|
|
| (160 intermediate revisions by 16 users not shown) |
| Line 1: |
Line 1: |
| − | Back to [[2008_Progress_Report]]
| + | This latest draft of this report has now been removed from the wiki and is available [[Media:2008_Namic_Progress_Report.doc|here]] in a MS word document for the final submission. If you have any changes to the last version of text, please send these to Tina. The final version will be posted back here by May 30th. If you really need to look at the last wiki version, please click on the history tab of this page and look at the last one edited on May 22nd. [[User:Tkapur|Tkapur]] 14:07, 23 May 2008 (EDT) |
| − | | |
| − | | |
| − | | |
| − | | |
| − | | |
| − | =Guidelines for preparation=
| |
| − | | |
| − | *[[2008_Progress_Report#Scientific Report Timeline]] - Main point is that May 15 is the date by which all sections below need to be completed. No extensions are possible.
| |
| − | *DBPs - If there is work outside of the roadmap projects that you would like to report, you are welcome to create a separate section for it under "Other".
| |
| − | *The outline for this report is similar to the 2007 report, which is provided here for reference: [[2007_Annual_Scientific_Report]].
| |
| − | *In preparing summaries for each of the 8 topics in this report, please leverage the detailed pages for projects provided here: [[NA-MIC_Internal_Collaborations]].
| |
| − | *Publications will be mined from the SPL publications database. All core PIs need to ensure that all NA-MIC publications are in the publications database by May 15.
| |
| − | | |
| − | =Introduction (Tannenbaum)=
| |
| − | The National Alliance for Medical Imaging Computing (NA-MIC) is now in its fourth year. This Center is comprised of a multi-institutional, interdisciplinary team of computer scientists, software engineers, and medical investigators who have come together to develop and apply computational tools for the analysis and visualization of medical imaging data. A further purpose of the Center is to provide infrastructure and environmental support for the development of computational algorithms and open source technologies, and to oversee the training and dissemination of these tools to the medical research community. The first driving biological projects (DBPs) three years for Center were inspired by schizophrenia research. In the fourth year new DBPs have been added. Three are centered around diseases of the brain: (a) brain lesion analysis in neuropschiatric systemic lupus erythematosus; (b) a study of cortical thickness for autism; and (c) stochastic tractography for VCFS. In an very new direction, we have added DBP on the prostate: brachytherapy needle positioning robot integration.
| |
| − | | |
| − | =Clinical Roadmap Projects=
| |
| − | ==Roadmap Project: Stochastic Tractography for VCFS (Kubicki)==
| |
| − | ===Overview (Kubicki)===
| |
| − | ===Algorithm Component (Golland)===
| |
| − | ===Engineering Component (Davis)===
| |
| − | ===Clinical Component (Kubicki)===
| |
| − | ===Additional Information===
| |
| − | Additional Information for this project is available [http://wiki.na-mic.org/Wiki/index.php/DBP2:Harvard:Brain_Segmentation_Roadmap here on the NA-MIC wiki].
| |
| − | ==Roadmap Project: Brachytherapy Needle Positioning Robot Integration (Fichtinger)==
| |
| − | ===Overview (Fichtinger)===
| |
| − | ===Algorithm Component (Tannenbaum)===
| |
| − | Currently we attack the segmentation of the prostate in two ways. The first way is a combination of Cellular Automata(CA also called Grow Cut) with Geodesic Active Contour(GAC) methods. While the second is using a ellipsoid to match the prostate in 3D image. More details are given below and at [[Projects:ProstateSegmentatio|More...]]
| |
| − | | |
| − | 1. CA algorithm is used to give a rough segmentation which is fed into GAC for finer tuning. Both algorithm are implemented in 3D. A ITK-Cellular Automata filter, dealing with N-D data, has already been completed and submitted into the NA-MIC SandBox.
| |
| − | | |
| − | 2. Prostate is usually modeled as an ellipsoid. We try using ellipsoid model, coupled with various local and global segmentation energy definition, to give an fully automatic segmentation.
| |
| − | | |
| − | ===Engineering Component (Hayes)===
| |
| − | ===Clinical Component (Fichtinger)===
| |
| − | ===Additional Information===
| |
| − | Additional Information for this project is available [http://wiki.na-mic.org/Wiki/index.php/DBP2:JHU:Roadmap here on the NA-MIC wiki].
| |
| − | ==Roadmap Project: Brain Lesion Analysis in Neuropsychiatric Systemic Lupus Erythematosus (Bockholt)==
| |
| − | ===Overview (Bockholt)===
| |
| − | ===Algorithm Component (Whitaker)===
| |
| − | ===Engineering Component (Pieper)===
| |
| − | ===Clinical Component (Bockholt)===
| |
| − | ===Additional Information===
| |
| − | Additional Information for this project is available [http://wiki.na-mic.org/Wiki/index.php/DBP2:MIND:Roadmap here on the NA-MIC wiki].
| |
| − | ==Roadmap Project: Cortical Thickness for Autism(Hazlett)==
| |
| − | ===Overview (Hazlett)===
| |
| − | ===Algorithm Component (Styner)===
| |
| − | ===Engineering Component (Miller, Vachet)===
| |
| − | ===Clinical Component (Hazlett)===
| |
| − | ===Additional Information===
| |
| − | Additional Information for this project is available [http://wiki.na-mic.org/Wiki/index.php/DBP2:UNC:Cortical_Thickness_Roadmap here on the NA-MIC wiki].
| |
| − | | |
| − | =Four Infrastructure Topics=
| |
| − | ==Diffusion Image Analysis (Gerig)==
| |
| − | ===Progress===
| |
| − | ===Key Investigators===
| |
| − | ===Additional Information===
| |
| − | Additional Information for this topic is available [http://wiki.na-mic.org/Wiki/index.php/NA-MIC_Internal_Collaborations:DiffusionImageAnalysis here on the NA-MIC wiki].
| |
| − | ==Structural Analysis(Tannenbaum)==
| |
| − | ===Progress===
| |
| − | Under Structural Analysis, the main topics of research for NAMIC are structural segmentation, registration techniques and shape analysis. These topics are correlated and research in one often finds application in another. For example, shape analysis can yield useful priors for segmentation, or segmentation and registration can provide structural correspondences for use in shape analysis and so on.
| |
| − | | |
| − | An overview of selected progress highlights under these broad topics follows.
| |
| − | | |
| − | Structural Segmentation
| |
| − | | |
| − | * Directional Based Segmentation
| |
| − | We have proposed a directional segmentation framework for Direction-weighted Magnetic Resonance imagery by augmenting the Geodesic Active Contour framework with directional information. The classical scalar conformal factor is replaced by a factor that incorporates directionality. We mathematically showed that the optimization problem is well-defined when the factor is a Finsler metric. The calculus of variations or dynamic programming may be used to find the optimal curves. This past year we have applied this methodology in extracting the anchor tract (or centerline) of neural fiber bundles. Further we have applied this in conjunction with the Bayes’ rule into volumetric segmentation for extracting the entire fiber bundles. We have also proposed a novel shape prior in the volumetric segmentation to extract tubular fiber bundles.
| |
| − | | |
| − | * Stochastic Segmentation
| |
| − | | |
| − | We have continued work this year on developing new stochastic methods for implementing curvature-driven flows for medical tasks like segmentation. We can now generalize our results to an arbitrary Riemannian surface which includes the geodesic active contours as a special case. We are also implementing the directional flows based on the anisotropic conformal factor described above using this stochastic methodology. Our stochastic snakes’ models are based on the theory of interacting particle systems. This brings together the theories of curve evolution and hydrodynamic limits, and as such impacts our growing use of joint methods from probability and partial differential in image processing and computer vision. We now have working code written in C++ for the two dimensional case and have worked out the stochastic model of the general geodesic active contour model.
| |
| − | | |
| − | * Statistical PDE Methods for Segmentation
| |
| − | | |
| − | Our objective is to add various statistical measures into our PDE flows for medical imaging. This will allow the incorporation of global image information into the locally defined PDE framework. This year, we developed flows which can separate the distributions inside and outside the evolving contour, and we have also been including shape information in the flows. We have completed a statistically based flow for segmentation using fast marching, and the code has been integrated into Slicer.
| |
| − | | |
| − | * Atlas Renormalization for Improved Brain MR Image Segmentation
| |
| − | | |
| − | Atlas-based approaches can automatically identify detailed brain structures from 3-D magnetic resonance (MR) brain images. However, the accuracy often degrades when processing data acquired on a different scanner platform or pulse sequence than the data used for the atlas training. In this project, we work to improve the performance of an atlas-based whole brain segmentation method by introducing an intensity renormalization procedure that automatically adjusts the prior atlas intensity model to new input data. Validation using manually labeled test datasets shows that the new procedure improves segmentation accuracy (as measured by the Dice coefficient) by 10% or more for several structures including hippocampus, amygdala, caudate, and pallidum. The results verify that this new procedure reduces the sensitivity of the whole brain segmentation method to changes in scanner platforms and improves its accuracy and robustness, which can thus facilitate multicenter or multisite neuroanatomical imaging studies.
| |
| − | | |
| − | *Multiscale Shape Segmentation Techniques
| |
| − | | |
| − | The goal of this project is to represent multiscale variations in a shape population in order to drive the segmentation of deep brain structures, such as the caudate nucleus or the hippocampus. Our technique defines a multiscale parametric model of surfaces belonging to the same population using a compact set of spherical wavelets targeted to that population. We derived a parametric active surface evolution using the multiscale prior coefficients as parameters for our optimization procedure to naturally include the prior for segmentation. Additionally, the optimization method can be applied in a coarse-to-fine manner. We applied our algorithm to the caudate nucleus, a brain structure of interest in the study of schizophrenia. Our validation shows that our algorithm is computationally efficient and outperforms the Active Shape Model (ASM) algorithm, by capturing finer shape details.
| |
| − | | |
| − | Registration
| |
| − | | |
| − | * Optimal Mass Transport Registration
| |
| − | The aim of this project is to provide a computationally efficient non-rigid/elastic image registration algorithm based on the Optimal Mass Transport theory. We use the Monge-Kantorovich formulation of the Optimal Mass Transport problem and implement the gradient flow PDE approach using multi-resolution and multi-grid techniques to speed up the convergence. We also leverage the computational power of general purpose graphics processing units available on standard desktop computing machines to exploit the inherent parallelism in our algorithm. We have implemented 2D and 3D multi-resolution registration using Optimal Mass Transport and are currently working on the registration of 3D datasets. | |
| − | | |
| − | * Diffusion Tensor Image Processing Tools
| |
| − |
| |
| − | We aim to provide methods for computing geodesics and distances between diffusion tensors. One goal is to provide hypothesis testing for differences between groups. This will involve interpolation techniques for diffusion tensors as weighted averages in the metric framework. We will also provide filtering and eddy current correction. This year, we developed a Slicer module for DT-MRI Rician noise removal, developed prototypes of DTI geometry and statistical packages, and began work on a general method for hypothesis testing between diffusion tensor groups.
| |
| − | | |
| − | * Point Set Rigid Registration
| |
| − | | |
| − | We propose a particle filtering scheme for the registration of 2D and 3D point set undergoing a rigid body transformation where we incorporate stochastic dynamics to model the uncertainty of the registration process. Typically, registration algorithms compute the transformations parameters by maximizing a metric given an estimate of the correspondence between points across the two sets of interest. This can be viewed as a posterior estimation problem, in which the corresponding distribution can naturally be estimated using a particle filter. In this work, we treat motion as a local variation in the pose parameters obtained from running a few iterations of the standard Iterative Closest Point (ICP) algorithm. Employing this idea, we introduce stochastic motion dynamics to widen the narrow band of convergence as well as provide a dynamical model of uncertainty. In contrast with other techniques, our approach requires no annealing schedule, which results in a reduction in computational complexity as well as maintains the temporal coherency of the state (no loss of information). Also, unlike most alternative approaches for point set registration, we make no geometric assumptions on the two data sets.
| |
| − | | |
| − | * Cortical Correspondence using Particle System
| |
| − | | |
| − | In this project, we want to compute cortical correspondence on populations, using various features such as cortical structure, DTI connectivity, vascular structure, and functional data (fMRI). This presents a challenge because of the highly convoluted surface of the cortex, as well as because of the different properties of the data features we want to incorporate together. We would like to use a particle based entropy minimizing system for the correspondence computation, in a population-based manner. This is advantageous because it does not require a spherical parameterization of the surface, and does not require the surface to be of spherical topology. It would also eventually enable correspondence computation on the subcortical structures and on the cortical surface using the same framework. To circumvent the disadvantage that particles are assumed to lie on local tangent planes, we plan to first ‘inflate’ the cortex surface. Currently, we are at testing stage using structural data, namely, point locations and sulcal depth (as computed by FreeSurfer).
| |
| − | | |
| − | * Multimodal Atlas
| |
| − | | |
| − | In this work, we propose and investigate an algorithm that jointly co-registers a collection of images while computing multiple templates. The algorithm, called iCluster for Image Clustering, is based on the following idea: given the templates, the co-registration problem becomes simple, reducing to a number of pairwise registration instances. On the other hand, given a collection of images that have been co-registered, an off-the shelf clustering or averaging algorithm can be used to compute the templates. The algorithm assumed a fixed and known number of template images. We formulate the problem as a maximum likelihood solution and employ a Generalized Maximum Likelihood algorithm to solve it. In the E-step, we compute membership probabilities. In the M-step, we update the template images as weighted averages of the images, where weights are the memberships and the template priors are updated, and then perform a collection of independent pairwise registration instances. The algorithm is currently implemented in the Insight ToolKit (ITK) and we next plan to integrate it into Slicer.
| |
| − | | |
| − | * Groupwise Registration
| |
| − | | |
| − | We aim at providing efficient groupwise registration algorithms for population analysis of anatomical structures. Here we extend a previously demonstrated entropy based groupwise registration method to include a free-form deformation model based on B-splines. We provide an efficient implementation using stochastic gradient descents in a multi-resolution setting. We demonstrate the method in application to a set of 50 MRI brain scans and compare the results to a pairwise approach using segmentation labels to evaluate the quality of alignment. Our results indicate that increasing the complexity of the deformation model improves registration accuracy significantly, especially at cortical regions.
| |
| − | | |
| − | Shape Analysis
| |
| − | | |
| − | * Shape Analysis Framework Using SPHARM-PDM
| |
| − | | |
| − | The UNC shape analysis is based on an analysis framework of objects with spherical topology, described by sampled spherical harmonics SPHARM-PDM. The input of the proposed shape analysis is a set of binary segmentations of a single brain structure, such as the hippocampus or caudate. Group tests can be visualized by P-values and by mean difference magnitude and vector maps, as well as maps of the group covariance information. The implementation has reached a stable framework and has been disseminated to several collaborating labs within NAMIC (BWH, Georgia Tech, Utah). The current development focuses on integrating the current command line tools into the Slicer (v3) via the Slicer execution model. The whole shape analysis pipeline is encapsulated and accessible to the trained clinical collaborator. The current toolset distribution (via NeuroLib) now also contains open data for other researchers to evaluate their shape analysis enhancements.
| |
| − | | |
| − | * Multiscale Shape Analysis
| |
| − | | |
| − | We present a novel method of statistical surface-based morphometry based on the use of non-parametric permutation tests and a spherical wavelet (SWC) shape representation. As an application, we analyze two brain structures, the caudate nucleus and the hippocampus. We show that the results nicely complement the results obtained with shape analysis using a sampled point representation (SPHARM-PDM). We used the UNC pipeline to pre-process the images, and for each triangulated SPHARM-PDM surface, a spherical wavelet description is computed. We then use the UNC statistical toolbox to analyze differences between two groups of surfaces described by the features of choice that is the 3D spherical wavelet coefficients. This year, we conducted statistical shape analysis of the two brain structures and compared the results obtained to shape analysis using a SPHARM-PDM representation.
| |
| − | | |
| − | * Population Analysis of Anatomical Variability
| |
| − | | |
| − | In contrast to shape-based segmentation that utilizes a statistical model of the shape variability in one population (typically based on Principal Component Analysis), we are interested in identifying and characterizing differences between two sets of shape examples. We use the discriminative framework to characterize the differences in shape by training a classifier function and studying its sensitivity to small perturbations in the input data. An additional benefit is that the resulting classifier function can be used to label new examples into one of the two populations, e.g., for early detection in population screening or prediction in longitudinal studies. We have implemented stand alone code for training a classifier, jackknifing and permutation testing, and are currently porting the software into ITK. We have also started exploring alternative, surface-based descriptors which are promising in improving our ability to detect and characterize subtle differences in the shape of anatomical structures due to diseases such as schizophrenia.
| |
| − | | |
| − | * Shape Analysis with Overcomplete Wavelets
| |
| − | | |
| − | In this work, we extend the Euclidean wavelets to the sphere. The resulting over-complete spherical wavelets are invariant to the rotation of the spherical image parameterization. We apply the over-complete spherical wavelet to cortical folding development and show significantly consistent results as well as improved sensitivity compared with the previously used bi-orthogonal spherical wavelet. In particular, we are able to detect developmental asymmetry in the left and right hemispheres.
| |
| − | | |
| − | *Shape based Segmentation and Registration
| |
| − | | |
| − | When there is little or no contrast along boundaries of different regions, standard image segmentation algorithms perform poorly and segmentation is done manually using prior knowledge of shape and relative location of underlying structures. We have proposed an automated approach guided by covariant shape deformations of neighboring structures, which is an additional source of prior knowledge. Captured by a shape atlas, these deformations are transformed into a statistical model using the logistic function. The mapping between atlas and image space, structure boundaries, anatomical labels, and image inhomogeneities are estimated simultaneously within an expectation-maximization formulation of the maximum a posteriori Probability (MAP) estimation problem. These results are then fed into an Active Mean Field approach, which views the results as priors to a Mean Field approximation with a curve length prior. Our method filters out the noise as compared to thresholding using initial likelihoods, and it captures multiple structures as in the brain (where both major brain compartments and subcortical structures are obtained) because it naturally evolves families of curves. The algorithm is currently implemented in 3D Slicer Version 2.6 and a beta version is available in 3D Slicer Version 3.
| |
| − | | |
| − | *Spherical Wavelets
| |
| − | | |
| − | In this project, we apply a spherical wavelet transformation to extract shape features of cortical surfaces reconstructed from magnetic resonance images (MRI) of a set of subjects. The spherical wavelet transformation can characterize the underlying functions in a local fashion in both space and frequency, in contrast to spherical harmonics that have a global basis set. We perform principal component analysis (PCA) on these wavelet shape features to study patterns of shape variation within normal population from coarse to fine resolution. In addition, we study the development of cortical folding in newborns using the Gompertz model in the wavelet domain, allowing us to characterize the order of development of large-scale and finer folding patterns independently. We develop an efficient method to estimate the regularized Gompertz model based on the Broyden–Fletcher–Goldfarb–Shannon (BFGS) approximation. Promising results are presented using both PCA and the folding development model in the wavelet domain. The cortical folding development model provides quantitative anatomical information regarding macroscopic cortical folding development and may be of potential use as a biomarker for early diagnosis of neurological deficits in newborns.
| |
| − | | |
| − | ===Key Investigators===
| |
| − | * MIT: Polina Golland, Kilian Pohl, Sandy Wells, Eric Grimson, Mert R. Sabuncu
| |
| − | * UNC: Martin Styner, Ipek Oguz, Xavier Barbero
| |
| − | * Utah: Ross Whitaker, Guido Gerig, Suyash Awate, Tolga Tasdizen, Tom Fletcher, Joshua Cates, Miriah Meyer
| |
| − | * GaTech: Allen Tannenbaum, John Melonakos, Vandana Mohan, Tauseef ur Rehman, Shawn Lankton, Samuel Dambreville, Yi Gao, Romeil Sandhu, Xavier Le Faucheur, James Malcolm
| |
| − | * Isomics: Steve Pieper
| |
| − | * GE: Bill Lorensen, Jim Miller
| |
| − | * Kitware: Luis Ibanez, Karthik Krishnan
| |
| − | * UCLA: Arthur Toga, Michael J. Pan, Jagadeeswaran Rajendiran
| |
| − | * BWH: Sylvain Bouix, Motoaki Nakamura, Min-Seong Koo, Martha Shenton, Marc Niethammer, Jim Levitt, Yogesh Rathi, Marek Kubicki, Steven Haker
| |
| − | | |
| − | ===Additional Information===
| |
| − | Additional Information for this topic is available [http://wiki.na-mic.org/Wiki/index.php/NA-MIC_Internal_Collaborations:StructuralImageAnalysis here on the NA-MIC wiki].
| |
| − | ==fMRI Analysis (Golland)==
| |
| − | ===Progress===
| |
| − | ===Key Investigators===
| |
| − | ===Additional Information===
| |
| − | Additional Information for this topic is available [http://wiki.na-mic.org/Wiki/index.php/NA-MIC_Internal_Collaborations:fMRIAnalysis here on the NA-MIC wiki].
| |
| − | ==NA-MIC Kit Theme (Schroeder)==
| |
| − | ===Progress===
| |
| − | ===Key Investigators===
| |
| − | ===Additional Information===
| |
| − | Additional Information for this topic is available [http://wiki.na-mic.org/Wiki/index.php/NA-MIC-Kit here on the NA-MIC wiki].
| |
| − | ==Other Projects==
| |
| − | Any Project(s) not covered by the 8 sections above
| |
| − | | |
| − | ==Highlights(Schroeder)==
| |
| − | ===EM Segmenter or TBD===
| |
| − | ===DTI progress or TBD===
| |
| − | ===Outreach (Gollub)===
| |
| − | | |
| − | ==Impact and Value to Biocomputing (Miller)==
| |
| − | ===Impact within the Center===
| |
| − | ===Impact within NIH Funded Research===
| |
| − | ===National and International Impact===
| |
| − | ==NA-MIC Timeline (Whitaker)==
| |
| − | | |
| − | ==Appendix A Publications (Kapur)==
| |
| − | These will be mined from the SPL publications database. All core PIs need to ensure that all NA-MIC publications are in the publications database by May 15.
| |
| − | | |
| − | ==Appendix B EAB Report and Response (Kapur)==
| |
| − | ===EAB Report===
| |
| − | ===Response to EAB Report===
| |
