Publications by Year: 2009

We propose an integrated registration and clustering algorithm, called "consistency clustering", that automatically constructs a probabilistic white-matter atlas from a set of multi-subject diffusion weighted MR images. We formulate the atlas creation as a maximum likelihood problem which the proposed method solves using a generalized Expectation Maximization (EM) framework. Additionally, the algorithm employs an outlier rejection and denoising strategy to produce sharp probabilistic maps of certain bundles of interest. We test this algorithm on synthetic and real data, and evaluate its stability against initialization. We demonstrate labeling a novel subject using the resulting spatial atlas and evaluate the accuracy of this labeling. Consistency clustering is a viable tool for completely automatic white-matter atlas construction for sub-populations and the resulting atlas is potentially useful for making diffusion measurements in a common coordinate system to identify pathology related changes or developmental trends.

An estimator of the Orientation Probability Density Function (OPDF) of fiber tracts in the white matter of the brain from High Angular Resolution Diffusion data is presented. Unlike Q-Balls, which use the Funk-Radon transform to estimate the radial projection of the 3D Probability Density Function, the Jacobian of the spherical coordinates is included in the Funk-Radon approximation to the radial integral. Thus, true angular marginalizations are computed, which allows a strict probabilistic interpretation. Extensive experiments with both synthetic and real data show the better capability of our method to characterize complex micro-architectures compared to other related approaches (Q-Balls and Diffusion Orientation Transform), especially for low values of the diffusion weighting parameter.

In this paper, we propose a new nonparametric Bayesian framework to cluster white matter fiber tracts into bundles using a hierarchical Dirichlet processes mixture (HDPM) model. The number of clusters is automatically learnt from data with a Dirichlet process (DP) prior instead of being manually specified. After the models of bundles have been learnt from training data without supervision, they can be used as priors to cluster/classify fibers of new subjects. When clustering fibers of new subjects, new clusters can be created for structures not observed in the training data. Our approach does not require computing pairwise distances between fibers and can cluster a huge set of fibers across multiple subjects without subsampling. We present results on multiple data sets, the largest of which has more than 120, 000 fibers.

The computed tomographic (CT) densities of imaged structures are a function of the CT scanning protocol, the structure size, and the structure density. For objects that are of a dimension similar to the scanner point spread function, CT will underestimate true structure density. Prior investigation suggests that this process, termed contrast reduction, could be used to estimate the strength of thin structures, such as cortical bone. In this investigation, we endeavored to exploit this process to provide a CT-based measure of airway disease that can assess changes in airway wall thickening and density that may be associated with the mural remodeling process in subjects with chronic obstructive pulmonary disease (COPD). An initial computer-based study using a range of simulated airway wall sizes and densities suggested that CT measures of airway wall attenuation could detect changes in both wall thickness and structure density. A second phantom-based study was performed using a series of polycarbonate tubes of known density. The results of this again demonstrated the process of contrast reduction and further validated the computer-based simulation. Finally, measures of airway wall attenuation, wall thickness, and wall area (WA) divided by total cross-sectional area, WA percent (WA%), were performed in a cohort of 224 subjects with COPD and correlated with spirometric measures of lung function. The results of this analysis demonstrated that wall attenuation is comparable to WA% in predicting lung function on univariate correlation and remain as a statistically significant correlate to the percent forced expiratory volume in 1 s predicted when adjusted for measures of both emphysema and WA%. These latter findings suggest that the quantitative assessment of airway wall attenuation may offer complementary information to WA% in characterizing airway disease in subjects with COPD.

Least Squares (LS) and its weighted version are standard techniques to estimate the Diffusion Tensor (DT) from Diffusion Weighted Images (DWI). They require to linearize the problem by computing the logarithm of the DWI. For the single-coil Rician noise model it has been shown that this model does not introduce a significant bias, but for multiple array coils and parallel imaging, the noise cannot longer be modeled as Rician. As a result the validity of LS approaches is not assured. An analytical study of noise statistics for a multiple coil system is carried out, together with the Weighted LS formulation and noise analysis for this model. Results show that the bias in the computation of the components of the DT may be comparable to their variance in many cases, stressing the importance of unbiased filtering previous to DT estimation.

One known issue in Q-Ball imaging is the blurring in the radial integral defining the Orientation Distribution Function of fiber bundles, due to the computation of the Funk-Radon Transform (FRT). Three novel techniques to overcome this problem are presented, all of them based upon different assumptions about the behavior of the attenuation signal outside the sphere densely sampled from HARDI data sets. A systematic study with synthetic data has been carried out to show that the FRT blurring is not as important as the error introduced by some unrealistic assumptions, and only one of the three techniques (the one with the less restrictive assumption) improves the accuracy of Q-Balls.

We present a modulated gradient spin-echo method, which uses a train of sinusoidally shaped gradient pulses separated by 180 degrees radio-frequency (RF) pulses. The RF pulses efficiently refocus chemical shifts and de-phasing due to susceptibility differences, resulting in undistorted, high-resolution diffusion weighted spectra. This allows for the simultaneous spectral characterization of the diffusion of several molecular species with different chemical shifts. The technique is robust against susceptibility artifacts, field inhomogeneity and imperfections in the gradient generating equipment. The feasibility of the technique is demonstrated by measuring the diffusion of water, oil, and water-soluble salt in a highly concentrated water-in-oil emulsion. The diffusion of water and salt reveal precise information about the droplet size distribution below the mum-range. Common droplet size distribution explains both the data for water with finite long-range diffusion and the data for salt with negligible long-range diffusion. The results of water diffusion show that the technique is efficient in deconvolving the effects of molecular exchange between droplets and restricted diffusion within droplets. The effects of water exchange suggest that droplets of different sizes are uniformly distributed within the sample.

A new pulsed-gradient spin-echo NMR protocol for assessing the local self-diffusion coefficient D(0) of water confined within living cells is proposed. Equations for the apparent mean-square displacement as a function of the effective diffusion time t(d) and the duration of the displacement-encoding gradient pulses delta are derived. The standard method of estimating D(0)—reducing t(d) until the influence of collisions between the water molecules and the plasma membrane can be neglected—often fails because of the small size of typical cells. As demonstrated here, the decrease of the apparent with increasing delta at constant t(d) can be utilized to measure D(0).

PURPOSE: To apply an intensity-based nonrigid registration algorithm to MRI-guided prostate brachytherapy clinical data and to assess its accuracy. MATERIALS AND METHODS: A nonrigid registration of preoperative MRI to intraoperative MRI images was carried out in 16 cases using a Basis-Spline algorithm in a retrospective manner. The registration was assessed qualitatively by experts’ visual inspection and quantitatively by measuring the Dice similarity coefficient (DSC) for total gland (TG), central gland (CG), and peripheral zone (PZ), the mutual information (MI) metric, and the fiducial registration error (FRE) between corresponding anatomical landmarks for both the nonrigid and a rigid registration method. RESULTS: All 16 cases were successfully registered in less than 5 min. After the nonrigid registration, DSC values for TG, CG, PZ were 0.91, 0.89, 0.79, respectively, the MI metric was -0.19 +/- 0.07 and FRE presented a value of 2.3 +/- 1.8 mm. All the metrics were significantly better than in the case of rigid registration, as determined by one-sided t-tests. CONCLUSION: The intensity-based nonrigid registration method using clinical data was demonstrated to be feasible and showed statistically improved metrics when compare to only rigid registration. The method is a valuable tool to integrate pre- and intraoperative images for brachytherapy.

Standard particle filtering technique have previously been applied to the problem of fiber tracking by Brun et al. [Brun, A., Bjornemo, M., Kikinis, R., Westin, C.F., 2002. White matter tractography using sequential importance sampling. In: Proceedings of the ISMRM Annual Meeting, p. 1131] and Bjornemo et al. [Bjornemo, M., Brun, A., Kikinis, R., Westin, C.F., 2002. Regularized stochastic white matter tractography using diffusion tensor MRI, In: Proc. MICCAI, pp. 435-442]. However, these previous attempts have not utilised the full power of the technique, and as a result the fiber paths were tracked in a goal directed way. In this paper, we provide an advanced technique by presenting a fast and novel probabilistic method for white matter fiber tracking in diffusion weighted MRI (DWI), which takes advantage of the weighting and resampling mechanism of particle filtering. We formulate fiber tracking using a non-linear state space model which captures both smoothness regularity of the fibers and the uncertainties in the local fiber orientations due to noise and partial volume effects. Global fiber tracking is then posed as a problem of particle filtering. To model the posterior distribution, we classify voxels of the white matter as either prolate or oblate tensors. We then construct the orientation distributions for prolate and oblate tensors separately. Finally, the importance density function for particle filtering is modeled using the von Mises-Fisher distribution on a unit sphere. Fast and efficient sampling is achieved using Ulrich-Wood’s simulation algorithm. Given a seed point, the method is able to rapidly locate the globally optimal fiber and also provides a probability map for potential connections. The proposed method is validated and compared to alternative methods both on synthetic data and real-world brain MRI datasets.