CORSICA: correction of structured noise in fMRI by automatic identification of ICA components

When applied to functional magnetic resonance imaging (fMRI) data, spatial independent component analysis (sICA), a data-driven technique that addresses the blind source separation problem, seems able to extract components specifically related to physiological noise and brain movements. These components should be removed from the data to achieve structured noise reduction and improve any subsequent detection and analysis of signal fluctuations related to neural activity. We propose a new automatic method called CORSICA (CORrection of Structured noise using spatial Independent Component Analysis) to identify the components related to physiological noise, using prior information on the spatial localization of the main physiological fluctuations in fMRI data. As opposed to existing spectral priors, which may be subject to aliasing effects for long-TR data sets (typically acquired with TR >1 s), such spatial priors can be applied to fMRI data, regardless of the TR of the acquisitions. By comparing the proposed automatic selection to a manual selection performed visually by a human operator, we first show that CORSICA is able to identify the noise-related components for long-TR data with a high sensitivity and a specificity of 1. On short-TR data sets, we validate that the proposed method of noise reduction allows a substantial improvement of the signal-to-noise ratio evaluated at the cardiac and respiratory frequencies, even in the gray matter, while preserving the main fluctuations related to neural activity.     

Measurement and characterization of the human spinal cord SEEP response using event-related spinal fMRI

Although event-related fMRI is able to reliably detect brief changes in brain activity and is now widely used throughout systems and cognitive neuroscience, there have been no previous reports of event-related spinal cord fMRI. This is likely attributable to the various technical challenges associated with spinal fMRI (e.g., imaging a suitable length of the cord, reducing image artifacts from the vertebrae and intervertebral discs, and dealing with physiological noise from spinal cord motion). However, with many of these issues now resolved, the largest remaining impediment for event-related spinal fMRI is a deprived understanding of the spinal cord fMRI signal time course. Therefore, in this study, we used a proton density-weighted HASTE sequence, with functional contrast based on signal enhancement by extravascular water protons (SEEP), and a motion-compensating GLM analysis to (i) characterize the SEEP response function in the human cervical spinal cord and (ii) demonstrate the feasibility of event-related spinal fMRI. This was achieved by applying very brief (1 s) epochs of 22°C thermal stimulation to the palm of the hand and measuring the impulse response function. Our results suggest that the spinal cord SEEP response (time to peak ≈8 s; FWHM ≈4 s; and probably lacking pre- and poststimulus undershoots) is slower than previous estimates of SEEP or BOLD responses in the brain, but faster than previously reported spinal cord BOLD responses. Finally, by detecting and mapping consistent signal-intensity changes within and across subjects, and validating these regions with a block-designed experiment, this study represents the first successful demonstration of event-related spinal fMRI.     

Magnetic resonance imaging of soft-tissue tumors: Comparison with computed tomography

Twenty-seven patients with soft-tissue tumors were examined with a Picker 0.15-tesla resistive magnet and by computed tomography (CT). In all but one patient, MRI was better than or equal to CT in defining the anatomic extent of the tumor. We could determine whether major vascular structures were engulfed by the tumor in 80% of the MRI examinations but only in 62% of the CT scans. MRI and CT were equally effective in determining the presence or absence of bony invasion. The MRI images of all the tumors showed increased signal intensity relative to normal muscle when spin-echo (SE) sulse sequences with long repeat times were used (SE: echo time [TE], 60 ms; repetition time [TR], 2,000 ms). When T1 weighted pulse sequences were used (SE: TE, 30 ms; TR, 500 ms or inversion recovery: inversion time, 500 ms; TE, 40 ms; TR, 2,000 ms) the malignant tumors showed decreased signal intensity compared to normal muscle. Only lipomas showed high signal intensity on both T1 and T2 weighted pulse sequences.     

Sources of functional magnetic resonance imaging signal fluctuations in the human brain at rest: a 7 T study

Signal fluctuations in functional magnetic resonance imaging (fMRI) can result from a number of sources that may have a neuronal, physiologic or instrumental origin. To determine the relative contribution of these sources, we recorded physiological (respiration and cardiac) signals simultaneously with fMRI in human volunteers at rest with their eyes closed. State-of-the-art technology was used including high magnetic field (7 T), a multichannel detector array and high-resolution (3 mm³) echo-planar imaging. We investigated the relative contribution of thermal noise and other sources of variance to the observed fMRI signal fluctuations both in the visual cortex and in the whole brain gray matter. The following sources of variance were evaluated separately: low-frequency drifts due to scanner instability, effects correlated with respiratory and cardiac cycles, effects due to variability in the respiratory flow rate and cardiac rate, and other sources, tentatively attributed to spontaneous neuronal activity. We found that low-frequency drifts are the most significant source of fMRI signal fluctuations (3.0% signal change in the visual cortex, TE=32 ms), followed by spontaneous neuronal activity (2.9%), thermal noise (2.1%), effects due to variability in physiological rates (respiration 0.9%, heartbeat 0.9%), and correlated with physiological cycles (0.6%). We suggest the selection and use of four lagged physiological noise regressors as an effective model to explain the variance related to fluctuations in the rates of respiration volume change and cardiac pulsation. Our results also indicate that, compared to the whole brain gray matter, the visual cortex has higher sensitivity to changes in both the rate of respiration and the spontaneous resting-state activity. Under the conditions of this study, spontaneous neuronal activity is one of the major contributors to the measured fMRI signal fluctuations, increasing almost twofold relative to earlier experiments under similar conditions at 3 T.     

Noninvasive observation of cervical spinal cord activity in children by functional MRI during cold thermal stimulation

Functional magnetic resonance imaging (fMRI) is a non-invasive neuroimaging tool that indirectly identifies areas of neural activity in the brain and more recently has been applied to the adult spinal cord (spinal fMRI). Spinal fMRI could clearly benefit pediatric populations as well. The purpose of this work was to characterize the response observed with spinal fMRI in the brainstem and cervical (C) spinal cord of awake, healthy children during thermal stimulation (17°C and 27°C) applied to the right hand. Functional MRI detected neuronal activity in the expected region of the spinal cord (C6 and C7) as well as in the brainstem and thalamus. The observed magnitudes of signal change of the responses to 17°C and 27°C were similar; however, the spatial distribution of active pixels was greater during 17°C stimulation. The results of this study indicate that fMRI can be used to assess activity in the spinal cords of children, with good sensitivity and reliability.     

Issues about the fMRI of the human spinal cord

Noninvasive functional studies on human spinal cord by means of magnetic resonance imaging (MRI) are gaining attention because of the promising applications in the study of healthy and injured central nervous system. The findings obtained are generally consistent with the anatomic knowledge based on invasive methods, but the origin and specificity of functional contrast is still debated. In this paper, a review of current knowledge and major issues about functional MRI (fMRI) in the human spinal cord is presented, with emphasis on the main methodological and technical problems and on forthcoming applications as clinical tool.     

Assessment of data acquisition parameters,and analysis techniques for noise reduction in spinal cord fMRI data

Purpose

The purpose of this work is to characterize the noise in spinal cord functional MRI, assess current methods aimed at reducing noise, and optimize imaging parameters.

Methods

Functional MRI data were acquired at multiple echo times and the contrast-to-noise ratio (CNR) was calculated. Independently, the repetition time was systematically varied with and without parallel imaging, to maximize BOLD sensitivity and minimize type I errors. Noise in the images was characterized by examining the frequency spectrum, and investigating whether autocorrelations exist. The efficacy of several physiological noise reduction methods in both null (no stimuli) and task (thermal pain paradigm) data was also assessed. Finally, our previous normalization methods were extended.

Results

The echo time with the highest functional CNR at 3 Tesla is at roughly 75 msec. Parallel imaging reduced the variance and the presence of autocorrelations, however the BOLD response in task data was more robust in data acquired without parallel imaging. Model-free based approaches further increased the detection of active voxels in the task data. Finally, inter-subject registration was improved.

Conclusions

Results from this study provide a rigorous characterization of the properties of the noise and assessment of data acquisition and analysis methods for spinal cord and brainstem fMRI.     

In vivo quantitative 1H MRS of cerebellum and evaluation of quantitation reproducibility by simulation of different levels of noise and spectral resolution

A quantitative analysis of cerebellar metabolites in normal subjects has been performed by proton MR spectroscopy (MRS) with relaxation time correction. Quantitation was carried out in seven healthy human subjects with the well-established LCModel program. The prior knowledge utilized for quantitation was obtained from solutions containing the major brain metabolites and MRS investigated under the same experimental conditions. The tissue water signal was used as an internal standard for the in vivo studies. Both in vitro (for the prior knowledge template) and in vivo data were acquired separately at 1.5 T by PRESS sequence (TR, 1500 ms; TE, 30 ms). The absolute concentration of main cerebellar metabolites was corrected for relaxation time effects. Different noise and line broadening conditions were considered and simulated in the spectral processing in order to evaluate the effect of spectral quality on the concentration estimates.     

Temporal and spatial MRI responses to subsecond visual activation

The temporal and spatial characteristics of oxygenation-sensitive MRI responses to very brief visual stimuli (five Hz reversing black and white checkerboard pattern versus darkness) were investigated (nine subjects) by means of serial single-shot gradient-echo echo-planar imaging (2.0 T, TR = 400 ms, mean TE = 54 ms, flip angle 30°). The use of a 0.2-s stimulus and a 90-s control phase resulted in an initial latency phase (about 2 s, no signal change), a positive MRI response (2.5% signal increase peaking at 5 s after stimulus onset), and a post-stimulus undershoot (1% signal decrease peaking at 15 s after stimulus onset) lasting for about 50–60 s. The finding that a subsecond visual stimulus elicits both a strong positive MRI response and a long-lasting undershoot provides further evidence for the neuronal origin of slow signal fluctuations seen in the absence of functional challenge and their utility for mapping functional connectivity. The additional observation that a reduction of the inter-stimulus control phase from 90 s to 9.8 s does not seem to affect the spatial extent of cortical activation in pertinent maps is of major relevance for the design and analysis of “event-related” MRI studies.     

Respiratory noise correction using phase information

Respiratory noise is a confounding factor in functional magnetic resonance imaging (MRI) data analysis. A novel method called Respiratory noise Correction using Phase information is proposed to retrospectively correct for the respiratory noise in functional MRI (fMRI) time series. It is demonstrated that the respiratory movement and the phase of functional MRI images are highly correlated in time. The signal fluctuation due to respiratory movements can be effectively estimated from the phase variation and removed from the functional MRI time series using a Wiener filtering technique. In our experiments, this new method is compared with RETROICOR, which requires recording respiration signal simultaneously in an fMRI experiment. The two techniques show comparable performance with respect to the respiratory noise correction for fMRI time series. However, this technique is more advantageous because there is no need for monitoring the subjects’ respiration or changing functional MRI protocols. This technique is also potentially useful for correcting respiratory noise from abnormal breathing or when the respiration is not periodic.     

The mechanical state of intracranial tissues in elderly subjects studied by imaging CSF and brain pulsations

The biomechanical properties of intracranial tissues influence the mechanical coupling of brain and CSF oscillations to the driving vascular pulsations. Dynamic phase contrast MRI was used to measure the transfer functions that characterize these couplings in normal elderly subjects and patients with Alzheimer's disease. The transfer functions of both groups were significantly different from the previously reported transfer functions of normal young subjects. The data show that vascular pulsations tend to cause greater spinal cord movements and smaller CSF oscillations in the older subjects than in the younger ones. These results are likely to be due to age-related changes in the mechanical state of intracranial tissues.     

Constrained modeling for spectroscopic measurement of bi-exponential spin-lattice relaxation of water in vivo

The (1)H NMR water signal from spectroscopic voxels localized in gray matter contains contributions from tissue and cerebral spinal fluid (CSF). A typically weak CSF signal at short echo times makes separating the tissue and CSF spin-lattice relaxation times (T(1)) difficult, often yielding poor precision in a bi-exponential relaxation model. Simulations show that reducing the variables in the T(1) model by using known signal intensity values significantly improves the precision of the T(1) measurement. The method was validated on studies on eight healthy subjects (four males and four females, mean age 21 +/- 2 years) through a total of twenty-four spectroscopic relaxation studies. Each study included both T(1) and spin-spin relaxation (T(2)) experiments. All volumes were localized along the Sylvian fissure using a stimulated echo localization technique with a mixing time of 10 ms. The T(2) experiment consisted of 16 stimulated echo acquisitions ranging from a minimum echo time (TE) of 20 ms to a maximum of 1000 ms, with a repetition time of 12 s. All T(1) experiments consisted of 16 stimulated echo acquisition, using a homospoil saturation recovery technique with a minimum recovery time of 50 ms and a maximum 12 s. The results of the T(2) measurements provided the signal intensity values used in the bi-exponential T(1) model. The mean T(1) values when the signal intensities were constrained by the T(2) results were 1055.4 ms +/- 7.4% for tissue and 5393.5 ms +/- 59% for CSF. When the signal intensities remained free variables in the model, the mean T(1) values were 1085 ms +/- 19.4% and 5038.8 ms +/- 113.0% for tissue and CSF, respectively. The resulting improvement in precision allows the water tissue T(1) value to be included in the spectroscopic characterization of brain tissue.     

Cardiac-induced physiological noise in 3D gradient echo brain imaging: effect of k-space sampling scheme

The physiological noise in 3D image acquisition is shown to depend strongly on the sampling scheme. Five sampling schemes are considered: Linear, Centric, Segmented, Random and Tuned. Tuned acquisition means that data acquisition at k-space positions k and -k are separated with a specific time interval. We model physiological noise as a periodic temporal oscillation with arbitrary spatial amplitude in the physical object and develop a general framework to describe how this is rendered in the reconstructed image. Reconstructed noise can be decomposed in one component that is in phase with the signal (parallel) and one that is 90° out of phase (orthogonal). Only the former has a significant influence on the magnitude of the signal. The study focuses on fMRI using 3D EPI. Each k-space plane is acquired in a single shot in a time much shorter than the period of the physiological noise. The above mentioned sampling schemes are applied in the slow k-space direction and noise propagates almost exclusively in this direction. The problem then, is effectively one-dimensional. Numerical simulations and analytical expressions are presented. 3D noise measurements and 2D measurements with high temporal resolution are conducted. The measurements are performed under breath-hold to isolate the effect of cardiac-induced pulsatile motion. We compare the time-course stability of the sampling schemes and the extent to which noise propagates from a localized source into other parts of the imaging volume. Tuned and Linear acquisitions perform better than Centric, Segmented and Random.     

Biophysical and neural basis of resting state functional connectivity: Evidence from non-human primates

Functional MRI (fMRI) has evolved from simple observations of regional changes in MRI signals caused by cortical activity induced by a task or stimulus, to task-free acquisitions of images in a resting state. Such resting state signals contain low frequency fluctuations which may be correlated between voxels, and strongly correlated regions are deemed to reflect functional connectivity within synchronized circuits. Resting state functional connectivity (rsFC) measures have been widely adopted by the neuroscience community, and are being used and interpreted as indicators of intrinsic neural circuits and their functional states in a broad range of applications, both basic and clinical. However, there has been relatively little work reported that validates whether inter-regional correlations in resting state fluctuations of fMRI (rsfMRI) signals actually measure functional connectivity between brain regions, or to establish how MRI data correlate with other metrics of functional connectivity. In this mini-review, we summarize recent studies of rsFC within mesoscopic scale cortical networks (100 μm–10 mm) within a well defined functional region of primary somatosensory cortex (S1), as well as spinal cord and brain white matter in non-human primates, in which we have measured spatial patterns of resting state correlations and validated their interpretation with electrophysiological signals and anatomic connections. Moreover, we emphasize that low frequency correlations are a general feature of neural systems, as evidenced by their presence in the spinal cord as well as white matter. These studies demonstrate the valuable role of high field MRI and invasive measurements in an animal model to inform the interpretation of human imaging studies.     

Acoustic noise during functional magnetic resonance imaging

Functional magnetic resonance imaging (fMRI) enables sites of brain activation to be localized in human subjects. For studies of the auditory system, acoustic noise generated during fMRI can interfere with assessments of this activation by introducing uncontrolled extraneous sounds. As a first step toward reducing the noise during fMRI, this paper describes the temporal and spectral characteristics of the noise present under typical fMRI study conditions for two imagers with different static magnetic field strengths. Peak noise levels were 123 and 138 dB re 20 microPa in a 1.5-tesla (T) and a 3-T imager, respectively. The noise spectrum (calculated over a 10-ms window coinciding with the highest-amplitude noise) showed a prominent maximum at 1 kHz for the 1.5-T imager (115 dB SPL) and at 1.4 kHz for the 3-T imager (131 dB SPL). The frequency content and timing of the most intense noise components indicated that the noise was primarily attributable to the readout gradients in the imaging pulse sequence. The noise persisted above background levels for 300-500 ms after gradient activity ceased, indicating that resonating structures in the imager or noise reverberating in the imager room were also factors. The gradient noise waveform was highly repeatable. In addition, the coolant pump for the imager's permanent magnet and the room air-handling system were sources of ongoing noise lower in both level and frequency than gradient coil noise. Knowledge of the sources and characteristics of the noise enabled the examination of general approaches to noise control that could be applied to reduce the unwanted noise during fMRI sessions.     

Failure to direct detect magnetic field dephasing corresponding to ERP generation

Functional magnetic resonance imaging (fMRI) has become the method of choice for mapping brain activity in human subjects and detects changes in regional blood oxygenation and volume associated with local changes in neuronal activity. While imaging based on blood oxygenation level dependent (BOLD) contrast has good spatial resolution and sensitivity, the hemodynamic signal develops relatively slowly and is only indirectly related to neuronal activity. An alternative approach termed magnetic source magnetic resonance imaging (msMRI) is based on the premise that neural activity may be mapped by magnetic resonance imaging (MRI) with greater temporal resolution by detecting the local magnetic field perturbations associated with local neuronal electric currents. We used a hybrid ms/BOLD MRI method to investigate whether msMRI could detect signal changes that occur simultaneously at the time of the production of well-defined event-related potentials, the P300 and N170, in regions that previously have been identified as generators of these electrical signals. Robust BOLD activations occurred after some seconds, but we were unable to detect any significant changes in the T2*-weighted signal in these locations that correlated temporally with the timings of the evoked response potentials (ERPs).     

Spatial normalization, bulk motion correction and coregistration for functional magnetic resonance imaging of the human cervical spinal cord and brainstem

Functional magnetic resonance imaging (fMRI) of the cortex is a powerful tool for neuroscience research, and its use has been extended into the brainstem and spinal cord as well. However, there are significant technical challenges with extrapolating the developments that have been achieved in the cortex to their use in the brainstem and spinal cord. Here, we develop a normalized coordinate system for the cervical spinal cord and brainstem, demonstrating a semiautomated method for spatially normalizing and coregistering fMRI data from these regions. fMRI data from 24 experiments in eight volunteers are normalized and combined to create the first anatomical reference volume, and based on this volume, we define a standardized region-of-interest (ROI) mask, as well as a map of 52 anatomical regions, which can be applied automatically to fMRI results. The normalization is demonstrated to have an accuracy of less than 2 mm in 93% of anatomical test points. The reverse of the normalization procedure is also demonstrated for automatic alignment of the standardized ROI mask and region-label map with fMRI data in its original (unnormalized) format. A reliable method for spatially normalizing fMRI data is essential for analyses of group data and for assessing the effects of spinal cord injury or disease on an individual basis by comparing with results from healthy subjects.     

Cortex-based independent component analysis of fMRI time series

The cerebral cortex is the main target of analysis in many functional magnetic resonance imaging (fMRI) studies. Since only about 20% of the voxels of a typical fMRI data set lie within the cortex, statistical analysis can be restricted to the subset of the voxels obtained after cortex segmentation. While such restriction does not influence conventional univariate statistical tests, it may have a substantial effect on the performance of multivariate methods.

Here, we describe a novel approach for data-driven analysis of single-subject fMRI time series that combines techniques for the segmentation and reconstruction of the cortical surface of the brain and the spatial independent component analysis (sICA) of the functional time courses (TCs). We use the mesh of the white matter/gray matter boundary, automatically reconstructed from high-spatial-resolution anatomical MR images, to limit the sICA decomposition of a coregistered functional time series to those voxels which are within a specified region with respect to the cortical sheet (cortex-based ICA, or cbICA). We illustrate our analysis method in the context of fMRI blocked and event-related experimental designs and in an fMRI experiment with perceptually ambiguous stimulation, in which an a priori specification of the stimulation protocol is not possible.

A comparison between cbICA and conventional hypothesis-driven statistical methods shows that cortical surface maps and component TCs blindly obtained with cbICA reliably reflect task-related spatiotemporal activation patterns. Furthermore, the advantages of using cbICA when the specification of a temporal model of the expected hemodynamic response is not straightforward are illustrated and discussed. A comparison between cbICA and anatomically unconstrained ICA reveals that — beside reducing computational demand — the cortex-based approach improves the fitting of the ICA model in the gray matter voxels, the separation of cortical components and the estimation of their TCs, particularly in the case of fMRI data sets with a complex spatiotemporal statistical structure.     

Separation of physiological very low frequency fluctuation from aliasing by switched sampling interval fMRI scans

Anesthetized children have dominant blood-oxygen-level-dependent (BOLD) signal sources presenting high-power fluctuations at very low frequencies (VLF <0.05 Hz). Aliasing of frequencies higher than critically sampled has been regarded as one probable origin of the VLF fluctuations. Aliased signal frequencies change when the sampling rate of the data is altered. In this study, the aliasing of VLF BOLD signal fluctuation was analysed by switching the repetition time (TR) of magnetic resonance (MR) images. Eleven anesthetized children were imaged at 1.5 T using TRs of 500 and 1200 ms. The BOLD signal sources were separated with independent component analysis (ICA). Occipital cortex signal sources had nonaliased VLF fluctuation ( approximately 0.03 Hz) in 9 of 11 subjects. Arterial signal sources failed to present stable power peaks at frequencies lower than 0.42 Hz presumably due to aliasing. Cerebrospinal fluid (CSF)-related signal sources showed nonaliased VLF in four subjects. In conclusion, the VLF BOLD signal fluctuation in the occipital cortex is a true physiological fluctuation, not a result of signal aliasing.