Fred Tam

Retrospective coregistration of functional magnetic resonance imaging data using external monitoring

Coregistration is essential for correcting head motion artifacts in functional magnetic resonance imaging (fMRI). Coregistration algorithms typically realign images through optimization of a similarity measure based on voxel signal intensities. However, coregistration can also be performed through external monitoring, whereby a tracking device measures head motion directly and independently of the imaging data. This paper describes development of external monitoring using fMRI‐compatible infrared cameras. Three subjects participated in block‐design fMRI experiments consisting of bilateral finger tapping alone and tapping combined with visuomotor tracking to produce controlled task‐correlated head motion. Functional MRI time‐series were coregistered using the external monitoring technique and a known image‐based algorithm for comparison. Over various performance characteristics, external monitoring and image‐based coregistration exhibited good agreement, in particular reducing signals correlated with millimeter task‐correlated motions by 50–100%, with a 5% difference between the two techniques. These results promise future applications and refinements of external monitoring in patient populations where head motion is especially problematic. Possibilities include 3D prospective coregistration during real‐time fMRI, coregistration of individual slices, and motion correction in anatomic MRI. Magn Reson Med 53:141–149, 2005.

Optimizing Preprocessing and Analysis Pipelines for Single-Subject FMRI. I. Standard Temporal Motion and Physiological Noise Correction Methods

Subject‐specific artifacts caused by head motion and physiological noise are major confounds in BOLD fMRI analyses. However, there is little consensus on the optimal choice of data preprocessing steps to minimize these effects. To evaluate the effects of various preprocessing strategies, we present a framework which comprises a combination of (1) nonparametric testing including reproducibility and prediction metrics of the data‐driven NPAIRS framework (Strother et al. [2002]: NeuroImage 15:747–771), and (2) intersubject comparison of SPM effects, using DISTATIS (a three‐way version of metric multidimensional scaling (Abdi et al. [2009]: NeuroImage 45:89–95). It is shown that the quality of brain activation maps may be significantly limited by sub‐optimal choices of data preprocessing steps (or “pipeline”) in a clinical task‐design, an fMRI adaptation of the widely used Trail‐Making Test. The relative importance of motion correction, physiological noise correction, motion parameter regression, and temporal detrending were examined for fMRI data acquired in young, healthy adults. Analysis performance and the quality of activation maps were evaluated based on Penalized Discriminant Analysis (PDA). The relative importance of different preprocessing steps was assessed by (1) a nonparametric Friedman rank test for fixed sets of preprocessing steps, applied to all subjects; and (2) evaluating pipelines chosen specifically for each subject. Results demonstrate that preprocessing choices have significant, but subject‐dependant effects, and that individually‐optimized pipelines may significantly improve the reproducibility of fMRI results over fixed pipelines. This was demonstrated by the detection of a significant interaction with motion parameter regression and physiological noise correction, even though the range of subject head motion was small across the group (≪ 1 voxel). Optimizing pipelines on an individual‐subject basis also revealed brain activation patterns either weak or absent under fixed pipelines, which has implications for the overall interpretation of fMRI data, and the relative importance of preprocessing methods. Hum Brain Mapp, 2012.

Magnetoencephalographic study of vibrotactile evoked transient and steady-state responses in human somatosensory cortex.

Somatosensory responses to vibrotactile stimulation applied to the index fingertip were recorded with whole-head MEG in eleven healthy young adult participants. Stimulus trains were produced by a pneumatically driven membrane oscillating at 22 Hz for a trial duration of 1 s, separated by interstimulus intervals (ISIs) of 0.5, 1.0, 3.0, and 7.0 s. Data analysis was performed in two frequency bands. Transient onset responses in the lower frequency band (<20 Hz) contained a clearly expressed P50 component. The higher frequency band (18-30 Hz) revealed a gamma-band response (GBR) within the first 200 ms followed by rhythmic activity at the stimulus frequency that continued throughout the stimulus duration, known as the steady-state response (SSR). Dipoles associated with the transient responses and SSRs were localized in two distinct regions within the primary somatosensory cortex (SI), with transient responses located on average 3 mm more medial and inferior than the SSRs. The transient and GBR peak amplitudes increased with ISI, whereas the SSR amplitude showed no ISI dependence. These results may reflect functionally and spatially distinct neural populations. Further investigations are required to assess the implications of these findings for probing the somatosensory system using other functional neuroimaging methods such as fMRI.

Brain activity during driving with distraction: an immersive fMRI study

Introduction: Non-invasive measurements of brain activity have an important role to play in understanding driving ability. The current study aimed to identify the neural underpinnings of human driving behavior by visualizing the areas of the brain involved in driving under different levels of demand, such as driving while distracted or making left turns at busy intersections. Materials and Methods: To capture brain activity during driving, we placed a driving simulator with a fully functional steering wheel and pedals in a 3.0 Tesla functional magnetic resonance imaging (fMRI) system. To identify the brain areas involved while performing different real-world driving maneuvers, participants completed tasks ranging from simple (right turns) to more complex (left turns at busy intersections). To assess the effects of driving while distracted, participants were asked to perform an auditory task while driving analogous to speaking on a hands-free device and driving. Results: A widely distributed brain network was identified, especially when making left turns at busy intersections compared to more simple driving tasks. During distracted driving, brain activation shifted dramatically from the posterior, visual and spatial areas to the prefrontal cortex. Conclusions: Our findings suggest that the distracted brain sacrificed areas in the posterior brain important for visual attention and alertness to recruit enough brain resources to perform a secondary, cognitive task. The present findings offer important new insights into the scientific understanding of the neuro-cognitive mechanisms of driving behavior and lay down an important foundation for future clinical research.

A new tablet for writing and drawing during functional MRI

Writing and drawing are understudied with fMRI, partly for lack of a device that approximates these behaviors well while supporting task feedback and quantitative behavioral logging in the confines of the magnet. Consequently, we developed a tablet based on touchscreen technology that is accurate, reliable, relatively inexpensive, and fMRI compatible. After confirming fMRI compatibility, we conducted preliminary fMRI experiments examining the neural correlates of a widely used pen‐and‐paper neuropsychological assessment, the trail making test. In two subjects, we found left hemisphere frontal lobe activations similar to the major results of a previous group study, and we also noted individual differences mostly in the right hemisphere. These results demonstrate the utility of the new tablet for adaptations of pen‐and‐paper tests and suggest possible uses of the tablet for longitudinal, within‐subjects studies of disease or therapy. We also discuss using the tablet for several other types of tests requiring many, continuous, or two‐dimensional responses that were previously very difficult to perform during fMRI. Hum Brain Mapp, 2011.

Functional MRI of working memory and selective attention in vibrotactile frequency discrimination

BackgroundFocal lesions of the frontal, parietal and temporal lobe may interfere with tactile working memory and attention. To characterise the neural correlates of intact vibrotactile working memory and attention, functional MRI was conducted in 12 healthy young adults. Participants performed a forced-choice vibrotactile frequency discrimination task, comparing a cue stimulus of fixed frequency to their right thumb with a probe stimulus of identical or higher frequency. To investigate working memory, the time interval between the 2 stimuli was pseudo-randomized (either 2 or 8 s). To investigate selective attention, a distractor stimulus was occasionally presented contralaterally, simultaneous to the probe.ResultsDelayed vibrotactile frequency discrimination, following a probe presented 8 s after the cue in contrast to a probe presented 2 s after the cue, was associated with activation in the bilateral anterior insula and the right inferior parietal cortex. Frequency discrimination under distraction was correlated with activation in the right anterior insula, in the bilateral posterior parietal cortex, and in the right middle temporal gyrus.ConclusionThese results support the notion that working memory and attention are organised in partly overlapping neural circuits. In contrast to previous reports in the visual or auditory domain, this study emphasises the involvement of the anterior insula in vibrotactile working memory and selective attention.

Optimizing Preprocessing and Analysis Pipelines for Single-Subject fMRI: 2. Interactions with ICA, PCA, Task Contrast and Inter-Subject Heterogeneity

A variety of preprocessing techniques are available to correct subject-dependant artifacts in fMRI, caused by head motion and physiological noise. Although it has been established that the chosen preprocessing steps (or “pipeline”) may significantly affect fMRI results, it is not well understood how preprocessing choices interact with other parts of the fMRI experimental design. In this study, we examine how two experimental factors interact with preprocessing: between-subject heterogeneity, and strength of task contrast. Two levels of cognitive contrast were examined in an fMRI adaptation of the Trail-Making Test, with data from young, healthy adults. The importance of standard preprocessing with motion correction, physiological noise correction, motion parameter regression and temporal detrending were examined for the two task contrasts. We also tested subspace estimation using Principal Component Analysis (PCA), and Independent Component Analysis (ICA). Results were obtained for Penalized Discriminant Analysis, and model performance quantified with reproducibility (R) and prediction metrics (P). Simulation methods were also used to test for potential biases from individual-subject optimization. Our results demonstrate that (1) individual pipeline optimization is not significantly more biased than fixed preprocessing. In addition, (2) when applying a fixed pipeline across all subjects, the task contrast significantly affects pipeline performance; in particular, the effects of PCA and ICA models vary with contrast, and are not by themselves optimal preprocessing steps. Also, (3) selecting the optimal pipeline for each subject improves within-subject (P,R) and between-subject overlap, with the weaker cognitive contrast being more sensitive to pipeline optimization. These results demonstrate that sensitivity of fMRI results is influenced not only by preprocessing choices, but also by interactions with other experimental design factors. This paper outlines a quantitative procedure to denoise data that would otherwise be discarded due to artifact; this is particularly relevant for weak signal contrasts in single-subject, small-sample and clinical datasets.

Hemispheric asymmetries of motor versus nonmotor processes during (visuo)motor control

Language and certain aspects of motor control are typically served by the left hemisphere, whereas visuospatial and attentional control are lateralized to the right. Here a (visuo)motor tracing task was used to identify hemispheric lateralization beyond the general, contralateral organization of the motor system. Functional magnetic resonance imaging (fMRI) was applied in 40 male right‐handers (19–30 yrs) during line tracing with dominant and nondominant hand, with and without visual guidance. Results revealed a network of areas activating more in the right than left hemisphere, irrespective of the effector. Inferior portions of frontal gyrus and parietal lobe overlapped largely with a previously described ventral attention network responding to unexpected or behaviourally relevant stimuli. This demonstrates a hitherto unreported functionality of this circuit that also seems to activate when spatial information is continuously exploited to adapt motor behaviour. Second, activation of left dorsal premotor and postcentral regions during tracing with the nondominant left hand was more pronounced than that in their right hemisphere homologues during tracing with the dominant right hand. These activation asymmetries of motor areas ipsilateral to the moving hand could not be explained by asymmetries in skill performance, the degree of handedness, or interhemispheric interactions. The latter was measured by a double‐pulse transcranial magnetic stimulation paradigm, whereby a conditioning stimulus was applied over one hemisphere and a test stimulus over the other. We propose that the left premotor areas contain action representations strongly related to movement implementation which are also accessed during movements performed with the left body side. Hum Brain Mapp, 2011.

Real‐time correction by optical tracking with integrated geometric distortion correction for reducing motion artifacts in functional MRI

Head motion artifacts are a major problem in functional MRI that limit its use in neuroscience research and clinical settings. Real‐time scan‐plane correction by optical tracking has been shown to correct slice misalignment and nonlinear spin‐history artifacts; however, residual artifacts due to dynamic magnetic field nonuniformity may remain in the data. A recently developed correction technique, Phase Labeling for Additional Coordinate Encoding, can correct for absolute geometric distortion using only the complex image data from two echo planar images with slightly shifted k‐space trajectories. An approach is presented that integrates Phase Labeling for Additional Coordinate Encoding into a real‐time scan‐plane update system by optical tracking, applied to a tissue‐equivalent phantom undergoing complex motion and an functional MRI finger tapping experiment with overt head motion to induce dynamic field nonuniformity. Experiments suggest that such integrated volume‐by‐volume corrections are very effective at artifact suppression, with potential to expand functional MRI applications. Magn Reson Med, 2013.

A novel method for integrating MEG and BOLD fMRI signals with the linear convolution model in human primary somatosensory cortex

Characterizing the neurovascular coupling between hemodynamic signals and their neural origins is crucial to functional neuroimaging research, even more so as new methods become available for integrating results from different functional neuroimaging modalities. We present a novel method to relate magnetoencephalography (MEG) and BOLD fMRI data from primary somatosensory cortex within the context of the linear convolution model. This model, which relates neural activity to BOLD signal change, has been widely used to predict BOLD signals but typically lacks experimentally derived measurements of neural activity. In this study, an fMRI experiment is performed using variable‐duration (≤1 s) vibrotactile stimuli applied at 22 Hz, analogous to a previously published MEG study (Nangini et al., [ 2006 ]: Neuroimage 33:252–262), testing whether MEG source waveforms from the previous study can inform the convolution model and improve BOLD signal estimates across all stimulus durations. The typical formulation of the convolution model in which the input is given by the stimulus profile is referred to as Model 1. Model 2 is based on an energy argument relating metabolic demand to the postsynaptic currents largely responsible for the MEG current dipoles, and uses the energy density of the estimated MEG source waveforms as input to the convolution model. It is shown that Model 2 improves the BOLD signal estimates compared to Model 1 under the experimental conditions implemented, suggesting that MEG energy density can be a useful index of hemodynamic activity. Hum Brain Mapp 2008.