David N. Kennedy

Whole Brain Segmentation: Automated Labeling of Neuroanatomical Structures in the Human Brain

We present a technique for automatically assigning a neuroanatomical label to each voxel in an MRI volume based on probabilistic information automatically estimated from a manually labeled training set. In contrast to existing segmentation procedures that only label a small number of tissue classes, the current method assigns one of 37 labels to each voxel, including left and right caudate, putamen, pallidum, thalamus, lateral ventricles, hippocampus, and amygdala. The classification technique employs a registration procedure that is robust to anatomical variability, including the ventricular enlargement typically associated with neurological diseases and aging. The technique is shown to be comparable in accuracy to manual labeling, and of sufficient sensitivity to robustly detect changes in the volume of noncortical structures that presage the onset of probable Alzheimers disease.

Acute Effects of Cocaine on Human Brain Activity and Emotion

We investigated brain circuitry mediating cocaine-induced euphoria and craving using functional MRI (fMRI). During double-blind cocaine (0.6 mg/kg) and saline infusions in cocaine-dependent subjects, the entire brain was imaged for 5 min before and 13 min after infusion while subjects rated scales for rush, high, low, and craving. Cocaine induced focal signal increases in nucleus accumbens/subcallosal cortex (NAc/SCC), caudate, putamen, basal forebrain, thalamus, insula, hippocampus, parahippocampal gyrus, cingulate, lateral prefrontal and temporal cortices, parietal cortex, striate/extrastriate cortices, ventral tegmentum, and pons and produced signal decreases in amygdala, temporal pole, and medial frontal cortex. Saline produced few positive or negative activations, which were localized to lateral prefrontal cortex and temporo-occipital cortex. Subjects who underwent repeat studies showed good replication of the regional fMRI activation pattern following cocaine and saline infusions, with activations on saline retest that might reflect expectancy. Brain regions that exhibited early and short duration signal maxima showed a higher correlation with rush ratings. These included the ventral tegmentum, pons, basal forebrain, caudate, cingulate, and most regions of lateral prefrontal cortex. In contrast, regions that demonstrated early but sustained signal maxima were more correlated with craving than with rush ratings; such regions included the NAc/SCC, right parahippocampal gyrus, and some regions of lateral prefrontal cortex. Sustained negative signal change was noted in the amygdala, which correlated with craving ratings. Our data demonstrate the ability of fMRI to map dynamic patterns of brain activation following cocaine infusion in cocaine-dependent subjects and provide evidence of dynamically changing brain networks associated with cocaine-induced euphoria and cocaine-induced craving.

A Bayesian model of shape and appearance for subcortical brain segmentation.

Automatic segmentation of subcortical structures in human brain MR images is an important but difficult task due to poor and variable intensity contrast. Clear, well-defined intensity features are absent in many places along typical structure boundaries and so extra information is required to achieve successful segmentation. A method is proposed here that uses manually labelled image data to provide anatomical training information. It utilises the principles of the Active Shape and Appearance Models but places them within a Bayesian framework, allowing probabilistic relationships between shape and intensity to be fully exploited. The model is trained for 15 different subcortical structures using 336 manually-labelled T1-weighted MR images. Using the Bayesian approach, conditional probabilities can be calculated easily and efficiently, avoiding technical problems of ill-conditioned covariance matrices, even with weak priors, and eliminating the need for fitting extra empirical scaling parameters, as is required in standard Active Appearance Models. Furthermore, differences in boundary vertex locations provide a direct, purely local measure of geometric change in structure between groups that, unlike voxel-based morphometry, is not dependent on tissue classification methods or arbitrary smoothing. In this paper the fully-automated segmentation method is presented and assessed both quantitatively, using Leave-One-Out testing on the 336 training images, and qualitatively, using an independent clinical dataset involving Alzheimers disease. Median Dice overlaps between 0.7 and 0.9 are obtained with this method, which is comparable or better than other automated methods. An implementation of this method, called FIRST, is currently distributed with the freely-available FSL package.

A Functional MRI Study of Subjects Recovered From Hemiparetic Stroke

BACKGROUND AND PURPOSE Stroke recovery mechanisms remain incompletely understood, particularly for subjects with cortical stroke, in whom limited data are available. We used functional magnetic resonance imaging to compare brain activations in normal controls and subjects who recovered from hemiparetic stroke. METHODS Functional magnetic resonance imaging was performed in ten stroke subjects with good recovery, five with deep, and five with cortical infarcts. Brain activation was achieved by index finger-tapping. Statistical parametric activation maps were obtained using a t test and a threshold of P < .001. In five bilateral motor regions, the volume of activated brain for each stroke subject was compared with the distribution of activation volumes among nine controls. RESULTS Control subjects activated several motor regions. During recovered hand finger-tapping, stroke subjects activated the same regions as controls, often in a larger brain volume. In the unaffected hemisphere, sensorimotor cortex activation was increased in six of nine stroke subjects compared with controls. Cerebellar hemisphere contralateral and premotor cortex ipsilateral to this region, as well as supplementary motor areas, also had increased activation. In the stroke hemisphere, activation exceeding controls was uncommon, except that three of five cortical strokes showed peri-infarct activation foci. During unaffected hand finger-tapping, increased activation by stroke subjects compared with controls was uncommon; however, decreased activation was seen in unaffected sensorimotor cortex, suggesting that this regions responsiveness increased to the ipsilateral hand and decreased to contralateral hand movements. Use of a different threshold for defining activation (P < .01) did not change the overall findings (kappa = .75). CONCLUSIONS Recovered finger-tapping by stroke subjects activated the same motor regions as controls but to a larger extent, particularly in the unaffected hemisphere. Increased reliance on these motor areas may represent an important component of motor recovery. Functional magnetic resonance imaging studies of subjects who recovered from stroke provide evidence for several processes that may be related to restoration of neurologic function.

Reliability in multi-site structural MRI studies: Effects of gradient non-linearity correction on phantom and human data

Longitudinal and multi-site clinical studies create the imperative to characterize and correct technological sources of variance that limit image reproducibility in high-resolution structural MRI studies, thus facilitating precise, quantitative, platform-independent, multi-site evaluation. In this work, we investigated the effects that imaging gradient non-linearity have on reproducibility of multi-site human MRI. We applied an image distortion correction method based on spherical harmonics description of the gradients and verified the accuracy of the method using phantom data. The correction method was then applied to the brain image data from a group of subjects scanned twice at multiple sites having different 1.5 T platforms. Within-site and across-site variability of the image data was assessed by evaluating voxel-based image intensity reproducibility. The image intensity reproducibility of the human brain data was significantly improved with distortion correction, suggesting that this method may offer improved reproducibility in morphometry studies. We provide the source code for the gradient distortion algorithm together with the phantom data.

Volumetric MRI analysis comparing subjects having attention-deficit hyperactivity disorder with normal controls

Article abstract-Objective: To test by MRI-based morphometry the a priori hypotheses that developmental anomalies exist in attention-deficit hyperactivity disorder (ADHD) in left caudate and right prefrontal/frontal/ and/or posterior parietal hemispheric regions, in accord with neurochemical, neuronal circuitry and attentional network hypotheses, and prior imaging studies. Design: Case-control study. Setting: Academic medical center. Participants: Fifteen male subjects with ADHD without comorbid diagnoses (aged 12.4 +/- 3.4 years) and 15 male normal controls (aged 14.4 +/- 3.4), group-matched for age, IQ, and handedness. Main outcome measures: Global and hemispheric regional volumes (in cm3) of cerebral hemispheres, cortex, white matter, ventricles, caudate, lenticulate, central gray nuclei, insula, amygdala, and hippocampus. Results: Despite similar hemispheric volumes, ADHD subjects had smaller volumes of (1) left total caudate and caudate head (p < 0.04), with reversed asymmetry (p < 0.03); (2) right anterior-superior (frontal) region en bloc (p < 0.03) and white matter (p < 0.01); (3) bilateral anterior-inferior region en bloc (p < 0.04); and (4) bilateral retrocallosal (parietal-occipital) region white matter (p < 0.03). Possible structural correlates of ADHD response to stimulants were noted in an exploratory analysis, with the smallest and symmetric caudate, and smallest left anterior-superior cortex volumes found in the responders, but reversed caudate asymmetry and the smallest retrocallosal white matter volumes noted in the nonresponders. Conclusions: This study is the first to report localized hemispheric structural anomalies in ADHD, which are concordant with theoretical models of abnormal frontal-striatal and parietal function, and with possible differing morphologic substrates of response to stimulant medication. NEUROLOGY 1997;48: 589-601

Acupuncture modulates the limbic system and subcortical gray structures of the human brain: evidence from fMRI studies in normal subjects.

Acupuncture, an ancient therapeutic technique, is emerging as an important modality of complementary medicine in the United States. The use and efficacy of acupuncture treatment are not yet widely accepted in Western scientific and medical communities. Demonstration of regionally specific, quantifiable acupuncture effects on relevant structures of the human brain would facilitate acceptance and integration of this therapeutic modality into the practice of modern medicine. Research with animal models of acupuncture indicates that many of the beneficial effects may be mediated at the subcortical level in the brain. We used functional magnetic resonance imaging (fMRI) to investigate the effects of acupuncture in normal subjects and to provide a foundation for future studies on mechanisms of acupuncture action in therapeutic interventions. Acupuncture needle manipulation was performed at Large Intestine 4 (LI 4, Hegu) on the hand in 13 subjects [Stux, 1997]. Needle manipulation on either hand produced prominent decreases of fMRI signals in the nucleus accumbens, amygdala, hippocampus, parahippocampus, hypothalamus, ventral tegmental area, anterior cingulate gyrus (BA 24), caudate, putamen, temporal pole, and insula in all 11 subjects who experienced acupuncture sensation. In marked contrast, signal increases were observed primarily in the somatosensory cortex. The two subjects who experienced pain instead of acupuncture sensation exhibited signal increases instead of decreases in the anterior cingulate gyrus (BA 24), caudate, putamen, anterior thalamus, and posterior insula. Superficial tactile stimulation to the same area elicited signal increases in the somatosensory cortex as expected, but no signal decreases in the deep structures. These preliminary results suggest that acupuncture needle manipulation modulates the activity of the limbic system and subcortical structures. We hypothesize that modulation of subcortical structures may be an important mechanism by which acupuncture exerts its complex multisystem effects. Hum Brain Mapp 9:13–25, 2000.

Localization of white matter volume increase in autism and developmental language disorder

Increased brain volume in autism appears to be driven mainly by an unexplained white matter enlargement, and we have reported a similar phenomenon in developmental language disorder (DLD). Localization of this enlargement would strongly guide research into its cause, tissue basis, and functional implications. We utilized a white matter parcellation technique that divides cerebral white matter into an outer zone containing the radiate compartment and an inner zone containing sagittal and bridging system compartments. In both high‐functioning autism and DLD, enlargement localized to the radiate white matter (all lobes in autism, all but parietal in DLD), whereas inner zone white matter compartments showed no volume differences from controls. Furthermore, in both autism and DLD, later or longer‐myelinating regions showed greater volume increases over controls. Neither group showed cerebral cortex, corpus callosum, or internal capsule volume differences from control. Radiate white matter myelinates later than deep white matter; this pattern of enlargement thus is consistent with striking postnatal head circumference percentile increases reported in autism. These findings suggest an ongoing postnatal process in both autism and DLD that is probably intrinsic to white matter, that primarily affects intrahemispheric and corticocortical connections, and that places these two disorders on the same spectrum.

The integrated response of the human cerebro-cerebellar and limbic systems to acupuncture stimulation at ST 36 as evidenced by fMRI

Clinical and experimental data indicate that most acupuncture clinical results are mediated by the central nervous system, but the specific effects of acupuncture on the human brain remain unclear. Even less is known about its effects on the cerebellum. This fMRI study demonstrated that manual acupuncture at ST 36 (Stomach 36, Zusanli), a main acupoint on the leg, modulated neural activity at multiple levels of the cerebro-cerebellar and limbic systems. The pattern of hemodynamic response depended on the psychophysical response to needle manipulation. Acupuncture stimulation typically elicited a composite of sensations termed deqi that is related to clinical efficacy according to traditional Chinese medicine. The limbic and paralimbic structures of cortical and subcortical regions in the telencephalon, diencephalon, brainstem and cerebellum demonstrated a concerted attenuation of signal intensity when the subjects experienced deqi. When deqi was mixed with sharp pain, the hemodynamic response was mixed, showing a predominance of signal increases instead. Tactile stimulation as control also elicited a predominance of signal increase in a subset of these regions. The study provides preliminary evidence for an integrated response of the human cerebro-cerebellar and limbic systems to acupuncture stimulation at ST 36 that correlates with the psychophysical response.

Language dominance determined by whole brain functional MRI in patients with brain lesions

Background: Functional MRI (fMRI) is of potential value in determining hemisphere dominance for language in epileptic patients. Objective: To develop and validate an fMRI-based method of determining language dominance for patients with a wide range of potentially operable brain lesions in addition to epilepsy. Methods: Initially, a within-subjects design was used with 19 healthy volunteers (11 strongly right-handed, 8 left-handed) to determine the relative lateralizing usefulness of three different language tasks in fMRI. An automated, hemispheric analysis of laterality was used to analyze whole brain fMRI data sets. To evaluate the clinical usefulness of this method, we compared fMRI-determined laterality with laterality determined by Wada testing or electrocortical stimulation mapping, or both, in 23 consecutive patients undergoing presurgical evaluation of language dominance. Results: Only the verb generation task was reliably lateralizing. fMRI, using the verb generation task and an automated hemispheric analysis method, was concordant with invasive measures in 22 of 23 patients (12 Wada, 11 cortical stimulation). For the single patient who was discordant, in whom a tumor involved one-third of the left hemisphere, fMRI became concordant when the tumor and its reflection in the right hemisphere were excluded from laterality analysis. No significant negative correlation was obtained between lesion size and strength of laterality for the patients with lesions involving the dominant hemisphere. Conclusion: This fMRI method shows potential for evaluating language dominance in patients with a variety of brain lesions.