Faculty Bibliography

Big data initiatives such as the Enhancing NeuroImaging Genetics through Meta-Analysis consortium (ENIGMA), combine data collected by independent studies worldwide to achieve more generalizable estimates of effect sizes and more reliable and reproducible outcomes. Such efforts require harmonized image analyses protocols to extract phenotypes consistently. This harmonization is particularly challenging for resting state fMRI due to the wide variability of acquisition protocols and scanner platforms; this leads to site-to-site variance in quality, resolution and temporal signal-to-noise ratio (tSNR). An effective harmonization should provide optimal measures for data of different qualities. We developed a multi-site rsfMRI analysis pipeline to allow research groups around the world to process rsfMRI scans in a harmonized way, to extract consistent and quantitative measurements of connectivity and to perform coordinated statistical tests. We used the single-modality ENIGMA rsfMRI preprocessing pipeline based on modelfree Marchenko-Pastur PCA based denoising to verify and replicate resting state network heritability estimates. We analyzed two independent cohorts, GOBS (Genetics of Brain Structure) and HCP (the Human Connectome Project), which collected data using conventional and connectomics oriented fMRI protocols, respectively. We used seed-based connectivity and dual-regression approaches to show that the rsfMRI signal is consistently heritable across twenty major functional network measures. Heritability values of 20-40% were observed across both cohorts.

OBJECTIVE: To assess the association between peripheral lipid/fat profiles and cerebral gray matter (GM) and white matter (WM) in healthy Old Order Amish (OOA). METHODS: Blood lipids, abdominal adiposity, liver lipid contents, and cerebral microstructure were assessed in OOA (N = 64, 31 males/33 females, ages 18-77). Orthogonal factors were extracted from lipid and imaging adiposity measures. GM assessment used the Human Connectome Project protocol to measure whole-brain average cortical thickness. Diffusion-weighted imaging was used to derive WM fractional anisotropy and kurtosis anisotropy measurements. RESULTS: Lipid/fat measures were captured by three orthogonal factors explaining 80% of the variance. Factor one loaded on cholesterol and/or low-density lipoprotein cholesterol measurements; factor two loaded on triglyceride/liver measurements; and factor three loaded on abdominal fat measurements. A two-stage regression including age/sex (first stage) and the three factors (second stage) examined the peripheral lipid/fat effects. Factors two and three significantly contributed to WM measures after Bonferroni corrections (P < 0.007). No factor significantly contributed to GM. Blood pressure (BP) inclusion did not meaningfully alter the lipid/fat-WM relationship. CONCLUSIONS: Peripheral lipid/fat indicators were significantly and negatively associated with cerebral WM rather than with GM, independent of age and BP level. Dissecting the fat/lipid components contributing to different brain imaging parameters may open a new understanding of the body-brain connection through lipid metabolism.

OBJECTIVE: Prior studies in prostate diffusion-weighted magnetic resonance imaging (MRI) have largely explored the impact of b-value and diffusion directions on estimated diffusion coefficient D. Here we suggest varying diffusion time, t, to study time-dependent D(t) in prostate cancer, thereby adding an extra dimension in the development of prostate cancer biomarkers. METHODS: Thirty-eight patients with peripheral zone prostate cancer underwent 3-T MRI using an external-array coil and a diffusion-weighted image sequence acquired for b = 0, as well as along 12 noncollinear gradient directions for b = 500 s/mm using stimulated echo acquisition mode (STEAM) diffusion tensor imaging (DTI). For this sequence, 6 diffusion times ranging from 20.8 to 350 milliseconds were acquired. Tumors were classified as low-grade (Gleason score [GS] 3 + 3; n = 11), intermediate-grade (GS 3 + 4; n = 16), and high-grade (GS >/=4 + 3; n = 11). Benign peripheral zone and transition zone were also studied. RESULTS: Apparent diffusion coefficient (ADC) D(t) decreased with increasing t in all zones of the prostate, though the rate of decay in D(t) was different between sampled zones. Analysis of variance and area under the curve analyses suggested better differentiation of tumor grades at shorter t. Fractional anisotropy (FA) increased with t for all regions of interest. On average, highest FA was observed within GS 3 + 3 tumors. CONCLUSIONS: There is a measurable time dependence of ADC in prostate cancer, which is dependent on the underlying tissue and Gleason score. Therefore, there may be an optimal selection of t for prediction of tumor grade using ADC. Controlling t should allow ADC to achieve greater reproducibility between different sites and vendors. Intentionally varying t enables targeted exploration of D(t), a previously overlooked biophysical phenomenon in the prostate. Its further microstructural understanding and modeling may lead to novel diffusion-derived biomarkers.

BACKGROUND:Segmentation of gliomas in multi-parametric (MP-)MR images is challenging due to their heterogeneous nature in terms of size, appearance and location. Manual tumor segmentation is a time-consuming task and clinical practice would benefit from (semi-) automated segmentation of the different tumor compartments. METHODS:We present a semi-automated framework for brain tumor segmentation based on non-negative matrix factorization (NMF) that does not require prior training of the method. L1-regularization is incorporated into the NMF objective function to promote spatial consistency and sparseness of the tissue abundance maps. The pathological sources are initialized through user-defined voxel selection. Knowledge about the spatial location of the selected voxels is combined with tissue adjacency constraints in a post-processing step to enhance segmentation quality. The method is applied to an MP-MRI dataset of 21 high-grade glioma patients, including conventional, perfusion-weighted and diffusion-weighted MRI. To assess the effect of using MP-MRI data and the L1-regularization term, analyses are also run using only conventional MRI and without L1-regularization. Robustness against user input variability is verified by considering the statistical distribution of the segmentation results when repeatedly analyzing each patient's dataset with a different set of random seeding points. RESULTS:Using L1-regularized semi-automated NMF segmentation, mean Dice-scores of 65%, 74 and 80% are found for active tumor, the tumor core and the whole tumor region. Mean Hausdorff distances of 6.1 mm, 7.4 mm and 8.2 mm are found for active tumor, the tumor core and the whole tumor region. Lower Dice-scores and higher Hausdorff distances are found without L1-regularization and when only considering conventional MRI data. CONCLUSIONS:Based on the mean Dice-scores and Hausdorff distances, segmentation results are competitive with state-of-the-art in literature. Robust results were found for most patients, although careful voxel selection is mandatory to avoid sub-optimal segmentation.

The time dependence of the diffusion coefficient is a hallmark of tissue complexity at the micrometer level. Here we demonstrate how biophysical modeling, combined with a specifically tailored diffusion MRI acquisition performing diffusion tensor imaging (DTI) for varying diffusion times, can be used to determine fiber size and membrane permeability of muscle fibers in vivo. We describe the random permeable barrier model (RPBM) and its assumptions, as well as the details of stimulated echo DTI acquisition, signal processing steps, and potential pitfalls. We illustrate the RPBM method on a few pilot examples involving human subjects (previously published as well as new), such as revealing myofiber size derived from RPBM increase after training in a calf muscle, and size decrease with atrophy in shoulder rotator cuff muscle. Finally, we comment on the potential clinical relevance of our results

Non-negative matrix factorization (NMF) has become a widely used tool for additive parts-based analysis in a wide range of applications. As NMF is a non-convex problem, the quality of the solution will depend on the initialization of the factor matrices. In this study, the successive projection algorithm (SPA) is proposed as an initialization method for NMF. SPA builds on convex geometry and allocates endmembers based on successive orthogonal subspace projections of the input data. SPA is a fast and reproducible method, and it aligns well with the assumptions made in near-separable NMF analyses. SPA was applied to multi-parametric magnetic resonance imaging (MRI) datasets for brain tumor segmentation using different NMF algorithms. Comparison with common initialization methods shows that SPA achieves similar segmentation quality and it is competitive in terms of convergence rate. Whereas SPA was previously applied as a direct endmember extraction tool, we have shown improved segmentation results when using SPA as an initialization method, as it allows further enhancement of the sources during the NMF iterative procedure.

Schizophrenia, a devastating psychiatric illness with onset in the late teens to early 20s, is thought to involve disrupted brain connectivity. Functional and structural disconnections of cortical networks may underlie various cognitive deficits, including a substantial reduction in the speed of information processing in schizophrenia patients compared with controls. Myelinated white matter supports the speed of electrical signal transmission in the brain. To examine possible neuroanatomical sources of cognitive deficits, we used a comprehensive diffusion-weighted imaging (DWI) protocol and characterized the white matter diffusion signals using diffusion kurtosis imaging (DKI) and permeability-diffusivity imaging (PDI) in patients (n = 74), their nonill siblings (n = 41), and healthy controls (n = 113). Diffusion parameters that showed significant patient-control differences also explained the patient-control differences in processing speed. This association was also found for the nonill siblings of the patients. The association was specific to processing-speed abnormality but not specific to working memory abnormality or psychiatric symptoms. Our findings show that advanced diffusion MRI in white matter may capture microstructural connectivity patterns and mechanisms that govern the association between a core neurocognitive measure-processing speed-and neurobiological deficits in schizophrenia that are detectable with in vivo brain scans. These non-Gaussian diffusion white matter metrics are promising surrogate imaging markers for modeling cognitive deficits and perhaps, guiding treatment development in schizophrenia.

We introduce and evaluate a post-processing technique for fast denoising diffusion-weighted MR images. By exploiting the intrinsic redundancy in diffusion MRI using universal properties of the eigenspectrum of random covariance matrices, we remove noise-only principal components, thereby enabling signal-to-noise ratio enhancements, yielding parameter maps of improved quality for visual, quantitative, and statistical interpretation. By studying statistics of residuals, we demonstrate that the technique suppresses local signal fluctuations that solely originate from thermal noise rather than from other sources such as anatomical detail. Furthermore, we achieve improved precision in the estimation of diffusion parameters and fiber orientations in the human brain without compromising the accuracy and/or spatial resolution.

PURPOSE: To estimate the spatially varying noise map using a redundant series of magnitude MR images. METHODS: We exploit redundancy in non-Gaussian distributed multidirectional diffusion MRI data by identifying its noise-only principal components, based on the theory of noisy covariance matrices. The bulk of principal component analysis eigenvalues, arising due to noise, is described by the universal Marchenko-Pastur distribution, parameterized by the noise level. This allows us to estimate noise level in a local neighborhood based on the singular value decomposition of a matrix combining neighborhood voxels and diffusion directions. RESULTS: We present a model-independent local noise mapping method capable of estimating the noise level down to about 1% error. In contrast to current state-of-the-art techniques, the resultant noise maps do not show artifactual anatomical features that often reflect physiological noise, the presence of sharp edges, or a lack of adequate a priori knowledge of the expected form of MR signal. CONCLUSIONS: Simulations and experiments show that typical diffusion MRI data exhibit sufficient redundancy that enables accurate, precise, and robust estimation of the local noise level by interpreting the principal component analysis eigenspectrum in terms of the Marchenko-Pastur distribution. Magn Reson Med, 2015. (c) 2015 Wiley Periodicals, Inc.

PURPOSE: To study and reduce the effect of Gibbs ringing artifact on computed diffusion parameters. METHODS: We reduce the ringing by extrapolating the k-space of each diffusion weighted image beyond the measured part by selecting an adequate regularization term. We evaluate several regularization terms and tune the regularization parameter to find the best compromise between anatomical accuracy of the reconstructed image and suppression of the Gibbs artifact. RESULTS: We demonstrate empirically and analytically that the Gibbs artifact, which is typically observed near sharp edges in magnetic resonance images, has a significant impact on the quantification of diffusion model parameters, even for infinitesimal diffusion weighting. We find the second order total generalized variation to be a good choice for the penalty term to regularize the extrapolation of the k-space, as it provides a parsimonious representation of images, a practically full suppression of Gibbs ringing, and the absence of staircasing artifacts typical for total variation methods. CONCLUSIONS: Regularized extrapolation of the k-space data significantly reduces truncation artifacts without compromising spatial resolution in comparison to the default option of window filtering. In particular, accuracy of estimating diffusion tensor imaging and diffusion kurtosis imaging parameters improves so much that unconstrained fits become possible. Magn Reson Med, 2015. (c) 2015 Wiley Periodicals, Inc.