We propose a new clustering algorithm. This algorithm performs clustering and manifold learning simultaneously by using a graph-theoretical approach to manifold learning. We apply this algorithm in order to cluster white matter ﬁber tracts obtained fromDiffusion TensorMRI (DT-MRI) through streamline tractography. Our algorithm is able perform clustering of these fiber tracts incorporating information about the shape of the ﬁber and a priori knowledge as the probability of the fiber belonging to known anatomical structures. This anatomical knowledge is incorporated as a volumetric white matter atlas, in this case LONI's ICBM DTI-81

Simultaneous Manifold Learning and Clustering: Grouping White Matter Fiber Tracts Using a Volumetric White Matter Atlas

The ﬁeld of computational anatomy has developed rigorous frameworks for analyzing anatomical shape, based on diffeomorphic transformations of a template. However, differences in algorithms used for template warping, in regularization parameters, and in the template itself, lead to different representations of the same anatomy. Variations of these parameters are considered as confounding factors. Recently, extensions of the conventional computational anatomy framework to account for such confounding variations has shown that learning the equivalence class derived from the multitude of representations can lead to improved and more stable morphological descriptors. Herein, we follow that approach, estimating the morphological appearance manifold obtained by varying parameters of the template warping procedure. Our approach parallels work in the computer vision ﬁeld, in which variations lighting, pose and other parameters leads to image appearance manifolds representing the exact same ﬁgure in different ways. The proposed framework is then used for groupwise registration and statistical analysis of biomedical images, by employing a minimum variance criterion to perform manifold-constrained optimization, i.e. to traverse each individual’s morphological appearance manifold until all individuals' representations come as close to each other as possible. Eﬀectively,
this process removes the aforementioned confounding eﬀects and potentially leads to morphological representations reﬂecting purely biological variations, instead of variations introduced by modeling assumptions and parameter settings. The nonlinearity of a morphological appearance manifold is treated via local approximations of the manifold via PCA.

Morphological Appearance Manifolds in Computational Anatomy: Groupwise Registration and Morphological Analysis

This paper describes the setup of a segmentation competition for the automatic extraction of Multiple Sclerosis (MS) lesions from brain Magnetic Resonance Imaging (MRI) data. This competition is one of three competitions that make up a comparison workshop at the 2008 Medical Image Computing and Computer Assisted Intervention (MICCAI) conference and was modeled after the successful comparison workshop on liver and caudate segmentation at the 2007 MICCAI conference. In this paper, the rationale for organizing the competition is discussed, the training and test data sets for both segmentation tasks are described and the scoring system used to evaluate the segmentation is presented.

3D Segmentation in the Clinic: A Grand Challenge II: MS lesion segmentation

In this work a minimum cost path approach is adopted to extract coronary artery centerlines from CTA data. The algorithm depends on the manual definition of the start and end point of the vessel. The cost image used in the minimal cost path approach is based on a vesselness measure and a smooth window function on intensity. In the majority of the cases the method was able to extract the centerlines successfully (overlap > 90%). Accuracy of the method is around two times the voxelsize of the datasets. To conclude, minimum cost path approaches have potential for coronary artery centerline extraction, but improvements, especially regarding the accuracy of the method, still need investigations.

Two Point Minimum Cost Path Approach for CTA Coronary Centerline Extraction

Several algorithms exist for correcting the nonuniform intensity in magnetic resonance images caused by field inhomogeneities. These algorithms constitute important preprocessing steps for subsequent image analysis tasks. One such algorithm, known as parametric bias field correction (PABIC), has already been implemented in ITK. Another popular algorithm is the nonuniform intensity normalization (N3) approach. A particularly salient advantage of this algorithm is that it does not require a prior tissue model for its application. In addition, the source code for N3 is publicly available at the McConnell Brain Imaging Centre (Montreal Neurological Institute, McGill University) which includes source code and the coordinating set of perl scripts. This submission describes an implementation of the N3 algorithm for the Insight Toolkit given as a single class, viz.	itk::N3MRIBiasFieldCorrectionImageFilter. We tried to maintain minimal difference between the publicly available MNI N3 implementation and our ITK im- plementation. The only intentional variation is the substitution of an earlier contribution, i.e. the class itk::BSplineScatteredDataPointSetToImageFilter, for the originally proposed least-squares approach for B-spline fitting used to model the bias field. In addition, we include a more extensive modification to the original N3 algorithm found in the class itk::N4MRIBiasFieldCorrectionImageFilter. The latter algorithm employs a multi-resolution approach, similar to FFD image registration strategies, and has a slightly modified iterative update scheme.

N4ITK: Nick's N3 ITK Implementation For MRI Bias Field Correction

 Branches of tubular structures (vasculature, trachea, neuron, etc.) in medical images are critical for the topology of these structures. In many applications, It is very helpful to be able to decompose tubular structures and identify every individual branch. For example, quantification of geometric vascular features, registration of trachea movement due to respiration, tracing of neuron path. However, manual decomposition can be tedious, time-consuming, and subject to operator bias. In this paper, we propose a novel method to decompose tubular structures automatically and describe how to implement it in ITK framework. The input is a 2D/3D binary image that can be obtained from any segmentation techniques, as well as the junctions, which can be generated automatically from our previously contributed ITK class: itk::JunctionDetectionFilter. The output will be branches with their labels and their connection. There are only two parameters which need to be set by the user. We provide here the implementation as a ITK class: itk::BranchDecompositionFilter.
Please cite the following paper if you are interested in our work.
G. Xiong, C. Chen, J. Chen, Y. Xie, and L. Xing, Tracking the Motion Trajectories of Junction Structures in 4D CT Images of the Lung, Vol. 57, No. 15, pp. 4905-4930, Physics in Medicine and Biology, 2012.

Automatic Branch Decomposition for Tubular Structures

An Insight Toolkit (ITK) processing framework for segmenting and tracking nuclei in time-lapse microscopy images using coupled active contours is presented in this paper. We implement the method of Dufour et al.[2] to segment and track cells in fluorescence microscopy images. The basic idea is to model the image as a constant intensity background with constant intensity foreground components. We utilizes our earlier submissions on the Chan and Vese algorithm [1] and its multiphase extension [5] to build our new tracking filter. The tracking filter itk::MultiphaseLevelSetTracking inputs a segmentation result (or a coarse estimate) from the previous time-point along with the feature image and generates a new segmentation output. By iteratively repeating this process across all time-points, real-time tracking is made possible. We include 2D/3D example code, parameter settings and show the results generated on a 2D zebrafish embryo image series.

Cell Tracking using Coupled Active Surfaces for Nuclei and Membranes

This document describes an implementation of David S. Tuch\'s numerical Q-Ball reconstruction algorithm implemented using the Insight Toolkit ITK (www.itk.org). The document is accompanied with the source code, input data, parameters and output data that were used for validating the implementation. 

Numerical Q-Ball Image Reconstruction: an ITK Implementation

This article provides an implementation of the symmetric log-domain diffeomorphic image registration algorithm, or symmetric demons algorithm for short. It generalizes Thirion's demons and the diffeomorphic demons algorithm. The main practical advantages of the symmetric demons with respect to the other demons variants is that is provides the inverse of the spatial transformation at no additional computational cost and ensures that the registration of image A to image B provides the inverse of the registration from image B to image A. The algorithm works completely in the log-domain, i.e. it uses a stationary velocity field to encode the spatial transformation as its exponential. Within the Insight Toolkit (ITK), the classical demons algorithm is implemented as part of the finite difference solver framework. Our code reuses and extends this generic framework. The source code is composed of a set of reusable ITK filters and classes together with their unit tests. We also provide a small example program that allows the user to compare the different variants of the demons algorithm. This paper gives an overview of the algorithm, an overview of its implementation and a small user guide to ease the use of the registration executable.

An ITK Implementation of the Symmetric Log-Domain Diffeomorphic Demons Algorithm

This document describes a contribution to the Insight Toolkit intended to support the process of registering two Meshes. The methods included here are restricted to Meshes with a Spherical geometry and topology, and with scalar values associated to their nodes.
This paper is accompanied with the source code, input data, parameters and output data that we used for validating the algorithm described in this paper. This adheres to the fundamental principle that scientific publications must facilitate reproducibility of the reported results.