No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Enhancement of Temporal Resolution and BOLD Sensitivity in Real-Time fMRI using Multi-Slab Echo-Volumar Imaging
Abstract
In this study, a new approach to high-speed fMRI using multi-slab echo-volumar imaging (EVI) is developed that minimizes geometrical image distortion and spatial blurring, and enables nonaliased sampling of physiological signal fluctuation to increase BOLD sensitivity compared to conventional echo-planar imaging (EPI). Real-time fMRI using whole brain 4-slab EVI with 286 ms temporal resolution (4mm isotropic voxel size) and partial brain 2-slab EVI with 136 ms temporal resolution (4×4×6 mm(3) voxel size) was performed on a clinical 3 Tesla MRI scanner equipped with 12-channel head coil. Four-slab EVI of visual and motor tasks significantly increased mean (visual: 96%, motor: 66%) and maximum t-score (visual: 263%, motor: 124%) and mean (visual: 59%, motor: 131%) and maximum (visual: 29%, motor: 67%) BOLD signal amplitude compared with EPI. Time domain moving average filtering (2s width) to suppress physiological noise from cardiac and respiratory fluctuations further improved mean (visual: 196%, motor: 140%) and maximum (visual: 384%, motor: 200%) t-scores and increased extents of activation (visual: 73%, motor: 70%) compared to EPI. Similar sensitivity enhancement, which is attributed to high sampling rate at only moderately reduced temporal signal-to-noise ratio (mean: -52%) and longer sampling of the BOLD effect in the echo-time domain compared to EPI, was measured in auditory cortex. Two-slab EVI further improved temporal resolution for measuring task-related activation and enabled mapping of five major resting state networks (RSNs) in individual subjects in 5 min scans. The bilateral sensorimotor, the default mode and the occipital RSNs were detectable in time frames as short as 75 s. In conclusion, the high sampling rate of real-time multi-slab EVI significantly improves sensitivity for studying the temporal dynamics of hemodynamic responses and for characterizing functional networks at high field strength in short measurement times.
... Recent advances in fMRI acquisition rates, using multi-band (simultaneous multi-slice) encoded EPI (MB-EPI) Setsompop et al., 2012), simultaneous image refocusing (Chen et al., 2015), magnetic resonance encephalography (MREG) (Hennig et al., 2020) and multi-slab echo-volumar imaging (MS-EVI) (Posse et al., 2012), have played an important role in improving sensitivity for mapping functional connectomics (Feinberg et al., 2010;Smith et al., 2012;Posse et al., 2013) and for detecting task-based and resting state fMRI signal changes at frequencies above 0.2 Hz (Lewis et al., 2016). In MS-EVI an entire 3D k-space with a slab is sampled following a single excitation (Mansfield et al., 1989), which accelerates the volume acquisition and must be accomplished in a time on the order of the T 2 * (~ 50 ms at 3 T) to minimize blurring and geometrical distortion. ...
... The stack of slabs to cover the volume of interest is acquired sequentially using an interleaved acquisition order to minimize signal saturation at slab edges. This approach combines the sampling efficiency of singleshot 3D encoding with the sensitivity advantage of multi-echo acquisitions (Posse et al., 2012;Puckett et al., 2018). It allows for significantly increased temporal resolution without the √R penalty incurred when using conventional parallel imaging methods combined with the sensitivity advantage of volumetric data acquisition. ...
... The development of MB-EVI was performed in multiple steps on scanners with different operating systems, which allowed to implement MB-EVI with increasingly higher spatial-temporal resolution and image reconstruction performance. The initial implementation of MB-EVI on a 3 T Siemens Trio scanner with VB17A operating system used a Siemens product EPI sequence that was modified based on custom multi-echo EPI (Speck and Hennig, 1998) and MS-EVI (Posse et al., 2012) pulse sequences to support up to 32 echo times. A k z dephasing gradient was implemented and k z encoding gradient blips were placed in front of the EPI readout modules for each echo time to encode k z steps -N/2-1 to N/2 ( Figure 1a). ...
Purpose
In this study we develop undersampled echo-volumar imaging (EVI) using multi-band/simultaneous multi-slab encoding in conjunction with multi-shot slab-segmentation to accelerate 3D encoding and to reduce the duration of EVI encoding within slabs. This approach combines the sampling efficiency of single-shot 3D encoding with the sensitivity advantage of multi-echo acquisition. We describe the pulse sequence development and characterize the spatial–temporal resolution limits and BOLD sensitivity of this approach for high-speed task-based and resting-state fMRI at 3 T. We study the feasibility of further acceleration using compressed sensing (CS) and assess compatibility with NORDIC denoising.
Methods
Multi-band echo volumar imaging (MB-EVI) combines multi-band encoding of up to 6 slabs with CAIPI shifting, accelerated EVI encoding within slabs using up to 4-fold GRAPPA accelerations, 2-shot kz-segmentation and partial Fourier acquisitions along the two phase-encoding dimensions. Task-based and resting-state fMRI at 3 Tesla was performed across a range of voxel sizes (between 1 and 3 mm isotropic), repetition times (118–650 ms), and number of slabs (up to 12). MB-EVI was compared with multi-slab EVI (MS-EVI) and multi-band-EPI (MB-EPI).
Results
Image quality and temporal SNR of MB-EVI was comparable to MS-EVI when using 2–3 mm spatial resolution. High sensitivity for mapping task-based activation and resting-state connectivity at short TR was measured. Online deconvolution of T2* signal decay markedly reduced spatial blurring and improved image contrast. The high temporal resolution of MB-EVI enabled sensitive mapping of high-frequency resting-state connectivity above 0.3 Hz with 3 mm isotropic voxel size (TR: 163 ms). Detection of task-based activation with 1 mm isotropic voxel size was feasible in scan times as short as 1 min 13 s. Compressed sensing with up to 2.4-fold retrospective undersampling showed negligible loss in image quality and moderate region-specific losses in BOLD sensitivity. NORDIC denoising significantly enhanced fMRI sensitivity without introducing image blurring.
Conclusion
Combining MS-EVI with multi-band encoding enables high overall acceleration factors and provides flexibility for maximizing spatial–temporal resolution and volume coverage. The high BOLD sensitivity of this hybrid MB-EVI approach and its compatibility with online image reconstruction enables high spatial–temporal resolution real-time task-based and resting state fMRI.
... Increasingly available high-speed fMRI methods, such as simultaneous multislice echo-planar imaging (SMS-EPI) (Feinberg et al., 2010;Setsompop et al., 2012), magnetic resonance encephalography (Zahneisen et al., 2011), and multislab echo-volumar imaging (MS-EVI) (Posse et al., 2012), enable unaliased measurement of physiological noise to increase sensitivity in resting-state fMRI. However, few realtime fMRI analysis tools offer the processing performance required for high-speed fMRI. ...
... MS-EVI (Posse et al., 2012) was acquired in five healthy controls using the following: TR/TE (echo time)-136/28 msec, flip angle-10°, number of scans-2276, 2 slabs, spatial matrix-64 · 64 · 8 · 2, voxel size-4 · 4 · 6 mm 3 , interslab gap-10%, generalized autocalibrating partial parallel acquisition acceleration-4, k y partial Fourier encoding-6/8, and oversampling along the slab-direction (7/8) resulting in 13 reconstructed slices, scan time: 5:16 min. The in-plane image reconstruction with parallel imaging reconstruction was performed on the scanner. ...
... The in-plane image reconstruction with parallel imaging reconstruction was performed on the scanner. The through-plane reconstruction of MS-EVI was performed online (Posse et al., 2012) on an external Linux workstation using our custom TurboFIRE (Turbo Functional Imaging in Real-time) fMRI analysis tool (Gembris et al., 2000;Posse et al., 2013). SMS-EPI was acquired in six healthy controls using the following: TR/TE-400/35 msec, flip angle-42°, number of scans: 900, number of slices: 32, spatial matrix-64 · 64 · 32, voxel size-3 · 3 · 3 mm 3 , interslice gap-0 mm, SMS factor-8, and scan time: 6:08 min. ...
Background / Introduction: There is considerable interest in using real-time fMRI for monitoring functional connectivity dynamics. To date, the majority of real-time resting-state fMRI studies have examined limited number of brain regions. This is in part due to the computational demands of traditional seed- and ICA-based methods, in particular when using increasingly available high-speed fMRI methods.
Methods:
This study describes a computationally efficient real-time seed-based resting-state fMRI analysis pipeline using moving averaged sliding-windows with partial correlations and regression of motion parameters and signals from white matter and cerebrospinal fluid.
Results:
Analytical and numerical analyses of averaged sliding-window correlation and sliding-window regression as a function of window width show selectable bandpass filter characteristics and effective suppression of artifactual correlations resulting from signal drifts and transients. The analysis pipeline is compatible with multi-slab echo-volumar imaging and simultaneous multi-slice echo-planar imaging with repetition times as short as 136 ms. High-speed resting-state fMRI data in healthy controls demonstrate the effectiveness of this approach for minimizing artifactual correlations in white and gray matter, which was comparable to conventional regression across the entire scan. Integrating sliding-window averaging (width: W1) within a 2nd level sliding-window (width: W2) enabled monitoring of intra- and inter-network correlation dynamics of up to 12 resting-state networks with bandpass filter characteristics determined by the 1st level sliding-window and temporal resolution W1+W2.
Conclusions:
The computational performance and confound tolerance make this seed-based resting-state fMRI approach suitable for real-time monitoring of data quality and resting-state connectivity dynamics in neuroscience and clinical research studies.
... These signals are in the range of 0.1 − 0.3 Hz and 1 − 2 Hz, respectively, and are aliased into the low frequency band for low sampling rates of 1 − 3 s used with conventional MRI. The direct measurement of these signals with fast fMRI sequences allows to employ some methodology to filter these signals out from the desired BOLD signal (Hennig et al., 2007;Posse et al., 2012;Lin et al., 2012a). Several methods were developed for a fast acquisition of fMRI BOLD data (Witzel et al., 2008;Feinberg et al., 2010;Lin et al., 2012b;Posse et al., 2012). ...
... The direct measurement of these signals with fast fMRI sequences allows to employ some methodology to filter these signals out from the desired BOLD signal (Hennig et al., 2007;Posse et al., 2012;Lin et al., 2012a). Several methods were developed for a fast acquisition of fMRI BOLD data (Witzel et al., 2008;Feinberg et al., 2010;Lin et al., 2012b;Posse et al., 2012). The fast fMRI sequence most widely used is simultaneous multislice echo planar imaging (SMS-EPI) with a spatial resolution of 3mm isotropic voxel size and a temporal resolution of down to 0.4 s for a 3D scan (Feinberg and Setsompop, 2013). ...
... However, using a vector autoregressive process to fit the data can be problematic as fMRI signals typically have a temporal resolution of 1 − 3 s (Lin et al., 2014), whereas characteristic time scales of neuronal processes are in the order of tens to hundreds of milliseconds. "Although MR acquisition sequences with faster temporal resolutions are becoming increasingly common (Feinberg et al., 2010;Posse et al., 2012;, neuronal processes still undergo considerable down-sampling in fMRI time series, affecting the reliability of GC estimates (Seth et al., 2013;Friston et al., 2014a). Moreover, temporal relationships between cerebral areas are confounded by the spatial variability of the haemodynamic response function (subsubsection 6.2.2.2 'resting-state fMRI data' in 'Results' and(Handwerker et al., 2004, 2012))." ...
The brain is a huge network with outstanding computational capabilities. Understanding the brain network and its principle of operation is of high interest. Therefore, brain activity can be measured to retrieve information about the brains functioning using functional magnetic resonance imaging, which is a great non-invasive tool to measure the neuronal activity. However, with this method the neuronal activity is measured only indirectly as the desired information is distorted by a specific low-pass filter, which differs in its form in different brain regions.
In this thesis a methodology is presented to estimate this low pass filter for every brain region. Thereby, a high variability of this filter function throughout the brain is found. With the knowledge about the filter the data can be corrected for relative timing offsets between brain regions to yield a good estimate of the underlying neuronal activity. Using the corrected data, further a novel methodology is presented to estimate the sparse directed brain connectivity, which is applied to resting-state fMRI data of healthy subjects, where it proves superior to conventional undirected connectivity analysis.
... A segmented multishot approach to 3D Cartesian readouts with high flexibility in the acquisition and acceleration schemes at the cost of increased physiological noise is 3D EPI. 5,6 Three-dimensional Cartesian single-shot readouts of 1 or multiple slabs per brain have been realized through echo volumar imaging, [7][8][9] which was shown to enhance BOLD sensitivity compared with EPI but offers less spatial resolution than simultaneous multislice EPI. In the range of sub-100 ms approaches, notable results have been achieved by inverse imaging, 10,11 which involves omitting of encoding lines in 1 dimension while compromising the point spread function of the imaging process, and MR encephalography. ...
... The high temporal resolution can enable unaliasing and correction of physiological fluctuations, allow for characterization of subject-dependent and region of interest (ROI)dependent hemodynamic response function waveforms, and leads to increased sensitivity for mapping task-based activation and functional connectivity as well as for detecting dynamic changes in connectivity over time. 1,8,16,17 The larger number of data points also holds the potential for faster and/ or more accurate detection of resting-state networks. 18,19 Moreover, BOLD signal changes have been shown to contain measurable fluctuations at frequencies up to 5 Hz, which are not accessible by conventional whole brain EPI. ...
... The simplification of the signal model expressed by Equation (8), however, does not consider nonconstant signal fluctuations within the ROIs. Resulting deviations in time series from calculating Equation (11) as opposed to the introduced Equation (5) are case-dependent and cannot be generally quantified (an application example is shown subsequently). ...
Purpose
A partial image reconstruction formalism is introduced for the targeted extraction of real‐time feedback from arbitrary trajectories when full image reconstruction in real time is computationally too demanding.
Methods
Explicit calculation and storage of linear combinations of lines of the reconstruction matrix by an incomplete basis change in spatial coordinates lead to translation of the expensive full reconstruction from a frame‐wise application to a region of interest (ROI)‐wise application. This step is independent from signal data and can be executed before the experiment. Subsequently, the results of the sum over fully reconstructed voxels can be evaluated directly. Data from a high‐speed fMRI acquisition was used to investigate the targeted partial reconstruction of a functional ROI atlas, incorporating an intravolume dephasing correction. The same data and ROIs were used for a comparison of the time series obtained with those obtained from already existing methods for compartment‐wise reconstruction. To examine real‐time feasibility, the reconstruction was implemented and tested for online reconstruction performance.
Results
The reconstruction yields results that are virtually identical to the standard reconstruction (i.e., the magnitude sums over the ROIs), with negligible discrepancies even after termination of the conjugate gradient algorithm at a feasible number of iterations. Notably, more discrepancies arise with existing compartment‐wise reconstructions. The online real‐time implementation evaluated 1 ROI within 2.8 ms in the case of a highly parallel 3D whole brain acquisition.
Conclusion
The high reconstruction fidelity and speed are satisfying for the exemplary application of real‐time functional feedback using a highly parallel 3D whole brain acquisition.
... To avoid adaptation, most olfactory fMRI studies employ either eventrelated designs with short odorous stimulation or block designs with short odorous pulses incorporated in each block of stimulation. Furthermore, short TR results in more accurate temporal resolution of the BOLD response, which has led to the development of novel acquisition techniques that achieve high temporal resolution by reducing the TR to <1 s (Dilharreguy et al. 2003;Lin et al. 2006;van der Zwaag et al. 2006;Feinberg et al. 2010;Posse et al. 2012;Witt et al. 2016). To our knowledge, the effects of fast sampling rate (short TR) on olfactory fMRI data have not been previously studied. ...
... peak time determination increases with short TR (Miezin et al. 2000;Dilharreguy et al. 2003). Moreover, short TR is associated with better resolution of heartbeat related physiological signal fluctuations and reduced sensitivity to intrascan head motion (Posse et al. 2012;Smith et al. 2013). Therefore, several novel acquisition techniques have been developed, aiming to increase the sampling rate of the BOLD signal and indicating significant benefits of collecting fMRI data with TR <1 s (Lin et al. 2006;van der Zwaag et al. 2006;Feinberg et al. 2010;Posse et al. 2012;Witt et al. 2016). ...
... Moreover, short TR is associated with better resolution of heartbeat related physiological signal fluctuations and reduced sensitivity to intrascan head motion (Posse et al. 2012;Smith et al. 2013). Therefore, several novel acquisition techniques have been developed, aiming to increase the sampling rate of the BOLD signal and indicating significant benefits of collecting fMRI data with TR <1 s (Lin et al. 2006;van der Zwaag et al. 2006;Feinberg et al. 2010;Posse et al. 2012;Witt et al. 2016). In particular, faster TRs achieved with multi-slice EPI sequences (as the sequence used in this study) can capture more information per time unit, allowing more accurate representation of the BOLD response (Chen et al. 2015). ...
Studying olfaction with functional magnetic resonance imaging (fMRI) poses various methodological challenges. This study aimed to investigate the effects of stimulation length and repetition time (TR) on the activation pattern of 4 olfactory brain regions: the anterior and the posterior piriform cortex, the orbitofrontal cortex, and the insula. Twenty-two healthy participants with normal olfaction were examined with fMRI, with 2 stimulation lengths (6 s and 15 s) and 2 TRs (0.901 s and 1.34 s). Data were analyzed using General Linear Model (GLM), Tensorial Independent Component Analysis (TICA), and by plotting the event-related time course of brain activation in the 4 olfactory regions of interest. The statistical analysis of the time courses revealed that short TR was associated with more pronounced signal increase and short stimulation was associated with shorter time to peak signal. Additionally, both long stimulation and short TR were associated with oscillatory time courses, whereas both short stimulation and short TR resulted in more typical time courses. GLM analysis showed that the combination of short stimulation and short TR could result in visually larger activation within these olfactory areas. TICA validated that the tested paradigm was spatially and temporally associated with a functionally connected network that included all 4 olfactory regions. In conclusion, the combination of short stimulation and short TR is associated with higher signal increase and shorter time to peak, making it more amenable to standard GLM-type analyses than long stimulation and long TR, and it should, thus, be preferable for olfactory fMRI.
... Alternative techniques for ultrafast EPI acquisitions have been proposed. These include inverse imaging (InI) (Lin et al., 2012), generalized inverse imaging (GIN) (Boyacioglu et al., 2013), and multislab echo-volumar imaging (multi-slab EVI or MEVI) (Posse et al., 2012(Posse et al., , 2013. These techniques offer a fast-sampling rate and reduced sensitivity to physiological noise. ...
... These techniques offer a fast-sampling rate and reduced sensitivity to physiological noise. However, they come with the trade-off of potential loss of spatial resolution or introduction of geometrical distortions (Lin et al., 2012;Posse et al., 2012Posse et al., , 2013Boyacioglu et al., 2013). Moreover, these studies have primarily utilized 3T MRI scanners. ...
The development of innovative non-invasive neuroimaging methods and biomarkers is critical for studying brain disease. Imaging of cerebrospinal fluid (CSF) pulsatility may inform the brain fluid dynamics involved in clearance of cerebral metabolic waste. In this work, we developed a methodology to characterize the frequency and spatial localization of whole brain CSF pulsations in humans. Using 7 Tesla (T) human magnetic resonance imaging (MRI) and ultrafast echo-planar imaging (EPI), in-vivo images were obtained to capture pulsations of the CSF signal. Physiological data were simultaneously collected and compared with the 7 T MR data. The primary components of signal pulsations were identified using spectral analysis, with the most evident frequency bands identified around 0.3, 1.2, and 2.4 Hz. These pulsations were mapped spatially and temporally onto the MR image domain and temporally onto the physiological measures of electrocardiogram and respiration. We identified peaks in CSF pulsations that were distinct from peaks in grey matter and white matter regions. This methodology may provide novel in vivo biomarkers of disrupted brain fluid dynamics.
... However, it should be noted that others have highlighted the need for higher sampling frequencies in fMRI (TR < 1 s) studies, regardless of fractal analysis, as they result in more accurate modeling of the hemodynamic response function (Dilharreguy et al., 2003), less fluctuations due to head movement during scans (Smith et al., 2013), and better identification of physiologic confounds (Posse et al., 2012). ...
... Given the vast benefits, there have been major advancements in acquisition techniques aimed at increasing the temporal sampling of BOLD fMRI data, such as echo-volumar imaging, inverse imaging, and multiplexed echo-planar imaging (Posse et al., 2012). While the continuous development of acquisition methods will advance the field of fMRI fractal analysis, individual efforts to increase the length of scans and sampling frequency should be made for optimal fractal analysis. ...
The following review will aid readers in providing an overview of scale-free dynamics and monofractal analysis, as well as its applications and potential in functional magnetic resonance imaging (fMRI) neuroscience and clinical research. Like natural phenomena such as the growth of a tree or crashing ocean waves, the brain expresses scale-invariant, or fractal, patterns in neural signals that can be measured. While neural phenomena may represent both monofractal and multifractal processes and can be quantified with many different interrelated parameters, this review will focus on monofractal analysis using the Hurst exponent (H). Monofractal analysis of fMRI data is an advanced analysis technique that measures the complexity of brain signaling by quantifying its degree of scale-invariance. As such, the H value of the blood oxygenation level-dependent (BOLD) signal specifies how the degree of correlation in the signal may mediate brain functions. This review presents a brief overview of the theory of fMRI monofractal analysis followed by notable findings in the field. Through highlighting the advantages and challenges of the technique, the article provides insight into how to best conduct fMRI fractal analysis and properly interpret the findings with physiological relevance. Furthermore, we identify the future directions necessary for its progression towards impactful functional neuroscience discoveries and widespread clinical use. Ultimately, this presenting review aims to build a foundation of knowledge among readers to facilitate greater understanding, discussion, and use of this unique yet powerful imaging analysis technique.
... However, this approach perturbs spins outside the selected intersection, limiting subsequent selections in other regions when multiple 3D zoomed volumes are desired, such as in 3D multi-slab imaging. [22][23][24] In addition, the reliance on two or more RF pulses makes the method incompatible with GRE-EPI for neurofunctional imaging. Another approach of achieving 3D rFOVI is outer-volume suppression (OVS). ...
... Our analysis showed that up to 25% reduction in SNR could occur when compared to the scenario of using a 90° flip angle. To improve SNR efficiency and increase spatial coverage, a 3D multi-slab technique (a hybrid between 2D and 3D imaging), [22][23][24] may be incorporated into 3D-rFOVI in future studies. Another limitation of the study is that the FOV reduction was demonstrated only along the phase-encoded and slab-selection directions. ...
Purpose
This study aimed at developing a 3D reduced field‐of‐view imaging (3D‐rFOVI) technique using a 2D radiofrequency (RF) pulse, and demonstrating its ability to achieve isotropic high spatial resolution and reduced image distortion in echo planar imaging (EPI).
Methods
The proposed 3D‐rFOVI technique takes advantage of a 2D RF pulse to excite a slab along the conventional slice‐selection direction (i.e., z‐direction) while limiting the spatial extent along the phase‐encoded direction (i.e., y‐direction) within the slab. The slab is phase‐encoded in both through‐slab and in‐slab phase‐encoded directions. The 3D‐rFOVI technique was implemented at 3T in gradient‐echo and spin‐echo EPI pulse sequences for functional MRI (fMRI) and diffusion‐weighted imaging (DWI), respectively. 3D‐rFOVI experiments were performed on a phantom and human brain to illustrate image distortion reduction, as well as isotropic high spatial resolution, in comparison with 3D full‐FOV imaging.
Results
In both the phantom and the human brain, image voxel dislocation was substantially reduced by 3D‐rFOVI when compared with full‐FOV imaging. In the fMRI experiment with visual stimulation, 3D isotropic spatial resolution of (2 × 2 × 2 mm³) was achieved with an adequate signal‐to‐noise ratio (81.5) and blood oxygen level‐dependent (BOLD) contrast (2.5%). In the DWI experiment, diffusion‐weighted brain images with an isotropic resolution of (1 × 1 × 1 mm³) was obtained without appreciable image distortion.
Conclusion
This study indicates that 3D‐rFOVI is a viable approach to 3D neuroimaging over a zoomed region.
... Cartesian single-shot 3D acquisition-echo volumar imaging or 3D-EPI-has been around a long time [27] and also been used for fMRI [28]. It doesn't quite reach the temporal performance of MREG, but in combination with controlled aliasing and multi-slab acquisition short repetition times down to 371 ms have been reached [29]. ...
... Yet, the proposed models and subsequent implementations in fMRI statistical analysis packages only considered temporal autocorrelations in conventional, low temporal resolution fMRI data. Initial activation studies using fast fMRI data thus reported very large statistical scores that were, however, somewhat inflated [15,28].We thus started to perform subsequent data analyses using the FMRISTAT toolbox, which is not widely used, but which allows modeling the noise as a spatially varying, high-order autoregressive process [54,55]. This provided a well-fitting model of MREG time series autocorrelations while reducing the inflated statistical scores, which were nevertheless still ~ 60% higher than the scores obtained with conventional EPI [52]. ...
Objective
This review article gives an account of the development of the MR-encephalography (MREG) method, which started as a mere ‘Gedankenexperiment’ in 2005 and gradually developed into a method for ultrafast measurement of physiological activities in the brain. After going through different approaches covering k-space with radial, rosette, and concentric shell trajectories we have settled on a stack-of-spiral trajectory, which allows full brain coverage with (nominal) 3 mm isotropic resolution in 100 ms. The very high acceleration factor is facilitated by the near-isotropic k-space coverage, which allows high acceleration in all three spatial dimensions.
Methods
The methodological section covers the basic sequence design as well as recent advances in image reconstruction including the targeted reconstruction, which allows real-time feedback applications, and—most recently—the time-domain principal component reconstruction (tPCR), which applies a principal component analysis of the acquired time domain data as a sparsifying transformation to improve reconstruction speed as well as quality.
Applications
Although the BOLD-response is rather slow, the high speed acquisition of MREG allows separation of BOLD-effects from cardiac and breathing related pulsatility. The increased sensitivity enables direct detection of the dynamic variability of resting state networks as well as localization of single interictal events in epilepsy patients. A separate and highly intriguing application is aimed at the investigation of the glymphatic system by assessment of the spatiotemporal patterns of cardiac and breathing related pulsatility.
Discussion
MREG has been developed to push the speed limits of fMRI. Compared to multiband-EPI this allows considerably faster acquisition at the cost of reduced image quality and spatial resolution.
... To avoid adaptation, most olfactory fMRI studies rely on either event-related designs with short odorous stimulation or block designs with short odorous pulses incorporated in each block of stimulation. Furthermore, fast sampling rate results in more accurate temporal resolution of the BOLD response, which has led to the development of novel acquisition techniques that achieve high temporal resolution by reducing the TR to <1 s [98,[102][103][104][105][106]. Therefore, several factors should be considered when designing an fMRI study for the investigation of neural activity within the olfactory network. ...
... showing that the accuracy of HRF peak time determination increases with short TR [98]. Several novel acquisition techniques have, therefore, been developed, aiming to increase the sampling rate of the BOLD signal and indicating significant benefits of collecting fMRI data with TR < 1 s [102][103][104][105][106]. In particular, faster TRs achieved with multi-slice echo planar imaging sequences (as the sequence used in this study) can capture more information per time unit, allowing more accurate representation of the BOLD response [120]. ...
... The past decade has seen the rapid development of fast fMRI sequences. By increasing the temporal resolution of the acquisition, these sequences improve the physiological noise correction and increase the statistical power of the fMRI data analyses (Jacobs et al., 2014a;Posse et al., 2012a). Techniques such as MR-Encephalography (Zahneisen et al., 2012), ...
... Although fast acquisitions are not necessary to sample the low frequency (<0.1 Hz) fluctuations associated with most of the BOLD activity, there is a fast growing literature on fast acquisitions which are primarily beneficial to correct aliased high frequency physiological noise (Hennig et al., 2007;Lin et al., 2012;Posse et al., 2012a).Furthermore, there is also relevant RSN information at higher frequencies (Boubela et al., 2013;Chen and Glover, 2015;Gohel and Biswal, 2015;Kalcher et al., 2014Kalcher et al., , 2014Lee et al., 2013a;Lin et al., 2015Lin et al., , 2015Niazy et al., 2011;Yuan et al., 2014) and the increased number of time points improves the sensitivity to map the RSNs compared to conventional sequences (Feinberg et al., 2010). ...
Die funktionelle Magnetresonanztomographie hat in den Bereichen Hirnbildgebung und Kognitionswissenschaften zunehmend an Bedeutung gewonnen. Die Verschiebung der zeitlichen Auflösung in Richtung von hundert Millisekunden ermöglicht die dynamische Untersuchung der Hirnnetzwerke. In dieser Dissertation wird ein ultraschnelles bildgebendes Verfahren namens Magnetresonanz-Enzephalographie (MREG) verwendet, um verschiedene Vorteile der erhöhten
zeitlichen Auflösung bei der Analyse von Ruhezustandsnetzwerken (RSN) zu untersuchen. Zunächst wurde die Fähigkeit der Netzwerkextraktion unter verschiedenen Scan-Dauern und die Anzahl der detektierten Netzwerke untersucht. Eine verbesserte Detektion von RSNs in kurzen Scanzeiten ermöglicht zuverlässige dynamische Konnektivitätsanalysen. Die Dynamik der
Hirnverbindungen wird durch die Varianz der funktionalen Konnektivität quantifiziert, die durch eine sliding window Analyse verfolgt wird. Durch diese Maßnahme werden einige transiente, aber konsistente Interaktionen der Hirnregionen identifiziert, die durch konventionelle Analysen nicht nachweisbar wären. In dieser Arbeit wird das Potential dieser Ergebnisse diskutiert und einige vorläufige klinische Anwendungen vorgestellt.
... With fast fMRI, researchers have been able to probe neural oscillations at frequencies well above the limits of conventional acquisitions (e.g., (Lee et al., 2013;Lewis et al., 2016)) and characterize brain temporal dynamics at much finer temporal resolutions (e.g., (Lewis et al., 2018;Lin et al., 2018;Smith et al., 2012)). Furthermore, additional temporal samples are achieved without increasing the scan duration, which is commonly thought to be advantageous for enhanced sensitivity to neural fluctuations (e.g., (Feinberg et al., 2010;Posse et al., 2012;Smith et al., 2013)). ...
... To examine whether sub-second TRs are superior to conventional long-TR acquisitions, several studies have empirically evaluated the performance of fast acquisitions in specific cases and analysis strategies, which led to somewhat conflicting observations. For instance, a few task-based studies showed that sub-second TRs could lead to enhanced sensitivity to neural activation, evidenced by higher statistical scores in general linear model (GLM)-based sensory task activation analyses McDowell and Carmichael, 2018;Posse et al., 2012), detection of additional task clusters when combined with multi-echo acquisition (Boyacioglu et al., 2015), enhanced classification accuracy of complex cognitive states Demetriou et al., 2018), and better correspondence with epileptic spikes identified by concurrent EEG recordings (Jacobs et al., 2014). However, a recent study assessed brain activation under a broad range of task types and showed that the benefits of fast acquisitions analyzed with GLM-based approaches were very modest (Demetriou et al., 2018). ...
Recent advances in parallel imaging and simultaneous multi-slice techniques have permitted whole-brain fMRI acquisitions at sub-second sampling intervals, without significantly sacrificing the spatial coverage and resolution. Apart from probing brain function at finer temporal scales, faster sampling rates may potentially lead to enhanced functional sensitivity, owing possibly to both cleaner neural representations (due to less aliased physiological noise) and additional statistical benefits (due to more degrees of freedom for a fixed scan duration). Accompanying these intriguing aspects of fast acquisitions, however, confusion has also arisen regarding (1) how to preprocess/analyze these fast fMRI data, and (2) what exactly is the extent of benefits with fast acquisitions, i.e., how fast is fast enough for a specific research aim? The first question is motivated by the altered spectral distribution and noise characteristics at short sampling intervals, while the second question seeks to reconcile the complicated trade-offs between the functional contrast-to-noise ratio and the effective degrees of freedom. Although there have been recent efforts to empirically approach different aspects of these two questions, in this work we discuss, from a theoretical perspective accompanied by some illustrative, proof-of-concept experimental in vivo human fMRI data, a few considerations that are rarely mentioned, yet are important for both preprocessing and optimizing statistical inferences for studies that employ acquisitions with sub-second sampling intervals. Several summary recommendations include concerns regarding advisability of relying on low-pass filtering to de-noise physiological contributions, employment of statistical models with sufficient complexity to account for the substantially increased serial correlation, and cautions regarding using rapid sampling to enhance functional sensitivity given that different analysis models may associate with distinct trade-offs between contrast-to-noise ratios and the effective degrees of freedom. As an example, we demonstrate that as TR shortens, the intrinsic differences in how noise is accommodated in general linear models and Pearson correlation analyses (assuming Gaussian distributed stochastic signals and noise) can result in quite different outcomes, either gaining or losing statistical power.
... However, the reliability of fast fMRI remains unclear. On the one hand, acquiring a higher number of timepoints without increasing the scanning duration enhances statistical power (Dowdle et al. 2021;Feinberg et al. 2010;Posse et al. 2012;Smith et al. 2013). On the other hand, speed comes at the cost of lower signal-to-noise ratio per time frame (Barth et al. 2016;Boubela et al. 2014;Edelstein et al. 1986;Feinberg and Setsompop 2013;Preibisch et al. 2015) with respect to conventional fMRI protocols (TR ~2-3 s), with the ultimate result of decreasing the statistical validity of inferences about intrinsic FC (Corbin et al. 2018). ...
Functional brain fingerprinting has emerged as an influential tool to quantify reliability in neuroimaging studies and to identify cognitive biomarkers in both healthy and clinical populations. Recent studies have revealed that brain fingerprints reside in the timescale‐specific functional connectivity of particular brain regions. However, the impact of the acquisition's temporal resolution on fingerprinting remains unclear. In this study, we examine for the first time the reliability of functional fingerprinting derived from resting‐state functional MRI (rs‐fMRI) with different whole‐brain temporal resolutions (TR = 0.5, 0.7, 1, 2, and 3 s) in a cohort of 20 healthy volunteers. Our findings indicate that subject identifiability within a fixed TR is successful across different temporal resolutions, with the highest identifiability observed at TR 0.5 and 3 s (TR(s)/identifiability(%): 0.5/64; 0.7/47; 1/44; 2/44; 3/56). We discuss this observation in terms of protocol‐specific effects of physiological noise aliasing. We further show that, irrespective of TR, associative brain areas make an higher contribution to subject identifiability (functional connections with highest mean ICC: within subcortical network [SUB; ICC = 0.0387], within default mode network [DMN; ICC = 0.0058]; between DMN and somato‐motor [SM] network [ICC = 0.0013]; between ventral attention network [VA] and DMN [ICC = 0.0008]; between VA and SM [ICC = 0.0007]), whereas sensory‐motor regions become more influential when integrating data from different TRs (functional connections with highest mean ICC: within fronto‐parietal network [ICC = 0.382], within dorsal attention network [DA; ICC = 0.373]; within SUB [ICC = 0.367]; between visual network [VIS] and DA [ICC = 0.362]; within VIS [ICC = 0.358]). We conclude that functional connectivity fingerprinting derived from rs‐fMRI holds significant potential for multicentric studies also employing protocols with different temporal resolutions. However, it remains crucial to consider fMRI signal's sampling rate differences in subject identifiability between data samples, in order to improve reliability and generalizability of both whole‐brain and specific functional networks' results. These findings contribute to a better understanding of the practical application of functional connectivity fingerprinting, and its implications for future neuroimaging research.
... For instance, simultaneous multi-slice (SMS) acquisitions (10) such as multiband techniques (11; 12) reach whole-brain coverage in 0.5-1s and with around 2mm isotropic spatial resolution (9). Often, acceleration techniques decrease the signal-to-noise ratio and increase the likelihood of degrading artifacts (13; 14), thereby limiting data quality in research and clinical applications. For example, a recently proposed acquisition approach based on a 3D radial sampling trajectory (15) improved BOLD sensitivity but still required compromises between temporal and spatial resolution. ...
Functional magnetic resonance imaging (fMRI) is a methodological cornerstone of neuroscience. Most studies measure blood-oxygen-level-dependent (BOLD) signal using echo-planar imaging (EPI), Cartesian sampling, and image reconstruction with a one-to-one correspondence between the number of acquired volumes and reconstructed images. However, EPI schemes are subject to trade-offs between spatial and temporal resolutions. We overcome these limitations by measuring BOLD with a gradient recalled echo (GRE) with 3D radial-spiral phyllotaxis trajectory at a high sampling rate (28.24ms) on standard 3T field-strength. The framework enables the reconstruction of 3D signal time courses with whole-brain coverage at simultaneously higher spatial (1mm 3) and temporal (up to 250ms) resolutions, as compared to optimized EPI schemes. Additionally, artifacts are corrected before image reconstruction; the desired temporal resolution is chosen after scanning and without assumptions on the shape of the hemodynamic response. By showing activation in the calcarine sulcus of 20 participants performing an ON-OFF visual paradigm, we demonstrate the reliability of our method for cognitive neuroscience research.
... A series of 2D/3D hybrid techniques have been proposed to reduce the TR and thereby improve the SNR efficiency, including 2D simultaneous multi-slice (SMS) (10)(11)(12)(13)(14)(15)(16)(17)(18), 3D multislab (9,(19)(20)(21)(22)(23)(24)(25)(26)(27), and simultaneous multi-slab (SMSlab) (28)(29)(30)(31) techniques. 2D SMS excites multiple slices simultaneously, encodes within each slice using conventional 2D k-space acquisitions, and separates slices in reconstruction by leveraging the coil sensitivities in the slice direction. ...
Purpose: This study aims to propose a model-based reconstruction algorithm for simultaneous multi-slab diffusion MRI acquired with blipped-CAIPI gradients (blipped-SMSlab), which can also incorporate distortion correction. Methods: We formulate blipped-SMSlab in a 4D k-space with kz gradients for the intra-slab slice encoding and km (blipped-CAIPI) gradients for the inter-slab encoding. Because kz and km gradients share the same physical axis, the blipped-CAIPI gradients introduce phase interference in the z-km domain while motion induces phase variations in the kz-m domain. Thus, our previous k-space-based reconstruction would need multiple steps to transform data back and forth between k-space and image space for phase correction. Here we propose a model-based hybrid-space reconstruction algorithm to correct the phase errors simultaneously. Moreover, the proposed algorithm is combined with distortion correction, and jointly reconstructs data acquired with the blip-up/down acquisition to reduce the g-factor penalty. Results: The blipped-CAIPI-induced phase interference is corrected by the hybrid-space reconstruction. Blipped-CAIPI can reduce the g-factor penalty compared to the non-blipped acquisition in the basic reconstruction. Additionally, the joint reconstruction simultaneously corrects the image distortions and improves the 1/g-factors by around 50%. Furthermore, through the joint reconstruction, SMSlab acquisitions without the blipped-CAIPI gradients also show comparable correction performance with blipped-SMSlab. Conclusion: The proposed model-based hybrid-space reconstruction can reconstruct blipped-SMSlab diffusion MRI successfully. Its extension to a joint reconstruction of the blip-up/down acquisition can correct EPI distortions and further reduce the g-factor penalty compared with the separate reconstruction.
... We used a relatively common sequence to estimate the cardiac pulse waveforms. Other rapid MRI sequences have been designed to sample BOLD signals at fast rates, including multi-slab echo-volumar imaging, (Posse et al., 2012(Posse et al., , 2013, ultrafast generalized inverse imaging (Boyacioglu et al., 2013), magnetic resonance inverse imaging (Lin et al., 2012), and MREG (Lee et al., 2013). These rapid sequences facilitate filtering out physiological noise and estimating the BOLD signal in resting-state networks. ...
Blood and cerebrospinal fluid (CSF) pulse and flow throughout the brain, driven by the cardiac cycle. These fluid dynamics, which are essential to healthy brain function, are characterized by several noninvasive magnetic resonance imaging (MRI) methods. Recent developments in fast MRI, specifically simultaneous multislice acquisition methods, provide a new opportunity to rapidly and broadly assess cardiac-driven flow, including CSF spaces, surface vessels and parenchymal vessels. We use these techniques to assess blood and CSF flow dynamics in brief (3.5 min) scans on a conventional 3 T MRI scanner in five subjects. Cardiac pulses are measured with a photoplethysmography (PPG) on the index finger, along with functional MRI (fMRI) signals in the brain. We, retrospectively, align the fMRI signals to the heartbeat. Highly reliable cardiac-gated fMRI temporal signals are observed in CSF and blood on the timescale of one heartbeat (test-retest reliability within subjects R2 > 50%). In blood vessels, a local minimum is observed following systole. In CSF spaces, the ventricles and subarachnoid spaces have a local maximum following systole instead. Slower resting-state scans with slice timing, retrospectively, aligned to the cardiac pulse, reveal similar cardiac-gated responses. The cardiac-gated measurements estimate the amplitude and phase of fMRI pulsations in the CSF relative to those in the arteries, an estimate of the local intracranial impedance. Cardiac aligned fMRI signals can provide new insights about fluid dynamics or diagnostics for diseases where these dynamics are important.
... C rtCNR in bilateral amygdala and dmPFC. Error bars denote standard deviation 250.5 ± 90.4 3.3 ± 3.9 3.0 ± 1.8 0.7 ± 0.5 0.6 ± 2.0 higher visual 239.6 ± 149.5 3.0 ± 3.4 3.5 ± 2.9 1.0 ± 1.2 0.8 ± 2.3 precuneus 191.6 ± 70.1 3.7 ± 3.7 3.1 ± 2.4 1.2 ± 0.8 0.4 ± 2.9 primary visual 170.8 ± 88.7 3.4 ± 4.1 3.8 ± 2.2 1.8 ± 1.9 1.1 ± 3.5 sensorimotor 289.6 ± 123.9 4.5 ± 3.8 4.1 ± 3.0 0.6 ± 0.5 0.7 ± 2.0 ventral DMN 259.6 ± 113.3 4.3 ± 4.0 2.5 ± 1.8 0.6 ± 0.4 0.5 ± 2.0 visuospatial 303.2 ± 124.7 4.1 ± 4.4 3.5 ± 3.1 0.5 ± 0.3 0.9 ± 2.1 TR = 2 s) (Posse et al., 2012), while considering different acquisition parameters and head coils (Triantafyllou et al., 2011;Welvaert & Rosseel, 2013). ...
Real-time quality assessment (rtQA) of functional magnetic resonance imaging (fMRI) based on blood oxygen level-dependent (BOLD) signal changes is critical for neuroimaging research and clinical applications. The losses of BOLD sensitivity because of different types of technical and physiological noise remain major sources of fMRI artifacts. Due to difficulty of subjective visual perception of image distortions during data acquisitions, a comprehensive automatic rtQA is needed. To facilitate rapid rtQA of fMRI data, we applied real-time and recursive quality assessment methods to whole-brain fMRI volumes, as well as time-series of target brain areas and resting-state networks. We estimated recursive temporal signal-to-noise ratio (rtSNR) and contrast-to-noise ratio (rtCNR), and real-time head motion parameters by a framewise rigid-body transformation (translations and rotations) using the conventional current to template volume registration. In addition, we derived real-time framewise (FD) and micro (MD) displacements based on head motion parameters and evaluated the temporal derivative of root mean squared variance over voxels (DVARS). For monitoring time-series of target regions and networks, we estimated the number of spikes and amount of filtered noise by means of a modified Kalman filter. Finally, we applied the incremental general linear modeling (GLM) to evaluate real-time contributions of nuisance regressors (linear trend and head motion). Proposed rtQA was demonstrated in real-time fMRI neurofeedback runs without and with excessive head motion and real-time simulations of neurofeedback and resting-state fMRI data. The rtQA was implemented as an extension of the open-source OpenNFT software written in Python, MATLAB and C++ for neurofeedback, task-based, and resting-state paradigms. We also developed a general Python library to unify real-time fMRI data processing and neurofeedback applications. Flexible estimation and visualization of rtQA facilitates efficient rtQA of fMRI data and helps the robustness of fMRI acquisitions by means of substantiating decisions about the necessity of the interruption and re-start of the experiment and increasing the confidence in neural estimates.
... [11][12][13][14][15][16][17][18][19] The 3D multislab technique divides the whole FOV into several 3D slabs. [20][21][22][23][24][25][26][27][28][29] Previous studies have demonstrated successful high-resolution diffusion imaging with 1-2-mm isotropic resolution using SMS or conventional single-band multislab. 20,[22][23][24][25] However, when the resolution is further increased, the total number of slabs may be increased to ensure the whole brain coverage. ...
Purpose
This study aims to propose a novel algorithm for slab boundary artifact correction in both single‐band multislab imaging and simultaneous multislab (SMSlab) imaging.
Theory and Methods
In image domain, the formation of slab boundary artifacts can be regarded as modulating the artifact‐free images using the slab profiles and introducing aliasing along the slice direction. Slab boundary artifact correction is the inverse problem of this process. An iterative algorithm based on convolutional neural networks (CNNs) is proposed to solve the problem, termed CNN‐enabled inversion for slab profile encoding (CPEN). Diffusion‐weighted SMSlab images and reference images without slab boundary artifacts were acquired in 7 healthy subjects for training. Images of 5 healthy subjects were acquired for testing, including single‐band multislab and SMSlab images with 1.3‐mm or 1‐mm isotropic resolution. CNN‐enabled inversion for slab profile encoding was compared with a previously reported method (i.e., nonlinear inversion for slab profile encoding [NPEN]).
Results
CNN‐enabled inversion for slab profile encoding reduces the slab boundary artifacts in both single‐band multislab and SMSlab images. It also suppresses the slab boundary artifacts in the diffusion metric maps. Compared with NPEN, CPEN shows fewer residual artifacts in different acquisition protocols and more significant improvements in quantitative assessment, and it also accelerates the computation by more than 35 times.
Conclusion
CNN‐enabled inversion for slab profile encoding can reduce the slab boundary artifacts in multislab acquisitions. It shows better slab boundary artifact correction capacity, higher robustness, and computation efficiency when compared with NPEN. It has the potential to improve the accuracy of multislab acquisitions in high‐resolution DWI and functional MRI.
... 23,24 A segmented multi-shot approach to 3D Cartesian readouts with high flexibility in the acquisition and acceleration schemes at the cost of increased physiological noise is 3D EPI. 25,26 The 3D Cartesian singleshot readouts of one or multiple slabs per brain have been realized through echo volumar imaging, 9,27,28 which was shown to enhance BOLD sensitivity compared to EPI but offers less spatial resolution than SMS-EPI. As for the approach that was chosen for this publication, segmented rotated stack-ofspiral acquisitions with high reduction factors were explored for high-resolution anatomical images before. ...
Purpose
Highly undersampled acquisitions have been proposed to push the limits of temporal resolution in functional MRI. This contribution is aimed at identifying parameter sets that let the user trade‐off between ultra‐high temporal resolution and spatial signal quality by varying the sampling densities. The proposed method maintains the synergies of a temporal resolution that enables direct filtering of physiological artifacts for highest statistical power, and 3D read‐outs with optimal use of encoding capabilities of multi‐coil arrays for efficient sampling and high signal‐to‐noise ratio (SNR).
Methods
One‐ to four‐shot interleaved spherical stack‐of‐spiral trajectories with repetition times from 96 to 352 ms at a nominal resolution of 3 mm using different sampling densities were compared for image quality and temporal SNR (tSNR). The one‐ and three‐shot trajectories were employed in a resting state study for functional characterization.
Results
Compared to a previously described single‐shot trajectory, denser sampled trajectories of the same type are shown to be less prone to blurring and off‐resonance vulnerability that appear in addition to the variable density artifacts of the point spread function. While the multi‐shot trajectories lead to a decrease in tSNR efficiency, the high SNR due to the 3D read‐out, combined with notable increases in image quality, leads to superior overall results of the three‐shot interleaved stack of spirals. A resting state analysis of 15 subjects shows significantly improved functional sensitivity in areas of high off‐resonance gradients.
Conclusion
Mild variable‐density sampling leads to excellent tSNR behavior and no increased off‐resonance vulnerability, and is suggested unless maximum temporal resolution is sought.
... Over the last decade, the development of fast fMRI techniques has led to a higher sensitivity and statistical power in detecting both neuronal activation after external stimulation and functional connectivity during the resting state. 18,19 Imaging techniques such as simultaneous multislice (SMS), 20 echo-volumar imaging (EVI), 21 or inverse imaging (InI) 22 have pushed the limits of temporal resolution to a few hundred milliseconds for threedimensional (3D) coverage of the brain. In particular, MR encephalography (MREG) combines single-shot 3D non-Cartesian gradient encoding with parallel imaging to reach repetition times of 100 ms or even less. ...
Purpose
Spin‐echo (SE) functional MRI (fMRI) can be highly advantageous compared to gradient‐echo (GE) fMRI with respect to magnetic field‐inhomogeneity artifacts. However, at 3T, the majority of blood oxygenation level‐dependent (BOLD) fMRI experiments are performed using T2∗‐weighted GE sequences because of their superior sensitivity compared to SE‐fMRI. The presented SE implementation of a highly accelerated GE pulse sequence therefore aims to improve the sensitivity of SE‐fMRI while profiting from a reduction of susceptibility‐induced signal dropout.
Methods
Spin‐echo MR encephalography (SE‐MREG) is compared with the more conventionally used spin‐echo echo‐planar imaging (SE‐EPI) and spin‐echo simultaneous multislice (SE‐SMS) at 3T in terms of capability to detect neuronal activations and resting‐state functional connectivity. For activation analysis, healthy subjects underwent consecutive SE‐MREG (pulse repetition time [TR] = 0.25 seconds), SE‐SMS (TR = 1.3 seconds), and SE‐EPI (TR = 4.4 seconds) scans in pseudorandomized order applied to a visual block design paradigm for generation of t‐statistics maps. For the investigation of functional connectivity, additional resting‐state data were acquired for 5 minutes and a seed‐based correlation analysis using Stanford’s FIND (Functional Imaging in Neuropsychiatric Disorders) atlas was performed.
Results
The increased sampling rate of SE‐MREG relative to SE‐SMS and SE‐EPI improves the sensitivity to detect BOLD activation by 33% and 54%, respectively, and increases the capability to extract resting‐state networks. Compared with a brain region that is not affected by magnetic field inhomogeneities, SE‐MREG shows 2.5 times higher relative signal strength than GE‐MREG in mesial temporal structures.
Conclusion
SE‐MREG offers a viable possibility for whole‐brain fMRI with consideration of brain regions that are affected by strong susceptibility‐induced magnetic field gradients.
... However, it is recently proved that there exist high-frequency BOLD oscillations (up to nearly 1 Hz) which reflects potentially relevant brain activity [58,59]. Besides, for better characterization of the hemodynamic response function (HRF) which is subtly different across subjects and cortical areas [60][61][62], separation of physiological noise [20,[63][64][65][66] and improving the statistical sensitivity [67][68][69][70], fast imaging techniques are of great practical significance [71,72]. ...
Fast functional magnetic resonance imaging (fMRI) technique, such as MR-Encephalography (MREG), provides a magical power to see the rapid activity within our brain. However, the main challenges of MREG are the extra-high computational cost and the strong sensitivity to off-resonance effects during reconstruction. Therefore, the core goal of this thesis is to solve these challenges.
To reduce the computational cost, a time-domain principal component reconstruction (tPCR) method is developed. It contains three steps: (i) decomposing the k-t-space fMRI datasets into time-domain principal component space using singular value decomposition, (ii) reconstructing each principal component with redistributed computation power according to their weights, (iii) combining the reconstructed principal components back to image-t-space. This operation significantly improves the reconstruction efficiency, allowing higher integrated reconstruction precision with much less computational cost when compared with the traditional reconstruction.
To reduce the dynamic off-resonance artifacts caused by physiological noise and motion, a dynamic field map estimation technique based on two-shot reversed-trajectory design and deep learning is developed. The field map is estimated from the two images with reversed artifacts. It is more difficult to estimate field maps in a non-Cartesian trajectory using analytical methods, so the deep learning technique is introduced. A convolutional neural network was trained using simulated data, then the field map was estimated using the trained network at each time point. With the corrected field map, both the image quality and the sensitivity for functional analysis are improved.
In conclusion, the two techniques in this thesis make MREG reconstruction more efficient and accurate, promoting its broader applications.
... FMRI detects brain activity by measuring BOLD signals. 1 Even though the BOLD signals change slowly and smoothly, which makes standard 2D EPI sampling with a 2-3 seconds temporal resolution generally sufficient, 2 rapid sampling is still highly beneficial for the following purposes: (1) detection of high-frequency BOLD oscillations (up to nearly 1Hz) reflecting potentially relevant brain activity 3,4 ; (2) better characterization of the hemodynamic response function (HRF), which is subtly different across subjects and cortical areas 5-7 ; (3) separation of physiological noise, such as breathing and cardiac pulsation [8][9][10][11][12] ; and (4) greater statistical power and sensitivity. [13][14][15][16] Thus, rapid fMRI techniques are of great practical significance and can markedly improve fMRI analyses. 17,18 Reducing the acquisition duration of individual images usually involves combinations of parallel imaging, [19][20][21] k-space undersampling 22 and/or optimized non-Cartesian sampling patterns. ...
Purpose
To improve the reconstruction efficiency (i.e., computational load) and stability of iterative reconstruction for non‐Cartesian fMRI when using high undersampling rates and/or in the presence of strong off‐resonance effects.
Theory and Methods
The magnetic resonance encephalography (MREG) sequence with 3D non‐Cartesian trajectory and 0.1s repetition time (TR) was applied to acquire fMRI datasets. Different from a conventional time‐point‐by‐time‐point sequential reconstruction (SR), the proposed time‐domain principal component reconstruction (tPCR) performs three steps: (1) decomposing the k‐t‐space fMRI datasets into time‐domain principal component space using singular value decomposition, (2) reconstructing each principal component with redistributed computation power according to their weights, and (3) combining the reconstructed principal components back to image‐t‐space. The comparison of reconstruction accuracy was performed by simulation experiments and then verified in real fMRI data.
Results
The simulation experiments showed that the proposed tPCR was able to significantly reduce reconstruction errors, and subsequent functional activation errors, relative to SR at identical computational cost. Alternatively, at fixed reconstruction accuracy, computation time was greatly reduced. The improved performance was particularly obvious for L1‐norm nonlinear reconstructions relative to L2‐norm linear reconstructions and robust to different regularization strength, undersampling rates, and off‐resonance effects intensity. By examining activation maps, tPCR was also found to give similar improvements in real fMRI experiments.
Conclusion
The proposed proof‐of‐concept tPCR framework could improve (1) the reconstruction efficiency of iterative reconstruction, and (2) the reconstruction stability especially for nonlinear reconstructions. As a practical consideration, the improved reconstruction speed promotes the application of highly undersampled non‐Cartesian fast fMRI.
... The subsequent implementation of EVI into the first WS module and integration of GRAPPA acceleration strongly improved signal intensity and image quality (image uniformity, image distortion and ghosting), which were comparable with our recently developed multislab EVI method. 20 Activation in visual cortex was detected after the first task block and the resulting map shows an average correlation coefficient of 0.63 with spatial extent of 71.1 cc and a peak correlation of 0.8 ( Figure 11A). A 4% signal change with a contrast-to-noise ratio of >10 was measured in visual cortex ( Figure 11B). ...
This study evaluated the utility of concurrent water signal acquisition as part of the water suppression in MR spectroscopic imaging (MRSI), to allow simultaneous water referencing for metabolite quantification, and to concurrently acquire functional MRI (fMRI) data. We integrated a spatial‐spectral binomial water excitation RF pulse and a short spatial‐spectral echo‐planar readout into the water suppression module of 2D and 3D proton‐echo‐planar‐spectroscopic‐imaging (PEPSI) with a voxel size as small as 4 x 4 x 6 mm3. Metabolite quantification in reference to tissue water was validated in healthy controls for different prelocalization methods (spin‐echo, PRESS and semi‐LASER) and the clinical feasibility of a 3‐minute 3D semi‐Laser PEPSI scan (TR/TE: 1250/32 ms) with water referencing in patients with brain tumors was demonstrated. Spectral quality, SNR, Cramer‐Rao‐lower‐bounds and water suppression efficiency were comparable with conventional PEPSI. Metabolite concentration values in reference to tissue water, using custom LCModel‐based spectral fitting with relaxation correction, were in the range of previous studies and independent of the prelocalization method used. Next, we added a phase‐encoding undersampled echo‐volumar imaging (EVI) module during water suppression to concurrently acquire metabolite maps with water referencing and fMRI data during task execution and resting state in healthy controls. Integration of multimodal signal acquisition prolongated minimum TR by less than 50 ms on average. Visual and motor activation in concurrent fMRI/MRSI (TR: 1250–1500 ms, voxel size: 4 x 4 x 6 mm3) was readily detectable in single‐task blocks with percent signal change comparable with conventional fMRI. Resting‐state connectivity in sensory and motor networks was detectable in 4 minutes. This hybrid water suppression approach for multimodal imaging has the potential to significantly reduce scan time and extend neuroscience research and clinical applications through concurrent quantitative MRSI and fMRI acquisitions.
... Increasingly available high-speed data acquisition methods (Zahneisen et al., 2011;Lin et al., 2010;Setsompop et al., 2012) have reduced sensitivity to physiological signal fluctuation and increased sensitivity for mapping RSNs (Smith et al., 2012a;Feinberg et al., 2010;Posse et al., 2012). However, high sensitivity to head movement, which obscures networks as well as create false-positive correlations (Van Dijk, Sabuncu, & Buckner, 2012;Satterthwaite et al., 2012) despite state-of-the-art motion "correction" in postprocessing, and the low frequency range of signal fluctuations and spatialtemporal nonstationarity, which is prominent in the resting-state (C. ...
Resting‐state functional magnetic resonance imaging (rsfMRI) is a promising task‐free functional imaging approach, which may complement or replace task‐based fMRI (tfMRI) in patients who have difficulties performing required tasks. However, rsfMRI is highly sensitive to head movement and physiological noise, and validation relative to tfMRI and intraoperative electrocortical mapping is still necessary. In this study, we investigate (a) the feasibility of real‐time rsfMRI for presurgical mapping of eloquent networks with monitoring of data quality in patients with brain tumors and (b) rsfMRI localization of eloquent cortex compared with tfMRI and intraoperative electrocortical stimulation (ECS) in retrospective analysis. Five brain tumor patients were studied with rsfMRI and tfMRI on a clinical 3T scanner using MultiBand(8)‐echo planar imaging (EPI) with repetition time: 400 ms. Moving‐averaged sliding‐window correlation analysis with regression of motion parameters and signals from white matter and cerebrospinal fluid was used to map sensorimotor and language resting‐state networks. Data quality monitoring enabled rapid optimization of scan protocols, early identification of task noncompliance, and head movement‐related false‐positive connectivity to determine scan continuation or repetition. Sensorimotor and language resting‐state networks were identifiable within 1 min of scan time. The Euclidean distance between ECS and rsfMRI connectivity and task‐activation in motor cortex, Broca's, and Wernicke's areas was 5–10 mm, with the exception of discordant rsfMRI and ECS localization of Wernicke's area in one patient due to possible cortical reorganization and/or altered neurovascular coupling. This study demonstrates the potential of real‐time high‐speed rsfMRI for presurgical mapping of eloquent cortex with real‐time data quality control, and clinically acceptable concordance of rsfMRI with tfMRI and ECS localization.
... Numerous studies have demonstrated the success of either SMS or 3D multi-slab in dMRI and fMRI, particularly for whole-brain studies. 3,4,7,[22][23][24][25][26] However, when pushing the spatial resolution to 1 mm isotropic or higher, neither multi-slab nor SMS can guarantee an optimal SNR efficiency. 15,16,24 Therefore, in this study, we aim to develop a new method to combine SMS and multi-slab to further increase the SNR efficiency. ...
Purpose
To propose a novel 3D k‐space Fourier encoding and reconstruction framework for simultaneous multi‐slab (SMSlab) acquisition and demonstrate its efficacy in high‐resolution imaging.
Methods
First, it is illustrated in theory how the inter‐slab gap interferes with the formation of the SMSlab 3D k‐space. Then, joint RF and gradient encoding are applied to remove the inter‐slab gap interference and form a SMSlab 3D k‐space. In vivo experiments are performed to validate the proposed theory. Acceleration in the proposed SMSlab 3D k‐space is also evaluated.
Results
High‐resolution (1.0 mm isotropic) images can be reconstructed using the proposed SMSlab 3D framework. Controlled aliasing in parallel imaging sampling and 2D GRAPPA reconstruction can also be applied in the SMSlab 3D k‐space. Compared with conventional multi‐slab acquisition, SMSlab exhibits better SNR maintainability (such as lower g‐factors), especially at high acceleration factors.
Conclusion
It is demonstrated that the joint application of RF and gradient encoding enables SMSlab within a 3D Fourier encoding framework. Images with high isotropic resolution can be reconstructed, and further acceleration is also applicable. The proposed SMSlab 3D k‐space can be valuable for both high‐resolution and high‐efficiency diffusion and functional MRI.
... High spatial resolution can improve blood oxygenation level dependent (BOLD) sensitivity by reducing within voxel susceptibility artifacts and partial volume effect (Frahm et al., 1993), and allows resolving hemodynamic responses from individual cortical columns (Duong et al., 2001;Harel et al., 2006) or cortical layers (Siero et al., 2011;Poplawsky et al., 2015;Goense et al., 2016), which may further be used to infer information flow (Huber et al., 2017). Although hemodynamic response is slow, it has been demonstrated that high temporal resolution (>1 Hz) helps not only to increase BOLD sensitivity and statistical power (Neggers et al., 2008;Posse et al., 2012;Smith et al., 2013), to distinguish/reduce physiological noise (Chuang and Chen, 2001;Hennig et al., 2007;Tong et al., 2014), but also allows the detection of high frequency neural activation (Lewis et al., 2016) and resting-state oscillation above the conventional 0.1 Hz range (Boubela et al., 2013;Lee et al., 2013;Lin et al., 2015;Chen and Glover, 2015). As the neural basis of such fast hemodynamic activity is unclear, animal studies are needed to understand its neurophysiological origins (Chuang and Nasrallah, 2017). ...
... However, these new possibilities also require more advanced preprocessing and statistical computations, such as handling auto-correlations across many volumes, to fulfill the requirements for the commonly used parametric statistical analyses (Fadili et al., 2003). Specific advancements in the field of functional magnetic resonance imaging (fMRI) provided new acquisition methods such as simultaneous multi-slice (SMS, also known as multi-band, MB) (Feinberg et al., 2010;Feinberg and Setsompop, 2013), blipped controlled aliasing (CAIPI) EPI blipped-CAIPI-EPI) Setsompop et al., 2012), echo volumar imaging EVI) (Rabrait et al., 2008), multi-slab EVI (Posse et al., 2012) or simultaneous multi-slice inverse imaging (SMS-InI) (Hsu et al., 2017), allowing to acquire fMRI data using repetition times (TR) in the range of hundreds of milliseconds and are also suitable for real-time applications. The method used in this article to perform fMRI is called MR-Encephalography (MREG) (Hennig et al., 2007). ...
Real-time functional magnetic resonance imaging (rt-fMRI) enables the update of various brain-activity measures during an ongoing experiment as soon as a new brain volume is acquired. However, the recorded Blood-oxygen-level dependent (BOLD) signal also contains physiological artifacts such as breathing and heartbeat, which potentially cause misleading false positive effects especially problematic in brain-computer interface (BCI) and neurofeedback (NF) setups. The low temporal resolution of echo planar imaging (EPI) sequences (which is in the range of seconds) prevents a proper separation of these artifacts from the BOLD signal. MR-Encephalography (MREG) has been shown to provide the high temporal resolution required to unalias and correct for physiological fluctuations and leads to increased specificity and sensitivity for mapping task-based activation and functional connectivity as well as for detecting dynamic changes in connectivity over time. By comparing a simultaneous multislice echo planar imaging (SMS-EPI) sequence and an MREG sequence using the same nominal spatial resolution in an offline analysis for three different experimental fMRI paradigms (perception of house and face stimuli, motor imagery, Stroop task), the potential of this novel technique for future BCI and NF applications was investigated. First, adapted general linear model pre-whitening which accounts for the high temporal resolution in MREG was implemented to calculate proper statistical results and be able to compare these with the SMS-EPI sequence. Furthermore, the respiration- and cardiac pulsation-related signals were successfully separated from the MREG signal using independent component analysis which were then included as regressors for a GLM analysis. Only the MREG sequence allowed to clearly separate cardiac pulsation and respiration components from the signal time course. It could be shown that these components highly correlate with the recorded respiration and cardiac pulsation signals using a respiratory belt and fingertip pulse plethysmograph. Temporal signal-to-noise ratios of SMS-EPI and MREG were comparable. Functional connectivity analysis using partial correlation showed a reduced standard error in MREG compared to SMS-EPI. Also, direct time course comparisons by down-sampling the MREG signal to the SMS-EPI temporal resolution showed lower variance in MREG. In general, we show that the higher temporal resolution is beneficial for fMRI time course modeling and this aspect can be exploited in offline application but also, is especially attractive, for real-time BCI and NF applications.
... 13 The temporal resolution of fMRI being on the order of diagnostic review 2-3 seconds means that the technique is not truly real-time. 18 Moreover, there is generally a trade-off between spatial and temporal resolution since it takes more time to acquire images with smaller voxels. 14 In addition, the interpretation of fMRI is statistically demanding. ...
... For real-time fMRI, data acquisition, preprocessing, and analysis need to be optimized for speed. General improvements in the field of fMRI, such as specific imaging sequences that allow to acquire high quality data within a very short time (Posse et al., 1999(Posse et al., , 2012Speck and Hennig, 1998;Weiskopf et al., 2004aWeiskopf et al., , 2005 and the availability of higher magnetic field strengths have dramatically increased sensitivity of present-day fMRI (Duyn, 2012;Hahn et al., 2013;Sladky et al., 2013Sladky et al., , 2018van der Zwaag et al., 2009;Yacoub et al., 2008) and real-time fMRI methods (Baecke et al., 2015;Grone et al., 2015;Koush et al., 2011Koush et al., , 2013Koush et al., , 2014. In addition, real-time fMRI data analysis benefits from steadily increasing computational power (Moore, 1965), from the optimization of real-time analysis algorithms (Hinds et al., 2011;Koush et al., 2012;Magland et al., 2011), and from the adaptation of sophisticated data analysis techniques for real-time purposes (Esposito et al., 2003;Hollmann et al., 2011;Koush et al., 2013;LaConte et al., 2007;Sitaram et al., 2011;Zilverstand et al., 2014). ...
As a consequence of recent technological advances in the field of functional magnetic resonance imaging (fMRI), results can now be made available in real-time. This allows for novel applications such as online quality assurance of the acquisition, intra-operative fMRI, brain-computer-interfaces, and neurofeedback. To that aim, signal processing algorithms for real-time fMRI must reliably correct signal contaminations due to physiological noise, head motion, and scanner drift. The aim of this study was to compare performance of the commonly used online detrending algorithms exponential moving average (EMA), incremental general linear model (iGLM) and sliding window iGLM (iGLM window ). For comparison, we also included offline detrending algorithms (i.e., MATLAB's and SPM8's native detrending functions). Additionally, we optimized the EMA control parameter, by assessing the algorithm's performance on a simulated data set with an exhaustive set of realistic experimental design parameters. First, we optimized the free parameters of the online and offline detrending algorithms. Next, using simulated data, we systematically compared the performance of the algorithms with respect to varying levels of Gaussian and colored noise, linear and non-linear drifts, spikes, and step function artifacts. Additionally, using in vivo data from an actual rt-fMRI experiment, we validated our results in a post hoc offline comparison of the different detrending algorithms. Quantitative measures show that all algorithms perform well, even though they are differently affected by the different artifact types. The iGLM approach outperforms the other online algorithms and achieves online detrending performance that is as good as that of offline procedures. These results may guide developers and users of real-time fMRI analyses tools to best account for the problem of signal drifts in real-time fMRI.
... The improvements achievable by SMS-EPI have been evaluated for task-based fMRI Posse et al., 2012;Todd et al., 2017Todd et al., , 2016. However, given the fundamentally different analysis approaches used for rs-fMRI these results are not necessarily applicable to rs-fMRI studies as well. ...
Background:
Recent advancements in simultaneous multi-slice (SMS) imaging techniques have enabled whole-brain resting-state fMRI (rs-fMRI) scanning at sub-second temporal resolution, providing spectral ranges much wider than the typically used range of 0.01-0.1 Hz. However, the advantages of this accelerated acquisition for rs-fMRI have not been evaluated.
New method:
In this study, we used SMS Echo Planar Imaging (EPI) to probe whole-brain functional connectivity with a short repetition time (TR = 350 ms) and compared it with standard EPI with a longer TR of 2000 ms. We determined the effect of scan length and investigated the temporal filtration strategies that optimize results based on metrics of signal-noise separation and test-retest reliability using both seed-based and independent component analysis (ICA).
Results:
We found that use of either the entire frequency range of 0.01-1.4 Hz or the entire frequency range with the exclusion of typical cardiac and respiratory frequency values tended to provide the best functional connectivity maps.
Comparison with existing methods:
We found that the SMS-acquired rs-fMRI scans had improved the signal-noise separation, while preserving the same level of test-retest reliability compared to conventional EPI, and enabled the detection of reliable functional connectivity networks with scan times as short as 3 min.
Conclusions:
Our findings suggest that whole-brain rs-fMRI studies may benefit from the increased temporal resolution enabled by the SMS-EPI acquisition, leading to drastic scan time reductions, which in turn should enable the more widespread use of rs-fMRI in clinical research protocols.
... Moreover, temporal relationships between cerebral areas are confounded by the spatial variability of the hemodynamic response function (Handwerker et al., 2004). Although MR acquisition sequences with faster temporal resolutions are becoming increasingly common (Feinberg et al., 2010;Posse et al., 2012;Akin et al., 2017;LeVan et al., 2017), neuronal processes still undergo considerable downsampling in fMRI time series, affecting the reliability of GC estimates (Seth et al., 2013;Friston et al., 2014a). ...
The brain consists of neurons which are interconnected by synapses. The connectivity of the networks formed by the neurons and synapses is a key feature for the function and dysfunction of the brain. In humans, studying the connectivity comes with various challenges. Thus, the access to connectivity in humans is limited. The aim of this dissertation is to introduce a novel method to estimate effective connectivity of neural populations from continuous recordings of the activity of these populations which overcomes limitations of existing methods. This method estimates the connectivity of neural populations based on the covariance of the measured activity. The key mechanism for doing so is a L1 -minimization via a gradient descent on the manifold of unitary matrices. The fact that the gradient of a matrix on the unitary manifold is skew-hermitian and with this in the corresponding Lie-Algebra, is exploited in the update step to project the gradient back on the manifold via the exponential map. As presented in this thesis, this method works reliably for sparse networks with more than 40 nodes and a sufficient network interaction. The method can be applied on zero-lag covariance matrices, hence there is no restriction on the sampling rate of the measurement. Although based on a linear interaction model, the method also achieves reasonable results for networks which interact non-linearly. A comparison with structural measures based on fMRI data shows better agreement than state-of-the-art methods. Also, the method is robust against noise, unobserved nodes and variable hemodynamics of BOLD signals. In this thesis, a novel method for the estimation of effective connectivity from covariances of neural activity is presented. The features of this method make it applicable on a broad range of data-types, including data based on the BOLD effect or electro physiologic data such as ECoG. Applications of fast fMRI data show plausible and coherent results.
... Moreover, temporal relationships between cerebral areas are confounded by the spatial variability of the hemodynamic response function (Handwerker et al., 2004). Although MR acquisition sequences with faster temporal resolutions are becoming increasingly common (Feinberg et al., 2010;Posse et al., 2012;Akin et al., 2017;LeVan et al., 2017), neuronal processes still undergo considerable downsampling in fMRI time series, affecting the reliability of GC estimates (Seth et al., 2013;Friston et al., 2014a). ...
In functional magnetic resonance imaging (fMRI), functional connectivity is conventionally characterized by correlations between fMRI time series, which are intrinsically undirected measures of connectivity. Yet, some information about the directionality of network connections can nevertheless be extracted from the matrix of pairwise temporal correlations between all considered time series, when expressed in the frequency-domain as a cross-spectral density matrix. Using a sparsity prior, it then becomes possible to determine a unique directed network topology that best explains the observed undirected correlations, without having to rely on temporal precedence relationships that may not be valid in fMRI. Applying this method on simulated data with 100 nodes yielded excellent retrieval of the underlying directed networks under a wide variety of conditions. Importantly, the method did not depend on temporal precedence to establish directionality, thus reducing susceptibility to hemodynamic variability. The computational efficiency of the algorithm was sufficient to enable whole-brain estimations, thus circumventing the problem of missing nodes that otherwise occurs in partial-brain analyses. Applying the method to real resting-state fMRI data acquired with a high temporal resolution, the inferred networks showed good consistency with structural connectivity obtained from diffusion tractography in the same subjects. Interestingly, this agreement could also be seen when considering high-frequency rather than low-frequency connectivity (average correlation: r = 0.26 for f < 0.3 Hz, r = 0.43 for 0.3 < f < 5 Hz). Moreover, this concordance was significantly better (p < 0.05) than for networks obtained with conventional functional connectivity based on correlations (average correlation r = 0.18). The presented methodology thus appears to be well-suited for fMRI, particularly given its lack of explicit dependence on temporal lag structure, and is readily applicable to whole-brain effective connectivity estimation.
... [3][4][5] The disadvantage of fMRI, however, is its poor temporal resolution limited to the range of seconds, which poses specific challenges to experimental design and taskrelated manipulations. 6,7 Preceding the statistical analysis of fMRI data from a single subject or a group of participants, the images are subjected to standardized processing steps aiming at spatial alignment within and across subject(s) over the entire duration of the experiment. Group studies require additionally the spatial transformation of data in standardized space, allowing to perform statistical analysis at the voxel level. ...
Despite more than a century of investigation into the cortical organization of motor function, the existence of motor somatotopy is still debated. We review functional magnetic resonance imaging (fMRI) studies examining motor somatotopy in the cerebral cortex. In spite of a substantial overlap of representations corresponding to different body parts, especially in non-primary motor cortices, geographic approaches are capable of revealing somatotopic ordering. From the iconic homunculus in the contralateral primary cortex to the subtleties of ipsilateral somatotopy and its relations with lateralization, we outline potential reasons for the lack of segregation between motor representations. Among these are the difficulties in distinguishing activity that arises from multiple muscular effectors, the need for flexible motor control and coordination of complex movements through functional integration and artefacts in fMRI. Methodological advances with regard to the optimization of experimental design and fMRI acquisition protocols as well as improvements in spatial registration of images and indices aiming at the quantification of the degree of segregation between different functional representations are inspected. Additionally, we give some hints as to how the functional organization of motor function might be related to various anatomical landmarks in brain morphometry.
Functional brain fingerprinting has emerged as an influential tool to quantify reliability in neuroimaging studies and to identify cognitive biomarkers in both healthy and clinical populations. Recent studies have revealed that brain fingerprints reside in the timescale-specific functional connectivity of particular brain regions. However, the impact of the acquisition′s temporal resolution on fingerprinting remains unclear. In this study, we examine for the first time the reliability of functional fingerprinting derived from resting-state functional MRI (rs-fMRI) with different whole-brain temporal resolutions (TR = 0.5, 0.7, 1, 2, and 3 s) in a cohort of 20 healthy volunteers. Our findings indicate that subject identifiability within a fixed TR is successful across different temporal resolutions, with the highest identifiability observed at TR 0.5 and 3 s. We discuss this observation in terms of protocol-specific effects of physiological noise aliasing. We further show that, irrespective of TR, associative brain areas make substantial contributions to subject identifiability, whereas sensory-motor regions become influential only when integrating data from different TRs. We conclude that functional connectivity fingerprinting derived from rs-fMRI holds significant potential for multicentric studies also employing protocols with different temporal resolutions. However, it remains crucial to consider fMRI signal′s sampling rate differences in subject identifiability between data samples, in order to improve reliability and generalizability of both whole-brain and specific functional networks′ results. These findings contribute to a better understanding of the practical application of functional connectivity fingerprinting, and its implications for future neuroimaging research.
Background
Conventional cerebrovascular reactivity (CVR) estimation has demonstrated that many brain diseases and/or conditions are associated with altered CVR. Despite the clinical potential of CVR, characterization of temporal features of a CVR challenge remains uncommon. This work is motivated by the need to develop CVR parameters that characterize individual temporal features of a CVR challenge.
Methods
Data were collected from 54 adults and recruited based on these criteria: (1) Alzheimer’s disease diagnosis or subcortical Vascular Cognitive Impairment, (2) sleep apnea, and (3) subjective cognitive impairment concerns. We investigated signal changes in blood oxygenation level dependent (BOLD) contrast images with respect to hypercapnic and normocapnic CVR transition periods during a gas manipulation paradigm. We developed a model-free, non-parametric CVR metric after considering a range of responses through simulations to characterize BOLD signal changes that occur when transitioning from normocapnia to hypercapnia. The non-parametric CVR measure was used to examine regional differences across the insula, hippocampus, thalamus, and centrum semiovale. We also examined the BOLD signal transition from hypercapnia back to normocapnia.
Results
We found a linear association between isolated temporal features of successive CO2 challenges. Our study concluded that the transition rate from hypercapnia to normocapnia was significantly associated with the second CVR response across all regions of interest (p < 0.001), and this association was highest in the hippocampus (R² = 0.57, p < 0.0125).
Conclusion
This study demonstrates that it is feasible to examine individual responses associated with normocapnic and hypercapnic transition periods of a BOLD-based CVR experiment. Studying these features can provide insight on between-subject differences in CVR.
This chapter introduces Simultaneous MultiSlice (SMS) Imaging by drawing on the connections with Hadamard and POMP methods that were introduced long before the advent of parallel imaging to improve efficiency of multislice imaging, and it outlines widely used SMS reconstruction algorithms as extensions of the parallel imaging framework. Special emphasis is provided in describing how to adapt these SMS reconstruction algorithms to echo-planar imaging for functional and diffusion MRI. This chapter also introduces reconstruction metrics that can be useful tools for evaluating and comparing the SMS reconstruction methods, and it discusses some practical design considerations and trade-offs for applying SMS technology in practice.
Purpose
We investigate the use of TURBINE, a 3D radial‐Cartesian acquisition scheme in which EPI planes are rotated about the phase‐encoding axis to acquire a cylindrical k‐space for high‐fidelity ultrahigh isotropic resolution fMRI at 7 Tesla with minimal distortion and blurring.
Methods
An improved, completely self‐navigated version of the TURBINE sampling scheme was designed for fMRI at 7 Telsa. To demonstrate the image quality and spatial specificity of the acquisition, thin‐slab visual and motor BOLD fMRI at 0.67 mm isotropic resolution (16 mm slab, TRvol = 2.32 s), and 0.8 × 0.8 × 2.0 mm (whole‐brain, TRvol = 2.4 s) data were acquired. To prioritize the high spatial fidelity, we employed a temporally regularized reconstruction to improve sensitivity without any spatial bias.
Results
TURBINE images provide high structural fidelity with almost no distortion, dropout, or T2* blurring for the thin‐slab acquisitions compared to conventional 3D EPI owing to the radial sampling in‐plane and the short echo train used. This results in activation that can be localized to pre‐ and postcentral gyri in a motor task, for example, with excellent correspondence to brain structure measured by a T1‐MPRAGE. The benefits of TURBINE (low distortion, dropout, blurring) are reduced for the whole‐brain acquisition due to the longer EPI train. We demonstrate robust BOLD activation at 0.67 mm isotropic resolution (thin‐slab) and also anisotropic 0.8 × 0.8 × 2.0 mm (whole‐brain) acquisitions.
Conclusion
TURBINE is a promising acquisition approach for high‐resolution, minimally distorted fMRI at 7 Tesla and could be particularly useful for fMRI in areas of high B0 inhomogeneity.
Functional magnetic resonance imaging (fMRI), a non-invasive and widely used human neuroimaging method, is most known for its spatial precision. However, there is a growing interest in its temporal sensitivity. This is despite the temporal blurring of neuronal events by the blood oxygen level dependent (BOLD) signal, the peak of which lags neuronal firing by 4 to 6 seconds. Given this, the goal of this review is to answer a seemingly simple question – “What are the benefits of increased temporal sampling for fMRI?”. To answer this, we have combined fMRI data collected at multiple temporal scales, from 323 to 1000 milliseconds, with a review of both historical and contemporary temporal literature. After a brief discussion of technological developments that have rekindled interest in temporal research, we next consider the potential statistical and methodological benefits. Most importantly, we explore how fast fMRI can uncover previously unobserved neuro-temporal dynamics – effects that are entirely missed when sampling at conventional 1 to 2 second rates. With the intrinsic link between space and time in fMRI, this temporal renaissance also delivers improvements in spatial precision. Far from producing only statistical gains, the array of benefits suggest that the continued temporal work is worth the effort.
Neural electrical interfaces are important tools for local neural stimulation and recording, which potentially have wide application in the diagnosis and treatment of neural diseases, as well as in the transmission of neural activity for brain-computer interface (BCI) systems. At the same time, magnetic resonance imaging (MRI) is one of the effective and non-invasive techniques for recording whole-brain signals, providing details of brain structures and also activation pattern maps. Simultaneous recording of extracellular neural signals and MRI combines two expressions of the same neural activity and is believed to be of great importance for the understanding of brain function. However, this combination makes requests on the magnetic and electronic performance of neural interface devices. MRI-compatibility refers here to a technological approach to simultaneous MRI and electrode recording or stimulation without artifacts in imaging. Trade-offs between materials magnetic susceptibility selection and electrical function should be considered. Herein, prominent trends in selecting materials of suitable magnetic properties are analyzed and material design, function and application of neural interfaces are outlined together with the remaining challenge to fabricate MRI-compatible neural interface.
In vivo mapping of cerebrovascular oscillations in the 0.05-0.15 Hz remains difficult. Oscillations in the cerebrospinal fluid (CSF) represent a possible avenue for noninvasively tracking these oscillations using resting-state functional MRI (rs-fMRI), and have been used to correct for vascular oscillations in rs-fMRI functional connectivity. However, the relationship between low-frequency CSF and vascular oscillations remains unclear. In this study, we investigate this relationship using fast simultaneous rs-fMRI and photoplethysmogram (PPG), examining the 0.1 Hz PPG signal, heart-rate variability (HRV), pulse-intensity ratio (PIR), and the second derivative of the PPG (SDPPG). The main findings of this study are: (a) signals in different CSF regions are not equivalent in their associations with vascular and tissue rs-fMRI signals; (b) the PPG signal is maximally coherent with the arterial and CSF signals at the cardiac frequency, but coherent with brain tissue at ~0.2 Hz; (c) PIR is maximally coherent with the CSF signal near 0.03 Hz; and (d) PPG-related vascular oscillations only contribute to ~15% of the CSF (and arterial) signal in rs-fMRI. These findings caution against averaging all CSF regions when extracting physiological nuisance regressors in rs-fMRI applications, and indicate the drivers of the CSF signal are more than simply cardiac. Our study is an initial attempt at the refinement and standardization of how the CSF signal in rs-fMRI can be used and interpreted. It also paves the way for using rs-fMRI in the CSF as a potential tool for tracking cerebrovascular health through, for instance, the potential relationship between PIR and the CSF signal.
Neurofeedback training using real‐time functional magnetic resonance imaging (rtfMRI‐NF) allows subjects voluntary control of localised and distributed brain activity. It has sparked increased interest as a promising non‐invasive treatment option in neuropsychiatric and neurocognitive disorders, although its efficacy and clinical significance are yet to be determined. In this work, we present the first extensive review of acquisition, processing and quality control methods available to improve the quality of the neurofeedback signal. Furthermore, we investigate the state of denoising and quality control practices in 128 recently published rtfMRI‐NF studies. We found: (a) that less than a third of the studies reported implementing standard real‐time fMRI denoising steps, (b) significant room for improvement with regards to methods reporting and (c) the need for methodological studies quantifying and comparing the contribution of denoising steps to the neurofeedback signal quality. Advances in rtfMRI‐NF research depend on reproducibility of methods and results. Notably, a systematic effort is needed to build up evidence that disentangles the various mechanisms influencing neurofeedback effects. To this end, we recommend that future rtfMRI‐NF studies: (a) report implementation of a set of standard real‐time fMRI denoising steps according to a proposed COBIDAS‐style checklist (https://osf.io/kjwhf/), (b) ensure the quality of the neurofeedback signal by calculating and reporting community‐informed quality metrics and applying offline control checks and (c) strive to adopt transparent principles in the form of methods and data sharing and support of open‐source rtfMRI‐NF software. Code and data for reproducibility, as well as an interactive environment to explore the study data, can be accessed at https://github.com/jsheunis/quality‐and‐denoising‐in‐rtfmri‐nf.
In this thesis I developed new techniques for diffuse optical tomography. In the first part, I developed a novel method to compute datatypes for diffuse optical tomography. With this new method a larger set of datatypes can be computed and noise is less correlated. Results show that better resolution in depth is obtained in comparison with the state-of-the-art. Moreover, quantification of absorption is improved significantly. In the second part, I developed total variation regularization method for diffuse optical tomography in irregular meshes. After, I performed brain motor cortex activation experiments in adult subjects with the collaboration of Politecnico di Milano. Previously developed algorithms were applied to that measurements obtaining time-series hemodynamic reconstructions of motor cortex. Finally, I coordinated the largest open dataset in diffuse optics composed by the measurements done within the BitMap network.
Objective: Epilepsy causes measurable irregularity over a range of brain signal frequencies, as well as autonomic nervous system functions that modulate heart and respiratory rate variability. Imaging dynamic neuronal signals utilizing simultaneously acquired ultra-fast 10 Hz magnetic resonance encephalography (MREG), direct current electroencephalography (DC-EEG), and near-infrared spectroscopy (NIRS) can provide a more comprehensive picture of human brain function. Spectral entropy (SE) is a nonlinear method to summarize signal power irregularity over measured frequencies. SE was used as a joint measure to study whether spectral signal irregularity over a range of brain signal frequencies based on synchronous multimodal brain signals could provide new insights in the neural underpinnings of epileptiform activity.
Methods: Ten patients with focal drug-resistant epilepsy (DRE) and ten healthy controls (HC) were scanned with 10 Hz MREG sequence in combination with EEG, NIRS (measuring oxygenated, deoxygenated, and total hemoglobin: HbO, Hb, and HbT, respectively), and cardiorespiratory signals. After pre-processing, voxelwise SEMREG was estimated from MREG data. Different neurophysiological and physiological subfrequency band signals were further estimated from MREG, DC-EEG, and NIRS: fullband (0-5 Hz, FB), near FB (0.08-5 Hz, NFB), brain pulsations in very-low frequency (0.009–0.08 Hz, VLFP), respiratory (0.12–0.4 Hz, RFP), and cardiac (0.7–1.6 Hz, CFP) frequency bands. Global dynamic fluctuations in MREG and NIRS were analyzed in windows of 2 minutes with 50% overlap.
Results: Right thalamus, cingulate gyrus, inferior frontal gyrus, and frontal pole showed significantly higher SEMREG in DRE patients compared to HC. In DRE patients, SE of cortical Hb was significantly reduced in FB (p=0.045), NFB (p=0.017), and CFP (p=0.038), while both HbO and HbT were significantly reduced in RFP (p=0.038, p=0.045, respectively). Dynamic SE of HbT was reduced in DRE patients in RFP during minutes 2 to 6 and increased in global MREG during minutes 5 to 8. Fitting to the frontal MREG and NIRS results, DRE patients showed a significant increase in SEEEG in FB in fronto-central and parieto-occipital regions, in VLFP in parieto-central region, accompanied with a significant decrease in RFP in frontal pole and parietal and occipital (O2, Oz) regions.
Conclusion: This is the first study to show altered spectral entropy from synchronous MREG, EEG, and NIRS in DRE patients. Higher SEMREG in DRE patients in anterior cingulate gyrus together with SEEEG and SENIRS results in 0.12-0.4 Hz can be linked to altered parasympathetic function and respiratory pulsation mechanisms in the brain. Higher SEMREG in thalamus in DRE patients is connected to disturbances in anatomical and functional connections in epilepsy. Our results in 2-minute time windows may be related to vigilance changes in epilepsy. Findings suggest that spectral irregularity of both electrophysiological and hemodynamic signals are altered in specific way depending on the physiological frequency range.
Independent component analysis (ICA) and seed-based analyses are widely used techniques for studying intrinsic neuronal activity in task-based or resting scans. In this work, we show there is a direct link between the two, and show that there are some important differences between the two approaches in terms of what information they capture. We developed an enhanced connectivity-matrix independent component analysis (cmICA) for calculating whole brain voxel maps of functional connectivity, which reduces the computational complexity of voxel-based connectivity analysis on performing many temporal correlations. We also show there is a mathematical equivalency between parcellations on voxel-to-voxel functional connectivity and simplified cmICA. Next, we used this cost-efficient data-driven method to examine the resting state fMRI connectivity in schizophrenia patients (SZ) and healthy controls (HC) on a whole brain scale and further quantified the relationship between brain functional connectivity and cognitive performances measured by the Measurement and Treatment Research to Improve Cognition in Schizophrenia (MATRICS) battery. Current results suggest that SZ exhibit a wide-range abnormality, primarily a decrease, in functional connectivity both between networks and within different network hubs. Specific functional connectivity decreases were associated with MATRICS performance deficits. In addition, we found that resting state functional connectivity decreases was extensively associated with aging regardless of groups. In contrast, there was no relationship between positive and negative symptoms in the patients and functional connectivity. In sum, we have developed a novel mathematical relationship between ICA and seed-based connectivity that reduces computational complexity, which has broad applicability, and showed a specific application of this approach to characterize connectivity changes associated with cognitive scores in SZ.
Ultrafast functional magnetic resonance imaging (fMRI) can measure blood oxygen level dependent (BOLD) signals with high sensitivity and specificity. Here we propose a novel method: simultaneous multi-slice inverse imaging (SMS-InI) — a combination of simultaneous multi-slice excitation, simultaneous echo refocusing (SER), blipped controlled aliasing in parallel imaging echo-planar imaging (EPI), and regularized image reconstruction. Using a 32-channel head coil array on a 3 T scanner, SMS-InI achieves nominal isotropic 5-mm spatial resolution and 10 Hz sampling rate at the whole-brain level. Compared with traditional inverse imaging, we found that SMS-InI has higher spatial resolution with lower signal leakage and higher time-domain signal-to-noise ratio with the optimized regularization parameter in the reconstruction. SMS-InI achieved higher effective resolution and higher detection power in detecting visual cortex activity than InI. SMS-InI also detected subcortical fMRI signals with the similar sensitivity and localization accuracy like EPI. The spatiotemporal resolution of SMS-InI was used to reveal that presenting visual stimuli with 0.2 s latency between left and right visual hemifield led to 0.2 s relative hemodynamic response latency between the left and right visual cortices. Together, these results indicate that SMS-InI is a useful tool in measuring cortical and subcortical hemodynamic responses with high spatiotemporal resolution.
Penalized least-squares iterative image reconstruction algorithms used for spatial resolution limited imaging, such as functional magnetic resonance imaging (fMRI), commonly use a quadratic roughness penalty to regularize the reconstructed images. When used for complex-valued images, the conventional roughness penalty regularizes the real and imaginary parts equally. However, these imaging methods sometimes benefit from separate penalties for each part. The spatial smoothness from the roughness penalty on the reconstructed image is dictated by the regularization parameter(s). One method to set the parameter to a desired smoothness level is to evaluate the full width at half maximum (FWHM) of the reconstruction method’s local impulse response. Previous work has shown that when using the conventional quadratic roughness penalty one can approximate the local impulse response using an FFT-based calculation. However, that acceleration method cannot be applied directly for separate real and imaginary regularization. This paper proposes a fast and stable calculation for this case that also uses FFT-based calculations to approximate the local impulse responses of the real and imaginary parts. This approach is demonstrated with a quadratic image reconstruction of fMRI data that uses separate roughness penalties for the real and imaginary parts.
The development of highly accelerated fMRI acquisition techniques has led to novel possibilities to monitor cerebral activity non-invasively and with unprecedented temporal resolutions. With the emergence of dynamic connectivity and its ability to provide a much richer characterization of brain function compared to static measures, fast fMRI may yet play a crucial role in tracking dynamically varying networks. In spite of the dominance of slow hemodynamic contributions to the BOLD signal, high temporal sampling rates nevertheless improve the measurement of physiological noise, yielding an exceptional sensitivity for the detection of periods of transient connectivity at time scales of a few tens of seconds. There is also evidence that relevant BOLD fluctuations are detectable at high frequencies, implying that the benefits of fast fMRI extend beyond the ability to sample nuisance confounds. Here we review the latest technological advancements that have established fast fMRI as an effective acquisition technique, as well as its current and future implications on the analysis of dynamic networks.
Introduction: As the number of receive array channels (N) increases, theoretical analysis predicts that it is possible to reduce the size of the individual elements and get increasing SNR gains near the elements without losing sensitivity further from the coils, in addition to encoding acceleration performance (1). These theoretical studies are usually performed assuming ideal decoupling between the elements and body load dominant conditions. However, a 90 channel head array that we constructed for 1.5 showed a central SNR which dropped significantly compared to a 23 channel array of similar geometry with fewer, larger coil elements (2). We attributed this to poor loading of the small surface coil elements which could be readily addressed with the increased loading at higher field. In this work we address additional loss mechanisms unique to large-N arrays and the effect of an increased number of coupling partners as N is increased We have evaluated two designs for a 96 channel head array for 3T with particular attention to sources of SNR loss in the small coils, including various sources of coil loading and SNR deterioration due to coupling between coil elements. Our overall goal was for the larger-N array of small coils to gain peripheral SNR and at least equal the central SNR of a 32 channel helmet of similar geometry (3). Design changes aimed at reducing coupling among the channels and reducing losses due to induced eddy currents in the copper of adjacent coil elements were found to be necessary to achieve this goal. The designs were evaluated in phantom and human SNR and imaging measurements Methods: The coil arrays were tested on a modified Tim Trio 32 channel 3T clinical scanner (Siemens Medical Solutions, Erlangen Germany). The system was expanded to 128 channels by the addition of 3 more sets of 32 receivers, fully integrated within the system architecture. The coil elements were critically overlapped to minimize inductive coupling, and arranged on a close-fitting helmet using the soccer-ball element geometry described previously for 1.5T (2). Coil element designs were evaluated for loaded and unloaded Q both alone and within the array. Two types of elements were evaluated, both nearly 5cm diameter: 1) "circuit board elements" constructed from 5mm width G10 copper clad circuit board (1 Oz / ft 2) with 6 or 2 capacitors per coil with a standard capacitive match and preamp decoupling implemented through control of the length of coax (37cm) between the coil element and the preamp, and 2) "wire elements" constructed of bent 16awg copper wire, with 2 capacitors per coil, a series match and separate detuning circuit, and a new design of preamp with built in cable trap (Siemens Medical Solutions, Erlangen Germany) attached to the coil with a short (5cm) length of semi rigid coax (Figs 1, 2). SNR comparisons were made using proton density gradient echo images (TR/TE/flip/slice = 200ms/4.07ms/20deg/3mm, 256x256, FoV=220mm) obtained in human scans using the prototype arrays and an in-house 32 channel head coil with identical helmet dimensions (3).
INTRODUCTION. Parallel MRI has been introduced has a method to accelerate the encoding process by sub-sampling k-space while maintaining the total extent [1-2]. The rationale for this sub-encoding scheme is that the coil sensitivity maps are very smooth and retrieve k-space information only from the neighborhood of the actual gradient-encoding point. New array coil designs with a large number of small elements [3-4] provide strong variation of the coil sensitivities which may result in residual aliasing artifacts in parallel MRI due to sensitivity variation within the image voxel [5]. In this work, a novel parallel MRI method known as superresolution SENSE (SURE-SENSE) is proposed as an alternative to standard SENSE using coils with strongly varying coil sensitivities. Acceleration is performed by acquiring the low spatial resolution representation of the object being imaged and the coil sensitivity maps are acquired with much higher target spatial resolution. The increase in spatial resolution will be determined by the degree of coil sensitivity variation within the low resolution voxel and the SNR in the low resolution object image. We show feasibility of the method for human brain imaging using receiver arrays with 32 and 96 RF elements. THEORY AND METHODS. The forward model y=Es for SURE-SENSE is constructed assuming that y is the multi-coil low spatial resolution representation of s. A conjugate gradient algorithm [6] is used to solve the computationally intensive inverse problem. The encoding matrix E for SURE-SENSE is poorly conditioned since the variation of the coil sensitivities within the acquired voxel is lower than across larger distances which is case for sub-sampled acquisitions. Pre-conditioning [7] is employed to regularize the conjugate gradient solution by solving the transformed system M -1 E H Es=M -1 E H y, where M is a matrix that approximates E H E, but it is easier to invert. For M we used the matrix E H E for the case of fully-encoding where we have a diagonal matrix with entries given by the sum of coil sensitivity squares. The transformation results in inverting a well-conditioned encoding matrix with all the singular values clustered around a single point. Pre-conditioning will enforce low values of g-factor at the expense of attenuating high spatial frequencies in the solution. The maximum spatial resolution is the limited by the g-factor reduction by regularization that is consistent with acceptable SNR in the reconstructed high resolution image. Human brain data were acquired with a 3 Tesla MR Siemens scanner using close-fitting helmet array coils with 32 [3] and 96 [4] elements. A GRE sequence was employed with a 256×256 spatial matrix and a FOV of 220×220 mm 2 , resulting in in-plane spatial resolution of 0.86 mm 2 . Low spatial resolution data was extracted from the central k-space region of the acquisition. Coil sensitivity maps were estimated from the fully-encoded acquisition using a 4 th order polynomial fitting optimized for the 32-channel array coil [2]. For comparison purposes, the fully-encoded and the sub-encoded data were conventionally reconstructed using FFT and sensitivity-weighted combination. Reconstruction at lower spatial resolution, which represents higher intra-voxel sensitivity variation, was simulated to compare the performance of standard SENSE and SURE-SENSE for the same acceleration factor. Fully-encoded low resolution data was obtained from the central 64×64 k-space matrix. Standard SENSE reconstruction was applied to sub-sampled data obtained by decimating the 64×64 data with a factor R=4×4. SURE-SENSE reconstruction was applied to the central 16×16 k-space matrix. The reconstruction results were interpolated to a 256×256 matrix by using zero-filling in k-space. RESULTS. Fig. 1 shows the reconstruction of human brain MRI data. The average error with respect to the fully-encoded reconstruction was 9.6% for the 32-channel data (R=16) and 10.7% for the 96-channel data (R=64). Note that the reconstruction using the 96-channel array is recovering more spatial features as suggested by the uniformity of the error image. This highlights the advantages of the stronger variation of the coil sensitivities with this array. Standard SENSE reconstruction with intra-voxel sensitivity variation presented residual aliasing artifacts, especially periphery where sensitivity variation is stronger (Fig. 2). Superresolution SENSE is free of residual aliasing artifacts at the expense of a small loss in the final spatial resolution. The average error for SENSE was 18.1% while for SURE-SENSE was 7.5%. Fig. 1: Conventional reconstruction of fully-encoded data and SURE-SENSE reconstruction of sub-encoded data for a) 32-channel array with R=4x4, and b) 96-channel array with R=8x8.
Our contribution briefly outlines the basics of the well-established technique in data mining, namely the principal component analysis (PCA), and a rapidly emerging novel method, that is, the independent component analysis (ICA). The performance of PCA singular value decomposition-based and stationary linear ICA in blind separation of artificially generated data out of linear mixtures was critically evaluated and compared. All our results outlined the superiority of ICA relative to PCA in faithfully retrieval of the original independent source components. Key words: principal component analysis (PCA), independent component analysis (ICA), blind source separation (BSS), and higher-order statistics.
We introduce a novel fast algorithm for independent component analysis, which can be used for blind source separation and feature extraction. We show how a neural network learning rule can be transformed into a fixedpoint iteration, which provides an algorithm that is very simple, does not depend on any user-defined parameters, and is fast to converge to the most accurate solution allowed by the data. The algorithm finds, one at a time, all nongaussian independent components, regardless of their probability distributions. The computations can be performed in either batch mode or a semiadaptive manner. The convergence of the algorithm is rigorously proved, and the convergence speed is shown to be cubic. Some comparisons to gradient-based algorithms are made, showing that the new algorithm is usually 10 to 100 times faster, sometimes giving the solution in just a few iterations.
Echo planar imaging (EPI) is an MRI technique of particular value to neuroscience, with its use for virtually all functional MRI (fMRI) and diffusion imaging of fiber connections in the human brain. EPI generates a single 2D image in a fraction of a second; however, it requires 2-3 seconds to acquire multi-slice whole brain coverage for fMRI and even longer for diffusion imaging. Here we report on a large reduction in EPI whole brain scan time at 3 and 7 Tesla, without significantly sacrificing spatial resolution, and while gaining functional sensitivity. The multiplexed-EPI (M-EPI) pulse sequence combines two forms of multiplexing: temporal multiplexing (m) utilizing simultaneous echo refocused (SIR) EPI and spatial multiplexing (n) with multibanded RF pulses (MB) to achieve m×n images in an EPI echo train instead of the normal single image. This resulted in an unprecedented reduction in EPI scan time for whole brain fMRI performed at 3 Tesla, permitting TRs of 400 ms and 800 ms compared to a more conventional 2.5 sec TR, and 2-4 times reductions in scan time for HARDI imaging of neuronal fibertracks. The simultaneous SE refocusing of SIR imaging at 7 Tesla advantageously reduced SAR by using fewer RF refocusing pulses and by shifting fat signal out of the image plane so that fat suppression pulses were not required. In preliminary studies of resting state functional networks identified through independent component analysis, the 6-fold higher sampling rate increased the peak functional sensitivity by 60%. The novel M-EPI pulse sequence resulted in a significantly increased temporal resolution for whole brain fMRI, and as such, this new methodology can be used for studying non-stationarity in networks and generally for expanding and enriching the functional information.
Independent component analysis (ICA) has proven its applicability in both standard and resting-state fMRI. While there is consensus on single-subject ICA methodology, the extension to group ICA is more complex and a number of approaches have been suggested. Currently, two software packages are most frequently used for ICA group analysis: (1) GIFT introduced by Calhoun et al., and (2) PICA, proposed by Beckmann et al.. Both methods are based on the assumption of statistical independence of the extracted component maps ("spatial ICA"). Group maps are estimated via ICA on pre-calculated group data sets.
In this study, we applied the two analysis approaches to a group of fMRI resting-state data sets obtained from twenty-eight healthy subjects. Default implementations were used and the number of components was restricted to 5, 10, 15, 20, 25, 30, and 35. The performance of GIFT and PICA was assessed with respect to the number of resting-state networks detected at different component estimation levels and computational load.
At low component estimation levels GIFT analysis resulted in more RSNs than PICA, while for individually determined component levels both approaches obtained the same RSNs. Although component maps show some variability across the two methods, spatial and temporal comparison using correlation coefficients resulted in no significant differences between the RSNs detected across the different analyses
Our results show that both approaches provide an adequate way of group ICA obtaining a comparable number of RSNs differing mainly in calculation times.
Echo-volumar imaging is a hyperfast technique capable of producing volumetric magnetic resonance images in times of the order of 100 ms. By increasing the gradient strengths and introducing real-time processing and display, we have been able to produce the first 16×16×16 voxel snapshot head images on volunteers.
A new functional imaging method has been developed and used to measure T2*- and initial intensity (I0)-parameter maps of multiple slices during activation of the cortex with high temporal resolution. Multiple echo-planar images (EPls) are read out after a single excitation of the spin system, leading to several images with increasing effective echo times. Changes induced by primary visual stimulation in T2* and I0 were measured in eight subjects. Although stimulation-induced increases in I0 only occurred at short repetition times, T2* increased from 57.3 to 60.9 ms on average. The method combines the high stability of a single shot EPI experiment with the high information content of a multiecho acquisition. In addition, stimulation-induced changes in inflow can be easily separated from true blood oxygenation level dependent (BOLD) signal changes.
The principles of echo-volumar imaging (EVI) are introduced and it is shown that full volume images can be obtained very rapidly by this means. Four variants of EVI are discussed: a four-shot technique, two snapshot methods and a zoomed snapshot method. Experimental results are presented for one snapshot method and the four-shot technique. These are relatively coarse images obtained with phantoms at 4.0 MHz and serve as a first experimental demonstration of the method.
A new method of spin-density imaging by NMR is described which allows simultaneous observation and differentiation of signals arising from spins distributed throughout a thin layer or plane within the specimen. The method, which is based on selective rf irradiation of the sample in switched magnetic field gradients, can produce visual pictures considerably faster than previously described line-scan imaging methods. Some simple examples of two-dimensional images obtained by the method are presented.
Functional MR imaging (fMRI) of activation in the brain is one of the more demanding applications required of an NMR imaging instrument. Since the signal changes in activation imaging are quite small, any instrumental variations can easily compromise the fMRI contrast. The authors describe a simple method for measuring these scanner instabilities, which is implementable on all scanners, to determine whether instability is degrading the fMRI contrast-to-noise. In this method, a long time series of images is acquired with identical imaging parameters to the chosen fMRI scans, the fluctuations are measured as a function of region-of-interest (ROI) size, and compared with the single image SNR. By plotting these fluctuations versus ROI size, and comparing them with the theoretically approachable value, the impact of these fluctuations can be assessed. We also present a simplified version of the test, which can be implemented with minimal calculations.
Real-time fMRI is a rapidly emerging methodology that enables monitoring changes in brain activity during an ongoing experiment. In this article we demonstrate the feasibility of performing single-event sensory, motor, and higher cognitive tasks in real-time on a clinical whole-body scanner. This approach requires sensitivity optimized fMRI methods: Using statistical parametric mapping we quantified the spatial extent of BOLD contrast signal changes as a function of voxel size and demonstrate that sacrificing spatial resolution and readout bandwidth improves the detection of signal changes in real time. Further increases in BOLD contrast sensitivity were obtained by using real-time multi-echo EPI. Real-time image analysis was performed using our previously described Functional Imaging in REal time (FIRE) software package, which features real-time motion compensation, sliding window correlation analysis, and automatic reference vector optimization. This new fMRI methodology was validated using single-block design paradigms of standard visual, motor, and auditory tasks. Further, we demonstrate the sensitivity of this method for online detection of higher cognitive functions during a language task using single-block design paradigms. Finally, we used single-event fMRI to characterize the variability of the hemodynamic impulse response in primary and supplementary motor cortex in consecutive trials using single movements. Real-time fMRI can improve reliability of clinical and research studies and offers new opportunities for studying higher cognitive functions. Hum. Brain Mapping 12:25–41, 2001. © 2001 Wiley-Liss, Inc.
In functional magnetic resonance imaging (fMRI) head motion can corrupt the signal changes induced by brain activation. This paper describes a novel technique called Prospective Acquisition CorrEction (PACE) for reducing motion-induced effects on magnetization history. Full three-dimensional rigid body estimation of head movement is obtained by image-based motion detection to a high level of accuracy. Adjustment of slice position and orientation, as well as regridding of residual volume to volume motion, is performed in real-time during data acquisition. Phantom experiments demonstrate a high level of consistency (translation < 40μm; rotation < 0.05°) for detected motion parameters. In vivo experiments were carried out and they showed a significant decrease of variance between successively acquired datasets compared to retrospective correction algorithms. Magn Reson Med 44:457–465, 2000. © 2000 Wiley-Liss, Inc.
New algorithms for correlation analysis are presented that allow the mapping of brain activity from functional MRI (fMRI) data in real time during the ongoing scan. They combine the computation of the correlation coefficients between measured fMRI time-series data and a reference vector with “detrending,” a technique for the suppression of non-stimulus-related signal components, and the “sliding-window technique.” Using this technique, which limits the correlation computation to the last N measurement time points, the sensitivity to changes in brain activity is maintained throughout the whole experiment. For increased sensitivity in activation detection a fast and robust optimization of the reference vector is proposed, which takes into account a realistic model of the hemodynamic response function to adapt the parameterized reference vector to the measured data. Based on the described correlation method, real-time fMRI experiments using visual stimulation paradigms have been performed successfully on a clinical MR scanner, which was linked to an external workstation for image analysis. Magn Reson Med 43:259–268, 2000. © 2000 Wiley-Liss, Inc.
The physiological noise in the resting brain, which arises from fluctuations in metabolic-linked brain physiology and subtle brain pulsations, was investigated in six healthy volunteers using oxygenation-sensitive dual-echo spiral MRI at 3.0 T. In contrast to the system and thermal noise, the physiological noise demonstrates a signal strength dependency and, unique to the metabolic-linked noise, an echo-time dependency. Variations of the MR signal strength by changing the flip angle and echo time allowed separation of the different noise components and revealed that the physiological noise at 3.0 T (1) exceeds other noise sources and (2) is significantly greater in cortical gray matter than in white matter regions. The SNR in oxygenation-sensitive MRI is predicted to saturate at higher fields, suggesting that noise measurements of the resting brain at 3.0 T and higher may provide a sensitive probe of functional information. Magn Reson Med 46:631–637, 2001. © 2001 Wiley-Liss, Inc.
PURPOSE: To combine parallel imaging with 3D single-shot acquisition (echo volumar imaging, EVI) in order to acquire high temporal resolution volumar functional MRI (fMRI) data. MATERIALS AND METHODS: An improved EVI sequence was associated with parallel acquisition and field of view reduction in order to acquire a large brain volume in 200 msec. Temporal stability and functional sensitivity were increased through optimization of all imaging parameters and Tikhonov regularization of parallel reconstruction. Two human volunteers were scanned with parallel EVI in a 1.5T whole-body MR system, while submitted to a slow event-related auditory paradigm. RESULTS: Thanks to parallel acquisition, the EVI volumes display a low level of geometric distortions and signal losses. After removal of low-frequency drifts and physiological artifacts, activations were detected in the temporal lobes of both volunteers and voxelwise hemodynamic response functions (HRF) could be computed. On these HRF different habituation behaviors in response to sentence repetition could be identified. CONCLUSION: This work demonstrates the feasibility of high temporal resolution 3D fMRI with parallel EVI. Combined with advanced estimation tools, this acquisition method should prove useful to measure neural activity timing differences or study the nonlinearities and nonstationarities of the BOLD response.
This paper presents a general approach to the analysis of functional MRI time-series from one or more subjects. The approach is predicated on an extension of the general linear model that allows for correlations between error terms due to physiological noise or correlations that ensue after temporal smoothing. This extension uses the effective degrees of freedom associated with the error term. The effective degrees of freedom are a simple function of the number of scans and the temporal autocorrelation function. A specific form for the latter can be assumed if the data are smoothed, in time, to accentuate hemodynamic responses with a neural basis. This assumption leads to an expedient implementation of a flexible statistical framework. The importance of this small extension is that, in contradistinction to our previous approach, any parametric statistical analysis can be implemented. We demonstrate this point using a multiple regression analysis that tests for effects of interest (activations due to word generation), while taking explicit account of some obvious confounds.
Physiological noise arising from a variety of sources can significantly degrade the detection of task-related activity in BOLD-contrast fMRI experiments. If whole head spatial coverage is desired, effective suppression of oscillatory physiological noise from cardiac and respiratory fluctuations is quite difficult without external monitoring, since traditional EPI acquisition methods cannot sample the signal rapidly enough to satisfy the Nyquist sampling theorem, leading to temporal aliasing of noise. Using a combination of high speed magnetic resonance inverse imaging (InI) and digital filtering, we demonstrate that it is possible to suppress cardiac and respiratory noise without auxiliary monitoring, while achieving whole head spatial coverage and reasonable spatial resolution. Our systematic study of the effects of different moving average (MA) digital filters demonstrates that a MA filter with a 2 s window can effectively reduce the variance in the hemodynamic baseline signal, thereby achieving 57%-58% improvements in peak z-statistic values compared to unfiltered InI or spatially smoothed EPI data (FWHM = 8.6 mm). In conclusion, the high temporal sampling rates achievable with InI permit significant reductions in physiological noise using standard temporal filtering techniques that result in significant improvements in hemodynamic response estimation. Hum Brain Mapp, 2011. © 2011 Wiley-Liss, Inc.
MR-Encephalography (MREG) is a technique that allows real time observation of functional changes in the brain that appears within 100 msec. The high sampling rate is achieved at the cost of some spatial resolution. The article describes a novel imaging method for fast three-dimensional-MR-encephalography whole brain coverage based on rosette trajectories and the use of multiple small receiver coils. The technique allows the observation of changes in brain physiology at very high temporal resolution. A highly undersampled three-dimensional rosette trajectory is chosen, to perform single shot acquisition of k-space data within 23 msec. By using a 32-channel head coil array and regularized nonuniform Fourier transformation reconstruction, the spatial resolution is sufficient to detect even subtle centers of activation (e.g. human MT+). The method was applied to visual block design paradigms and compared with echo planar imaging-based functional MRI. As a proof-of-principle of the method's ability to detect local differences in the hemodynamic response functions, the analyzed MR-encephalography data revealed a spatially dependent delay of the arrival of the blood oxygenation level dependent response within the visual cortex.
Here we report early cross-sensory activations and audiovisual interactions at the visual and auditory cortices using magnetoencephalography (MEG) to obtain accurate timing information. Data from an identical fMRI experiment were employed to support MEG source localization results. Simple auditory and visual stimuli (300-ms noise bursts and checkerboards) were presented to seven healthy humans. MEG source analysis suggested generators in the auditory and visual sensory cortices for both within-modality and cross-sensory activations. fMRI cross-sensory activations were strong in the visual but almost absent in the auditory cortex; this discrepancy with MEG possibly reflects the influence of acoustical scanner noise in fMRI. In the primary auditory cortices (Heschl's gyrus) the onset of activity to auditory stimuli was observed at 23 ms in both hemispheres, and to visual stimuli at 82 ms in the left and at 75 ms in the right hemisphere. In the primary visual cortex (Calcarine fissure) the activations to visual stimuli started at 43 ms and to auditory stimuli at 53 ms. Cross-sensory activations thus started later than sensory-specific activations, by 55 ms in the auditory cortex and by 10 ms in the visual cortex, suggesting that the origins of the cross-sensory activations may be in the primary sensory cortices of the opposite modality, with conduction delays (from one sensory cortex to another) of 30-35 ms. Audiovisual interactions started at 85 ms in the left auditory, 80 ms in the right auditory and 74 ms in the visual cortex, i.e., 3-21 ms after inputs from the two modalities converged.
We present a method for detecting event-related responses in functional magnetic resonance imaging (fMRI). The occurrence of time-locked activations is formulated in terms of the general linear model, i.e., multiple linear regression. This permits the use of established statistical techniques that correct for multiple comparisons in the context of spatially smooth and serially correlated data. Responses are modelled using event-related temporal basis functions. Inferences are then made about all components of the model, using the F-ratio at all voxels in the image, to produce a statistical parametric map (SPM{F}). This method allows for the experimental design to relate the timing of events to the acquisition of data to give a temporal resolution (with respect to the event-related response) far better than the scanning repeat time.
Functional MRI (fMRI) most commonly employs 2D echo-planar imaging (EPI). The advantages for fMRI brought about by the increasingly popular ultra-high field strengths are best exploited in high-resolution acquisitions, but here 2D EPI becomes impractical for several reasons, including the very long volume acquisitions times. In this study at 7 T, a 3D EPI sequence with full parallel and partial Fourier imaging capability along both phase encoding axes was implemented and used to evaluate the sensitivity of 3D and corresponding 2D EPI acquisitions at four different spatial resolutions ranging from small to typical voxel sizes (1.5-3.0 mm isotropic). Whole-brain resting state measurements (N=4) revealed a better, or at least comparable sensitivity of the 3D method for gray and white matter. The larger vulnerability of 3D to physiological effects was outweighed by the much shorter volume TR, which moreover allows whole-brain coverage at high resolution within fully acceptable limits for event-related fMRI: TR was only 3.07 s for 1.5 mm, 1.88 s for 2.0 mm, 1.38 s for 2.5 mm and 1.07 s for 3.0 mm isotropic resolution. In order to investigate the ability to detect and spatially resolve BOLD activation in the visual cortex, functional 3D EPI experiments (N=8) were performed at 1 mm isotropic resolution with parallel imaging acceleration of 3x3, resulting in a TR of only 3.2 s for whole-brain coverage. From our results, and several other practical advantages of 3D over 2D EPI found in the present study, we conclude that 3D EPI provides a useful alternative for whole-brain fMRI at 7 T, not only when high-resolution data are required.
Using simultaneous measurements from multiple channels of a radio-frequency coil array, magnetic resonance inverse imaging (InI) can achieve ultra-fast dynamic functional imaging of the human with whole-brain coverage and a good spatial resolution. Mathematically, the InI reconstruction is a generalization of parallel MRI (pMRI), which includes image space and k-space reconstructions. Because of the auto-calibration technique, the pMRI k-space reconstruction offers more robust and adaptive reconstructions compared to the image space algorithm. Here we present the k-space InI (K-InI) reconstructions to reconstruct the highly accelerated BOLD-contrast fMRI data of the human brain to achieve 100 ms temporal resolution. Simulations show that K-InI reconstructions can offer 3D image reconstructions at each time frame with reasonable spatial resolution, which cannot be obtained using the previously proposed image space minimum-norm estimates (MNE) or linear constraint minimum variance (LCMV) spatial filtering reconstructions. The InI reconstructions of in vivo BOLD-contrast fMRI data during a visuomotor task show that K-InI offer 3 to 5 fold more sensitive detection of the brain activation than MNE and a comparable detection sensitivity to the LCMV reconstructions. The group average of the high temporal resolution K-InI reconstructions of the hemodynamic response also shows a relative onset timing difference between the visual (first) and somatomotor (second) cortices by 400 ms (600 ms time-to-peak timing difference). This robust and sensitive K-InI reconstruction can be applied to dynamic MRI acquisitions using a large-n coil array to improve the spatiotemporal resolution.
This work describes a novel method for highly undersampled projection imaging using constrained reconstruction by Tikhonov-Phillips regularization and its application for high temporal resolution functional MRI (fMRI) at a repetition time of 80 ms. The high-resolution reference image used as in vivo coil sensitivity is acquired in a separate acquisition using otherwise identical parameters. Activation studies using a standard checkerboard activation paradigm demonstrate the inherent high sensitivity afforded by the possibility to separate activation-related effects from "physiological noise.". In this first proof-of-principle of the constrained reconstruction based on regularization using arbitrary projections (COBRA) technique, experiments are performed in a single-slice mode, which allows for a comparison with fast single-slice echo-planar imaging (EPI) at equal temporal resolution. The COBRA method can be extended to three-dimensional (3D) encoding without severe penalty in temporal performance. Analysis of the global signal change demonstrates the excellent reproducibility of COBRA compared to standard EPI. Activation analysis is considerably improved by the possibility to remove electrocardiogram (ECG)-related and breathing-related signal fluctuations by physiological correction of each individual breathing and ECG cycle, respectively.
Standard parallel magnetic resonance imaging (MRI) techniques suffer from residual aliasing artifacts when the coil sensitivities vary within the image voxel. In this work, a parallel MRI approach known as Superresolution SENSE (SURE-SENSE) is presented in which acceleration is performed by acquiring only the central region of k-space instead of increasing the sampling distance over the complete k-space matrix and reconstruction is explicitly based on intra-voxel coil sensitivity variation. In SURE-SENSE, parallel MRI reconstruction is formulated as a superresolution imaging problem where a collection of low resolution images acquired with multiple receiver coils are combined into a single image with higher spatial resolution using coil sensitivities acquired with high spatial resolution. The effective acceleration of conventional gradient encoding is given by the gain in spatial resolution, which is dictated by the degree of variation of the different coil sensitivity profiles within the low resolution image voxel. Since SURE-SENSE is an ill-posed inverse problem, Tikhonov regularization is employed to control noise amplification. Unlike standard SENSE, for which acceleration is constrained to the phase-encoding dimension/s, SURE-SENSE allows acceleration along all encoding directions--for example, two-dimensional acceleration of a 2D echo-planar acquisition. SURE-SENSE is particularly suitable for low spatial resolution imaging modalities such as spectroscopic imaging and functional imaging with high temporal resolution. Application to echo-planar functional and spectroscopic imaging in human brain is presented using two-dimensional acceleration with a 32-channel receiver coil.
The fundamental limit for NMR imaging is set by an intrinsic signal-to-noise ratio (SNR) for a particular combination of rf antenna and imaging subjects. The intrinsic SNR is the signal from a small volume of material in the sample competing with electrical noise from thermally generated, random noise currents in the sample. The intrinsic SNR has been measured for a number of antenna-body section combinations at several different values of the static magnetic field and is proportional to B0. We have applied the intrinsic and system SNR to predict image SNR and have found satisfactory agreement with measurements on images. The relationship between SNR and pixel size is quite different in NMR than it is with imaging modalities using ionizing radiation, and indicates that the initial choice of pixel size is crucial in NMR. The analog of "contrast-detail-dose" plots for ionizing radiation imaging modalities is the "contrast-detail-time" plot in NMR, which should prove useful in choosing a suitable pixel array to visualize a particular anatomical detail for a given NMR receiving antenna.
A method is described for the correction of geometric distortions occurring in echo planar images. The geometric distortions are caused in large part by static magnetic field inhomogeneities, leading to pixel shifts, particularly in the phase encode direction. By characterizing the field inhomogeneities from a field map, the image can be unwarped so that accurate alignment to conventionally collected images can be made. The algorithm to perform the unwarping is described, and results from echo planar images collected at 1.5 and 4 Tesla are shown.
Two single-shot volume imaging techniques are described. The first, single-echo echo-volume imaging, is essentially the echo-volume imaging (EVI) sequence suggested by Mansfield (J. Phys. C. 10, L55 (1977)). The second is a multi-spin-echo approach in which one plane of k-space is collected during each spin echo. In both techniques, phase encoding gradients are applied in the z direction, and three-dimensional k-space is filled by a raster pattern in Cartesian coordinates. Spatial saturation is used to avoid aliasing in the y direction, and a selctive pulse is applied to excite the desired slab of tissue and eliminate aliasing in z. The average echo-times, measured from the center of the 90 degrees pulse to the center of the acquisition k-space (kx = ky = kz = 0), were 45 and 104 ms for single echo and multi-echo methods, respectively. Images of the human brain using both sequences are shown.
To perform true three-dimensional activation experiments in the human brain, dedicated localized echo-volume imaging (L-EVI) methods were developed. Three-dimensional acquisition allows generation of activation maps with minimal vascular enhancement related to inflow effects. The rapid acquisition of the L-EVI (approximately 100 msec) reduces signal instabilities caused by motion, facilitating the detection of the small intensity changes expected with brain activation. Single-shot L-EVI was performed on normal volunteers at 1.5 T, imaging a three-dimensional predefined volume (240 x 45 x 45 mm3) in the superior portion of the brain with a spatial resolution of 3.75 x 5 x 5 mm3. Increased brain coverage was achieved with a multi-volume imaging (three-shot) version, which simultaneously achieved effective suppression of signals from cerebrospinal fluid. In addition, both asymmetric spin-echo (ASE) and spin-echo (SE) versions of the technique were used to detect blood oxygenation level dependent (BOLD) signal changes in the motor cortex with a finger-tapping paradigm. Images obtained by the L-EVI sequence were qualitatively comparable to standard multislice two-dimensional echo-planar images. Both ASE and SE functional MRI (fMRI) experiments showed consistent activation in the contralateral primary sensorimotor cortex. Furthermore, significant differences in location and magnitude of activation was observed between the two methods, confirming theoretical predictions.
A new functional imaging method has been developed and used to measure T2*- and initial intensity (I0)-parameter maps of multiple slices during activation of the cortex with high temporal resolution. Multiple echo-planar images (EPIs) are read out after a single excitation of the spin system, leading to several images with increasing effective echo times. Changes induced by primary visual stimulation in T2* and I0 were measured in eight subjects. Although stimulation-induced increases in I0 only occurred at short repetition times, T2* increased from 57.3 to 60.9 ms on average. The method combines the high stability of a single shot EPI experiment with the high information content of a multiecho acquisition. In addition, stimulation-induced changes in inflow can be easily separated from true blood oxygenation level dependent (BOLD) signal changes.
Improved data acquisition and processing strategies for blood oxygenation level-dependent (BOLD)-contrast functional magnetic resonance imaging (fMRI), which enhance the functional contrast-to-noise ratio (CNR) by sampling multiple echo times in a single shot, are described. The dependence of the CNR on T2*, the image encoding time, and the number of sampled echo times are investigated for exponential fitting, echo summation, weighted echo summation, and averaging of correlation maps obtained at different echo times. The method is validated in vivo using visual stimulation and turbo proton echoplanar spectroscopic imaging (turbo-PEPSI), a new single-shot multi-slice MR spectroscopic imaging technique, which acquires up to 12 consecutive echoplanar images with echo times ranging from 12 to 213 msec. Quantitative T2*-mapping significantly increases the measured extent of activation and the mean correlation coefficient compared with conventional echoplanar imaging. The sensitivity gain with echo summation, which is computationally efficient provides similar sensitivity as fitting. For all data processing methods sensitivity is optimum when echo times up to 3.2 T2* are sampled. This methodology has implications for comparing functional sensitivity at different magnetic field strengths and between brain regions with different magnetic field inhomogeneities.
Functional magnetic resonance imaging in real time is an emerging tool for the assessment of dynamic changes in brain activation. The short response latency (tens of seconds) renders the technique more sensitive to motion artifacts. Motion correction in real time requires computationally efficient algorithms which can be executed on a complete 3D data set within a single time of repetition cycle. In this study, a method to evaluate motion and realign functional images in real time implemented on standard imaging hardware is introduced. The detection of activity in correlation maps is improved, and artifactual edge enhancements are reduced. As the estimation of large movements is stable, this algorithm is attractive for clinical studies with uncooperative patients. Magn Reson Med 45:167-171, 2001.
Real-time fMRI is a rapidly emerging methodology that enables monitoring changes in brain activity during an ongoing experiment. In this article we demonstrate the feasibility of performing single-event sensory, motor, and higher cognitive tasks in real-time on a clinical whole-body scanner. This approach requires sensitivity optimized fMRI methods: Using statistical parametric mapping we quantified the spatial extent of BOLD contrast signal changes as a function of voxel size and demonstrate that sacrificing spatial resolution and readout bandwidth improves the detection of signal changes in real time. Further increases in BOLD contrast sensitivity were obtained by using real-time multi-echo EPI. Real-time image analysis was performed using our previously described Functional Imaging in REal time (FIRE) software package, which features real-time motion compensation, sliding window correlation analysis, and automatic reference vector optimization. This new fMRI methodology was validated using single-block design paradigms of standard visual, motor, and auditory tasks. Further, we demonstrate the sensitivity of this method for online detection of higher cognitive functions during a language task using single-block design paradigms. Finally, we used single-event fMRI to characterize the variability of the hemodynamic impulse response in primary and supplementary motor cortex in consecutive trials using single movements. Real-time fMRI can improve reliability of clinical and research studies and offers new opportunities for studying higher cognitive functions.
Independent component analysis (ICA) is a promising analysis method that is being increasingly applied to fMRI data. A principal advantage of this approach is its applicability to cognitive paradigms for which detailed models of brain activity are not available. Independent component analysis has been successfully utilized to analyze single-subject fMRI data sets, and an extension of this work would be to provide for group inferences. However, unlike univariate methods (e.g., regression analysis, Kolmogorov-Smirnov statistics), ICA does not naturally generalize to a method suitable for drawing inferences about groups of subjects. We introduce a novel approach for drawing group inferences using ICA of fMRI data, and present its application to a simple visual paradigm that alternately stimulates the left or right visual field. Our group ICA analysis revealed task-related components in left and right visual cortex, a transiently task-related component in bilateral occipital/parietal cortex, and a non-task-related component in bilateral visual association cortex. We address issues involved in the use of ICA as an fMRI analysis method such as: (1) How many components should be calculated? (2) How are these components to be combined across subjects? (3) How should the final results be thresholded and/or presented? We show that the methodology we present provides answers to these questions and lay out a process for making group inferences from fMRI data using independent component analysis.
The physiological noise in the resting brain, which arises from fluctuations in metabolic-linked brain physiology and subtle brain pulsations, was investigated in six healthy volunteers using oxygenation-sensitive dual-echo spiral MRI at 3.0 T. In contrast to the system and thermal noise, the physiological noise demonstrates a signal strength dependency and, unique to the metabolic-linked noise, an echo-time dependency. Variations of the MR signal strength by changing the flip angle and echo time allowed separation of the different noise components and revealed that the physiological noise at 3.0 T (1) exceeds other noise sources and (2) is significantly greater in cortical gray matter than in white matter regions. The SNR in oxygenation-sensitive MRI is predicted to saturate at higher fields, suggesting that noise measurements of the resting brain at 3.0 T and higher may provide a sensitive probe of functional information.
We propose a method for the statistical analysis of fMRI data that seeks a compromise between efficiency, generality, validity, simplicity, and execution speed. The main differences between this analysis and previous ones are: a simple bias reduction and regularization for voxel-wise autoregressive model parameters; the combination of effects and their estimated standard deviations across different runs/sessions/subjects via a hierarchical random effects analysis using the EM algorithm; overcoming the problem of a small number of runs/session/subjects using a regularized variance ratio to increase the degrees of freedom.
In this study, a novel partially parallel acquisition (PPA) method is presented which can be used to accelerate image acquisition using an RF coil array for spatial encoding. This technique, GeneRalized Autocalibrating Partially Parallel Acquisitions (GRAPPA) is an extension of both the PILS and VD-AUTO-SMASH reconstruction techniques. As in those previous methods, a detailed, highly accurate RF field map is not needed prior to reconstruction in GRAPPA. This information is obtained from several k-space lines which are acquired in addition to the normal image acquisition. As in PILS, the GRAPPA reconstruction algorithm provides unaliased images from each component coil prior to image combination. This results in even higher SNR and better image quality since the steps of image reconstruction and image combination are performed in separate steps. After introducing the GRAPPA technique, primary focus is given to issues related to the practical implementation of GRAPPA, including the reconstruction algorithm as well as analysis of SNR in the resulting images. Finally, in vivo GRAPPA images are shown which demonstrate the utility of the technique.
An algorithm using an orthogonalization procedure to estimate the coefficients of general linear models (GLM) for functional magnetic resonance imaging (fMRI) calculations is described. The idea is to convert the basis functions or explanatory variables of a GLM into orthogonal functions using the usual Gram-Schmidt orthogonalization procedure. The coefficients associated with the orthogonal functions, henceforth referred to as auxiliary coefficients, are then easily estimated by applying the orthogonality condition. The original GLM coefficients are computed from these estimates. With this formulation, the estimates can be updated when new image data become available, making the approach applicable for real-time estimation. Since the contribution of each image data is immediately incorporated into the estimated values, storing the data in memory during the estimation process becomes unnecessary, minimizing the memory requirements of the estimation process. By employing Cholesky decomposition, the algorithm is a factor of two faster than the standard recursive least-squares approach. Results of the analysis of an fMRI study using this approach showed the algorithm's potential for real-time application.
Balanced steady-state free precession (SSFP) imaging sequences require short repetition times (TRs) to avoid off-resonance artifacts. The use of slab-selective excitations is common, as this can improve imaging speed by limiting the field of view (FOV). However, the necessarily short-duration excitations have poor slab profiles. This results in unusable slices at the slab edge due to significant flip-angle variations or aliasing in the slab direction. Variable-rate selective excitation (VERSE) is a technique by which a time-varying gradient waveform is combined with a modified RF waveform to provide the same excitation profile with different RF power and duration characteristics. With the use of VERSE, it is possible to design short-duration pulses with dramatically improved slab profiles. These pulses achieve high flip angles with only minor off-resonance sensitivity, while meeting SAR limits at 1.5 T. The improved slab profiles will enable more rapid 3D imaging of limited volumes, with more consistent image contrast across the excited slab.
Echo-planar imaging (EPI) is an ultrafast magnetic resonance (MR) imaging technique prone to geometric distortions. Various correction techniques have been developed to remedy these distortions. Here improvements of the point spread function (PSF) mapping approach are presented, which enable reliable and fully automated distortion correction of echo-planar images at high field strengths. The novel method is fully compatible with EPI acquisitions using parallel imaging. The applicability of parallel imaging to further accelerate PSF acquisition is shown. The possibility of collecting PSF data sets with total acceleration factors higher than the number of coil elements is demonstrated. Additionally, a new approach to visualize and interpret distortions in the context of various imaging and reconstruction methods based on the PSF is proposed. The reliable performance of the PSF mapping technique is demonstrated on phantom and volunteer scans at field strengths of up to 4 T.
Previous studies have shown that under some conditions, noise fluctuations in an fMRI time-course are dominated by physiological modulations of the image intensity with secondary contributions from thermal image noise and that these two sources scale differently with signal intensity, susceptibility weighting (TE) and field strength. The SNR of the fMRI time-course was found to be near its asymptotic limit for moderate spatial resolution measurements at 3 T with only marginal gains expected from acquisition at higher field strengths. In this study, we investigate the amplitude of image intensity fluctuations in the fMRI time-course at magnetic field strengths of 1.5 T, 3 T, and 7 T as a function of image resolution, flip angle and TE. The time-course SNR was a similar function of the image SNR regardless of whether the image SNR was modulated by flip angle, image resolution, or field strength. For spatial resolutions typical of those currently used in fMRI (e.g., 3 x 3 x 3 mm(3)), increases in image SNR obtained from 7 T acquisition produced only modest increases in time-course SNR. At this spatial resolution, the ratio of physiological noise to thermal image noise was 0.61, 0.89, and 2.23 for 1.5 T, 3 T, and 7 T. At a resolution of 1 x 1 x 3 mm(3), however, the physiological to thermal noise ratio was 0.34, 0.57, and 0.91 for 1.5 T, 3 T and 7 T for TE near T2*. Thus, by reducing the signal strength using higher image resolution, the ratio of physiologic to image noise could be reduced to a regime where increased sensitivity afforded by higher field strength still translated to improved SNR in the fMRI time-series.