal from the larger vasculature. Based on time-to-peak and full-width-at-half-maximum BOLD characteristics of different vascular compartments identified at 3T, we characterized the spatio-temporal properties of the BOLD response at 7T in the visual cortex using an event-related fMRI paradigm with short visual stimuli, high sampling rate, and multiple spatial resolutions. For the smallest voxelsize a high time-to-peak spatial heterogeneity of the BOLD response was observed with fast responses localized in parenchyma. This opens the possibility to use TTP to probe layer specific BOLD responses in the human brain.

We use the combination of MEG and fMRI to study the neural basis for BOLD non-linearity. Both nonlinear neural responses to stimuli and nonlinear vascular responses to neural activity may contribute to BOLD non-linearity and the relative contribution of these two effects remains poorly understood. We extended the study of non-linear neural response to both phase locked evoked response and time related beta oscillations in somatosensory cortex. The N20 peak amplitudes show non-linearities with ISI of 0.25-2s, influenced by beta power at the time of stimulation, suggesting that best oscillation should also be considered for BOLD convolution.

Studying changes in cerebral hemodynamics is possible via MR, including blood oxygen level dependent (BOLD) imaging. We used near infrared spectroscopy (NIRS) to sample oxy- and deoxyhemoglobin directly, and utilized a nitrogen challenge to change FiO2 and thus cause measurable changes in blood oxygenation. We have observed good correlation between the BOLD and NIRS signals, with higher correlation in the gray matter than in the white matter. In the near future, we will use this paradigm to study the limited autoregulation of cerebral blood flow in preterm neonates.

Changes in CBF/CBV/arterial-CBV(CBV_a)/post-arterial-CBV(CBV_pa) were measured in human brain during breath-hold and visual stimulation. δCBV/CBV was larger during breath-hold (54.9+/-5.8%) than visual stimulation (28.2+/-5.2%), a difference primarily originating from δCBV_pa/CBV_pa (54.5+/-4.9% vs. 22.2+/-3.8%); δCBV_a/CBV_a (53+/-6%) and δCBF/CBF (61+/-7%) were comparable in both tasks. During breath-hold, vasodilation distributed proportionally among arterial and post-arterial compartments, whereas, during visual stimulation, relative change in CBV_a was greater than that in CBV_pa. Our data indicate that the coupling between arterial-CBV and CBF was largely preserved during both tasks (rCBVa=rCBF^0.86+/-0.05), while the relationship between total-CBV and CBF was substantially different between breath-hold (rCBV=rCBF^0.90+/-0.05) and visual (rCBV=rCBF^0.52+/-0.04) stimulation.

To increase the spatial specificity of the BOLD signal, a large ratio R of microvascular to macrovascular BOLD signal is desirable. This can be be achieved using spin echo sequences. To simplify the calculation the extravascular BOLD signal (EV) is often approximated by a mono-exponential decay (MEA). To investigate the effect of the MEA we calculated R for multiple B0 and TE without using this approximation. The parameter range for an optimal R was considerable different to the results obtained using the MEA. This is mainly due to an vessel radius dependent delay of the EV which is not reproduced by the MEA.

We present a realistic vascular model based on Monte-Carlo modeling of diffusion and the finite element method to compute the background magnetic field of partly oxygenated finite venules exposed to up to 16.4 T. Our data show that the realistic vasculature data set is necessary to account for the effects due to finite-sized vessels. The venule data herein stems from 2 photon microscopy of the rat brain. Results show that the infinite vessel model is prone to error so that the use of realistic vascular data sets is necessary to get precise results. However, for a better understanding more realistic vascular data sets should be examined in future work.

To study the intravascular BOLD mechanism, we used a physiologically controlled blood perfusion system at 9.4T under oxygenated conditions for a series of hematocrits. Previous studies have shown that, at such high fields, the two-site (eryhtrocyte-plasma) fast exchange model can not describe oxygenation-based relaxation changes properly in that it gives incorrect lifetimes for water in erythrocytes (1-3ms). We show that, for the physiological range of hematocrits, a general two-site exchange model (including slow, fast and intermediate regimes) can appropriately describe blood relaxation in oxygenated blood and provides an erythrocyte lifetime of 12.2±3.7ms, in agreement with literature values

Knowledge of blood T1 and T2 values are important for many MRI studies that include BOLD modeling, high spatial specificity BOLD fMRI, blood flow MRI using arterial spin labeling (ASL), and blood volume MRI using vascular space occupancy (VASO) techniques. The purpose of the present study was to determine blood T1 and T2 values at 11.7T as a function of oxygenation level (Y), temperature, hematocrit fraction (Hct) and field strength (B0).

A quantitative ER-fMRI study requires to measuring hemodynamic response function (HRF) both accurately and precisely. A periodic ER-fMRI design can produce a high accuracy of HRF measurement but a low precision. Utilizing the approximate linearity of the HRF, a rapid-presentation (RP) ER-fMRI design can improve the precision by shorting intersitimulus interval (ISI). Nevertheless, hemodynamic response is non-linear and its corresponding effect on the estimated HRF increases with decreasing ISI, rendering the estimated HRF inaccurate for small ISI values. Accordingly, as demonstrated in this preliminary study, an optimal RP ER-fMRI design should maximize both accuracy and precision of HRF measurements.

Oxidative metabolism can be estimated from the BOLD signal following a calibration manipulation to determine a factor M. M is the maximum possible BOLD signal change. Carbogen inhalation was used here with intense visual stimulation to test whether an asymptote in BOLD signal could be reached. Results show a convergence of percent BOLD changes around 9% in visual cortex for 10% carbogen alone, and 5-10% carbogen breathing plus visual stimulation. The diminishing incremental response from visual stimulation at high carbogen concentrations suggests that these manipulations approach BOLD levels close to the saturation plateau.

Calibrated fMRI is a new technique that allows quantitative estimates of the relative changes in cerebral metabolic rate of oxygen (ÄCMRO2) and cerebral blood flow (ÄCBF) that accompany neural activation. In this research we extend our previous work to study changes in neurovascular coupling over an age range during a cognitive Stroop task.

37 volunteers (age range 20-70) were scanned using 3T MRI. We found BOLD response to the Stroop task increases with increasing age, calibration constant A was found to reduce with age, a trend to reduced ÄCMRO2 with increasing age, and globally ÄCBF did not change with age.
1135. Error Propagation in CMRO2 Derivations Using CBF and BOLD Imaging

Hsiao-Wen Chung¹, Wen Chau Wu

¹Electrical Engineering, National Taiwan University, Taipei, Taiwan, Taiwan

The purpose of this study is therefore to investigate the issues error propagation in CMRO2 estimations under different CNR. Results from our error propagation study suggest that CMRO2 estimations using CBF and BOLD are valid only when the CNR for CBF measurements is sufficiently large, or when the underlying changes in CMRO2 and CBF are sufficiently large as in hypercapnic experiments. Validity of current instantaneous CMRO2 measurements for resting-state brain functional studies is therefore in some doubt.

Hypercapnia creates global changes in cerebral blood flow, volume, and oxygenation that can be measured with fMRI and used for calibration. Breath-holding is a simple way to induce hypercapnia, but it may alter thoracic chest pressure and include a ValSalva effect. We alter chest pressure while keeping the hold duration constant to see how the BOLD signal changes. The initial BOLD undershoot and following peak scale with pressure. Because the precise contrast mechanisms behind these changes are not fully understood, they may be a confound in calibration studies, or a novel way to rapidly induce calibration-useful global BOLD signal changes.

Many studies tried to understand how neural activity change vascular parameters, but little attention was received to whether gas content changes in blood would reversely alter neural activity. To investigate such an effect, we used a recently developed MRI technique to quantify global cerebral metabolic rate of oxygen (CMRO2) under hypoxia and hyperoxia. Our data suggest that a change in arterial oxygen content can modulate brain metabolism in a dose-dependent manner, with hypoxia increasing CMRO2 and hyperoxia decreasing it. Therefore, in addition to the well-known “forward” neurovascular coupling, the “reverse” coupling may be important in the regulation of brain function.

Manual HC calibration depends on intrinsically low signal-to-noise perfusion imaging and individual vascular architecture, with resulting calibration (M)-values prone to large intra- and inter-subject variations that may bias oxygen metabolism studies. We thereby sough to investigate HO as a calibration alternative under rigorous control of end-tidal partial pressures of CO₂ (PetCO₂) and O₂ (PetO₂). Our findings suggest the viability of precisely controlling HO stimulation to provide more precise per-subject and per-brain-region M-estimates, based on high SNR PaO₂ measurements and the removal of the confound of vascular variation in population observed under HC-calibration.