Automated Detection of 3D Landmarks for the Elimination of Non-Biological Variation in Geometric Morphometric Analyses.

Landmark-based morphometric analyses are used by anthropologists, developmental and evolutionary biologists to understand shape and size differences (eg. in the cranioskeleton) between groups of specimens. The standard, labor intensive approach is for researchers to manually place landmarks on 3D image datasets. As landmark recognition is subject to inaccuracies of human perception, digitization of landmark coordinates is typically repeated (often by more than one person) and the mean coordinates are used. In an attempt to improve efficiency and reproducibility between researchers, we have developed an algorithm to locate landmarks on CT mouse hemi-mandible data. The method is evaluated on 3D meshes of 28-day old mice, and results compared to landmarks manually identified by experts. Quantitative shape comparison between two inbred mouse strains demonstrate that data obtained using our algorithm also has enhanced statistical power when compared to data obtained by manual landmarking.

The ontology of craniofacial development and malformation for translational craniofacial research.

We introduce the Ontology of Craniofacial Development and Malformation (OCDM) as a mechanism for representing knowledge about craniofacial development and malformation, and for using that knowledge to facilitate integrating craniofacial data obtained via multiple techniques from multiple labs and at multiple levels of granularity. The OCDM is a project of the NIDCR-sponsored FaceBase Consortium, whose goal is to promote and enable research into the genetic and epigenetic causes of specific craniofacial abnormalities through the provision of publicly accessible, integrated craniofacial data. However, the OCDM should be usable for integrating any web-accessible craniofacial data, not just those data available through FaceBase. The OCDM is based on the Foundational Model of Anatomy (FMA), our comprehensive ontology of canonical human adult anatomy, and includes modules to represent adult and developmental craniofacial anatomy in both human and mouse, mappings between homologous structures in human and mouse, and associated malformations. We describe these modules, as well as prototype uses of the OCDM for integrating craniofacial data. By using the terms from the OCDM to annotate data, and by combining queries over the ontology with those over annotated data, it becomes possible to create "intelligent" queries that can, for example, find gene expression data obtained from mouse structures that are precursors to homologous human structures involved in malformations such as cleft lip. We suggest that the OCDM can be useful not only for integrating craniofacial data, but also for expressing new knowledge gained from analyzing the integrated data.

A new tool for quantifying and characterizing asymmetry in bilaterally paired structures.

This paper introduces a new tool to quantify and characterize asymmetry in bilaterally paired structures. This method uses deformable registration to produce a dense vector field describing the point correspondences between two images of bilaterally paired structures. The deformation vector field properties are clustered to detect and describe regions of relevant asymmetry. Three methods are provided to analyze the asymmetries: the global asymmetry score uses cluster features to quantify overall asymmetry, the local asymmetry score quantifies asymmetry in user-defined regions of interest, and the asymmetry similarity measure quantifies pairwise similarity of individual asymmetry. The scores and image distances generated by this tool are shown to correlate highly with asymmetry ratings assigned by an expert.

A landmark-free framework for the detection and description of shape differences in embryos.

This paper introduces a new method to quantify and characterize shape changes during early facial development without the use of landmarks. Landmarks are traditionally used in morphometric analysis, but very few can be identified reliably across all stages of embryonic development. This method uses deformable registration to produce a dense vector field describing the point correspondences between two images. Low and mid-level features are extracted from the deformable vector field to find regions of organized differences that are biologically relevant. These methods are shown to detect regions of difference when evaluated on chick embryo images warped with small magnitude deformations in regions critical to midfacial development.

Three-dimensional head shape quantification for infants with and without deformational plagiocephaly.

OBJECTIVE: The authors developed and tested three-dimensional (3D) indices for quantifying the severity of deformational plagiocephaly (DP). DESIGN: The authors evaluated the extent to which infants with and without DP (as determined by clinic referral and two experts' ratings) could be correctly classified. PARTICIPANTS: Infants aged 4 to 11 months, including 154 with diagnosed DP and 100 infants without a history of DP or other craniofacial condition. After excluding participants with discrepant expert ratings, data from 90 infants with DP and 50 infants without DP were retained. MEASUREMENTS: Two-dimensional (2D) histograms of surface normal vector angles were extracted from 3D mesh data and used to compute the severity scores. OUTCOME MEASURES: Left posterior flattening score (LPFS), right posterior flattening score (RPFS), asymmetry score (AS), absolute asymmetry score (AAS), and an approximation of a previously described 2D measure, the oblique cranial length ratio (aOCLR). Two-dimensional histograms localized the posterior flatness for each participant. ANALYSIS: The authors fit receiver operating characteristic curves and calculated the area under the curves (AUC) to evaluate the relative accuracy of DP classification using the above measures. RESULTS: The AUC statistics were AAS = 91%, LPFS = 97%, RPFS = 91%, AS = 99%, and aOCLR = 79%. CONCLUSION: Novel 3D-based plagiocephaly posterior severity scores provided better sensitivity and specificity in the discrimination of plagiocephalic and typical head shapes than the 2D measurements provided by a close approximation of OCLR. These indices will allow for more precise quantification of the DP phenotype in future studies on the prevalence of this condition, which may lead to improved clinical care.

A Bayesian hierarchical model for classifying craniofacial malformations from CT imaging.

Single-suture craniosynostosis is a condition of the sutures of the infant's skull that causes major craniofacial deformities and is associated with an increased risk of cognitive deficits and learning/language disabilities. In this paper we adapt to classification of synostostic head shapes a Bayesian methodology that overcomes the limitations of our previously published shape representation and classification techniques. We evaluate our approach in a series of large-scale experiments and show performance superior to those of standard approaches such as Fourier descriptors, cranial spectrum, and Euclidian-distance-based analyses.

A model browser for biosimulation.

The complexities of biological simulation present difficulties with modeling and experimenting. Simulators process models represented as code, whereas biologists think about abstract models. Our ModelBrowser addresses this difficulty through interactive visualization. Variables and equations appear as a directed graph of nodes and edges, and the user can search and browse this graph by performing queries on metadata associated with the variables and the connectivity of the edges. The browser also supports a hierarchical categorization of the variables, such as by an ontology. We believe that the ModelBrowser will help biologists reason about code in the context of the abstract model, so that they can understand and modify others' code and debug their own.

Predicting neuropsychological development from skull imaging.

Craniosynostosis is a serious and common pediatric disease caused by the premature fusion of sutures of the skull. Although studies have shown an increase in risk for cognitive deficits in patients with isolated craniosynostosis, the causal basis for this association is still unclear. It is hypothesized that an abnormally shaped skull produces a secondary deformation of the brain that results in the disruption of normal neuropsychological development. In this paper, we conduct a comparative analysis of our newly developed shape descriptors in an attempt to understand the impact of skull deformations on neurobehavior. In particular, we show that our scaphocephaly severity indices and symbolic shape signatures are predictive of mental ability and psychomotor functions, respectively, which suggests the possibility that secondary deformation could influence neuro-developmental status.

Symbolic shape descriptors for classifying craniosynostosis deformations from skull imaging.

Craniosynostosis is a serious condition of childhood, caused by the early fusion of the sutures of the skull. The resulting abnormal skull development can lead to severe deformities, increased intra-cranial pressure, as well as vision, hearing and breathing problems. In this work we develop a novel approach to accurately classify deformations caused by metopic and isolated sagittal synostosis. Our method combines a novel set of symbolic shape descriptors and off-the-shelf classification tools to model morphological variations that characterize the synostotic skull. We demonstrate the efficacy of our methodology in a series of large-scale classification experiments that contrast the performance of our proposed symbolic descriptors to those of traditional numeric descriptors, such as clinical severity indices, Fourier-based descriptors and cranial image quantifications.

The digital anatomist structural abstraction: a scheme for the spatial description of anatomical entities.

In this paper, we propose a generalized scheme for the symbolic description of the spatial attributes of anatomical entities. The power of the scheme lies in the ability to model the spatial objects at the highest level of granularity: information can be obtained at the desired level of detail needed for a given application. This scheme uses the topological classes of point, line, surface, and volume to represent zero-D, one-D, two-D and three-D objects. A spatial object participates as a node in three complementary networks; the topology network, the part-of network, and the spatial associations network. The topology network describes a spatial object in terms of its boundaries, the part-of network describes a spatial object in terms of its parts, and the spatial associations network describes the spatial object in terms of its relationships to other spatial objects. All three of the networks can be used in combination or alone to answer queries to the spatial information system. The Digital Anatomist Structural Abstraction together with the other components of the Digital Anatomist Foundational Model will provide the information for describing and reasoning about anatomical entities.

The digital anatomist foundational model: principles for defining and structuring its concept domain.

We define a foundational model as an abstraction of a body of knowledge that explicitly declares the principles and concepts necessary for coherently and consistently modelling a knowledge domain. Principles for a foundational model of anatomy are defined and used to specify the components of such a model. These components include an anatomy ontology (Ao), an anatomical structural abstraction (ASA), an anatomical transformation abstraction (ATA) and metaknowledge (Mk), which comprises the rules for representing relationships in the other three components of the model. The foundational model Fm is therefore specified as the four-tuple Fm = (Ao,ASA,ATA,Mk). We hypothesize that this abstraction captures the information that is sufficient and necessary for describing the anatomy of any physical entity that constitutes the body, as well as that of the body itself.

Molecular heterogeneity of adult Philadelphia chromosome-positive acute lymphoblastic leukemia.

The (9;22) translocation which produces the Philadelphia (Ph1) chromosome activates the abl oncogene from chromosome 9 by recombination with the bcr gene from chromosome 22. This fusion gene is transcribed into a new 8.5-kilobase chimeric mRNA which is translated into a novel Mr 210,000 fusion protein which has a protein tyrosine kinase activity that is greatly increased in comparison to the activity of the normal abl protein. Studies from this laboratory and others have shown that virtually all patients with chronic myelogenous leukemia have this new bcr/abl fusion gene. In contrast to these findings in chronic myelogenous leukemia, a small number of patients with Ph1(+) acute lymphoblastic leukemia (ALL) have been studied and were found to lack the bcr/abl fusion gene [bcr(-)], but to have a new activation of abl, by recombination with an as yet undetermined region on chromosome 22. In this study, nine adults with Ph1(+)-ALL have been examined for evidence of a bcr/abl fusion gene. Of the nine patients, five have a bcr/abl recombination, whereas the remaining four patients do not. In contrast, the children studied to date have all been bcr(-). These data suggest that adults with Ph1(+)-ALL are a more heterogeneous group on a molecular level than are children, and that further studies will be required to determine the spectrum of molecular defects in patients with Ph1(+)-ALL, and the relationship of these various molecular defects to the clinical disease state of the individuals.

Relational matching.

The problem of computer vision is to give a computer a representation of a picture of a scene and have it figure out what objects are in the scene and in what spatial relationships. This involves low-level vision (image processing), midlevel vision (feature extraction and measurement), and high-level vision (interpretation). One important part of high-level vision is relational matching, the process of matching two relational descriptions of objects, often for the purpose of object identification. In this paper, we describe several kinds of relational matching, give sequential algorithms for solving relational matching problems, and briefly discuss parallel algorithm possibilities.

A metric for comparing relational descriptions.

Relational models are frequently used in high-level computer vision. Finding a correspondence between a relational model and an image description is an important operation in the analysis of scenes. In this paper the process of finding the correspondence is formalized by defining a general relational distance measure that computes a numeric distance between any two relational descriptions-a model and an image description, two models, or two image descriptions. The distance measure is proved to be a metric, and is illustrated with examples of distance between object models. A variant measure used in our past studies is shown not to be a metric.

Identification of space curves from two-dimensional perspective views.

This paper describes a new method to be used for matching three-dimensional objects with curved surfaces to two-dimensional perspective views. The method requires for each three-dimensional object a stored model consisting of a closed space curve representing some characteristic connected curved edges of the object. The input is a two-dimensional perspective projection of one of the stored models represented by an ordered sequence of points. The input is converted to a spline representation which is sampled at equal intervals to derive a curvature function. The Fourier transform of the curvature function is used to represent the shape. The actual matching is reduced to a minimization problem which is handled by the Levenberg-Marquardt algorithm [3].

Organization of relational models for scene analysis.

Relational models are commonly used in scene analysis systems. Most such systems are experimental and deal with only a small number of models. Unknown objects to be analyzed are usually sequentially compared to each model. In this paper, we present some ideas for organizing a large database of relational models. We define a simple relational distance measure, prove it is a metric, and using this measure, describe two organizational/access methods: clustering and binary search trees. We illustrate these methods with a set of randomly generated graphs.

Structural descriptions and inexact matching.

In this paper we formally define the structural description of an object and the concepts of exact and inexact matching of two structural descriptions. We discuss the problems associated with a brute-force backtracking tree search for inexact matching and develop several different algorithms to make the tree search more efficient. We develop the formula for the expected number of nodes in the tree for backtracking alone and with a forward checking algorithm. Finally, we present experimental results showing that forward checking is the most efficient of the algorithms tested.

A structural model of shape.

Shape description and recognition is an important and interesting problem in scene analysis. Our approach to shape description is a formal model of a shape consisting of a set of primitives, their properties, and their interrelationships. The primitives are the simple parts and intrusions of the shape which can be derived through the graph-theoretic clustering procedure described in [31]. The interrelationships are two ternary relations on the primitives: the intrusion relation which relates two simple parts that join to the intrusion they surround and the protrusion relation which relates two intrusions to the protrusion between them. Using this model, a shape matching procedure that uses a tree search with look-ahead to find mappings from a prototype shape to a candidate shape has been developed. An experimental Snobol4 implementation has been used to test the program on hand-printed character data with favorable results.

The Consistent Labeling Problem: Part II.

In this second part of a two-part paper, we explore the power and complexity of the g=fKP and g=vKP class of look-ahead operators which can be used to speed up the tree search in the consistent labeling problem. For a specified K and P we show that the fixedpoint power of g=fKP and g=vKP is the same, that g=fKP+1 is at least as powerful as g=fKP, and that g=vK+1p is at least as powerful at g=fKP. Finally, we define a minimal compatibility relation and show how the standard tree search procedure for finding all the consistent labelings is quicker for a minimal relation. This leads to the concept of grading the complexity of compatibility relations according to how much look-ahead work it requires to reduce them to minimal relations and suggests that the reason look-ahead operators, such as Waltz filtering, work so well is that the compatibility relations used in practice are not very complex and are reducible to minimal or near minimal relations by a g=fKP or g=vKP look-ahead operator with small value for parameter P.

Decomposition of two-dimensional shapes by graph-theoretic clustering.

This paper describes a technique for transforming a twodimensional shape into a binary relation whose clusters represent the intuitively pleasing simple parts of the shape. The binary relation can be defined on the set of boundary points of the shape or on the set of line segments of a piecewise linear approximation to the boundary. The relation includes all pairs of vertices (or segments) such that the line segment joining the pair lies entirely interior to the boundary of the shape. The graph-theoretic clustering method first determines dense regions, which are local regions of high compactness, and then forms clusters by merging together those dense regions having high enough overlap. Using this procedure on handdrawn colon shapes copied from an X-ray and on handprinted characters, the parts determined by the clustering often correspond well to decompositions that a human might make.