Article

Deformable M-Reps for 3D Medical Image Segmentation

Medical Image Display & Analysis Group, University of North Carolina, Chapel Hill.
International Journal of Computer Vision (Impact Factor: 3.81). 11/2003; 55(2-3):85-106. DOI: 10.1023/A:1026313132218
Source: PubMed

ABSTRACT

M-reps (formerly called DSLs) are a multiscale medial means for modeling and rendering 3D solid geometry. They are particularly well suited to model anatomic objects and in particular to capture prior geometric information effectively in deformable models segmentation approaches. The representation is based on figural models, which define objects at coarse scale by a hierarchy of figures - each figure generally a slab representing a solid region and its boundary simultaneously. This paper focuses on the use of single figure models to segment objects of relatively simple structure. A single figure is a sheet of medial atoms, which is interpolated from the model formed by a net, i.e., a mesh or chain, of medial atoms (hence the name m-reps), each atom modeling a solid region via not only a position and a width but also a local figural frame giving figural directions and an object angle between opposing, corresponding positions on the boundary implied by the m-rep. The special capability of an m-rep is to provide spatial and orientational correspondence between an object in two different states of deformation. This ability is central to effective measurement of both geometric typicality and geometry to image match, the two terms of the objective function optimized in segmentation by deformable models. The other ability of m-reps central to effective segmentation is their ability to support segmentation at multiple levels of scale, with successively finer precision. Objects modeled by single figures are segmented first by a similarity transform augmented by object elongation, then by adjustment of each medial atom, and finally by displacing a dense sampling of the m-rep implied boundary. While these models and approaches also exist in 2D, we focus on 3D objects. The segmentation of the kidney from CT and the hippocampus from MRI serve as the major examples in this paper. The accuracy of segmentation as compared to manual, slice-by-slice segmentation is reported.

Download full-text

Full-text

Available from: Sarang C Joshi
  • Source
    • "It is reported that the best three ranked segmentation algorithms on livers are based on SSMs in the contest held in 2007 [2]. According to our knowledge, several SSMs have been proposed, which are point distribution models (PDMs) [3], m-rep [4], SPHARMs [5, 6], and so on. We focus on PDMs in this paper since it is the most widely used SSMs. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Minimum description length (MDL) based group-wise registration was a state-of-the-art method to determine the corresponding points of 3D shapes for the construction of statistical shape models (SSMs). However, it suffered from the problem that determined corresponding points did not uniformly spread on original shapes, since corresponding points were obtained by uniformly sampling the aligned shape on the parameterized space of unit sphere. We proposed a particle-system based method to obtain adaptive sampling positions on the unit sphere to resolve this problem. Here, a set of particles was placed on the unit sphere to construct a particle system whose energy was related to the distortions of parameterized meshes. By minimizing this energy, each particle was moved on the unit sphere. When the system became steady, particles were treated as vertices to build a spherical mesh, which was then relaxed to slightly adjust vertices to obtain optimal sampling-positions. We used 47 cases of (left and right) lungs and 50 cases of livers, (left and right) kidneys, and spleens for evaluations. Experiments showed that the proposed method was able to resolve the problem of the original MDL method, and the proposed method performed better in the generalization and specificity tests.
    Full-text · Article · Jun 2013 · Computational and Mathematical Methods in Medicine
    • "Automated pattern recognition algorithms have been applied to biological visual data (Shamir et al., 2010). A number of shape representations, key in the segmentation of images in these algorithms (Trinh and Kimia, 2007), have been proposed, including contours (Olson et al., 1980; Kass et al., 1988), deformations from a template shape (Cootes et al., 1995), graph-based representations (Joshi et al., 2002; Pizer et al., 2003), more abstract representations based on topological closed surfaces that model the external shape of an organism (Isaeva et al., 2006), and statistical shape models (Heimann and Meinzer, 2009). These shape formalisms represent great advances in the task of describing precisely (as close as possible to reality) the shapes of biological objects. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Understanding the mechanisms governing the regeneration capabilities of many organisms is a fundamental interest in biology and medicine. An ever-increasing number of manipulation and molecular experiments are attempting to discover a comprehensive model for regeneration, with the planarian flatworm being one of the most important model species. Despite much effort, no comprehensive, constructive, mechanistic models exist yet, and it is now clear that computational tools are needed to mine this huge dataset. However, until now, there is no database of regenerative experiments, and the current genotype-phenotype ontologies and databases are based on textual descriptions, which are not understandable by computers. To overcome these difficulties, we present here Planform (Planarian formalization), a manually curated database and software tool for planarian regenerative experiments, based on a mathematical graph formalism. The database contains more than a thousand experiments from the main publications in the planarian literature. The software tool provides the user with a graphical interface to easily interact with and mine the database. The presented system is a valuable resource for the regeneration community and, more importantly, will pave the way for the application of novel artificial intelligence tools to extract knowledge from this dataset. Availability: The database and software tool are freely available at http://planform.daniel-lobo.com.
    No preview · Article · Feb 2013 · Bioinformatics
    • "Automated pattern recognition algorithms have been applied to biological visual data (Shamir et al., 2010). A number of shape representations, key in the segmentation of images in these algorithms (Trinh and Kimia, 2007), have been proposed, including contours (Olson et al., 1980; Kass et al., 1988), deformations from a template shape (Cootes et al., 1995), graph-based representations (Joshi et al., 2002; Pizer et al., 2003), more abstract representations based on topological closed surfaces that model the external shape of an organism (Isaeva et al., 2006), and statistical shape models (Heimann and Meinzer, 2009). These shape formalisms represent great advances in the task of describing precisely (as close as possible to reality) the shapes of biological objects. "
    [Show abstract] [Hide abstract]
    ABSTRACT: The mechanisms underlying the regenerative abilities of certain model species are of central importance to the basic understanding of pattern formation. Complex organisms such as planaria and salamanders exhibit an exceptional capacity to regenerate complete body regions and organs from amputated pieces. However, despite the outstanding bottom-up efforts of molecular biologists and bioinformatics focused at the level of gene sequence, no comprehensive mechanistic model exists that can account for more than one or two aspects of regeneration. The development of computational approaches that help scientists identify constructive models of pattern regulation is held back by the lack of both flexible morphological representations and a repository for the experimental procedures and their results (altered pattern formation). No formal representation or computational tools exist to efficiently store, search, or mine the available knowledge from regenerative experiments, inhibiting fundamental insights from this huge dataset. To overcome these problems, we present here a new class of ontology to encode formally and unambiguously a very wide range of possible morphologies, manipulations, and experiments. This formalism will pave the way for top-down approaches for the discovery of comprehensive models of regeneration. We chose the planarian regeneration dataset to illustrate a proof-of-principle of this novel bioinformatics of shape; we developed a software tool to facilitate the formalization and mining of the planarian experimental knowledge, and cured a database containing all of the experiments from the principal publications on planarian regeneration. These resources are freely available for the regeneration community and will readily assist researchers in identifying specific functional data in planarian experiments. More importantly, these applications illustrate the presented framework for formalizing knowledge about functional perturbations of morphogenesis, which is widely applicable to numerous model systems beyond regenerating planaria, and can be extended to many aspects of functional developmental, regenerative, and evolutionary biology.
    No preview · Article · Feb 2013 · Biology Open
Show more