Publications by Year: 2008

Simon K Warfield, Kelly H Zou, and William M Wells III. 7/2008. “Validation of Image Segmentation by Estimating Rater Bias and Variance.” Philos Trans A Math Phys Eng Sci, 366, 1874, Pp. 2361-75.Abstract

The accuracy and precision of segmentations of medical images has been difficult to quantify in the absence of a 'ground truth' or reference standard segmentation for clinical data. Although physical or digital phantoms can help by providing a reference standard, they do not allow the reproduction of the full range of imaging and anatomical characteristics observed in clinical data. An alternative assessment approach is to compare with segmentations generated by domain experts. Segmentations may be generated by raters who are trained experts or by automated image analysis algorithms. Typically, these segmentations differ due to intra-rater and inter-rater variability. The most appropriate way to compare such segmentations has been unclear. We present here a new algorithm to enable the estimation of performance characteristics, and a true labelling, from observations of segmentations of imaging data where segmentation labels may be ordered or continuous measures. This approach may be used with, among others, surface, distance transform or level-set representations of segmentations, and can be used to assess whether or not a rater consistently overestimates or underestimates the position of a boundary.

Eli Hershkovits, Allen Tannenbaum, and Rina Tannenbaum. 2008. “Adsorption of Block Copolymers from Selective Solvents on Curved Surfaces.” Macromolecules, 41, 9, Pp. 3190-3198.Abstract
We have investigated the adsorption of asymmetric poly(styrene-b-methyl methacrylate) block copolymers (PS-PMMA) from a selective solvent onto alumina (Al(2)O(3)) particles having variable and controllable radii. The solvent used was a bad solvent for the PS block (block A) and a good solvent for the PMMA block (block B), which has a higher affinity of the surface. Such a case represents a new class of adsorption, where both blocks compete for the adsorption sites of the metallic surface. Two theoretical models, the modified drops model and the perforated film model, have been evaluated as appropriate representation of such an adsorption scenario. The experimental results indicated that the adsorption of the PS-PMMA block copolymer generated a patterned surface comprised of a homogeneous melt layer of the PS block perforated with holes having a variable PMMA structure, depending on the distance from the bottom of the hole (alumina surface) and the distance from walls of the hole. The density gradient of the PMMA moiety in the hole reverted to the classical brush morphology at a critical distance from the surface of the hole.
Gudrun Rosenberger, Marek Kubicki, Paul G Nestor, Erin Connor, Georgia B Bushell, Douglas Markant, Margaret Niznikiewicz, Carl-Fredrik Westin, Ron Kikinis, Andrew J Saykin, Robert W McCarley, and Martha E Shenton. 2008. “Age-related deficits in fronto-temporal connections in schizophrenia: a diffusion tensor imaging study.” Schizophr Res, 102, 1-3, Pp. 181-8.Abstract
OBJECTIVE: Impairment of white matter connecting frontal and temporal cortices has been reported in schizophrenia. Yet, not much is known about the effects of age on fibers connecting these brain regions. Using diffusion tensor imaging tractography, we investigated the relationship between age and fiber integrity in patients with schizophrenia vs. healthy adults. METHODS: DTI tractography was used to create 3D reconstructions of the cingulum, uncinate and inferior occipito-frontal fasciculi in 27 patients with schizophrenia and 34 healthy volunteers (23-56 years of age, group-matched on age). Fractional anisotropy (FA), describing fiber integrity, was then calculated along the entire length of these tracts, and correlated with subjects' age. RESULTS: Patients revealed a significant decline in FA with age in both the cingulum and uncinate, but not in the inferior occipito-frontal fasciculi. No statistically significant correlations were found in these fiber bundles in controls. CONCLUSIONS: These results suggest an age-associated reduction of frontal-temporal connectivity in schizophrenia, but not in healthy controls.
Oleg Michailovich and Allen Tannenbaum. 2008. “On approximation of smooth functions from samples of partial derivatives with application to phase unwrapping.” Signal Processing, 88, 2, Pp. 358-374.Abstract
This paper addresses the problem of approximating smooth bivariate functions from the samples of their partial derivatives. The approximation is carried out under the assumption that the subspace to which the functions to be recovered are supposed to belong, possesses an approximant in the form of a principal shift-invariant (PSI) subspace. Subsequently, the desired approximation is found as the element of the PSI subspace that fits the data the best in the (2)-sense. In order to alleviate the ill-posedness of the process of finding such a solution, we take advantage of the discrete nature of the problem under consideration. The proposed approach allows the explicit construction of a projection operator which maps the measured derivatives into a stable and unique approximation of the corresponding function. Moreover, the paper develops the concept of discrete PSI subspaces, which may be of relevance for several practical settings where one is given samples of a function instead of its continuously defined values. As a final point, the application of the proposed method to the problem of phase unwrapping in homomorphic deconvolution is described.
Matthew Jolley, Jeroen Stinstra, Steve Pieper, Rob MacLeod, Dana H Brooks, Frank Cecchin, and John K Triedman. 2008. “A computer modeling tool for comparing novel ICD electrode orientations in children and adults.” Heart Rhythm, 5, 4, Pp. 565-72.Abstract
BACKGROUND: Use of implantable cardiac defibrillators (ICDs) in children and patients with congenital heart disease is complicated by body size and anatomy. A variety of creative implantation techniques has been used empirically in these groups on an ad hoc basis. OBJECTIVE: To rationalize ICD placement in special populations, we used subject-specific, image-based finite element models (FEMs) to compare electric fields and expected defibrillation thresholds (DFTs) using standard and novel electrode configurations. METHODS: FEMs were created by segmenting normal torso computed tomography scans of subjects ages 2, 10, and 29 years and 1 adult with congenital heart disease into tissue compartments, meshing, and assigning tissue conductivities. The FEMs were modified by interactive placement of ICD electrode models in clinically relevant electrode configurations, and metrics of relative defibrillation safety and efficacy were calculated. RESULTS: Predicted DFTs for standard transvenous configurations were comparable with published results. Although transvenous systems generally predicted lower DFTs, a variety of extracardiac orientations were also predicted to be comparably effective in children and adults. Significant trend effects on DFTs were associated with body size and electrode length. In many situations, small alterations in electrode placement and patient anatomy resulted in significant variation of predicted DFT. We also show patient-specific use of this technique for optimization of electrode placement. CONCLUSION: Image-based FEMs allow predictive modeling of defibrillation scenarios and predict large changes in DFTs with clinically relevant variations of electrode placement. Extracardiac ICDs are predicted to be effective in both children and adults. This approach may aid both ICD development and patient-specific optimization of electrode placement. Further development and validation are needed for clinical or industrial utilization.
Boon Thye Thomas Yeo, Wanmei Ou, and Polina Golland. 2008. “On the construction of invertible filter banks on the 2-sphere.” IEEE Trans Image Process, 17, 3, Pp. 283-300.Abstract
The theories of signal sampling, filter banks, wavelets, and "overcomplete wavelets" are well established for the Euclidean spaces and are widely used in the processing and analysis of images. While recent advances have extended some filtering methods to spherical images, many key challenges remain. In this paper, we develop theoretical conditions for the invertibility of filter banks under continuous spherical convolution. Furthermore, we present an analogue of the Papoulis generalized sampling theorem on the 2-Sphere. We use the theoretical results to establish a general framework for the design of invertible filter banks on the sphere and demonstrate the approach with examples of self-invertible spherical wavelets and steerable pyramids. We conclude by examining the use of a self-invertible spherical steerable pyramid in a denoising experiment and discussing the computational complexity of the filtering framework.
Peter Kazanzides, Tian Xia, Clint Baird, George Jallo, Kathryn Hayes, Nobuyuki Nakajima, and Nobuhiko Hata. 2008. “A cooperatively-controlled image guided robot system for skull base surgery.” Stud Health Technol Inform, 132, Pp. 198-203.Abstract
We created an image-guided robot system to assist with skull base drilling by integrating a robot, a commercial navigation system, and an open source visualization platform. The objective of this procedure is to create a cavity in the skull base to allow access for neurosurgical interventions. The motivation for introducing an image-guided robot is to improve safety by preventing the surgeon from accidentally damaging critical structures during the drilling procedure. Our approach is to attach the cutting tool to the robot end-effector and operate the robot in a cooperative control mode, where robot motion is determined from the forces and torques applied by the surgeon. We employ "virtual fixtures" to constrain the motion of the cutting tool so that it remains in the safe zone that was defined on a preoperative CT scan. This paper presents the system design and the results of phantom and cadaveric experiments. Both experiments have demonstrated the feasibility of the system, with average overcut error at about 1 mm and maximum errors at 2.5 mm.
Mert R Sabuncu, Serdar K Balci, and Polina Golland. 2008. “Discovering modes of an image population through mixture modeling.” Med Image Comput Comput Assist Interv, 11, Pt 2, Pp. 381-9.Abstract
We present iCluster, a fast and efficient algorithm that clusters a set of images while co-registering them using a parameterized, nonlinear transformation model. The output is a small number of template images that represent different modes in a population. This is in contrast with traditional approaches that assume a single template to construct atlases. We validate and explore the algorithm in two experiments. First, we employ iCluster to partition a data set of 416 whole brain MR volumes of subjects aged 18-96 years into three sub-groups, which mainly correspond to age groups. The templates reveal significant structural differences across these age groups that confirm previous findings in aging research. In the second experiment, we run iCluster on a group of 30 patients with dementia and 30 age-matched healthy controls. The algorithm produced three modes that mainly corresponded to a sub-population of healthy controls, a sub-population of patients with dementia and a mixture group that contained both types. These results suggest that the algorithm can be used to discover sub-populations that correspond to interesting structural or functional "modes".
Danial Lashkari, Ed Vul, Nancy Kanwisher, and Polina Golland. 2008. “Discovering structure in the space of activation profiles in fMRI.” Med Image Comput Comput Assist Interv, 11, Pt 1, Pp. 1016-24.Abstract
We present a method for discovering patterns of activation observed through fMIRI in experiments with multiple stimuli/tasks. We introduce an explicit parameterization for the profiles of activation and represent fMRI time courses as such profiles using linear regression estimates. Working in the space of activation profiles, we design a mixture model that finds the major activation patterns along with their localization maps and derive an algorithm for fitting the model to the fMRI data. The method enables functional group analysis independent of spatial correspondence among subjects. We validate this model in the context of category selectivity in the visual cortex, demonstrating good agreement with prior findings based on hypothesis-driven methods.
Wanmei Ou, Polina Golland, and Matti Hämäläinen. 2008. “A distributed spatio-temporal EEG/MEG inverse solver.” Med Image Comput Comput Assist Interv, 11, Pt 1, Pp. 26-34.Abstract
We propose a novel l1l2-norm inverse solver for estimating the sources of EEG/MEG signals. Based on the standard l1-norm inverse solver, the proposed sparse distributed inverse solver integrates the l1-norm spatial model with a temporal model of the source signals in order to avoid unstable activation patterns and "spiky" reconstructed signals often produced by the original solvers. The joint spatio-temporal model leads to a cost function with an l1l2-norm regularizer whose minimization can be reduced to a convex second-order cone programming problem and efficiently solved using the interior-point method. Validation with simulated and real MEG data shows that the proposed solver yields source time course estimates qualitatively similar to those obtained through dipole fitting, but without the need to specify the number of dipole sources in advance. Furthermore, the l1l2-norm solver achieves fewer false positives and a better representation of the source locations than the conventional l2 minimum-norm estimates.
Oleg Michailovich and Allen Tannenbaum. 2008. “Dynamic denoising of tracking sequences.” IEEE Trans Image Process, 17, 6, Pp. 847-56.Abstract
In this paper, we describe an approach to the problem of simultaneously enhancing image sequences and tracking the objects of interest represented by the latter. The enhancement part of the algorithm is based on Bayesian wavelet denoising, which has been chosen due to its exceptional ability to incorporate diverse a priori information into the process of image recovery. In particular, we demonstrate that, in dynamic settings, useful statistical priors can come both from some reasonable assumptions on the properties of the image to be enhanced as well as from the images that have already been observed before the current scene. Using such priors forms the main contribution of the present paper which is the proposal of the dynamic denoising as a tool for simultaneously enhancing and tracking image sequences. Within the proposed framework, the previous observations of a dynamic scene are employed to enhance its present observation. The mechanism that allows the fusion of the information within successive image frames is Bayesian estimation, while transferring the useful information between the images is governed by a Kalman filter that is used for both prediction and estimation of the dynamics of tracked objects. Therefore, in this methodology, the processes of target tracking and image enhancement "collaborate" in an interlacing manner, rather than being applied separately. The dynamic denoising is demonstrated on several examples of SAR imagery. The results demonstrated in this paper indicate a number of advantages of the proposed dynamic denoising over "static" approaches, in which the tracking images are enhanced independently of each other.
Thomas BT Yeo, Mert R Sabuncu, Rahul Desikan, Bruce Fischl, and Polina Golland. 2008. “Effects of registration regularization and atlas sharpness on segmentation accuracy.” Med Image Anal, 12, 5, Pp. 603-15.Abstract
In non-rigid registration, the tradeoff between warp regularization and image fidelity is typically determined empirically. In atlas-based segmentation, this leads to a probabilistic atlas of arbitrary sharpness: weak regularization results in well-aligned training images and a sharp atlas; strong regularization yields a "blurry" atlas. In this paper, we employ a generative model for the joint registration and segmentation of images. The atlas construction process arises naturally as estimation of the model parameters. This framework allows the computation of unbiased atlases from manually labeled data at various degrees of "sharpness", as well as the joint registration and segmentation of a novel brain in a consistent manner. We study the effects of the tradeoff of atlas sharpness and warp smoothness in the context of cortical surface parcellation. This is an important question because of the increasingly availability of atlases in public databases, and the development of registration algorithms separate from the atlas construction process. We find that the optimal segmentation (parcellation) corresponds to a unique balance of atlas sharpness and warp regularization, yielding statistically significant improvements over the FreeSurfer parcellation algorithm. Furthermore, we conclude that one can simply use a single atlas computed at an optimal sharpness for the registration-segmentation of a new subject with a pre-determined, fixed, optimal warp constraint. The optimal atlas sharpness and warp smoothness can be determined by probing the segmentation performance on available training data. Our experiments also suggest that segmentation accuracy is tolerant up to a small mismatch between atlas sharpness and warp smoothness.
James Malcolm, Yogesh Rathi, Anthony Yezzi, and Allen Tannenbaum. 2008. “Fast approximate surface evolution in arbitrary dimension.” Proc SPIE Int Soc Opt Eng, 6914.Abstract
The level set method is a popular technique used in medical image segmentation; however, the numerics involved make its use cumbersome. This paper proposes an approximate level set scheme that removes much of the computational burden while maintaining accuracy. Abandoning a floating point representation for the signed distance function, we use integral values to represent the signed distance function. For the cases of 2D and 3D, we detail rules governing the evolution and maintenance of these three regions. Arbitrary energies can be implemented in the framework. This scheme has several desirable properties: computations are only performed along the zero level set; the approximate distance function requires only a few simple integer comparisons for maintenance; smoothness regularization involves only a few integer calculations and may be handled apart from the energy itself; the zero level set is represented exactly removing the need for interpolation off the interface; and evolutions proceed on the order of milliseconds per iteration on conventional uniprocessor workstations. To highlight its accuracy, flexibility and speed, we demonstrate the technique on intensity-based segmentations under various statistical metrics. Results for 3D imagery show the technique is fast even for image volumes.
Clare Poynton, Mark Jenkinson, Stephen Whalen, Alexandra J Golby, and William Wells. 2008. “Fieldmap-free retrospective registration and distortion correction for EPI-based functional imaging.” Med Image Comput Comput Assist Interv, 11, Pt 2, Pp. 271-9.Abstract
We describe a method for correcting the distortions present in echo planar images (EPI) and registering the EPI to structural MRI. A fieldmap is predicted from an air / tissue segmentation of the MRI using a perturbation method and subsequently used to unwarp the EPI data. Shim and other missing parameters are estimated by registration. We obtain results that are similar to those obtained using fieldmaps, however neither fieldmaps, nor knowledge of shim coefficients is required.
John Melonakos, Eric Pichon, Sigurd Angenent, and Allen Tannenbaum. 2008. “Finsler active contours.” IEEE Trans Pattern Anal Mach Intell, 30, 3, Pp. 412-23.Abstract
In this paper, we propose an image segmentation technique based on augmenting the conformal (or geodesic) active contour framework with directional information. In the isotropic case, the Euclidean metric is locally multiplied by a scalar conformal factor based on image information such that the weighted length of curves lying on points of interest (typically edges) is small. The conformal factor which is chosen depends only upon position and is in this sense isotropic. While directional information has been studied previously for other segmentation frameworks, here we show that if one desires to add directionality in the conformal active contour framework, then one gets a well-defined minimization problem in the case that the factor defines a Finsler metric. Optimal curves may be obtained using the calculus of variations or dynamic programming based schemes. Finally we demonstrate the technique by extracting roads from aerial imagery, blood vessels from medical angiograms, and neural tracts from diffusion-weighted magnetic resonance imagery.
Samuel Dambreville, Yogesh Rathi, and Allen Tannenbaum. 2008. “A framework for image segmentation using shape models and kernel space shape priors.” IEEE Trans Pattern Anal Mach Intell, 30, 8, Pp. 1385-99.Abstract
Segmentation involves separating an object from the background in a given image. The use of image information alone often leads to poor segmentation results due to the presence of noise, clutter or occlusion. The introduction of shape priors in the geometric active contour (GAC) framework has proved to be an effective way to ameliorate some of these problems. In this work, we propose a novel segmentation method combining image information with prior shape knowledge, using level-sets. Following the work of Leventon et al., we propose to revisit the use of PCA to introduce prior knowledge about shapes in a more robust manner. We utilize kernel PCA (KPCA) and show that this method outperforms linear PCA by allowing only those shapes that are close enough to the training data. In our segmentation framework, shape knowledge and image information are encoded into two energy functionals entirely described in terms of shapes. This consistent description permits to fully take advantage of the Kernel PCA methodology and leads to promising segmentation results. In particular, our shape-driven segmentation technique allows for the simultaneous encoding of multiple types of shapes, and offers a convincing level of robustness with respect to noise, occlusions, or smearing.
Marc Niethammer, Patricio A Vela, and Allen Tannenbaum. 2008. “Geometric observers for dynamically evolving curves.” IEEE Trans Pattern Anal Mach Intell, 30, 6, Pp. 1093-108.Abstract
This paper proposes a deterministic observer framework for visual tracking based on non-parametric implicit (level-set) curve descriptions. The observer is continuous-discrete, with continuous-time system dynamics and discrete-time measurements. Its state-space consists of an estimated curve position augmented by additional states (e.g., velocities) associated with every point on the estimated curve. Multiple simulation models are proposed for state prediction. Measurements are performed through standard static segmentation algorithms and optical-flow computations. Special emphasis is given to the geometric formulation of the overall dynamical system. The discrete-time measurements lead to the problem of geometric curve interpolation and the discrete-time filtering of quantities propagated along with the estimated curve. Interpolation and filtering are intimately linked to the correspondence problem between curves. Correspondences are established by a Laplace-equation approach. The proposed scheme is implemented completely implicitly (by Eulerian numerical solutions of transport equations) and thus naturally allows for topological changes and subpixel accuracy on the computational grid.
James Malcolm, Yogesh Rathi, and Allen Tannenbaum. 2008. “A Graph Cut Approach to Image Segmentation in Tensor Space.” Proc IEEE Comput Soc Conf Comput Vis Pattern Recognit, Pp. 1-8.Abstract
This paper proposes a novel method to apply the standard graph cut technique to segmenting multimodal tensor valued images. The Riemannian nature of the tensor space is explicitly taken into account by first mapping the data to a Euclidean space where non-parametric kernel density estimates of the regional distributions may be calculated from user initialized regions. These distributions are then used as regional priors in calculating graph edge weights. Hence this approach utilizes the true variation of the tensor data by respecting its Riemannian structure in calculating distances when forming probability distributions. Further, the non-parametric model generalizes to arbitrary tensor distribution unlike the Gaussian assumption made in previous works. Casting the segmentation problem in a graph cut framework yields a segmentation robust with respect to initialization on the data tested.
Jinsong Ouyang, Georges El Fakhri, and Stephen C Moore. 2008. “Improved activity estimation with MC-JOSEM versus TEW-JOSEM in 111In SPECT.” Med Phys, 35, 5, Pp. 2029-40.Abstract
We have previously developed a fast Monte Carlo (MC)-based joint ordered-subset expectation maximization (JOSEM) iterative reconstruction algorithm, MC-JOSEM. A phantom study was performed to compare quantitative imaging performance of MC-JOSEM with that of a triple-energy-window approach (TEW) in which estimated scatter was also included additively within JOSEM, TEW-JOSEM. We acquired high-count projections of a 5.5 cm3 sphere of 111In at different locations in the water-filled torso phantom; high-count projections were then obtained with 111In only in the liver or only in the soft-tissue background compartment, so that we could generate synthetic projections for spheres surrounded by various activity distributions. MC scatter estimates used by MC-JOSEM were computed once after five iterations of TEW-JOSEM. Images of different combinations of liver/background and sphere/background activity concentration ratios were reconstructed by both TEW-JOSEM and MC-JOSEM for 40 iterations. For activity estimation in the sphere, MC-JOSEM always produced better relative bias and relative standard deviation than TEW-JOSEM for each sphere location, iteration number, and activity combination. The average relative bias of activity estimates in the sphere for MC-JOSEM after 40 iterations was -6.9%, versus -15.8% for TEW-JOSEM, while the average relative standard deviation of the sphere activity estimates was 16.1% for MC-JOSEM, versus 27.4% for TEW-JOSEM. Additionally, the average relative bias of activity concentration estimates in the liver and the background for MC-JOSEM after 40 iterations was -3.9%, versus -12.2% for TEW-JOSEM, while the average relative standard deviation of these estimates was 2.5% for MC-JOSEM, versus 3.4% for TEW-JOSEM. MC-JOSEM is a promising approach for quantitative activity estimation in 111In SPECT.
Tian Xia, Clint Baird, George Jallo, Kathryn Hayes, Nobuyuki Nakajima, Nobuhiko Hata, and Peter Kazanzides. 2008. “An integrated system for planning, navigation and robotic assistance for skull base surgery.” Int J Med Robot, 4, 4, Pp. 321-30.Abstract
BACKGROUND: We developed an image-guided robot system to provide mechanical assistance for skull base drilling, which is performed to gain access for some neurosurgical interventions, such as tumour resection. The motivation for introducing this robot was to improve safety by preventing the surgeon from accidentally damaging critical neurovascular structures during the drilling procedure. METHODS: We integrated a Stealthstation navigation system, a NeuroMate robotic arm with a six-degree-of-freedom force sensor, and the 3D Slicer visualization software to allow the robotic arm to be used in a navigated, cooperatively-controlled fashion by the surgeon. We employed virtual fixtures to constrain the motion of the robot-held cutting tool, so that it remained in the safe zone that was defined on a preoperative CT scan. RESULTS: We performed experiments on both foam skull and cadaver heads. The results for foam blocks cut using different registrations yielded an average placement error of 0.6 mm and an average dimensional error of 0.6 mm. We drilled the posterior porus acusticus in three cadaver heads and concluded that the robot-assisted procedure is clinically feasible and provides some ergonomic benefits, such as stabilizing the drill. We obtained postoperative CT scans of the cadaver heads to assess the accuracy and found that some bone outside the virtual fixture boundary was cut. The typical overcut was 1-2 mm, with a maximum overcut of about 3 mm. CONCLUSIONS: The image-guided cooperatively-controlled robot system can improve the safety and ergonomics of skull base drilling by stabilizing the drill and enforcing virtual fixtures to protect critical neurovascular structures. The next step is to improve the accuracy so that the overcut can be reduced to a more clinically acceptable value of about 1 mm.