Traditional poster

We acquired high resolution multiple phase-cycled bSSFP fMRI datasets in rat brains at 9.4T and separated FID and main echo components based on Fourier analysis. The FID component showed stronger fMRI signals than the main echo component, but it showed stronger correlation with both intracortical veins and cortical surface veins. The fMRI signal from the main echo component could contribute up to half of that from the FID component, implying that the main echo component should not be neglected in interpreting pass-band bSSFP fMRI signals.

The use of multi-echo imaging techniques for contrast enhancement in BOLD fMRI is gaining momentum, particularly in real-time fMRI applications. A novel method is presented for combining multi-echo signals in weighted summation using principal component analysis (PCA) derived weights. The method is evaluated on human volunteers performing a simple motor task at 1.5 T, and is compared to other reported weighting schemes. The data driven PCA weighting method is demonstrated to produce time-series that achieve high relative contrast-to-noise ratio gain, without requiring additional data collection.

We propose a novel variable-density (VD) spiral k-space trajectory for brain fMRI. The trajectory consists of an Archimedean spiral from the origin out to an arbitrary radius k1, extending beyond k1 with a spiral in which the sampling density decreases as the k-space radius increases. Thus, it allows for a reduction in readout time at the expense of undersampling only high spatial frequencies. We implemented the VD spiral in a single-shot 2D spiral-in/out sequence for high (128x128) resolution, and demonstrate improved activation in a sensory-motor task compared to conventional (fully Archimedean) single-shot and interleaved sequences.

High resolution BOLD fMRI data on visual cortex were acquired using mHASTE, a novel GRAPPA accelerated single shot TSE technique, and the results were compared with those of GE- and SE-EPI. Higher SNR was obtained with mHASTE than both EPIs, and increased functional activation was detected by mHASTE but not by EPI when going from low resolution to high resolution. mHASTE was also found to have greater activation than both EPI in some cases,especially at high resolution, suggesting a more robust BOLD contrast mechanism for mHASTE in high resolution fMRI.

A method to improve spatial and temporal resolutions in fMRI using PROPELLER-EPI. First results are shown which demonstrate that a sliding window reconstruction of high resolution long-axis propeller (LAP) data is suitable for simple fMRI experiments. Additionally the results achieved by the LAP measurements are compared to the standard 64x64 EPI sequence which is usually used in fMRI. From there it is shown that the activation maps created from the LAP scans are better localized along the cortex.

The functional contrast in passband balanced steady-state free precession (pbSSFP) with 3D segmented EPI readout is compared to that of GRE-EPI and SE-EPI at 3T. For pbSSFP, TR is varied from 6.5 ms to 45 ms. Standard flickering checkerboard paradigm is used. We find that the best functional contrast is obtained at TR = 33 ms with corresponding EPI-factor of 40. At this TR, the functional contrast in pbSSFP is approximately half that of GRE-EPI and twice that of SE-EPI with otherwise comparable scan parameters. False detections due to banding artefacts are present in pbSSFP.

For isotropic high resolution fMRI at ultra-high field strength, susceptibility effects and T2* decay must be properly addressed. A combination of reduced FOV imaging (zoomed imaging) and parallel imaging is optimized here, achieving acceleration factors of up to 5.5. The high acceleration reduces distortions and image blurring, while incurring no other image artifacts. With this approach, high quality single-shot EPI acquisitions can be obtained with an isotropic resolution of 0.65 mm and sufficient coverage for e.g. fMRI in the visual cortex of the human brain.

Echo planar imaging (EPI) is the most widely used method for functional MRI. However, functional images are often distorted because EPI is highly sensitive to field inhomogeneities, eddy currents, and gradient delays. Functional and neuro-anatomical registration is complicated by these distortions and by the fact that functional and anatomical images are usually obtained with different imaging sequences. This work investigates the use of an optimized 3D flyback EPI trajectory with echo time shifting to obtain functional and anatomical images that have minimal distortions and are inherently co-registered.

Spiral pulse sequences are commonly used in fMRI, and spiral-in is known to be considerably better than spiral-out at signal recovery in regions with strong susceptibility field gradients. Previously proposed theories in the literature do not address the probability of signal displacement or fully explain all of the differences in signal recovery between spiral-out and spiral-in. In the current work we demonstrate that the difference in image intensity is not due to differences in signal displacement between spiral-in and spiral-out, but rather the increased phase coherence of the displaced pixels when using spiral-in.

A model for image variance due to motion is developed and validated. It can be used minimize motion effects by optimizing EPI sequence parameters. In general, variance is minimized by minimizing the partial derivative of the steady-state magnetization along the slice axis. In particular, sidelobes contribute much of the noise at high flip angles; an optimum flip angle exists for a specified degree of motion and can be computed; and inter-slice gaps increase variance due to motion rather than decrease it.

Although 2D single-shot EPI is common in BOLD fMRI, recent studies have suggested that 3D multi-shot sequences such as PRESTO-SENSE may offer superior BOLD CNR through improved temporal efficiency. A four-way comparison was performed between 2D and 3D acquisition sequences at two voxel resolutions (1.19x1.19x2 mm³ and 2.19x2.19x2 mm³) at 7T. The quality of fMRI data was evaluated via independent and unbiased metrics of prediction and reproducibility using NPAIRS. Results suggest that EPI provides higher prediction and reproducibility for this study. Future work will investigate within-subject optimization, and further compare EPI with PRESTO-SENSE for an fMRI study requiring whole-brain coverage.

The sensitivity to physiological noises of the reconstructed BOLD MR images is altered by the employed parallel imaging strategies. In this study, we mapped and compared the physiological noise sensitivity of BOLD fMRI data acquired with and without employing parallel imaging at two different spatial resolutions. Using higher spatial resolution reduces the signal strength and the relative sensitivity to physiological noise. This can be of SNR advantage particularly for time series fMRI data acquired at higher magnetic field.

Complex-valued tailored pulses have been used in an EPI sequence to excite two or more slices in human brain using a whole-brain transmit coil. One or more members of a receive-coil array have been used to acquire multi-channel image data. Complex-valued coil profiles have been used to recover slices, and parallel image formation has been demonstrated. Acceleration by a factor of 4 has been achieved with an eight-channel head coil.

Single-shot half k-space GR-EPI sequence was used to push the BOLD imaging resolution to 300 micron cubic voxel and unique single column cortical activation in the sensory cortex was detected when stimulating the middle phalange of all 8 digits of rat.

We compare single-shot gradient-echo 2D-EPI, multi-shot gradient-echo 3D-FFE and multi-shot gradient-echo 3D-PRESTO in a polar angle retinotopic mapping experiment at four isotropic resolutions (1.12mm3, 1.67mm3, 2mm3, and 3mm3) at 7 Tesla. Retinotopic maps in agreement with literature were obtained at all resolutions. The 3D sequences provided similar BOLD sensitivity and significantly less distortion compared to 2D-EPI. In addition, 3D-PRESTO provided much higher temporal resolution than 2D-EPI. Our findings suggest rich potential for high-resolution 3D imaging sequences in retinotopic mapping and other functional MRI experiments at high field.

To present auditory stimuli in the absence of scanner noise, the sparse temporal sampling (STS) approach was introduced. The interleaved-silent steady-state (ISSS) technique is combining the idea of splitting image acquisition and stimulus presentation with a better sampling of the fMRI signal. We performed an auditory experiment with pleasant and unpleasant stimuli using four fMRI sessions: STS, ISSS, and simultaneous stimulus presentation and image acquisition with axial and sagittal scanning. The total acquisition time was the same in all four sessions. The best sensitivity for detecting activations of sub-cortical regions (such as the amygdala) was found for ISSS.

It is expected that BOLD MRI should be sensitive to tumor vascular oxygenation. In lung tumors BOLD MRI is challenging due to potential image artifacts from motion, blood flow, and susceptibility. The goal of this preliminary study was to optimize the BOLD imaging technique at 3T in patients with untreated lung cancer. A respiratory-gated multi-echo gradient-echo technique is demonstrated as a feasible method to quantify T2* values of lung tumors. The response of tumor T2* measurements to an oxygen-breathing challenge should be sensitive to tumor hypoxia and could therefore serve as a prognostic indicator before therapy.

The present 2D respiratory-triggered interleaved T1- and T2*-weighted sequence provides a promising means to study TOLD and BOLD response simultaneously without the need for registration and with good temporal resolution (less than 30 seconds).

If the BOLD response to EPI acoustic noise changes over the time course of a standard fMRI experiment, both Type I and Type II errors can be made in fMRI group analysis. We used a pulse sequence based on single-voxel functional spectroscopy to silently measure the BOLD response induced by EPI-like scanner noise over about 40 minutes. No significant habituation or facilitation with respect to the scanner noise was found over that time. This result eliminates a possible confound for auditory and speech neuroimaging studies, especially those involving learning.

fMRI data can often be affected by issues such as registration and scanner system errors. These issues affect interpretation of data and is especially relevant in longitudinal studies depending on accurate reproducibility of data. This study compared different analysis techniques of fMRI data to establish the most accurate means of overcoming these issues by focusing on fMRI data from the somatosensory system in the human brain.

Where real time fMRI needs solely a digital (e.g., task vs rest or task 1 vs task 2) output, answer blocks can provide a means of circumventing the drift that is known to affect results. Using answer blocks, we improve performance compared to averaging over a block using classifiers from 53% to 84%, and using ROI techniques from 53% to 64%. This also gives a means for presenting probabilistic outputs to class membership, which are vital when dealing with impaired patients, for whom this kind of technique may be their only communication channel.

Correlation analysis of low-frequency fluctuations in blood-oxygen level dependent (BOLD) fMRI data is known to yield functional connectivity maps. The procedure, also referred to as ‘resting-state connectivity’, has previously been applied to optical tomography (OT) data using dense probe arrays. Here, we assess whether a sparser topographical sampling still yields results that are comparable to the ‘gold standard’ of resting-state network assessment, i.e. fMRI. In a first step, we used a subset of optical fibers (2-3cm inter-optode distance) covering both motor cortices and combined OT with concurrent fMRI measurements to cross-validate our resting-state data analysis.

MR venography using susceptibility weighted imaging (SWI) has been utilized for the study of venous morphology and the venous vasculature in disease states. However, SWI has poor intrinsic temporal resolution, thus it is not able to provide information on high temporal dynamic changes within a few seconds that occur during neural stimulation. In this paper, we proposed a technique, which we labeled functional MR venography (fMRV), to investigate the venous dynamic response to external stimulation using 7T MRI. The presented result suggests that this technique may provide important, more precise information regarding the venous response and its role in the overall hemodynamic response to neural activity with high spatial (0.5 isotropic) and temporal (3 seconds) resolution.

The signal loss by the susceptibility artifact makes it difficult analyze the fMRI data of the orbitofrontal region of the brain. In addition, an EPI sequence, which is used in many fMRI experiments, has the geometric distortion and Nyqust ghosts. For the fMRI experiment with correction of these artifact and distortion, we used a flat RF pulse which provides nearly constant signal intensity against the magnetic susceptibility and a gradient echo sequence. In the result, it is shown that the signal loss in the orbitofrontal region of the brain is recovered without geometric distortion and Nyqust ghosts and the activation in that region was analyzed successfully.

Anodal/cathodal tDCS have facilitatory/inhibitory effects, respectively, on the stimulated cortical networks. Here we used concurrent tDCS-fMRI to test whether anodal/cathodal tDCS result in BOLD-fMRI signal changes during a resting condition. Furthermore, we examined tDCS-effects on brain activation during voluntary finger tapping. Anodal/cathodal tDCS over left M1 induced no detectable BOLD signal change. However, anodal/cathodal tDCS combined with finger tapping resulted in a decreased BOLD response in SMA, but not M1, in comparison to voluntary finger tapping without stimulation. This suggests that in contrast to the rest condition the combination of neuronal polarization and motor activation induces inhibition in remote brain areas.

In a high field MRI, improvements in SNR and image quality are very noticeable, even though there are some drawbacks such as increased field inhomogeneity and relatively short T₂^*. For EPI-based methods, these drawbacks are major challenges in performing fMRI. To reduce such problems, we employed the HGS-Parallel technique as an fMRI imaging method which used the conventional gradient echo. This sequence is relatively robust to field inhomogeneity and the T₂^* decay than the EPI. With the HGS-Parallel technique, we performed an fMRI experiment at a 7T for mapping the finger somatosensory area in a high quality and resolution form.

Biofeedback based on a real-time fMRI is a promising tool especially in clinical research. In this study we show feasibility of controlling the left middle frontal gyrus by using mental imagery strategy.

We performed high-resolution pass-band bSSFP fMRI at multiple phase cycling angles on rat brain at 9.4T. Activation foci in fMRI maps shifted as a function of phase cycling angle and the location of the foci was correlated with hyperintense regions in corresponding baseline transition-band bSSFP, some of which were also correlated with cortical surface veins or intracortical veins. The results indicated that there is spatial heterogeneity in signal sources (T2 or T2*) of pass band bSSFP fMRI. Baseline transition band bSSFP could be used to predict outcomes of corresponding pass-band bSSFP fMRI maps.

We have recently demonstrated the technical feasibility and the potential advantages of combining transcranial magnetic stimulation (TMS) with continuous arterial spin labeling (CASL) imaging. Here, we use this novel approach to assess the effects of repetitive TMS applied to the left dorsal premotor cortex (PMd) on rCBF (regional cerebral blood flow) during different motor states. The state-dependent effects of left PMd rTMS on rCBF, is in concordance with previous results using BOLD imaging and a different task. As a next step, we will analyze the time dependence of the observed TMS effects across the different experimental blocks of one run.

There has been a longstanding need to develop techniques that improve data quality in fMRI by suppressing motion artifact. Head motion exceeding a few millimetres remains problematic and high interest participants including motor stroke patients often exceed this threshold. Here, a new technique is described that attempts to reduce participant head motion through visual feedback training in an fMRI simulator. Results from three stroke patients show that simulator training had a significant effect in suppressing head motion: (1) 11.25 mm before, 0.83 mm after; (2) 1.63 mm before, 0.67 mm after; (3) 4.47 mm before, 0.51 mm after.

Long durations in fMRI are typical, but that is unfeasible for specific populations. Scan-time reduction is possible if one could capitalize on the increased sensitivity afforded by high field strength or multiple channel phased arrays in the high-resolution regime. We evaluated this using a 32-channel coil at 3T with the n-back task on 18 subjects. Compared to 12-channel coil, working memory activation was significantly more (paired t-test) with two-thirds of the 32-channel data. Combination of 32-channel coil and high-resolution could imply lesser sample size, prevent additional data collection and enable studies that would otherwise be impossible due to time restrictions.

This study involved a comprehensive evaluation of commonly-used techniques like EPI and spiral-out, as well as techniques designed to recover signal in SFG regions using BOLD methods (spiral-in, spiral-in/out, spiral-in/in and ASE spiral) and non-BOLD methods (FAIR and spin-echo spiral-in/out) at 4.0 T. A cognitive task used to evaluate temporal lobe epilepsy patients was presented to elicit activation in the inferior temporal lobe (as well as other brain regions). Notably, this work allowed us to examine the differing effects that the contrast and signal recovery mechanisms have on fMRI activation in both SFG and non-SFG regions.

Until now the temporal resolution in fMRI was mostly restricted by the used TR. Employing a new method to enhance the time resolution in fMRI below 100 ms we are able to trace neuronal network pathways with extremely short reaction times. The temporal dynamics of somatosensory processing could be measured and correspond with the known values from electrophysiological measures. With this new approach BOLD-fMRI enables to study the temporal dynamics of cortical processing with a temporal resolution of 10 ms.

Feraheme is a newly FDA-approved drug for treating chronic iron anemia in clinical populations. The approved iron dose of 510 mg falls within a weight-normalized range of 5-10 mg/kg for human subjects within the range 50-100 kg. To evaluate this drug as a potential contrast-enhancing agent for human clinical fMRI, we performed experiments in awake non-human primates at 3 Tesla to validate theoretical calculations. Results suggest that this agent could enhance the CNR ratio of clinical fMRI by factors of 5 and 2.5 and 1.5 at 3 Tesla, respectively.

Recently, FRActional Signal mapping from InvErsion Recovery (FRASIER) was proposed to map T1 and tissue fractional volume in the brain. In this study we incorporated FRASIER into an fMRI protocol and assessed the reproducibility of the technique. Using FRASIER, 15 slice T1 and fractional volume maps were acquired in every 10 seconds with 64¡¿64 matirx size. Standard deviations of the T1 and fractional volume maps was within 37ms and 3.5% in this study, demonstrating feasibility of FRASIER in fMRI settings.

This study is the first, to our knowledge, that tests the feasibility of fMRI on an open MRI system, with a magnetic field strength of 1.0 T, and compares the results with fMRI on a state-of-the-art 3.0 T MRI scanner. The optimal echo time for fMRI on an open 1.0 T MRI system was found to be around 70 ms. Results show that fMRI on an open 1.0 T MRI scanner is feasible for studies that are designed to analyze data at a group level, though not optimal for studies on single subjects.