Archip N, Rohling R, Cooperberg P, Tahmasebpour H. Ultrasound image segmentation using spectral clustering. Ultrasound Med Biol. 2005;31 (11) :1485-97.Abstract
Segmentation of ultrasound images is necessary in a variety of clinical applications, but the development of automatic techniques is still an open problem. Spectral clustering techniques have recently become popular for data and image analysis. In particular, image segmentation has been proposed via the normalized cut (NCut) criterion. This article describes an initial investigation to determine the suitability of such segmentation techniques for ultrasound images. The adaptation of the NCut technique to ultrasound is described first. Segmentation is then performed on simulated ultrasound images. Tests are also performed on abdominal and fetal images with the segmentation results compared to manual segmentation. The success of the segmentation on these test cases warrants further research into NCut-based segmentation of ultrasound images.
Maddah M, Mewes AUJ, Haker S, Grimson EWL, Warfield SK. Automated atlas-based clustering of white matter fiber tracts from DTMRI. Med Image Comput Comput Assist Interv. 2005;8 (Pt 1) :188-95.Abstract
A new framework is presented for clustering fiber tracts into anatomically known bundles. This work is motivated by medical applications in which variation analysis of known bundles of fiber tracts in the human brain is desired. To include the anatomical knowledge in the clustering, we invoke an atlas of fiber tracts, labeled by the number of bundles of interest. In this work, we construct such an atlas and use it to cluster all fiber tracts in the white matter. To build the atlas, we start with a set of labeled ROIs specified by an expert and extract the fiber tracts initiating from each ROI. Affine registration is used to project the extracted fiber tracts of each subject to the atlas, whereas their B-spline representation is used to efficiently compare them to the fiber tracts in the atlas and assign cluster labels. Expert visual inspection of the result confirms that the proposed method is very promising and efficient in clustering of the known bundles of fiber tracts.
Pichon E, Westin C-F, Tannenbaum AR. A Hamilton-Jacobi-Bellman approach to high angular resolution diffusion tractography. Med Image Comput Comput Assist Interv. 2005;8 (Pt 1) :180-7.Abstract
This paper describes a new framework for white matter tractography in high angular resolution diffusion data. A direction-dependent local cost is defined based on the diffusion data for every direction on the unit sphere. Minimum cost curves are determined by solving the Hamilton-Jacobi-Bellman using an efficient algorithm. Classical costs based on the diffusion tensor field can be seen as a special case. While the minimum cost (or equivalently the travel time of a particle moving along the curve) and the anisotropic front propagation frameworks are related, front speed is related to particle speed through a Legendre transformation which can severely impact anisotropy information for front propagation techniques. Implementation details and results on high angular diffusion data show that this method can successfully take advantage of the increased angular resolution in high b-value diffusion weighted data despite lower signal to noise ratio.
Yang Y, Zhu L, Haker S, Tannenbaum AR, Giddens DP. Harmonic skeleton guided evaluation of stenoses in human coronary arteries. Med Image Comput Comput Assist Interv. 2005;8 (Pt 1) :490-7.Abstract
This paper presents a novel approach that three-dimensionally visualizes and evaluates stenoses in human coronary arteries by using harmonic skeletons. A harmonic skeleton is the center line of a multi-branched tubular surface extracted based on a harmonic function, which is the solution of the Laplace equation. This skeletonization method guarantees smoothness and connectivity and provides a fast and straightforward way to calculate local cross-sectional areas of the arteries, and thus provides the possibility to localize and evaluate coronary artery stenosis, which is a commonly seen pathology in coronary artery disease.
Clatz O, Delingette H, Talos I-F, Golby AJ, Kikinis R, Jolesz FA, Ayache N, Warfield SK. Hybrid formulation of the model-based non-rigid registration problem to improve accuracy and robustness. Med Image Comput Comput Assist Interv. 2005;8 (Pt 2) :295-302.Abstract
We present a new algorithm to register 3D pre-operative Magnetic Resonance (MR) images with intra-operative MR images of the brain. This algorithm relies on a robust estimation of the deformation from a sparse set of measured displacements. We propose a new framework to compute iteratively the displacement field starting from an approximation formulation (minimizing the sum of a regularization term and a data error term) and converging toward an interpolation formulation (least square minimization of the data error term). The robustness of the algorithm is achieved through the introduction of an outliers rejection step in this gradual registration process. We ensure the validity of the deformation by the use of a biomechanical model of the brain specific to the patient, discretized with the finite element method. The algorithm has been tested on six cases of brain tumor resection, presenting a brain shift up to 13 mm.
Zhu L, Haker S, Tannenbaum A. Mass preserving registration for heart MR images. Med Image Comput Comput Assist Interv. 2005;8 (Pt 2) :147-54.Abstract
This paper presents a new algorithm for non-rigid registration between two doubly-connected regions. Our algorithm is based on harmonic analysis and the theory of optimal mass transport. It assumes an underlining continuum model, in which the total amount of mass is exactly preserved during the transformation of tissues. We use a finite element approach to numerically implement the algorithm.
Nain D, Haker S, Bobick A, Tannenbaum AR. Multiscale 3D shape analysis using spherical wavelets. Med Image Comput Comput Assist Interv. 2005;8 (Pt 2) :459-67.Abstract
Shape priors attempt to represent biological variations within a population. When variations are global, Principal Component Analysis (PCA) can be used to learn major modes of variation, even from a limited training set. However, when significant local variations exist, PCA typically cannot represent such variations from a small training set. To address this issue, we present a novel algorithm that learns shape variations from data at multiple scales and locations using spherical wavelets and spectral graph partitioning. Our results show that when the training set is small, our algorithm significantly improves the approximation of shapes in a testing set over PCA, which tends to oversmooth data.
Friman O, Westin C-F. Uncertainty in white matter fiber tractography. Med Image Comput Comput Assist Interv. 2005;8 (Pt 1) :107-14.Abstract
In this work we address the uncertainty associated with fiber paths obtained in white matter fiber tractography. This uncertainty, which arises for example from noise and partial volume effects, is quantified using a Bayesian modeling framework. The theory for estimating the probability of a connection between two areas in the brain is presented, and a new model of the local water diffusion profile is introduced. We also provide a theorem that facilitates the estimation of the parameters in this diffusion model, making the presented method simple to implement.
Pohl KM, Fisher J, Levitt JJ, Shenton ME, Kikinis R, Grimson EWL, Wells WM. A unifying approach to registration, segmentation, and intensity correction. Med Image Comput Comput Assist Interv. 2005;8 (Pt 1) :310-8.Abstract
We present a statistical framework that combines the registration of an atlas with the segmentation of magnetic resonance images. We use an Expectation Maximization-based algorithm to find a solution within the model, which simultaneously estimates image inhomogeneities, anatomical labelmap, and a mapping from the atlas to the image space. An example of the approach is given for a brain structure-dependent affine mapping approach. The algorithm produces high quality segmentations for brain tissues as well as their substructures. We demonstrate the approach on a set of 22 magnetic resonance images. In addition, we show that the approach performs better than similar methods which separate the registration and segmentation problems.
O'Donnell L, Westin C-F. White matter tract clustering and correspondence in populations. Med Image Comput Comput Assist Interv. 2005;8 (Pt 1) :140-7.Abstract
We present a novel method for finding white matter fiber correspondences and clusters across a population of brains. Our input is a collection of paths from tractography in every brain. Using spectral methods we embed each path as a vector in a high dimensional space. We create the embedding space so that it is common across all brains, consequently similar paths in all brains will map to points near each other in the space. By performing clustering in this space we are able to find matching fiber tract clusters in all brains. In addition, we automatically obtain correspondence of tractographic paths across brains: by selecting one or several paths of interest in one brain, the most similar paths in all brains are obtained as the nearest points in the high-dimensional space.
Wittek A, Kikinis R, Warfield SK, Miller K. Brain shift computation using a fully nonlinear biomechanical model. Med Image Comput Comput Assist Interv. 2005;8 (Pt 2) :583-90.Abstract
In the present study, fully nonlinear (i.e. accounting for both geometric and material nonlinearities) patient specific finite element brain model was applied to predict deformation field within the brain during the craniotomy-induced brain shift. Deformation of brain surface was used as displacement boundary conditions. Application of the computed deformation field to align (i.e. register) the preoperative images with the intraoperative ones indicated that the model very accurately predicts the displacements of gravity centers of the lateral ventricles and tumor even for very limited information about the brain surface deformation. These results are sufficient to suggest that nonlinear biomechanical models can be regarded as one possible way of complementing medical image processing techniques when conducting nonrigid registration. Important advantage of such models over the linear ones is that they do not require unrealistic assumptions that brain deformations are infinitesimally small and brain tissue stress-strain relationship is linear.
Archip N, Rohling R, Cooperberg P, Tahmasebpour H, Warfield SK. Spectral clustering algorithms for ultrasound image segmentation. Med Image Comput Comput Assist Interv. 2005;8 (Pt 2) :862-9.Abstract
Image segmentation algorithms derived from spectral clustering analysis rely on the eigenvectors of the Laplacian of a weighted graph obtained from the image. The NCut criterion was previously used for image segmentation in supervised manner. We derive a new strategy for unsupervised image segmentation. This article describes an initial investigation to determine the suitability of such segmentation techniques for ultrasound images. The extension of the NCut technique to the unsupervised clustering is first described. The novel segmentation algorithm is then performed on simulated ultrasound images. Tests are also performed on abdominal and fetal images with the segmentation results compared to manual segmentation. Comparisons with the classical NCut algorithm are also presented. Finally, segmentation results on other types of medical images are shown.
Cosman ER, Wells III WM. Bayesian Population Modeling of Effective Connectivity. Inf Process Med Imaging. 2005;19 :39-51.Abstract

A hierarchical model based on the Multivariate Autoregessive (MAR) process is proposed to jointly model neurological time-series collected from multiple subjects, and to characterize the distribution of MAR coefficients across the population from which those subjects were drawn. Thus, inference about effective connectivity between brain regions may be generalized beyond those subjects studied. The posterior on population- and subject-level connectivity parameters are estimated in a Variational Bayesian (VB) framework, and structural model parameters are chosen by the corresponding evidence criteria. The significance of resulting connectivity statistics are evaluated by permutation-based approximations to the null distribution. The method is demonstrated on simulated data and on actual multi-subject neurological time-series.

Wiegand LC, Warfield SK, Levitt JJ, Hirayasu Y, Salisbury DF, Heckers S, Dickey CC, Kikinis R, Jolesz FA, McCarley RW, et al. Prefrontal cortical thickness in first-episode psychosis: a magnetic resonance imaging study. Biol Psychiatry. 2004;55 (2) :131-40.Abstract
BACKGROUND: Findings from postmortem studies suggest reduced prefrontal cortical thickness in schizophrenia; however, cortical thickness in first-episode schizophrenia has not been evaluated using magnetic resonance imaging (MRI). METHODS: Prefrontal cortical thickness was measured using MRI in first-episode schizophrenia patients (n = 17), first-episode affective psychosis patients (n = 17), and normal control subjects (n = 17); subjects were age-matched within 2 years and within a narrow age range (18-29 years). A previous study using the same subjects reported reduced prefrontal gray matter volume in first-episode schizophrenia. Manual editing was performed on those prefrontal segmentations before cortical thickness was measured. RESULTS: Prefrontal cortical thickness was not significantly different among groups. Prefrontal gray matter volume and thickness were, however, positively correlated in both schizophrenia and control subjects. The product of boundary complexity and thickness, an alternative measure of volume, was positively correlated with volume for all three groups. Finally, age and age at first medication were negatively correlated with prefrontal cortical thickness only in first-episode schizophrenia. CONCLUSIONS: This study demonstrates the potential usefulness of MRI for the study of cortical thickness abnormalities in schizophrenia. Correlations between cortical thickness and age and between cortical thickness and age at first medication suggest that the longer the schizophrenic process has been operative, the thinner the prefrontal cortex, although this needs confirmation in a longitudinal study.
Kubicki M, Maier SE, Westin C-F, Mamata H, Ersner-Hershfield H, Estepar R, Kikinis R, Jolesz FA, McCarley RW, Shenton ME. Comparison of single-shot echo-planar and line scan protocols for diffusion tensor imaging. Acad Radiol. 2004;11 (2) :224-32.Abstract
RATIONALE AND OBJECTIVES: Both single-shot diffusion-weighted echo-planar imaging (EPI) and line scan diffusion imaging (LSDI) can be used to obtain magnetic resonance diffusion tensor data and to calculate directionally invariant diffusion anisotropy indices, ie, indirect measures of the organization and coherence of white matter fibers in the brain. To date, there has been no comparison of EPI and LSDI. Because EPI is the most commonly used technique for acquiring diffusion tensor data, it is important to understand the limitations and advantages of LSDI relative to EPI. MATERIALS AND METHODS: Five healthy volunteers underwent EPI and LSDI diffusion on a 1.5 Tesla magnet (General Electric Medical Systems, Milwaukee, WI). Four-mm thick coronal sections, covering the entire brain, were obtained. In addition, one subject was tested with both sequences over four sessions. For each image voxel, eigenvectors and eigenvalues of the diffusion tensor were calculated, and fractional anisotropy (FA) was derived. Several regions of interest were delineated, and for each, mean FA and estimated mean standard deviation were calculated and compared. RESULTS: Results showed no significant differences between EPI and LSDI for mean FA for the five subjects. When intersession reproducibility for one subject was evaluated, there was a significant difference between EPI and LSDI in FA for the corpus callosum and the right uncinate fasciculus. Moreover, errors associated with each FA measure were larger for EPI than for LSDI. CONCLUSION: Results indicate that both EPI- and LSDI-derived FA measures are sufficiently robust. However, when higher accuracy is needed, LSDI provides smaller error and smaller inter-subject and inter-session variability than EPI.
Tolsa CB, Zimine S, Warfield SK, Freschi M, Sancho Rossignol A, Lazeyras F, Hanquinet S, Pfizenmaier M, Huppi PS. Early alteration of structural and functional brain development in premature infants born with intrauterine growth restriction. Pediatr Res. 2004;56 (1) :132-8.Abstract
Placental insufficiency with fetal intrauterine growth restriction (IUGR) is an important cause of perinatal mortality and morbidity and is subsequently associated with significant neurodevelopmental impairment in cognitive function, attention capacity, and school performance. The underlying biologic cause for this association is unclear. Twenty-eight preterm infants (gestational age 32.5 +/- 1.9 wk) were studied by early and term magnetic resonance imaging (MRI). An advanced quantitative volumetric three-dimensional MRI technique was used to measure brain tissue volumes in 14 premature infants with placental insufficiency, defined by abnormal antenatal Doppler measurements and mean birth weights <10(th) percentile (1246 +/- 299 g) (IUGR) and in 14 preterm infants matched for gestational age with normal mean birth weights 1843 +/- 246 g (control). Functional outcome was measured at term in all infants by a specialized assessment scale of preterm infant behavior. Premature infants with IUGR had a significant reduction in intracranial volume (mean +/- SD: 253.7 +/- 29.9 versus 300.5 +/- 43.5 mL, p < 0.01) and in cerebral cortical gray matter (mean +/- SD: 77.2 +/- 16.3 versus 106.8 +/- 24.6 mL, p < 0.01) when measured within the first 2 wk of life compared with control premature infants. These findings persisted at term with intracranial volume (mean +/- SD: 429.3 +/- 47.9 versus 475.9 +/- 53.4 mL, p < 0.05) and cerebral cortical gray matter (mean +/- SD: 149.3 +/- 29.2 versus 189 +/- 34.2 mL, p < 0.01). Behavioral assessment at term showed a significantly less mature score in the subsystem of attention-interaction availability in IUGR infants (p < 0.01). Cerebral cortical gray matter volume at term correlated with attention-interaction capacity measured at term (r = 0.45, p < 0.05). These results suggest that placental insufficiency with IUGR have specific structural and functional consequences on cerebral cortical brain development. These findings may provide insight into the structural-functional correlate for the developmental deficits associated with IUGR.
Als H, Duffy FH, McAnulty GB, Rivkin MJ, Vajapeyam S, Mulkern RV, Warfield SK, Huppi PS, Butler SC, Conneman N, et al. Early experience alters brain function and structure. Pediatrics. 2004;113 (4) :846-57.Abstract
OBJECTIVE: To investigate the effects of early experience on brain function and structure. METHODS: A randomized clinical trial tested the neurodevelopmental effectiveness of the Newborn Individualized Developmental Care and Assessment Program (NIDCAP). Thirty preterm infants, 28 to 33 weeks' gestational age (GA) at birth and free of known developmental risk factors, participated in the trial. NIDCAP was initiated within 72 hours of intensive care unit admission and continued to the age of 2 weeks, corrected for prematurity. Control (14) and experimental (16) infants were assessed at 2 weeks' and 9 months' corrected age on health status, growth, and neurobehavior, and at 2 weeks' corrected age additionally on electroencephalogram spectral coherence, magnetic resonance diffusion tensor imaging, and measurements of transverse relaxation time. RESULTS: The groups were medically and demographically comparable before as well as after the treatment. However, the experimental group showed significantly better neurobehavioral functioning, increased coherence between frontal and a broad spectrum of mainly occipital brain regions, and higher relative anisotropy in left internal capsule, with a trend for right internal capsule and frontal white matter. Transverse relaxation time showed no difference. Behavioral function was improved also at 9 months' corrected age. The relationship among the 3 neurodevelopmental domains was significant. The results indicated consistently better function and more mature fiber structure for experimental infants compared with their controls. CONCLUSIONS: This is the first in vivo evidence of enhanced brain function and structure due to the NIDCAP. The study demonstrates that quality of experience before term may influence brain development significantly.
Wei X, Yoo S-S, Dickey CC, Zou KH, Guttmann CRG, Panych LP. Functional MRI of auditory verbal working memory: long-term reproducibility analysis. Neuroimage. 2004;21 (3) :1000-8.Abstract
Although functional MRI (fMRI) has shown to be a tool with great potential to study the normal and diseased human brain, the large variability in the detected hemodynamic responses across sessions and across subjects hinders a wider application. To investigate the long-term reproducibility of fMRI activation of verbal working memory (WM), eight normal subjects performed an auditory version of the 2-back verbal WM task while fMRI images were acquired. The experiment was repeated nine times with the same settings for image acquisition and fMRI task. Data were analyzed using SPM99 program. Single-session activation maps and multi-subject session-specific activation maps were generated. Regions of interest (ROIs) associated to specific components of verbal WM were defined based on the voxels' coordinates in Talairach space. Visual observation of the multi-subject activation maps showed similar activation patterns, and quantitative analysis showed small coefficients of variance of activation within ROIs over time, suggesting small longitudinal variability of activation. Visual observation of the activation maps of individual sessions demonstrated striking variation of activation across sessions and across subjects, and quantitative analysis demonstrated larger contribution from between-subject variation to overall variation than that from within-subject variation. We concluded that by multi-subject analysis of data from a relatively small number of subjects, reasonably reproducible activation for the 2-back verbal WM paradigm can be achieved. The level of reproducibility encourages the application of this fMRI paradigm to the evaluation of cognitive changes in future investigations. The quantitative estimation of the proportions of within-subject and between-subject variabilities in the overall variability may be helpful for the design of future studies.
Grau V, Mewes AUJ, Alcañiz M, Kikinis R, Warfield SK. Improved watershed transform for medical image segmentation using prior information. IEEE Trans Med Imaging. 2004;23 (4) :447-58.Abstract
The watershed transform has interesting properties that make it useful for many different image segmentation applications: it is simple and intuitive, can be parallelized, and always produces a complete division of the image. However, when applied to medical image analysis, it has important drawbacks (oversegmentation, sensitivity to noise, poor detection of thin or low signal to noise ratio structures). We present an improvement to the watershed transform that enables the introduction of prior information in its calculation. We propose to introduce this information via the use of a previous probability calculation. Furthermore, we introduce a method to combine the watershed transform and atlas registration, through the use of markers. We have applied our new algorithm to two challenging applications: knee cartilage and gray matter/white matter segmentation in MR images. Numerical validation of the results is provided, demonstrating the strength of the algorithm for medical image segmentation.
Park H-J, Levitt J, Shenton ME, Salisbury DF, Kubicki M, Kikinis R, Jolesz FA, McCarley RW. An MRI study of spatial probability brain map differences between first-episode schizophrenia and normal controls. Neuroimage. 2004;22 (3) :1231-46.Abstract
We created a spatial probability atlas of schizophrenia to provide information about the neuroanatomic variability of brain regions of patients with the disorder. Probability maps of 16 regions of interest (ROIs) were constructed by taking manually parcellated ROIs from subjects' magnetic resonance images (MRIs) and linearly transforming them into Talairach space using the Montreal Neurological Institute (MNI) template. ROIs included temporal, parietal, and prefrontal cortex subregions, with a principal focus on temporal lobe structures. Subject Ns ranged from 11 to 28 for the different ROIs. Our global measure of the spatial distribution of the transformed ROI was the sum of voxels with 50% overlap among subjects. The superior temporal gyrus (STG) and fusiform gyrus (FG) had lower values for schizophrenic subjects than for normal controls, suggestive of greater spatial variability for these ROIs in schizophrenic subjects. For the computation of statistical significance of group differences in portions of the ROI, we used voxel-wise comparisons and Fisher's exact test. First-episode schizophrenic patients compared with controls showed lower probability (P < 0.05) at dorso-posterior areas of planum temporale and Heschl's gyrus, lateral and anterior regions in the left hippocampus (HIPP), and dorsolateral regions of fusiform gyrus. Importantly, most ROIs of schizophrenic subjects showed a significantly lower spatial overlap than controls, even after nonlinear spatial normalization, suggesting a greater heterogeneity in the spatial distribution of ROIs. There is consequently a need for caution in neuroimaging studies where data from schizophrenic subjects are normalized to a particular stereotaxic coordinate system based on healthy controls. Apparent group differences in activation may simply reflect a greater heterogeneity of spatial distribution in schizophrenia.