Publications

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 Alzheimer’s disease.

Autism is a neurodevelopmental disorder affecting cognitive, language, and social functioning. Although language and social communication abnormalities are characteristic, prior structural imaging studies have not examined language-related cortex in autistic and control subjects. Subjects included 16 boys with autism (aged 7-11 years), with nonverbal IQ greater than 80, and 15 age- and handedness-matched controls. Magnetic resonance brain images were segmented into gray and white matter; cerebral cortex was parcellated into 48 gyral-based divisions per hemisphere. Asymmetry was assessed a priori in language-related inferior lateral frontal and posterior superior temporal regions and assessed post hoc in all regions to determine specificity of asymmetry abnormalities. Boys with autism had significant asymmetry reversal in frontal language-related cortex: 27% larger on the right in autism and 17% larger on the left in controls. Only one additional region had significant asymmetry differences on post hoc analysis: posterior temporal fusiform gyrus (more left-sided in autism), whereas adjacent fusiform gyrus and temporooccipital inferior temporal gyrus both approached significance (more right-sided in autism). These inferior temporal regions are involved in visual face processing. In boys with autism, language and social/face processing-related regions displayed abnormal asymmetry. These structural abnormalities may relate to language and social disturbances observed in autism.

BACKGROUND: Previous studies suggest that the impact of early insults predisposing to schizophrenia may have differential consequences by sex. We hypothesized that brain regions found to be structurally different in normal men and women (sexual dimorphisms) and abnormal in schizophrenia would show significant sex differences in brain abnormalities, particularly in the cortex, in schizophrenia. METHODS: Forty outpatients diagnosed as having schizophrenia by DSM-III-R were systematically sampled to be comparable within sex with 48 normal comparison subjects on the basis of age, ethnicity, parental socioeconomic status, and handedness. A comprehensive assessment of the entire brain was based on T1-weighted 3-dimensional images acquired from a 1.5-T magnet. Multivariate general linear models for correlated data were used to test for sex-specific effects regarding 22 hypothesized cortical, subcortical, and cerebrospinal fluid brain volumes, adjusted for age and total cerebrum size. Sex x group interactions were also tested on asymmetries of the planum temporale, Heschl’s gyrus, and superior temporal gyrus, additionally controlled for handedness. RESULTS: Normal patterns of sexual dimorphisms were disrupted in schizophrenia. Sex-specific effects were primarily evident in the cortex, particularly in the frontomedial cortex, basal forebrain, cingulate and paracingulate gyri, posterior supramarginal gyrus, and planum temporale. Normal asymmetry of the planum was also disrupted differentially in men and women with schizophrenia. There were no significant differential sex effects in subcortical gray matter regions or cerebrospinal fluid. CONCLUSION: Factors that produce normal sexual dimorphisms may be associated with modulating insults producing schizophrenia, particularly in the cortex.

BACKGROUND: Clues to the causes of schizophrenia can be derived from studying first-degree relatives because they are genetically related to an ill family member. Abnormalities observed in nonpsychotic relatives are indicators of possible genetic vulnerability to illness, independent of psychosis. We tested 4 hypotheses: (1) that hippocampal volume is smaller in nonpsychotic relatives than in controls, particularly in the left hemisphere; (2) that hippocampi will be smaller in multiplex relatives as compared with simplex relatives, and both will be smaller than in controls; (3) that hippocampal volumes and verbal declarative memory function will be positively correlated; and (4) that hippocampi will be smaller in patients with schizophrenia than in their nonpsychotic relatives or in controls.

New medical imaging modalities offering multi-valued data, such as phase contrast MRA and diffusion tensor MRI, require general representations for the development of automated algorithms. In this paper we propose a unified framework for the registration of medical volumetric multi-valued data using local matching. The paper extends the usual concept of similarity between two pieces of data to be matched, commonly used with scalar (intensity) data, to the general tensor case. Our approach to registration is based on a multiresolution scheme, where the deformation field estimated in a coarser level is propagated to provide an initial deformation in the next finer one. In each level, local matching of areas with a high degree of local structure and subsequent interpolation are performed. Consequently, we provide an algorithm to assess the amount of structure in generic multi-valued data by means of gradient and correlation computations. The interpolation step is carried out by means of the Kriging estimator, which provides a novel framework for the interpolation of sparse vector fields in medical applications. The feasibility of the approach is illustrated by results on synthetic and clinical data.

This paper presents processing and visualization techniques for Diffusion Tensor Magnetic Resonance Imaging (DT-MRI). In DT-MRI, each voxel is assigned a tensor that describes local water diffusion. The geometric nature of diffusion tensors enables us to quantitatively characterize the local structure in tissues such as bone, muscle, and white matter of the brain. This makes DT-MRI an interesting modality for image analysis. In this paper we present a novel analytical solution to the Stejskal-Tanner diffusion equation system whereby a dual tensor basis, derived from the diffusion sensitizing gradient configuration, eliminates the need to solve this equation for each voxel. We further describe decomposition of the diffusion tensor based on its symmetrical properties, which in turn describe the geometry of the diffusion ellipsoid. A simple anisotropy measure follows naturally from this analysis. We describe how the geometry or shape of the tensor can be visualized using a coloring scheme based on the derived shape measures. In addition, we demonstrate that human brain tensor data when filtered can effectively describe macrostructural diffusion, which is important in the assessment of fiber-tract organization. We also describe how white matter pathways can be monitored with the methods introduced in this paper. DT-MRI tractography is useful for demonstrating neural connectivity (in vivo) in healthy and diseased brain tissue.

OBJECTIVE: A major shortcoming of image-guided navigational systems is the use of preoperatively acquired image data, which does not account for intraoperative changes in brain morphology. The occurrence of these surgically induced volumetric deformations ("brain shift") has been well established. Maximal measurements for surface and midline shifts have been reported. There has been no detailed analysis, however, of the changes that occur during surgery. The use of intraoperative magnetic resonance imaging provides a unique opportunity to obtain serial image data and characterize the time course of brain deformations during surgery. METHODS: The vertically open intraoperative magnetic resonance imaging system (SignaSP, 0.5 T; GE Medical Systems, Milwaukee, WI) permits access to the surgical field and allows multiple intraoperative image updates without the need to move the patient. We developed volumetric display software (the 3D Slicer) that allows quantitative analysis of the degree and direction of brain shift. For 25 patients, four or more intraoperative volumetric image acquisitions were extensively evaluated. RESULTS: Serial acquisitions allow comprehensive sequential descriptions of the direction and magnitude of intraoperative deformations. Brain shift occurs at various surgical stages and in different regions. Surface shift occurs throughout surgery and is mainly attributable to gravity. Subsurface shift occurs during resection and involves collapse of the resection cavity and intraparenchymal changes that are difficult to model. CONCLUSION: Brain shift is a continuous dynamic process that evolves differently in distinct brain regions. Therefore, only serial imaging or continuous data acquisition can provide consistently accurate image guidance. Furthermore, only serial intraoperative magnetic resonance imaging provides an accurate basis for the computational analysis of brain deformations, which might lead to an understanding and eventual simulation of brain shift for intraoperative guidance.

In order to enhance 3D image data from magnetic resonance angiography (MRA), a novel method based on the theory of multidimensional adaptive filtering has been developed. The purpose of the technique is to suppress image noise while enhancing important structures. The method is based on local structure estimation using six 3D orientation selective filters, followed by an adaptive filtering step controlled by the local structure information. The complete filtering procedure requires approximately 3 minutes of computational time on a standard workstation for a 256 x 256 x 64 data set. The method has been evaluated using a mathematical vessel model and in vivo MRA data (both phase contrast and time of flight (TOF)). 3D adaptive filtering results in a better delineation of small blood vessels and efficiently reduces the high-frequency noise. Depending on the data acquisition and the original data type, contrast-to-noise ratio (CNR) improvements of up to 179% (8.9 dB) were observed. 3D adaptive filtering may provide an alternative to prolonging the scan time or using contrast agents in MRA when the CNR is low.

The vasculature is of utmost importance in neurosurgery. Direct visualization of images acquired with current imaging modalities, however, cannot provide a spatial representation of small vessels. These vessels, and their branches which show considerable variations, are most important in planning and performing neurosurgical procedures. In planning they provide information on where the lesion draws its blood supply and where it drains. During surgery the vessels serve as landmarks and guidelines to the lesion. The more minute the information is, the more precise the navigation and localization of computer guided procedures. Beyond neurosurgery and neurological study, vascular information is also crucial in cardiovascular surgery, diagnosis, and research. This paper addresses the problem of automatic segmentation of complicated curvilinear structures in three-dimensional imagery, with the primary application of segmenting vasculature in magnetic resonance angiography (MRA) images. The method presented is based on recent curve and surface evolution work in the computer vision community which models the object boundary as a manifold that evolves iteratively to minimize an energy criterion. This energy criterion is based both on intensity values in the image and on local smoothness properties of the object boundary, which is the vessel wall in this application. In particular, the method handles curves evolving in 3D, in contrast with previous work that has dealt with curves in 2D and surfaces in 3D. Results are presented on cerebral and aortic MRA data as well as lung computed tomography (CT) data.

A novel method for detecting neural activity in functional magnetic resonance imaging (fMRI) data is introduced. It is based on canonical correlation analysis (CCA), which is a multivariate extension of the univariate correlation analysis widely used in fMRI. To detect homogeneous regions of activity, the method combines a subspace modeling of the hemodynamic response and the use of spatial relationships. The spatial correlation that undoubtedly exists in fMR images is completely ignored when univariate methods such as as t-tests, F-tests, and ordinary correlation analysis are used. Such methods are for this reason very sensitive to noise, leading to difficulties in detecting activation and significant contributions of false activations. In addition, the proposed CCA method also makes it possible to detect activated brain regions based not only on thresholding a correlation coefficient, but also on physiological parameters such as temporal shape and delay of the hemodynamic response. Excellent performance on real fMRI data is demonstrated. Magn Reson Med 45:323-330, 2001.