A kernel machine-based fMRI physiological noise removal method

Functional magnetic resonance imaging (fMRI) technique with blood oxygenation level dependent (BOLD) contrast is a powerful tool for noninvasive mapping of brain function under task and resting states. The removal of cardiac- and respiration-induced physiological noise in fMRI data has been a significant challenge as fMRI studies seek to achieve higher spatial resolutions and characterize more subtle neuronal changes. The low temporal sampling rate of most multi-slice fMRI experiments often causes aliasing of physiological noise into the frequency range of BOLD activation signal. In addition, changes of heartbeat and respiration patterns also generate physiological fluctuations that have similar frequencies with BOLD activation. Most existing physiological noise-removal methods either place restrictive limitations on image acquisition or utilize filtering or regression based post-processing algorithms, which cannot distinguish the frequency-overlapping BOLD activation and the physiological noise. In this work, we address the challenge of physiological noise removal via the kernel machine technique, where a nonlinear kernel machine technique, kernel principal component analysis, is used with a specifically identified kernel function to differentiate BOLD signal from the physiological noise of the frequency. The proposed method was evaluated in human fMRI data acquired from multiple task-related and resting state fMRI experiments. A comparison study was also performed with an existing adaptive filtering method. The results indicate that the proposed method can effectively identify and reduce the physiological noise in fMRI data. The comparison study shows that the proposed method can provide comparable or better noise removal performance than the adaptive filtering approach.     

BOLD signal compartmentalization based on the apparent diffusion coefficient

Functional MRI (fMRI) can detect blood oxygenation level dependent (BOLD) hemodynamic responses secondary to neuronal activity. The most commonly used method for detecting fMRI signals is the gradient-echo echo-planar imaging (EPI) technique because of its sensitivity and speed. However, it is generally believed that a significant portion of these signals arises from large veins, with additional contribution from the capillaries and parenchyma. Early experiments using diffusion-weighted gradient-echo EPI have suggested that intra-voxel incoherent motion (IVIM) weighting inherent in the sequence can selectively attenuate contributions from different vessels based on the differences in the mobility of the blood within them. In the present study, we used similar approach to characterize the apparent diffusion coefficient (ADC) distribution within the activated areas of BOLD contrast. It is shown that the voxel values of the ADCs obtained from this technique can infer various vascular contributions to the BOLD signal.     

In contrast to BOLD: signal enhancement by extravascular water protons as an alternative mechanism of endogenous fMRI signal change

Despite the popularity and widespread application of functional magnetic resonance imaging (fMRI) in recent years, the physiological bases of signal change are not yet fully understood. Blood oxygen level-dependant (BOLD) contrast — attributed to local changes in blood flow and oxygenation, and therefore magnetic susceptibility — has become the most prevalent means of functional neuroimaging. However, at short echo times, spin-echo sequences show considerable deviations from the BOLD model, implying a second, non-BOLD component of signal change. This has been dubbed “signal enhancement by extravascular water protons” (SEEP) and is proposed to result from proton-density changes associated with cellular swelling. Given that such changes are independent of magnetic susceptibility, SEEP may offer new and improved opportunities for carrying out fMRI in regions with close proximity to air–tissue and/or bone–tissue interfaces (e.g., the prefrontal cortex and spinal cord), as well as regions close to large blood vessels, which may not be ideally suited for BOLD imaging. However, because of the interdisciplinary nature of the literature, there has yet to be a thorough synthesis, tying together the various and sometimes disparate aspects of SEEP theory. As such, we aim to provide a concise yet comprehensive overview of SEEP, including recent and compelling evidence for its validity, its current applications and its future relevance to the rapidly expanding field of functional neuroimaging. Before presenting the evidence for a non-BOLD component of endogenous functional contrast, and to enable a more critical review for the nonexpert reader, we begin by reviewing the fundamental principles underlying BOLD theory.     

Component structure of event-related fMRI responses in the different neurovascular compartments

In most functional magnetic resonance imaging (fMRI) studies, brain activity is localized by observing changes in the blood oxygenation level-dependent (BOLD) signal that are believed to arise from capillaries, venules and veins in and around the active neuronal population. However, the contribution from veins can be relatively far downstream from active neurons, thereby limiting the ability of BOLD imaging methods to precisely pinpoint neural generators. Hemodynamic measures based on apparent diffusion coefficients (ADCs) have recently been used to identify more upstream functional blood flow changes in the capillaries, arterioles and arteries. In particular, we recently showed that, due to the complementary vascular sensitivities of ADC and BOLD signals, the voxels conjointly activated by both measures may identify the capillary networks of the active neuronal areas. In this study, we first used simultaneously acquired ADC and BOLD functional imaging signals to identify brain voxels activated by ADC only, by both ADC and BOLD and by BOLD only, thereby delineating voxels relatively dominated by the arterial, capillary, and draining venous neurovascular compartments, respectively. We then examined the event-related fMRI BOLD responses in each of these delineated neurovascular compartments, hypothesizing that their event-related responses would show different temporal componentries. In the regions activated by both the BOLD and ADC contrasts, but not in the BOLD-only areas, we observed an initial transient signal reduction (an initial dip), consistent with the local production of deoxyhemoglobin by the active neuronal population. In addition, the BOLD-ADC overlap areas and the BOLD-only areas showed a clear poststimulus undershoot, whereas the compartment activated by only ADC did not show this component. These results indicate that using ADC contrast in conjunction with BOLD imaging can help delineate the various neurovascular compartments, improve the localization of active neural populations, and provide insight into the physiological mechanisms underlying the hemodynamic signals.     

The influence of spatial patterns of capillary networks on transverse relaxation

Tissue-inherent relaxation parameters offer valuable information about the arrangement of capillaries: in an external field, capillaries act as magnetic perturbers to generate local inhomogeneous fields due to the susceptibility difference of deoxygenated blood and the surrounding tissue. These field inhomogeneities influence the free induction decay in a characteristic way, and, conversely, the above tissue parameters can be recovered by multi-parametric fits of adequate theoretical models to experimentally sampled free induction decays. In this work we study the influence of different spatial patterns of capillary positions on the free induction decay. Starting from the standard single capillary approximation (Krogh cylinder) for a symmetric array of capillaries, the free induction decay is analyzed for increasingly random capillary positions, using a previously described Gibbs point field model. The effects of diffusion are implemented with a flexible and fast random walk simulation. We find that the asymmetric form of the obtained frequency distribution is more robust against variations of capillary radii than against shifts of capillary positions, and further that, for an inclusion of diffusion effects, the single capillary approximation models the uniform alignment of capillaries in the hexagonal lattice to great accuracy. An increase in randomization of capillary positions then leads to a significant change in relaxation times. This effect, however, is found less pronounced than that of changes in the off-resonance field strengths which are controlled by the oxygen extraction fraction, thus indicating that observed changes in BOLD imaging are more likely to be attributed to changes in oxygenation than to capillary alignment.     

Comparison of functional MR-venography and EPI-BOLD fMRI at 1.5 t

Functional magnetic resonance imaging (fMRI) of the brain using blood oxygenation level dependent (BOLD) contrast relies on the changes of paramagnetic deoxyhemoglobin concentration, which affects brain parenchyma and draining venous vessels. These changes in deoxyhemoglobin concentration in venous vessels can also be monitored using a high-resolution susceptibility-based MR-venography technique. Four volunteers participated in the study in which functional MR-venograms were compared with conventional echo-planar imaging (EPI)-BOLD-fMRI. In all cases, small venous vessels could be identified close to the areas of activation detected by conventional fMRI. In the venograms, task performance (finger tapping) resulted in a loss of venous vessel contrast compared to the resting state, which is consistent with a local decrease of deoxyhemoglobin concentration. MR-venography allows a direct visualization of the BOLD-effect at high spatial resolution. In combination with conventional fMRI, this technique may help to separate the contribution of brain parenchyma and venous vessels in fMRI studies.     

Influence of baseline hematocrit and hemodilution on BOLD fMRI activation.

Current understanding of blood oxygenation level dependent (BOLD) fMRI physiology predicts a close relationship between BOLD signal and blood hematocrit level. However, neither this relationship nor its effect on BOLD percent activation (BPA) has been empirically examined in man. To that end, BPA in primary visual cortex in response to photic stimulation was determined in a group of 24 normal subjects. A positive linear relationship between BPA and hematocrit was seen, particularly in men. To evaluate the effect of change in hematocrit on BPA, 9 men were studied before and following isotonic saline hemodilution, resulting in an average 6% reduction in hematocrit and an 8-31% reduction in BPA. No significant change in the number of activated pixels was seen. A model of predicted BPA as a function of hematocrit and vessel size was developed, and results from this model closely mirrored the empiric data. These results suggest that hematocrit significantly influences the magnitude of BPA and that such baseline factors should be accounted for when comparing BOLD data across groups of subjects, particularly in the many instances in which hematocrit may vary systematically. Such instances include several disease states as well as studies involving sex differences, drug administration, stress and other factors. Finally, the robust agreement between predicted and empiric data serves to validate a semiquantitative approach to the analysis of BOLD fMRI data.     

Neural activity-induced modulation of BOLD poststimulus undershoot independent of the positive signal

Despite intense research on the blood oxygenation level-dependent (BOLD) signal underlying functional magnetic resonance imaging, our understanding of its physiological basis is far from complete. In this study, it was investigated whether the so-called poststimulus BOLD signal undershoot is solely a passive vascular effect or actively induced by neural responses. Prolonged static and flickering black-white checkerboard stimulation with isoluminant grey screen as baseline condition were employed on eight human subjects. Within the same region of interest, the positive BOLD time courses for static and flickering stimuli were identical over the entire stimulus duration. In contrast, the static stimuli exhibited no poststimulus BOLD signal undershoot, whereas the flickering stimuli caused a strong BOLD poststimulus undershoot. To ease the interpretation, we performed an additional study measuring both BOLD signal and cerebral blood flow (CBF) using arterial spin labeling. Also for CBF, a difference in the poststimulus period was found for the two stimuli. Thus, a passive blood volume effect as the only contributor to the poststimulus undershoot comes short in explaining the BOLD poststimulus undershoot phenomenon for this particular experiment. Rather, an additional active neuronal activation or deactivation can strongly modulate the BOLD poststimulus behavior. In summary, the poststimulus time course of BOLD signal could potentially be used to differentiate neuronal activity patterns that are otherwise indistinguishable using the positive evoked response.     

Linear response functions of a quasi-one-dimensional electron gas

The predictions for density-density-type response functions of a quasi-one dimensional electron gas are compared in detail for various models (Hubbard and Tomonaga model) and for various approximations (Monte-Carlo calculations, RPA and extended RPA). It is shown that RPA does not lead to intrinsic contradictions for electron densities larger than 1.5 inverse exciton Bohr radii. Even in this high-density regime local field corrections enhance spin- and density susceptibilities significantly.     

Effect of magnetic field gradients induced by microvasculature on NMR measurements of molecular self-diffusion in biological tissues

Presence of induced mesoscopic gradients of magnetic field in magnetically heterogeneous samples affects the measured value of apparent diffusion coefficient. This effect is investigated theoretically in the context of diffusion measurements in perfused biological tissues with blood as the paramagnetic compartment. It is shown that the apparent diffusion coefficient is sensitive to mutual correlations in vessel positions. Neglect of these correlations results in a failure of the commonly used model of microvasculature in which vessels are described as independently placed cylinders. The model is modified to account for intervessel correlations. The results indicate an underestimation of apparent diffusion coefficient in proportion to the magnetic susceptibility of intravascular compartment in agreement with published experimental data. The proportionality coefficient depends on the microvascular architecture. Comparison with experimental data yields a numerical value for a new model parameter that characterises the correlation in mutual positions of blood vessels.     

Decoding neural events from fMRI BOLD signal: A comparison of existing approaches and development of a new algorithm

Neuroimaging methodology predominantly relies on the blood oxygenation level dependent (BOLD) signal. While the BOLD signal is a valid measure of neuronal activity, variances in fluctuations of the BOLD signal are not only due to fluctuations in neural activity. Thus, a remaining problem in neuroimaging analyses is developing methods that ensure specific inferences about neural activity that are not confounded by unrelated sources of noise in the BOLD signal. Here, we develop and test a new algorithm for performing semiblind (i.e., no knowledge of stimulus timings) deconvolution of the BOLD signal that treats the neural event as an observable, but intermediate, probabilistic representation of the system's state. We test and compare this new algorithm against three other recent deconvolution algorithms under varied levels of autocorrelated and Gaussian noise, hemodynamic response function (HRF) misspecification and observation sampling rate. Further, we compare the algorithms' performance using two models to simulate BOLD data: a convolution of neural events with a known (or misspecified) HRF versus a biophysically accurate balloon model of hemodynamics. We also examine the algorithms' performance on real task data. The results demonstrated good performance of all algorithms, though the new algorithm generally outperformed the others (3.0% improvement) under simulated resting-state experimental conditions exhibiting multiple, realistic confounding factors (as well as 10.3% improvement on a real Stroop task). The simulations also demonstrate that the greatest negative influence on deconvolution accuracy is observation sampling rate. Practical and theoretical implications of these results for improving inferences about neural activity from fMRI BOLD signal are discussed.     

血液对流换热的数值计算

在血管壁施加第三类边界条件是计算血液与组织间对流换热的一种近似计算方法.为分析其可行性,用有限元数值模拟方法计算血管分支结构中血液与组织的对流换热,得到不同流速和半径下分支血管内血液的截面平均Nu数沿管长的变化曲线.结果表明,血管树中分支血管的Nu数变化幅度不大,且趋于稳定值的速度很快.以相同边界条件下包含简单血管系统的舌体为例,分别用近似方法和完全耦合计算方法,进行血液流场和舌体温度场模拟.通过比较计算结果,得出两种方法得到的温度场分布趋势基本相同;用完全耦合计算方法得到的舌体温度略高于用近似方法得到的舌体温度,两者差值小于0.2℃.     

Visual cortex reactivity in sedated children examined with perfusion MRI (FAIR)

Sleeping and sedated children can respond to visual stimulation with a decrease in blood oxygenation level dependent (BOLD) functional MRI signal response. The contribution of metabolic and hemodynamic parameters to this inverse signal response is incompletely understood. It has been hypothesized that it is caused by a relatively greater increase of oxygen consumption compared to rCBF (regional cerebral blood flow) increase. We studied the rCBF changes during visual stimulation in four sedated children, aged 4-71 months, and four alert adults, with an arterial water spin labeling technique (FAIR) and BOLD fMRI in a 1.5T MR scanner. In the children, FAIR signal decreased by a mean of 0.96% (range 0.77-1.05) of the baseline periods of the non-selective images, while BOLD signal decreased by 2.03% (range 1.99-2.93). In the adults, FAIR and BOLD signal increased by 0.88% (range 0.8-0.99) and 2.63% (range 1.99-2.93), respectively. Thus, in the children, an rCBF increase could not be detected by perfusion MRI, but indications of a FAIR signal decrease were found. An rCBF decrease in the primary visual cortex during stimulation has not been reported previously, but it is a possible explanation for the negative BOLD response. Future studies will have to address if this response pattern is a consequence of age or sleep/sedation.     

BOLD functional MRI in disease and pharmacological studies: room for improvement?

In the past decade the use of blood oxygen level-dependent (BOLD) fMRI to investigate the effect of diseases and pharmacological agents on brain activity has increased greatly. BOLD fMRI does not measure neural activity directly, but relies on a cascade of physiological events linking neural activity to the generation of MRI signal. However, most of the disease and pharmacological studies performed so far have interpreted changes in BOLD fMRI as "brain activation," ignoring the potential confounds that can arise through drug- or disease-induced modulation of events downstream of the neural activity. This issue is especially serious in diseases (like multiple sclerosis, brain tumours and stroke) and drugs (like anaesthetics or those with a vascular action) that are known to influence these physiological events. Here we provide evidence that, to extract meaningful information on brain activity in patient and pharmacological BOLD fMRI studies, it is important to identify, characterise and possibly correct these influences that potentially confound the results. We suggest a series of experimental measures to improve the interpretability of BOLD fMRI studies. We have ranked these according to their potential information and current practical feasibility. First-line, necessary improvements consist of (1) the inclusion of one or more control tasks, and (2) the recording of physiological parameters during scanning and subsequent correction of possible between-group differences. Second-line, highly recommended important aim to make the results of a patient or drug BOLD study more interpretable and include the assessment of (1) baseline brain perfusion, (2) vascular reactivity, (3) the inclusion of stimulus-related perfusion fMRI and (4) the recording of electrophysiological responses to the stimulus of interest. Finally, third-line, desirable improvements consist of the inclusion of (1) simultaneous EEG-fMRI, (2) cerebral blood volume and (3) rate of metabolic oxygen consumption measurements and, when relevant, (4) animal studies investigating signalling between neural cells and blood vessels.     

A nonlinear BOLD model accounting for refractory effect by applying the longitudinal relaxation in NMR to the linear BOLD model

A mathematical model to regress the nonlinear blood oxygen level-dependent (BOLD) fMRI signal has been developed by incorporating the refractory effect into the linear BOLD model of the biphasic gamma variate function. The refractory effect was modeled as a relaxation of two separate BOLD capacities corresponding to the biphasic components of the BOLD signal in analogy with longitudinal relaxation of magnetization in NMR. When tested with the published fMRI data of finger tapping, the nonlinear BOLD model with the refractory effect reproduced the nonlinear BOLD effects such as reduced poststimulus undershoot and saddle pattern in a prolonged stimulation as well as the reduced BOLD signal for repetitive stimulation.     

Blood vessel enhancement for DSA images based on adaptive multi-scale filtering

Digital subtraction angiography (DSA) plays a significant role in the diagnosis, treatment planning and assessment of diseases. However, because of the geometrical complexity and fine characteristics of blood vessel structures, accurate and robust detection of blood vessels still remains a problem. In this paper, a blood vessel enhancement algorithm is proposed. The main purpose of this work is to improve the visual quality of blood vessels in DSA images. The new blood vessel enhancement algorithm is based on the multi-scale space theory and Hessian matrix. Not only the eigenvalues of Hessian matrix but also the angles between eigenvectors are utilized for the blood vessel enhancement of DSA. The filter parameters and scale factors are decided adaptively. Eigenvalues of the Hessian matrix are also used for the noise elimination. Experimental results show that the proposed algorithm has a good performance in blood vessel enhancement of DSA images. The proposed algorithm filters image background and non-vascular structure effectively. The deformation of blood vessels occurred in the enhancement process is avoided and more small blood vessels are visible in DSA images.     

Spontaneous low-frequency blood oxygenation level-dependent fluctuations and functional connectivity analysis of the 'resting' brain

Functional magnetic resonance imaging techniques using the blood oxygenation level-dependent (BOLD) contrast are widely used to map human brain function by relating local hemodynamic responses to neuronal stimuli compared to control conditions. There is increasing interest in spontaneous cerebral BOLD fluctuations that are prominent in the low-frequency range (<0.1 Hz) and show intriguing spatio-temporal correlations in functional networks. The nature of these signal fluctuations remains unclear, but there is accumulating evidence for a neural basis opening exciting new avenues to study human brain function and its connectivity at rest. Moreover, an increasing number of patient studies report disease-dependent variation in the amplitude and spatial coherence of low-frequency BOLD fluctuations (LFBF) that may afford greater diagnostic sensitivity and easier clinical applicability than standard fMRI. The main disadvantage of this emerging tool relates to physiological (respiratory, cardiac and vasomotion) and motion confounds that are challenging to disentangle requiring thorough preprocessing. Technical aspects of functional connectivity fMRI analysis and the neuroscientific potential of spontaneous LFBF in the default mode and other resting-state networks have been recently reviewed. This review will give an update on the current knowledge of the nature of LFBF, their relation to physiological confounds and potential for clinical diagnostic and pharmacological studies.     

Quantification of the power changes in BOLD signals using Welch spectrum method during different single-hand motor imageries

Motor imagery is an experimental paradigm implemented in cognitive neuroscience and cognitive psychology. To investigate the asymmetry of the strength of cortical functional activity due to different single-hand motor imageries, functional magnetic resonance imaging (fMRI) data from right handed normal subjects were recorded and analyzed during both left-hand and right-hand motor imagery processes. Then the average power of blood oxygenation level-dependent (BOLD) signals in temporal domain was calculated using the developed tool that combines Welch power spectrum and the integral of power spectrum approach of BOLD signal changes during motor imagery. Power change analysis results indicated that cortical activity exhibited a stronger power in the precentral gyrus and medial frontal gyrus with left-hand motor imagery tasks compared with that from right-hand motor imagery tasks. These observations suggest that right handed normal subjects mobilize more cortical nerve cells for left-hand motor imagery. Our findings also suggest that the approach based on power differences of BOLD signals is a suitable quantitative analysis tool for quantification of asymmetry of brain activity intensity during motor imagery tasks.